4.1. The South African Scoring System version 5

The South African Scoring System version 5 (SASS5) is a rapid bioassessment index based on the presence or absence of selected families of aquatic macroinvertebrates and their perceived sensitivity or tolerance to deteriorating water quality [24]. In SASS5, macroinvertebrate families are awarded scores in the range of 1–15 in increasing order of sensitivity to deteriorating water quality. Families considered sensitive are awarded high scores and those considered tolerant low scores. The results are expressed both as an index score, that is, SASS5 score, and as an average score per recorded taxon (ASPT) value. The SASS5 score is calculated by summing the scores of all recorded families, while the ASPT value is obtained by dividing the total SASS5 score by the number of families recorded. In addition to being a useful water quality assessment index, SASS5 is used to assess emerging water quality problems, development impacts, ecological state and spatio-temporal trends of biological assemblages.

### 4.2. The study area

single-species acute and chronic toxicity tests have been standardised and are widely use in the ecosystem-based approach to managing water quality [30]. The strengths of the laboratory single-species tests include (i) precision: they are conducted in a highly regulated environment, where external influences are isolated, so that there is a high level of precision with regard to the toxicant effects on the organism; (ii) repeatability: single-species, laboratory-based experiments are easily reproducible and repeatable, provided that sets, guidelines and protocols are followed; (iii) high level of acceptance in the regulatory circle: these tests still form the cornerstone of risk assessments of harmful chemicals in the environment; (iv) simplicity: these test

Although the single-species laboratory-based tests are widely used in managing aquatic ecosystems, they are unable to provide direct community or ecosystem-level effects. They rely heavily on laboratory to field extrapolations by applying safety assessment factors or the species sensitivity distribution (SSD) approach [31]. Reducing uncertainties requires using more ecologically relevant and realistic assessments that employ multi-species in experimental settings that are closer to the natural field conditions. The multi-species model-stream ecosystem approach occupies an intermediate space between field biomonitoring studies and the traditional single-species laboratory-based approach. If reasonably controlled, manipulated and replicated, they can simulate community and even ecosystem effects [32]. While the single-species approach offers high degree of precision, repeatability and simplicity, modelstream ecosystems represent a compromise between these factors, and their high environmen-

Model-stream ecosystems are termed mesocosms or microcosms depending on their sizes and locations [33–37]. For example, Odum [33] defined mesocosms as outdoor experimental streams bounded and partially closed, which closely simulate the natural conditions. Buikema and Voshell [34] use volume as a factor for differentiating between microcosms and mesocosms, referring to microcosms as experimental streams (usually indoor) with a volume equal or less than 10 m3 and mesocosms as those (usually outdoor) having a volume greater

long and microcosms as those that are shorter. However, Belanger [36] review revealed that increased physical sizes of experimental streams did not correspond to increased biological complexity. Since the goal of a multi-species model-stream ecosystem is to achieve an adequate ecological realism irrespective of size, the terms 'microcosm' and 'mesocosm' are actually inappropriate. Instead, the appropriate terminologies should be a 'model-stream ecosystem',

Model-stream ecosystems have some advantages over conventional single-species toxicity tests. They enable the simulation of natural conditions, offering a high degree of environmental realism and enabling complex biophysical interactions. They enable the researcher to evaluate direct effects of pollutants at higher biological organisation such as population, community and even ecosystem levels [37]. Moreover, they enable the study of biotic interaction and community dynamics and measurement of indirect ecosystem effects. Their shortcomings are that they are not easily reproducible, have low precision and are not

. Hill et al. [35] defined mesocosms as experimental streams that are more than 15 m

are usually very simple to undertake, hence their appeal in regulatory circles.

tal realisms [32].

8 Water Quality

than 10 m<sup>3</sup>

simple to undertake.

'experimental streams or artificial streams'.

The Swartkops River originates in the foothills of the Groot Winterhoek Mountains and then meanders through the towns of Uitenhage, Despatch and Perseverance before discharging into the Indian Ocean at Algoa Bay, near the city of Port Elizabeth (Figure 1). Climate in the catchment is warm and temperate, and rainfalls vary between the upper and lower regions. The upper region usually receives higher rainfall than the lower region. The catchment geology is mainly of marine, estuarine and fluvial origin. Soils in the upper catchment are not deep and are unsuitable for agriculture. Those in the low-lying floodplain region are deep and well suited for agriculture. The dominant vegetation in the catchment is bushveld and succulent thicket.

Although the river is an important ecological and socio-economic asset, serving as a home to important bird and fish species, and providing water for small-scale irrigation, the health and functionality of the entire system are being threatened by deteriorating water quality. Several sources of pollution including raw sewage run-off from informal settlements, treated wastewater effluent discharges from municipal treatment works, agricultural farmlands, surrounding road and rail networks, and industrial sites were all influencing the water quality of the river and hence the need to assess its health using the SASS5.

### 4.3. Sampling sites and macroinvertebrates sampling

Four sites within the same ecoregion were selected for the study. Site 1 (33°45′08.4″ S, 25°20′ 32.6″ E), situated in the upper reaches of the river was the least impacted and thus was chosen as the control site. It has a well diverse range of macroinvertebrate sampling habitats. Site 2 (33°47′29.0″ S, 25°24′26.4″ E) was in the industrial town of Uitenhage, where surrounding impacts include run-off from roads and informal settlements, free-ranging livestock and other agricultural practices. All macroinvertebrate sampling biotopes were adequately represented at the site. Site 2 is situated upstream of the discharge point of the Kelvin Jones wastewater treatment work (WWTW) in the town of Uitenhage. Site 3 (33°47′11.8″ S, 25°25′53.97″ E) is further downstream, but also within the industrial town of Uitenhage, where surrounding impacts include industrial and wastewater effluent discharges, run-off from road and rail networks, and agricultural activities. The Kelvin Jones WWTW is the main pollution source at Site 3. Macroinvertebrate sampling biotopes at Site 3 were also adequate. Site 4 (33°47′34.0″ S, 25°27′58.7″ E) further downstream of Site 3 was situated in the residential town of Despatch. Municipal run-off, sand and gravel mining on the riparian zone were the main impacts at Site 4. Although Site 4 was not as polluted as Site 3, it would have been good to select another site further downstream to monitor for potential system recovery. However, the tidal limit at Perseverance between the estuary and the freshwater section is only a short distance downstream of Site 4. Consequently, it was not possible to select a fifth site further downstream because of likely estuarine effects.

Figure 1. Map of the Swartkops River showing the sampling sites and the relative position of the Kelvin Jones Wastewater Treatment Works.

Macroinvertebrates were sampled using the SASS5 protocol. At each site, over a period of three years, between late August 2009 and September 2012, samples were collected seasonally. A total of eight sampling events were conducted over the sampling period. Macroinvertebrates were collected using a kick net (300 × 300 mm frame, 1000 μm mesh) from three distinct biotope groups: stones (stones-in-and-out-of-current), vegetation (marginal and aquatic vegetation) and sediment (gravel, sand and mud, GSM) as prescribed in the SASS5 protocol.

Sampled macroinvertebrates were tipped into a white rectangular tray, half-filled with river water, and macroinvertebrate families identified on site using identification keys by Gerber and Gabriel [38]. The identified families were recorded on a SASS5 sheet together with their abundance estimates. SASS5 scores, number of taxa and ASPT values were calculated and then interpreted as described in the following section. Time spent on field identification adhered strictly to recommendation in the SASS5 protocol.
