Five sampling sites were selected along the Mukuvisi River. The Seke Road Bridge was the first site. At this site, industrial effluent (from Graniteside and City Centre) and leachate from landfills are presumed to be discharged into the river. Site 2 was the Magaba bend industry. Effluent is expected to be discharged from the informal industry in the Magaba area into the river. Site 3 was Glen Norah where most of the effluent from the Southert on industry is discharged into the river. Site 4 was the Amalinda farm. Effluent from urban cultivation such as fertilizers and pesticides are leached into the river. Site 5 was pension farm where effluent from the Firle Sewage Treatment Plant is presumed to be discharged into the river. The sampled sites in the Mukuvisi River are shown in Figure 1.

Sediment and fish were sampled at five selected sites in the Mukuvisi River once after every two months (from September 2008 to April 2009). Sampling occurred between 1000 and 1400 hours on all occasions to minimize variation due to climate factors. Sediment samples were collected using an acid-washed polyethene corer to a depth of 3 cm. At each sampling site three replicate samples were collected at an interval distance of 15 m. The sediments were stored in sterilized polyethene bags and returned to the laboratory in sealed containers containing ice packs for metal analysis. Live fish samples were collected at each site using 89 mm meshfyke nets and a DC electro fisher (Smith-Root Type IV-A). Electro fishing was done for 10 minutes at each sampling station.

Temperature, pH, conductivity and dissolved oxygen (DO) were measured in situ. Water temperature and DO were measured using a HACH oxygen 330i meter; pH and conductivity were measured using a HACH pH meter and WTW 330i conductivity meter, respectively. Water samples were analysed in the laboratory for total and reactive phosphorus (orthophosphate), total nitrogen, nitrate and ammonia. The nutrients were measured with a

HACH water analysis kit (HACH DR/2010 portable data logging spectrophotometer) using filtered water samples filtered through 0.45 μm Whatman GF 47 mm filters, except for total nitrogen and phosphorus analysis. The methods are described in detail in [21] .

The two step acid digestion process used for the sediments is a strong digestion method and was chosen so that the total metal content could be estimated. Metals in the sediments were analysed using flame atomic absorption spectrophotometry (FAAS) [22] after acid digestion to extract the metals from the sediments. This procedure consisted of two parts:

1) Digestion: The sediments were oven-dried at 180˚C in a muffle furnace and large aggregates were broken up and then 20 ml nitric acid and 5 ml perchloric acid were added to 5 g of the oven-dried sample. The mixture was heated on a hot plate until fumes were produced and allowed to cool to room temperature.

2) Acidification: 20 ml of 50% hydrochloric acid was added to the mixture from the first digestion. The acidified mixture was heated until boiling and then was cooled to room temperature. The acidified mixture was filtered and distilled water was added to the filtrate in a volumetric flask up to the 100 ml mark. The digested sediments were then analysed for metal amounts in an atomic absorption spectrometer [22] .

In the laboratory, fish were dissected with clean autoclaved stain less teel instruments. One gram (wet weight) of gill, liver, kidney and muscle tissue of C. gariepinus was measured using an electronic balance. The tissues were placed into Petri dishes to dry at 120˚C until they reached a constant weight. Drygill, liver, kidney and muscle tissue were placed in digestion flasks to which 5 ml perchloric acid and 10 ml nitric acid were added. All acids used were of analytical grade quality. The digestion flasks were placed in an oven at 130˚C until all materials dissolved. The prepared tissue samples were analysed using flame atomic absorption spectrophotometry (FAAS).

Spatial differences in limnochemical aspects of the water among sites sampled in the Mukuvisi River were investigated using Kruskall Wallis ANOVA at 5% level of significance respectively using Paleontological Statistics (PAST) software Version 190 [23] . The bioconcentration factor (BCF) was calculated after metal concentrations were determined for sediment and fish tissues. The BCF is defined in this study as the concentration of a metal in a fish tissue in relation to the concentration of that metal in the sediment surrounding that tissue, and was calculated using the formula [24] :

BCF values greater than 1000 are considered high and those under 250 low. Those between these extremes are classified as moderate. Differences in bioconcentration factors were assessed using the Kruskall Anova at 5% significance level.

Canonical correspondence of the metal concentration in the sediment and fish tissue to water limnochemistry was analysed using CANOCCO 5 [25] .

The pH of the water at all sites sampled in the Mukuvisi River was slightly acidic to alkaline (6.1 - 7.3) during the period of study, only site 1 had a pH below the lower limit accepted for effluent and river water (Table 1). The mean temperature varied with the time of sampling ranging from 20.8˚C to 22.1^°C. Conductivity readings were above 540 µS∙cm⁻¹ at all sites. Dissolved oxygen was highest at site 2 (5.0 mg∙l⁻¹) and lowest at site 5(1.1 mg∙l⁻¹). Phosphate levels exceeded the WHO limits (0.5 mg∙l⁻¹), whilst nitrate levels remained within the acceptable limits of 10.0 mg∙l⁻¹. The highest levels of ammonia were recorded at site 3 (0.18 mg∙l⁻¹) whilst the lowest levels were recorded at site 1 (0.03 mg∙l⁻¹). There were significant differences (Kruskall ANOVA; p < 0.05) in the concentration of D.O, RP and TP among sites sampled in the Mukuvisi River.

All the three metals (Al, Mn and Co) were recorded in sediments at all sites (Table 2). Aluminium had the highest concentrations in the sediments followed by manganese. The levels of aluminium ranged from 17.20 - 215.00 mg∙kg⁻¹ in sediments. Manganese levels ranged from1.44 - 72.40 mg∙kg⁻¹, whilst the levels of cobalt ranged from 0.20 - 2.85 mg∙kg⁻¹. No fish were collected from sites 1 and 5 (Table 2). Aluminium was not detected in gills and kidney on site 3, whilst manganese was not detected in fish muscle and liver. Manganese was not detected in the kidney on sites 2 and 3. Cobalt was not detected in the liver on sites 2 and 3. Aluminium ranges in the fish tissues were; muscle (0.27 - 1.18 µg∙g⁻¹), gill (ND-1.35 µg∙g⁻¹), kidney (ND-2.7 µg∙g⁻¹), liver (0.06 - 1.13 µg∙g⁻¹). Manganese ranges in the fish tissues were; gill (0.15 - 0.85 µg∙g⁻¹), kidney (0.10 - 0.30 µg∙g⁻¹), whilst cobalt ranges were; muscle (0.01 - 0.1 µg∙g⁻¹), gill (0.02 - 0.13 µg∙g⁻¹), kidney (0.01 - 0.08 µg∙g⁻¹) and liver (0.03 - 0.05 µg∙g⁻¹). The prominence (hierarchical ranking) order of mean concentration of metals in the sediments was; Al > Mn > Co. Prominence order of metals in the muscles and liver of the catfish was; Al > Co >Mn, whilst in the gills and kidneys it was; Al >Mn> Co (Table 3).

Metal accumulation pattern (bioconcentration factors) differed for each organ analysed and comprised of the following pattern: Aluminium: Liver > Gill > Muscle > Kidney, Manganese: Kidney > Gill > Liver > Muscle and for Cobalt: Liver > Muscle = Kidney > Gill (Table 4). Aluminium had the highest bioconcentration factors

Note: ND: not detected, Rs: ranges, M: mean.

Note: ND: Not detected.

in all organs of the catfish analysed except in the kidney where Mn had a significantly higher bioconcentration (Table 4). There was a significant difference (ANOVA, p < 0.05) in the bioconcentration the three metals from the sediments into the catfish stomach muscles and liver.

Canonical correspondence analyses show that there was a strong association in aluminum concentration in the kidney, liver and gill of the catfish with the levels of DO and TN in water in Mukuvisi River particularly at site 2, whilst there is a strong association of Al concentration in the catfish muscles with nitrate levels in the water at Mukuvisi site 1. The first two axes explain 87.61% of the variation in Al concentration in the catfish tissues sampled (Figure 2). A narrow angle of attachment shows that temperature; conductivity and ammonia values were closely linked in the water phase in Mukuvisi River. The correspondence analysis for manganese was not valid as the number of ND non-detectable values in the catfish tissues which were treated as missing figures or zeros exceeded the possible explanatory variables (water limnochemical parameters). Figure 3 shows that the cobalt concentration in the muscle and gill of the catfish had a strong association with the nitrate, ammonia and DO values in the Mukuvisi River. Cobalt concentration in the kidney and liver of the catfish sampled in the Mukuvisi River had a strong association with the RP values in the water. The first two axes explain only 64.13% of the variation in the concentration of Co concentration in the catfish tissues analysed in this study. A narrow angle of attachment shows that temperature, conductivity and TN values are closely related in the water phase in Mukuvisi River (Figure 3).