Introduction

Widespread anthropogenic pressures on aquatic ecosystems caused by rapid urbanization, industrial expansion and agricultural activities have resulted in increased loads of inorganic and organic materials and, consequently, adverse effects on aquatic environments. Water quality assessment is principally a biological problem in that its primary effect is on living organisms (Gaufin 1973). The quality of water is generally reflected in the indigenous communities of aquatic organisms in terms of the composition and diversity of the species present, their population densities and their physiological condition (Weber 1973). Although over the last few decades a number of biological monitoring methods have been developed for the assessment of the condition of freshwater environments, these have rarely been applied in Africa. In the few African countries where regular monitoring of the quality of the aquatic environment is practiced, the tendency has been to concentrate on physical and chemical water quality determinants.

Diatoms (Class Bacillariophyceae) are abundant in almost all aquatic habitats, constituting an estimated 25% of the Earth’s primary production (Werner 1977). They are increasingly being used to assess water quality as they are generally species-rich, relatively easy to sample, have quick responses to changes in physical and chemical characteristics and are sensitive to subtle changes in environmental conditions or disturbances that may affect other communities only at greater levels of disturbance (e.g., Dixit et al. 1992; Bahls 1993; Lowe and Pan 1996; Stevenson and Bahls 1999). A number of studies have shown that diatoms are excellent biological indicators for many types of pollution in aquatic systems (Patrick and Palavage 1994; Kelly et al. 1995). However, the use of diatoms in water quality programs in African countries has been limited largely due to a dearth of basic phycological research.

Lake Kariba is a large, minimally polluted and oligotrophic man-made lake, although some areas presently are becoming polluted by increased human activities along its shores. The objective of this study was to determine the response of the periphytic diatom community associated with Vallisneria aethiopica, which was the most common and dominant submerged macrophyte during the study period, to human activities or impacts along the shores. It is part of a more comprehensive study aimed at developing a suite of biotic indicators for the assessment of ecological integrity in the shallow zones of Lake Kariba using diatoms and macroinvertebrates.

Materials and methods

Lake Kariba was formed in 1958 by the damming of the Zambezi River. The Zambezi River, with a catchment area of about 1,193,500 km2, is southern Africa’s largest river and is composed of three ecologically distinct zones, the Upper (1078 km), Middle (853 km) and Lower Zambezi (563 km). Lake Kariba lies within the Middle Zambezi zone, between latitudes 16°28′ S and 18°06′ S, and longitudes 26°40′ E and 29°03′ E. The lake has a surface area of 4364 km2 at the normal operation level of 484 m a.s.I, a length of 276 km, an average width of 19 km and an average depth of 29 m. The lake is shared between Zambia and Zimbabwe and has over the years developed a variety of other uses, such as tourism, aquaculture and inshore and pelagic fishery industries.

The study was carried out in the Sanyati Basin of Lake Kariba (Fig. 1), the easternmost of the five major basins forming the lake. For the purposes of the study, the shoreline was divided into areas with minimal human activities, peri-urban areas and urban areas. The areas of minimal disturbance are located in the southern part of the basin adjacent to Matusadona National Park. Ten sites were selected from these areas in a targeted manner to cover the full range of human impacts known to occur along the shores of the basin. Water and periphytic diatom samples were collected monthly at the ten sites between May and July of 2005.

Fig. 1
figure 1

The location of sampling sites (1–10) along the shoreline of the Sanyati Basin of Lake Kariba. Insert The location of Lake Kariba between Zambia and Zimbabwe

Full size image

Notable human activities or developments within 500 m of the shoreline include residential areas, commercial aquacultural activities, maintenance of boats and boating activities, hotels and other tourist accommodation, roads and vehicles, urban development, industrial development, construction activities, sewage effluent disposal, domestic activities (washing and bathing) and effluent disposal from farming activities. To estimate relative degrees of disturbance among the sites, we scored the level of impact of each activity or development at each site qualitatively on a scale of 0 (not occurring at the site) through 1 (low) and 2 (medium) to 3 (high). An overall total human disturbance score was thus obtained, and each site was subsequently categorized as belonging to one of four categories: no evident disturbance (0), little disturbance (1–11), moderate disturbance (12–22) and much disturbance (23–33).

Five water samples (2-l bottles) were collected at each sampling site from different points along a stretch of about 100 m. All samples from each site were put into a 20-l bucket from which a 2-l sub-sample was taken for analysis of the concentrations of ammonium, nitrates, phosphates, total phosphorus (P) and sulphates. Water temperature, pH, dissolved oxygen, conductivity and turbidity were measured in situ. Temperature, pH and dissolved oxygen were measured with a mercury thermometer, a WTW 330i pH meter (Geotech Environmental Equipment, Denver, Colo.) and WTW Oxi 330 oxygen meter (Geotech Environmental Equipment), respectively. The concentrations of ammonium, nitrates, phosphates, total P and sulphates were obtained using the spectrophotometric methods described by Madera et al. (1982). Analysis of variance (ANOVA) was used to test for differences between sites in the physical and chemical variables measured.

To sample for epiphytic diatoms, five plants of the submerged macrophyte Vallisneria aethiopica were randomly chosen from different points along a stretch of about 100 m at each site. The plants were cut underwater with scissors and gently brought to the surface, taking care not to disturb the biofilm on the plants. Three portions of leaf were obtained from each plant. The diatoms were separated from the leaf fragments by vigorously shaking the fragments for 1 min in 1-l of tap water and then lightly scraping each segment using a toothbrush after which a 100 ml sub-sample was preserved in Lugol’s solution. In the laboratory, diatom valves were prepared for light microscopy using acid cleaning following the procedures of Biggs and Kilroy (2000). Two permanent Naphrax slide mounts were prepared for each sample. Diatoms were viewed under at a magnification of 1000× using immersion oil. A minimum of 300 diatom valves was counted from one of the slides prepared from each site. The diatoms were identified to the genus level using standard taxonomic texts.

The number of diatom genera obtained in each sample was recorded. The percentage relative abundance of each taxon in each sample was calculated as

$$ \hbox{Relative abundance }\% = (n_{\rm i}/N) * 100$$

where n i refers to the number of individuals of taxa i in the sample, and N refers to the total number of diatoms counted in the sample. Spearman rank correlations were evaluated using STATISTICA ver. 7 (StatSoft 2004) to determine the relationship between diatom abundance, human disturbance and water physical and chemical variables.

Diatom community diversity was calculated for each sample as Shannon’s Diversity Index H′ (Shannon and Weaver 1949). The mathematical equation for the Shannon-Wiener index is

$$ H^{\prime} = - \Sigma^{S} n_{\rm i}/n \ln n_{\rm i}/n $$

where H′ is the community diversity, n is the total number of organisms in the sample, n i is the number of individuals per taxon and S is the total number of taxa counted per sample.

An evenness index, the Pielou index (Krebs 1989; Clarke and Warwick 1994), was calculated for each sample using the formula

$$ J^{\prime}= H^{\prime}/ \hbox{Log }S $$

in which H′ is the Shannon-Wiener function and S refers to the number of taxa counted per sample. Kruskal-Wallis ANOVA was used to test for differences in diversity and evenness between sites. Spearman rank correlations were evaluated to determine the relationship (if any) that existed between the number of genera per site and the diversity and evenness indices and human disturbance and water physical and chemical variables.

We also calculated 16 diatom index of biotic integrity (DIBI) metrics using the environmental preferences of diatom species taken from published literature (van Dam et al. 1994; Hill et al. 2000) (Table 1). The mean environmental ratings for species within a genus were used as the environmental preferences of the diatom genera. The metrics were scored from 0 to 10, with decreasing scores corresponding to increasing human disturbance. The DIBI for each sample was determined by calculating the sum of the 16 metrics. The individual metrics and the total DIBI were then evaluated against human disturbance scores and water physical and chemical variables using the Spearman correlation.

Table 1 The metrics included in the diatom index of biotic integrity (DIBI)
Full size table

Results

None of the sites was categorized as being highly disturbed (Table 2). Sites 1, 2 and 3 were categorized as having no human disturbance; Site 10 had a low human disturbance score, and the other six sites were moderately disturbed. The highest human disturbance score was obtained for Site 8 (Table 2).

Table 2 Criteria for the human disturbance scoresa within 500 m of the shore zone of Sanyati Basin
Full size table

The mean values for the physical and chemical variables at the ten sites are shown in Table 3. The only variables that differed significantly between the sites were pH and turbidity. The mean pH at Sites 2, 3 and 6 was significantly greater than that at Sites 4, 5, 7, 9 and 10, with the pH at Site 3 also being significantly greater than that of Site 8 (ANOVA, F 9,17 =  4.03, P < 0.05). The turbidity at Site 2 was significantly greater than that at Sites 4 and 5, while that of Site 6 was significantly greater than that at Sites 3, 7, 8, 9 and 10 (ANOVA, F 9,17 = 2.55, P < 0.05).

Table 3 Mean (±SE) levels of selected physical and chemical variables of the water at the ten sites (n = 3 for variables at each site except where indicated otherwise)
Full size table

A total of 9993 diatoms were counted and 40 genera identified (Table 4). Nine genera, Achnanthidium, Cocconeis, Diatoma, Encyonema, Fragilaria, Gomphonema, Navicula, Nitzschia, and Synedra occurred at all ten sites. The two most abundant genera were Achnanthidium and Gomphonema, which comprised 23.4 and 42.9% of the total number of diatoms counted, respectively. Achnanthidium was most abundant at Sites 1 and 3, making up 63.1 and 41.1%, respectively, of the total number of diatoms at those sites and was generally found in relatively high numbers at all remote sites with no obvious human disturbance. Gomphonema dominated at all of the other eight sites, making up more than 50% of the total number of diatoms at Sites 4, 5, 6 and 8 (Table 4), all of which had moderate human disturbance scores.

Table 4 Percentage relative abundances of periphytic diatoms associated with Vallisneria aethiopica in the shallow marginal waters of Sanyati Basin
Full size table

The highest mean number of genera was obtained at Site 3 (20 ± 1.5) and the lowest at Site 4 (7.7 ± 1.5) (Table 5). Site 4 was the only site with a mean of fewer than ten diatom genera. The Shannon diversity index ranged from a high of 1.9 to a low of 0.9. Sites 2, 3, 10 and 11 showed relatively high levels of diatom diversity, while Sites 4 and 8 showed the lowest diversity of diatoms as well as the lowest evenness (Table 5). Taxa evenness ranged from 0.4 at Site 4 to 0.7 at Sites 3 and 9. There were no significant differences in the number of genera per site, the Shannon diversity index and the Pielou evenness index among the ten sites (Kruskal-Wallis ANOVA, P > 0.05). The number of genera per site and Shannon diversity was positively and significantly correlated with dissolved oxygen, while the number of genera was also significantly and negatively correlated with ammonium–nitrate concentration (Spearman correlation, P < 0.05).

Table 5 Mean number of genera, diversity and evenness (±SE) at ten sites along the shores of Sanyati Basin
Full size table

The relative abundances of Achnanthidium, Denticula, Pinnularia, Rhopalodia and Stauroneis were negatively and significantly correlated with the disturbance score (Spearman Rank correlation, P < 0.05) (Table 6), suggesting that increases in human activities in the shallow marginal waters of the basin resulted in reduced abundance of these genera. The abundance of Gomphonema was significantly and positively correlated with the disturbance score (Spearman Rank correlation, P < 0.05) (Table 6), suggesting that Gomphonema was more tolerant of human impacts within the shallow marginal waters than other genera. The abundances of five genera (Achnanthes, Cymbella, Denticula, Encyonema and Pinnularia) were negatively and significantly correlated to total P, those of two genera (Bacillaria and Mastogloia) with nitrate–N (NO −3 ), those of two genera (Diatoma and Navicula) with ammonium–N (NH +4 ) and those of two genera (Eunotia and Mastogloia) with turbidity. Thus, the relative abundances of 13 diatom genera associated with V. aethiopica decreased with increasing disturbance score in the shallow waters of the Sanyati Basin.

Table 6 Spearman correlation coefficientsa for the relationship between percentage relative abundances of the periphytic diatoms, disturbance scores and selected physical and chemical characteristics of the water in the shallow marginal zones of Sanyati Basin
Full size table

Of the 16 DIBI metrics, 11 were significantly correlated with at least one of the 12 environmental variables (Table 7); 9 of the 16 metrics used in calculating the DIBI were significantly correlated with the DIBI (Table 7). Among the environmental variables the disturbance score was the only one that was significantly and inversely correlated with the DIBI (Table 7).

Table 7 Spearman correlation coefficientsa for the relationship between the DIBI and its component metrics
Full size table

The DIBI scores ranged between 113.8 (Site 4) and 145.9 (Site1) (Table 8). There were significant differences in DIBI scores among the sites (ANOVA, F 9, 17 = 4.21, P = 0.005). The DIBI scores at Sites 1 and 3 were significantly greater than those at Sites 2, 4, 5, 6, 7 and 8. Site 4 also had DIBI scores that were significantly lower than those at sites 7, 9 and 10. Thus, in general, remote sites with lower human activities had higher DIBI scores than those located where human activities were high. These data suggest that increases in human disturbances result in a decrease in the biological integrity of the diatom community associated with V. aethiopica in the shallow marginal water of Sanyati Basin.

Table 8 DIBI scores obtained from diatoms associated with V. aethiopica along the shallow marginal waters of Sanyati Basin
Full size table

Discussion

This study is the first to examine the responses of periphytic diatoms to human activities in a water body in Zimbabwe. Diatoms have been used successfully elsewhere to determine the ecological conditions of water bodies (Squires et al. 1979; Duncan and Blinn 1989; Bahls et al. 1992; Hardwick et al. 1992; Biggs and Hickey 1994; van Dam et al. 1994; Hill et al. 2000; Blinn and Bailey 2001; Köster and Hübener 2001). Most of the work on the use of diatoms has largely concentrated on epilithic diatoms in lotic systems. The present study has shown that periphytic diatoms on V. aethiopica may be used as tools to assess the impact of human activities on shallow inshore areas of a lentic system, Lake Kariba.

The most abundant genus at remote sites with no human disturbance was Achnanthidium, while at those sites with increased human activities Gomphonema dominated the diatom assemblage. The study showed that Achnanthes, Cymbella, Denticula, Encyonema and Pinnularia were sensitive to increased concentration of total P, Bacillaria and Mastogloia to an increase in nitrate–N, Diatoma and Navicula to an increase in ammonium–N and Eunotia and Mastogloia to turbidity. A number of other studies have also shown that diatom communities are affected by changes in physical and chemical variables, including changes in the concentrations of nutrients (Bahls et al. 1984; Charles 1985; Anderson et al. 1990).

The diatom assemblages from most of the sites located in remote areas and subjected to minimal human impact had relatively higher numbers of diatom genera and higher indices of diversity and community evenness than sites located in areas with relatively higher levels of human disturbance. The use of biotic indices is a water quality assessment approach that has been developed to avoid shortcomings associated with using the indicator species and community structure (richness, diversity, evenness) approaches (Hill et al. 2000). We found that although the number of genera, diversity and evenness tended to be higher at remote areas with minimal human disturbance, there were no significant differences among sites; in addition, the number of genera, diversity and evenness showed no significant correlations with the level of human disturbance. Unlike the indices based on community structure, the DIBI was more responsive to disturbance and showed a significant and inverse correlation with human disturbance.

In this study generic rather than species level resolution was used largely due to the limited information available on both the taxonomy of diatoms and diatoms in general in the Southern Africa region as well as the need to develop an easy-to-use, cost-effective diatom-based monitoring system for the lake that can be used by non-experts in algal taxonomy. The study revealed that human impact on the shallow water zones of Lake Kariba could potentially be evaluated using genus level taxonomic resolution of diatoms. Although there is some debate on the use of genus versus species level identification for assessing the quality of freshwater environments, a number of studies have found that while some information is lost, genus-based indices generally give acceptable indications of environmental quality (Kelly et al. 1995; Hill et al. 2001; Wunsan et al. 2002). Periphytic diatoms associated with V. aethiopica are therefore a potential tool that may be used to determine the ecological conditions of the shallow water margins in Sanyati Basin, and there is need for further exploration of their use for routine monitoring of the impact of anthropogenic activities, especially the development of a composite DIBI.

Conclusion

We report here the preliminary results of an ongoing study on the use of diatoms and macroinvertebrates as a tool for assessing ecological conditions within the shallow marginal waters of a large manmade reservoir, Lake Kariba. Although further work is necessary, this study shows that the diatom communities associated with V. aethiopica respond to differences and variations in human-induced effects in the shallow waters, thereby making them candidates for further development as freshwater biomonitoring tools in Zimbabwe.