Department of Water and Sanitation
	Institute for Water Quality Studies

A FISH KILL IN THE HARTBEESPOORT DAM,
SOUTH AFRICA, OCTOBER 1999.

REPORT NUMBER: N/A210/02/DEQ/4299

PROJECT: Eutrophication - Fish kills

STATUS OF REPORT: Final

DATE: November 1999

The evidence of a fish kill early in October 1999 in the Hartbeespoort Dam, as well as complaints about serious filter clogging from the water treatment works initiated this study to determine what factors could have contributed to the fish kill in the Hartbeespoort Dam.

A fish kill was reported to the IWQS by the samplers of the IWQS Technical Group early in October 1999. About 20 dead carp (Cyprinus carpio) were seen near the Kosmos area on the banks of the Hartbeespoort Dam. The presence of a strange brown colouring in the Hartbeespoort Dam was also reported, as well as the fact that this same brown colouring has been seen before in the deeper depth samples taken by the IWQS samplers. This initiated the once-off sampling of the Hartbeespoort Dam on 14 October 1999 to determine the cause of the fish kill and the strange brown colouring of the water. The source of the brown colour was determined through the analysis of the algal identification samples taken on the 12^th (preliminary sample) and the 14^th October 1999. The dominant and almost only phytoplankton species in the impoundment (near the dam wall) was Ceratium hirundinella (O.F. Muller).

C. hirundinella is a common and widespread freshwater dinoflagellate. It has been found in numerous man-made impoundments in South Africa (SHILLINGLAW, 1981; TRUTER 1987; and VAN GINKEL unpublished data), however the species has not been found in previous studies in the Hartbeespoort Dam (NIWR 1985; and VAN GINKEL unpublished data). It has also not previously been associated with bloom forming conditions in South Africa, although it has been found to be periodically dominant in other South African impoundments (e.g. the Klipvoor Dam, the Kosterrivier Dam, the Boskop Dam and the Bronkhorstspruit Dam). According to STARMACH (1974), PADISAK (1985) and BUCKA & ZUREK (1994), C. hirundinella is an indicator of clean waters, as it avoids water rich in organic compounds.

The Hartbeespoort Dam, situated on the confluence of the Crocodile and the Magalies Rivers (Figure 1) that drain highly populated urban areas, namely Pretoria, Johannesburg and Krugersdorp, has been subject to algal blooms (Microcystis auriginosa) (NIWR 1985) and massive aquatic weed (Eichornia crassipes) development. The catchment was identified as a hyper-eutrophic system and the promulgation of the 1 mg/l Phosphate (P) effluent Standard was to be applied in the catchment (DWA 1988, ANON 1988a, ANON 1988b). Because of financial constraints in the metropolitan areas, sewage treatments works do not always comply with the 1mg/l P effluent standard. There are also numerous informal settlements in the catchment, as well as intensive farming activities, all of which contribute to the extent of eutrophication within the catchment.

The Hartbeespoort Dam is sampled bi-weekly at Site 1 (See Figure 1) by the IWQS. Major inorganic chemical variables, total nutrients (total phosphorus and total nitrogen (KN & NO3 + NO2)), biological variables (namely chlorophyll a, algal species composition, suspended solids and phaeophytin a) and physical variables (temperature and oxygen profiles, Secchi disc depth and pH) are measured and/or analysed for.

The evidence of a fish kill early in October 1999 in the Hartbeespoort Dam initiated a once-off sampling survey in the inflowing rivers (Sites 2, 3, 4 & 5) and at the main water extraction point for the local village (Site 6). This survey was conducted on the 14^th October 1999 to determine the water quality situation and phytoplankton composition in the inflowing rivers and at the extraction point.

The routine sampling site (Site 1) is situated close to the dam wall (Site 1 in Figure 1). Depth samples were taken at 0 m, 1 m, 2 m, 5 m, 10 m, 15 m and 20 m with a Van Doorn sampler. These water samples were tested for chlorophyll a, major inorganic chemical constituents and total nutrient concentrations.

The analysis for chlorophyll a concentrations and algal compositions (Figure 2a) indicated that a bloom of C. hirundinella occurred in the main basin of the Hartbeespoort Dam. The chlorophyll a concentrations in the main basin at Site 1 reached the extremely high concentrations of 617.4 µg/l . However, C. hirundinella is believed to be indicative of clean water. The chlorophyll a concentration at Site 6, the intake site for the Schoemansville Water Care Works, has lead to serious problems of filter clogging.

Very low chlorophyll a concentrations occurred at all the inflowing tributaries (Sites 2, 3, 4 & 5 At Site 3 and 5 the chlorophyll a concentrations were lower than 1 µg/l , showing that there was little live chlorophyll a present at these sites. All four inflowing sites did, however, have decomposing phytoplankton (Figure 2a). The phaeophytin a concentration indicate the extent of decomposing phytoplankton material. This indicates that large quantities of phytoplanktonic matter are being decomposed at all of these sites. Decomposition at Site 1 was the highest with phaeophytin a concentrations of up to 29.1 µg/l . This decomposition of phytoplankton material varied between 8 µg/l and 29 µg/l at Sites 1, 2, 5 and 6.

Figure 3. The water quality variables at the six sampling sites in the Hartbeespoort Dam that might have influenced the fish kill during October 1999. a) Nitrogen species (NO3+NO2-N; NH4-N and KN-N) concentrations in mg/l . b) Phosphorus species (PO4-P & TP) concentrations in mg/l . c) The total dissolved salts concentrations (mg/l ). d) The pH values at each site. e) The surface temperatures (°C) and surface dissolved oxygen concentrations (mg/l ).

It is interesting to note that there is quite a high variability in the number of species and the species composition within the phytoplankton population at the different sites in the Hartbeespoort Dam. No cyanobacteria were present during sampling. Of the inflowing rivers, Site 5 had the lowest number of species. Major bacteriological activity at Site 5 was identified

by the analysis of the algal identification samples. This gives an indication of the extent of additional organic decomposition that was taking place at Site 5.

The once-off sampling survey on 14 October 1999 (Figure 2a & b) showed clearly that very low chlorophyll a concentrations and low numbers of C. hirundinella were found in the inflowing streams (Sites 2, 3, 4 & 5). This phenomenon might be explained by the findings of PADISAK (1985) who noted low abundance of C. hirundinella at sites where strong currents are present.

In Figure 3 the nutrients (a & b), TDS (c), pH (d) and surface water temperature and surface dissolved oxygen (e) are indicated.

During the sampling survey the nutrient concentrations in the inflowing rivers and at the dam wall were unacceptably high (Mesotrophic TP levels is < 47 µg/l ), except at Site 4. In the Crocodile River (Site 2) the 238 µg/l PO4-P indicate that highly enriched water are brought into the impoundment. The Magalies River (Site 5) with 36 µg/l PO4-P is much lower. However, the TP concentration at Site 5 was similar to the TP concentration at Site 2. Site 1, near the dam wall had the highest TP concentration. These results indicate that the Hartbeespoort Dam is highly enriched and that the highest load is most probably coming through the Crocodile River.

Figure 3a indicates that at Sites 2, 3, 4 & 6 the available form of nitrogen is nitrates and nitrites (NO3+NO2-N), while the main available form of nitrogen at Sites 1 & 5 is ammonium (NH4-N). The ammonium concentration at Site 5 is much higher than at Site 1 and indicates the presence of material with a high organic loading at this site, as NH4-N accumulates from the decomposition of organic matter from sediments under anaerobic conditions (WETZEL 1981). The high temperature combined with the high pH of 8.1, at Site 5, present the situation that up to 15 per cent of the ammonium available is in the toxic free ammonia form (DWAF 1996b). The Target Water Quality Range (TWQR) for un-ionised ammonia is <= 7 µg/l and the Acute Effect Value (AEV) is 100 µg/l . Taking this in consideration the un-ionised ammonia at Site 5 was 129 µg/l and, thus, lethal to the aquatic ecosystem. Another major difference in water quality, however, between Site 1 and Site 5 is the extent of available oxygen, where dissolved oxygen at Site 5 is less than 1 mg/l DO. The acute toxicity of ammonia to fish increases as dissolved oxygen decreases. Ideal conditions existed thus for the occurrence of a fish kill.

No great variability at the different sites was present for total dissolved salts (Figure 3c), pH (Figure 3d) or temperature (Figure 3e). The dissolved oxygen concentrations (Figure 3e) indicate clearly the total lack of oxygen at Site 5, which might be due to the high bacteriological activity (as was noted during the analysis of this sample) at this site.

The results in Table 1 indicate that there was high a zinc concentration in the Crocodile River inflow (Site 2) as compared to the target water quality range (< 0.002 mg/l ) for the aquatic ecosystem. In the Magalies River inflow (Site 5) the manganese concentration and the iron concentration exceeded acceptable levels for the target water quality range for aquaculture (Mn: < 0.1 mg/l ; Fe: < 0.01 mg/l ). Manganese is an essential micronutrient involved in proteoglycan synthesis in vertebrates, but is neurotoxic in excessive amounts. Manganese is mobilised from the sediment under anaerobic conditions, as was the case at Site 5. The manganese concentration of 0.185 mg/l falls within the sub-lethal category (0.1 – 0.5 mg/l ). The iron concentration at Site 5 of 0.016 mg/l is not lethal according to DWAF (1996a). Both manganese and iron do not seem to be very soluble under alkaline conditions therefore their toxicity is reduced.

This was not a major fish kill as only 20 dead carp was seen on the 12^th October and only 2 dead carp during the once-off sampling on the 14^th October 1999. It is, however indicative that a problem existed in the Hartbeespoort Dam area. The fish kill was caused by a large influx of organic material at the Magalies River side of the impoundment. The decomposition of the organic matter depleted the site of oxygen and released the highly toxic ammonia nitrogen under high temperature and pH conditions. Elevated manganese and iron concentrations, probably due to the anaerobic conditions, might have contributed to the lethal situation at Site 5.

1. The origin (natural or will full pollution) of the organic material should be traced to prevent similar incidents from happening again.
2. The nutrient contributions of specifically the Crocodile River should be monitored closely, as the highest nutrient load is entering the system at this point.
3. All stakeholders in the catchment should be made aware of the results of pollution incidents and asked to co-operate in future to implement the best management practices for diminishing nutrient influx into the system.
4. The C. hirundinella bloom is being investigated to determine the time period and extent of the bloom.

WETZEL, R.G. 1983. Limnology 2^nd Edition. Saunders College Publishing, Philadelphia. USA.