The deteriorating nutrient status of the Berg River, South Africa

The upper catchment area of the Berg River in the Western Cape, South Africa, supplies most of Cape Town and its suburbs with freshwater, in addition to providing water for irrigation purposes along the middle and lower reaches of the river. This study investigates the nutrient status of the Berg River and long-term trends therein. It is shown that inorganic nitrogen and phosphorus levels increase downstream by a factor of more than 10, in response to anthropogenic inputs. Similarly, nutrient levels fluctuate seasonally by more than an order of magnitude, in response to input from diffuse and point sources of pollution. These changes of more than 1 000% far exceed the 15% maximum change stipulated by the South African water quality guidelines for aquatic ecosystems. Total phosphorus levels indicate that hypertrophic conditions prevail at least episodically at all of the Berg River monitoring stations and most of the time at some of them. Additionally, river water phosphate levels show a dramatic increase over the past 20 years. There is also strong evidence that the trophic status of the Berg River is very sensitive to reduced river runoff. The implication is that the construction of the new Berg River Dam in the upper catchment area of the Berg River will exacerbate the existing situation, threatening ecosystem services, human health and lucrative agricultural activities.


Introduction
Eutrophication, excessive plant growth in response to nutrient enrichment, is considered to be one of the most serious problems facing freshwater ecosystems, globally (Vitousek et al., 1997;Carpenter et al., 1998;Galloway and Cowling, 2002;Camargo and Alonso, 2006;Mainstone and Parr, 2002). The major nutrients that contribute to eutrophication are phosphorus as phosphate ions (PO 4 3-) and nitrogen as nitrate (NO 3 -), nitrite (NO 2 -) and ammonium (NH 4 + ) ions. Nutrient levels of many freshwater ecosystems have increased dramatically, by a factor of 4 at least, over the last couple of decades in response to widespread agricultural intensification and increased discharge of domestic wastes (Vitousek et al., 1997;Galloway and Cowling, 2002). A particular problem facing developing countries such as South Africa is the significant increase in urban runoff and increasingly so from overloaded or dysfunctional municipal water treatment plants and un-sewered human settlements (Barnes, 2003;Bere, 2007;Mtetwa and Schutte, 2003;Luger and Brown, 2003;Van Vuuren, 2005). All of these are potentially significant sources of nutrients and other pollutants to river and groundwater reservoirs The South African water quality guidelines (DWAF, 1996a) stipulate Target Water Quality Range (TWQR) values for 7 different water-use sectors: domestic, recreational, industrial, irrigation, stock watering, aquaculture and aquatic ecosystems. TWQR is defined as 'the range of concentrations or levels at which the presence of the constituent would have no known adverse or anticipated effect on the fitness of the water assuming long-term continuous use, and for safeguarding the health of aquatic ecosystems.' Aquatic ecosystems are unique amongst the different types of water users, in that aquatic plant and animal species have very different water quality requirements and tolerances, depending on locality. As a result, there are no stipulated TWQR nutrient values for aquatic ecosystems, but rather a recommendation that 'a TWQR should be derived only after case-and site-specific studies' (DWAF, 1996b). Additionally, 'inorganic nitrogen (and phosphorus) concentrations should not be changed by more than 15% from that of the water body under local unimpacted conditions at any time of the year.' There is no documented evidence, however, that such 'case-and site-specific studies' have been carried out for any of South Africa's freshwater ecosystems. It is also not clear that the development of such site-specific TWQR values for nutrients is one of the objectives of the relatively new National Eutrophication Monitoring Programme, NEMP (DWAF, 2002). As a result, classification of the trophic status of South Africa's aquatic ecosystems is presently restricted to the use of 4 broad categories: oligotrophic, mesotrophic, eutrophic and hypertrophic (Table 1), with no allowances made for diverse ecosystem requirements.
This study provides a detailed investigation of the nutrient status of the Berg River, located in the Western Cape Province in South Africa (Fig. 1). The Berg River provides the bulk of the water for household and industrial use in the Cape Town metropole and greater Cape Peninsula area, in addition to irrigation water for extensive cultivation along the length of the river. A combination of recent dry spells, population growth and a fast growing local economy has put severe pressure on water resources within this system. The construction of an additional dam in the Groot Drakenstein Mountains near Franschhoek (Berg River Dam) will provide some relief for Cape Town's water supply problems, but has also raised serious concerns about the implications for water quality along the lower reaches of the river. The new NEMP 'has not been implemented' in the Berg Water Management Area (NEMP, 2003), despite the known eutrophic status of the Misverstand Dam (http://www.dwaf.gov. za/iwqs/eutrophication/NEMP/TrofieseStatus2003.pdf). Some of the manifestations of 'eutrophic' conditions along the Berg River are the increasingly problematic presence of water hyacinth along the middle and lower reaches of the Berg River and dramatic declines in fish catches. Additionally, bacterial counts indicative of sewage pollution (Barnes, 2003;Paulse et al., 2007) and heavy metal concentrations  exceeding recommended DWAF guidelines have been reported for the Berg River.
The most recent sources of information on the nutrient water quality status of the Berg River (Görgens and De Clercq, 2005;Quibell, 1993) are based on DWAF water quality monitoring data prior to 1998. It has been noted (Görgens and De Clercq, 2005) that 'an increase in phosphate concentrations at all stations can be clearly seen', but these trends or more recent DWAF water quality data for the Berg River system have not been examined in great detail. This study examines DWAF water quality data up to 2005 at 9 monitoring stations on the Berg River to determine downstream, seasonal and long-term trends in the following nutrient parameters: nitrate and nitrite (NO 3 and NO 2 -, or NO x ), ammonium (NH 4 + ), total phosphorus (TP) and orthophosphate (PO 4 3-).

Study area and database
The Berg River (300 km long) rises in the Groot Drakenstein Mountains near the town of Franschhoek, drains a catchment area of about 9 000 km 2 , and enters the sea on the west coast at Velddrif (Fig. 1). The geology of the catchment area is dominated by sandstone and quartzites of the Cape Supergroup in the upper reaches, Cape granites in the middle reaches and recent sediments near the coast. The catchment is therefore characterised by nutrient-poor lithologies. Almost 50% of the catchment area is cultivated agricultural land, mainly vineyards, fruit trees and wheat fields. River flow peaks during the winter rainy season, from June to August. Although evaporation exceeds precipitation throughout the catchment, the river water budget is dominated by runoff.
DWAF water quality monitoring data is available at 9 sites along the Berg River: LB1, LB2 and B1 to B7 (Fig. 1, Table  2), from as early as 1967 at one of the stations and from the 1970s at most of the stations. With due consideration of differences in length and completeness of time series data between stations, only data from 1985 onwards were considered in this study. Sampling frequency also varies between sites, from almost weekly to monthly. Where more than one water quality data point was available in a given month, an average monthly value was calculated to provide time-series data at a monthly resolution. This also provides compatibility with DWAF's total monthly water flow records.
The DWAF database contains data for the dissolved inorganic nitrogen species [NO 2 -+ NO 3 -] and [NH 4 + ] (all expressed as µg N/ℓ, with 20 to 40 µg N/ℓ detection limits) and dissolved total phosphorus (TP) and soluble reactive phosphate (SRP measured as PO 4 3-, expressed as µg P/ℓ, with reported 3 to 5 µg P/ℓ detection) limits. Data for [NH 4 + ] and total dissolved phosphorus (TP) were available at only some of the stations or sections of the record. As a result median, mean and maximum TP and [NH 4 + ] values reported (Table 2) represent shorter data periods.
For data evaluation purposes nutrient levels in the Berg River are compared to both South African trophic status guideline values (Table 1) and more detailed international water quality guidelines. The latter, for the protection of aquatic animals, are 2 000 to 3 600 µg NO 3 --N/ℓ for the NO 2 --NO 3 forms of inorganic nitrogen (Camargo et al., 2005;CCME, 2003) and between 20 and 100 µg P/ℓ for soluble reactive phosphorus (Mainstone and Parr, 2002). Un-ionised ammonia (NH 3 ) is the most toxic form of inorganic nitrogen to aquatic animals and water quality criteria ranging from 50 to 350 µg NH 3 -N/ℓ for short-term exposures and 10 to 20 µg NH 3 -N/ℓ for long-term exposures have been recommended (Camargo and Alonso, 2006; Constable et al.,  Environment Canada, 2001;USPA, 1986). NH 3 concentrations are not directly measured in the Berg River, but measured levels of NH 4 + combined with pH values of 6 to 8 predict very low to negligible levels of NH 3 . Recommended dissolved inorganic nitrogen and phosphorus levels for the prevention of eutrophication are lower than those for aquatic animals. Levels higher than 30 µg TP/ℓ are generally considered conducive to eutrophication, provided that inorganic nitrogen or other nutrients are not limiting (Camargo and Alonso, 2006;Swedish EPA, 2000). Plants require nitrogen and phosphorus in a ratio of between 7 and 8 (weight/weight) and concomitant dissolved values of > 400 µg total N/ℓ and > 30 µg total P/ℓ are generally considered favourable for eutrophication in freshwater systems.
Annual nutrient fluxes at each of the monitoring stations were calculated for 2 time periods (1985-1994 and 1995-2004) as follows: for each 10-year period, monthly averaged river flow,

Downstream trends in river water nutrient levels
Long-term monthly median, mean and maximum Berg River nutrient values are listed in Table 2 (Table 2). Towards the coast nutrient levels decrease again, to values approximating those observed in the upper reaches of the catchment, possibly indicating the consumption of nutrients by algal and macrophyte productivity within the river system. Ammonium levels are relatively constant downstream and represent a minor fraction of the total inorganic nitrogen pool, indicative of a well-aerated system (Table 2).
According to the NEMP trophic status classification scheme (DWAF, 2002), based on long-term mean [TP] levels, all of the stations for which TP monitoring data are available are eutrophic with the exception of B3, which is hypertrophic. Along almost the entire length of the Berg River, from B2 to B6 (Dal Josafat in Paarl to Hermon; Fig. 1) long-term mean [NO 3 -+ NO 2 -] values exceed the 400 µg/ℓ recommended international guideline for aquatic plant life. Additionally, at all the stations where TP data are available, the long-term mean value exceeds the 30 µg TP/ℓ recommended international guideline for aquatic plant life, by a factor of 2 to almost 10 ( Table 2). It is instructive to note that nutrient levels at the Misverstand Dam (B5), earmarked as the NEMP monitoring site on the Berg River, are more than a factor of two lower than further upstream (B3 at Hermon).
Even very brief episodes of nutrient enrichment can be exceedingly detrimental to aquatic plant and animal life, there-    (Figs. 2a and b). The highest values and most pronounced trend in [PO 4 3-] levels are observed at B3 (Fig 3c), where baseline values have more than doubled, from below 60 to more than 120 µg P/ℓ over the past 20 years. The middle reaches of the Berg River, therefore, are approaching a state of permanent hypertrophic conditions. At all the monitoring stations baseline [PO 4 3-] is either already exceeding the 30 µg TP/ ℓ recommended value for aquatic plant life, or approaching it. Additionally, the international water quality guidelines for aquatic animal life (~ 4 000 µg N/ℓ for [NO 3 -+ NO 2 -] and 20 to 100 µg P/ℓ for soluble reactive phosphorus (Mainstone and Parr, 2002) are exceeded at least episodically for [NO 3 -+ NO 2 -] at B3 and B6, and at all the monitoring stations in the case of phosphorus.

Seasonal fluctuations and long-term trends in river water nutrient levels and fluxes
Evaluation of long-term nutrient levels (Figs. 2 and 3) demonstrates well-defined seasonal changes in [NO 3 -+ NO 2 -] and that the amplitude of this seasonal cycle has remained fairly constant since the 1980s at all stations except B3, where it has increased by a factor of ~ 2, and B4, where it has increased by ~ 50% (Fig. 2c). Even at the stations where the seasonal amplitude of change in [NO 3 -+ NO 2 -] levels has remained fairly constant, however, the magnitude of seasonal variability is at least an order of magnitude (Fig. 2). This translates into a seasonal change resulting from anthropogenic factors of at least 1 000%, compared to the less than 15% change stipulated by the South African water quality guidelines for aquatic ecosystems (DWAF, 1996b). The long-term [PO 4 3-] records demonstrate less well-defined seasonal cycles compared to [NO x ], but a dramatic increase in levels and the magnitude of intra-annual variability over time (Fig. 3). During the 1985-1994 period [PO 4 3-] levels were relatively constant throughout the year at all stations in the catchment area. During the past 10 years, however, concentration levels have almost doubled throughout the year at all stations and a more pronounced seasonal cycle has emerged.
Representative seasonal runoff and nutrient concentration profiles were constructed for the periods 1985-1994 and 1995-2004, to yield insight into changing nutrient dynamics and the relative roles of diffuse and point sources of nutrients. Typical Berg River seasonal profiles are illustrated for B3 (Fig. 4) and the total annual fluxes derived from the combined run-  1985-1994 1995-2004

Figure 4
Comparative river flow,  and [PO 4 3-] monthly averaged data for the periods 1985-1994 and 1995-2004, at (Table 3). Runoff (Fig. 4a) peaks during the winter, consistent with the winter rainfall location of the Berg River catchment area. An important observation is that the 1995-2005 period was drier than 1985-1994, with average runoff for the month of June reduced by as much as 60% at Station B3 (Table 4). Seasonal [NO x ] profiles peak during high runoff conditions, consistent with a diffuse source such as agricultural runoff (Fig. 4b). The 1995 ] profile also indicates increased levels compared to 1985-1994, throughout the year (Table 4). Increased values during drier conditions are indicative of a concentration effect, i.e. reduced dilution of anthropogenic inputs. The most pronounced impact of this concentration effect during reduced runoff conditions is an increase of almost 400% in NO x values during the summer and 54 to 60% during the winter at monitoring station B3 (Table 4).
Seasonal [PO 4 3-] profiles demonstrate a dramatic change in seasonality over time (Fig. 4c), in addition to 67 to 373% increases in average monthly concentrations over time (Table  4). An absence of seasonality in [PO 4 3-] during the 1985-1994 period has been replaced by a seasonal profile that exhibits a peak in [PO 4 3-] values coinciding with the onset of increased river flow during late spring/early winter (Fig. 4c).
Evaluation of the average annual NO x flux during 1995-2004 compared to the 1985-1994 period reveals a flux reduction of only 13%, compared to an almost 40% reduction in runoff (Table 3).  PO 4 3fluxes, in contrast to NO x , have increased during the 1995-2004 period by as much as 50%, despite the reduced runoff (Fig.  4c). NO x fluxes/catchment area values peak in the middle section of the Berg River at station B2, coincident with peak concentration levels (Table 3). PO 4 3fluxes/catchment area values however, peak in the upper Berg River catchment at station B1 (Table 3).
A reduction in the annual NO x flux between B4 and B5 and in the annual PO 4 3flux between B3 and B5 suggests in situ consumption of nutrients, most probably assimilation by plants and algae and adsorption by sediments (Table 3). There is a reduction in runoff at B5 compared to B4, however, attributable to water extraction just upstream of the B5 monitoring station, that contribute to the nutrient flux reductions observed between B4 and B5 (Table 3).

Discussion
The two most likely anthropogenic sources of nutrients along the Berg River are agricultural runoff and effluent from overloaded municipal sewage works and un-sewered communities. Both sources are expected to peak in magnitude along the middle section of the Berg River, between Paarl and Hermon (B3 to B4), the most heavily cultivated and most populated area along the river. This includes informal human settlements that have developed along the banks of the river.
Diffuse nutrient sources, such as agricultural runoff, produce seasonal concentration profiles coincident with river runoff, i.e. concentrations that peak during high runoff conditions. In contrast, point sources such as sewage effluent from municipal water treatment plants generally result in seasonal concentration profiles that have no relation to runoff, i.e. relatively constant input throughout the year, or an inverse relation to river runoff.
The positive relationship between NO x levels and fluxes with runoff, i.e. peaks during the rainy winter season, is consistent with a diffuse source such as agricultural runoff being the most likely source of NO x enrichment (Fig. 4b). Additionally, the smaller NO x flux reduction during the past 10 years, compared to the 40% reduction in runoff, implies one of two scenarios or a combination thereof: • Increased fertiliser application during the latter 10 years • An increase in a different source, such as sewage effluent.
Evidence for increased NO x levels during low runoff conditions supports an increased point-source scenario. It is also suggested that overloading of water treatment plants during high runoff conditions or flooding of informal human settlements during winter storm events may result in nutrient enrichment during