Report No: KV138/02

October 2002



This report is based on data collected in the Western Cape during two previous Water Research Commission (WRC)-funded projects. The first focused on the "effects of water quality variables on riverine ecosystems" (Dallas & Day 1993, Dallas et al. 1995) and the second on "the development of tools for assessing regional water quality guidelines" (Dallas et al. 1998). During these projects, research into and development of the key bioassessment tool used in South Africa, namely SASS4, was undertaken (Dallas 1995, Dallas 1997). A subsequent project, commissioned by the Department of Water Affairs and Forestry (DWAF), focused on the "derivation of ecological reference conditions for riverine macroinvertebrates" (Dallas 2000a, 2000b, Dallas & Fowler 2000, Fowler et al. 2000).

This report, therefore, represents an amalgamation and analysis of data from the WRC and DW AF projects, addressing specific objectives related to aquatic bioassessment and defining ecological reference conditions for riverine macroinvertebrates. The current project duration was from January 2001 until 31 December 2001. The greater part of the report formed the basis of a thesis submitted in fulfilment of the requirements for the Degree of Doctor of Philosophy, Department of Zoology, University of Cape Town (December 2001).


The protection of water resources in order to ensure their long term sustainability and the utilisation of these resources in the most efficient and effective manner, within the constraints set by the requirements for their protection, are the key interdependent components of the South African National Water Act (Act No 36 of 1998). This integrated approach to resource protection requires that measurable and verifiable resource quality objectives (RQOs), that clearly define acceptable levels of protection for water resources, be established. The RQOs have four components: requirements for water quantity (water level and flow), requirements for water quality (chemical, physical and biological characteristics of the water), requirements for habitat integrity (of instream and riparian habitats), and requirements for biotic integrity (health, assemblage structure and distribution). These components highlight the complexity and interactive nature of aquatic ecosystems. An alteration in anyone of the independent components, namely water quantity, water quality or habitat integrity will invariably lead to a change in biotic integrity. Biological assessment, or bioassessment, is a tool that integrates the effects of these components. Its utility in assessing environmental condition, especially water quality and general river condition, and in defining reference conditions for macroinvertebrates in river ecosystems, forms the basis of this report.

In particular, the extent of spatial and temporal variability in macroinvertebrate assemblages in South Africa and the implications of this variability for bioassessment and defining reference conditions, are examined in this report. Briefly, a reference condition is the condition that is representative of a group of minimally-disturbed sites, i.e. reference site, organised by selected physical, chemical and biological characteristics (Reynoldson et al. 1997) and that enables the degree of degradation or deviation from natural conditions to be ascertained. Two key questions addressed in this report are:

  1. to what extent is spatial heterogeneity a feature of lotic ecosystems in South Africa, and is it possible to partition intrinsic spatial variability in such a way that defining reference conditions based on several similar reference sites is feasible?
  2. to what extent is temporal heterogeneity a feature of lotic ecosystems in South Africa, and is it possible to account for may be detected when comparing it to a derived reference condition?


General aim

Central to this report is the question of whether ecological reference conditions are realistic and attainable entities, or whether intrinsic spatial and temporal heterogeneity of and variability in lotic systems are such that establishing reference conditions is not possible. The key questions posed, therefore, relate to the extent to which macroinvertebrate assemblages vary spatially and temporally, and the implications of this variability to bioassessment and defining reference conditions. The question has been addressed by examining regional variability of macroinvertebrate assemblages within the context of assessing the utility of the spatial framework for regional classification of reference sites; by examining variability at the level of habitat; by examining temporal variability; and by identifying the environmental variables contributing to the variability in macroinvertebrate assemblages. To answer these questions patterns of spatial and temporal heterogeneity were examined in two distinct geographic regions, and at the level of individual taxa, macroinvertebrate assemblages and the derived biotic index, i.e. SASS scores.

Specific aims are:


E4.1 The protocol for deriving reference conditions

The protocol developed in Dallas (2000b) formed a sound basis for data analyses when applied to another region, i.e. the Western Cape. Each of the steps described in the protocol are important when reference conditions are established. Of significance are the regional differences in the relative importance of biotopes, biotope preferences of individual taxa, and biotope and seasonal differences in macroinvertebrate assemblages. In the Western Cape, data limitations prevented the calculation of ratios. Instead absolute values were used and biological bands were derived based on the relationship of ASPT to SASS4 Score, This proved to be a useful means for data interpretation and subsequent detection of disturbance at a monitoring site.

E4.2 Spatial variability in macro invertebrate assemblages at the regional level

In general, a priori regional classification of sites using the hierarchical spatial framework developed in South Africa provided a useful framework for preliminary classification of reference sites. Within-class variability (i.e. within a bioregion, ecoregion or bio-subregion etc.) was always lower than between-class variability (i.e. between bioregions, ecoregions, bio-subregions, etc.). Groups of sites based on a posteriori analysis of macroinvertebrate data, however, provided a more robust classification than any of the regional classifications.

Spatial classifications therefore offer geographic partitions within which to expect somewhat similar conditions and regional reference sites selected within the context of the hierarchical spatial framework are likely to be more representative of specific river types than those selected without using the spatial framework. Some variability within both regional classes and groups of sites with similar macroinvertebrate assemblages could not be accounted for at the regional or subregional levels, suggesting the presence of additional factors acting at a lower scale such as site or habitat.

The need for additional partitioning of variability at a lower scale is thus highlighted, as is the need for the classification of sites to be an iterative process that allows for subjective a priori regional classifications to be modified on the basis of independent, objective a posteriori classification of biological assemblages. The lack of distinctiveness in macroinvertebrate assemblages from mountain streams and cobble-bed foothills, both of which are upland subregions, suggests that, from a practical perspective, and within the confines of bioassessment, mountain stream and foothill-cobble bed sites may be grouped together.

E4.3 Spatial variability in macro invertebrate assemblages at the habitat level

Spatial variability at the level of habitat, specifically SASS-biotopes, revealed that several taxa exhibited a degree of biotope specificity, with some taxa recorded more frequently in one biotope rather than another. The relative importance of a biotope as a habitat for macroinvertebrates, as a reflection of both its availability and its utilisation by aquatic organisms, also varied regionally. The importance of hydraulic condition coupled with substrate type became apparent with differences in taxa observed within a biotope-group, e.g. stones-in-current versus stones-out-of-current. Seasonal differences in the distinctiveness of biotopes were observed in the Western Cape, with distinctiveness more pronounced in autumn, under low-flow conditions, in comparison with less pronounced biotope specificity in spring.

In terms of SASS Scores, stones-in-current/stones-out-of-current (SIC/SOOC) was shown to be the most important SASS biotope-group and taxa associated with it contributed the highest percentage to SASS Scores calculated at the site level. SIC/SOOC was also the most consistent in terms of its associated macroinvertebrate assemblage. There was a significant positive relationship between SASS4 Score and number of taxa with number of SASS-biotopes sampled and a negative correlation between ASPT and number of SASS- biotopes sampled.

The importance of sampling SASS-biotopes separately is clearly demonstrated. This enables SASS data to be interpreted on a "per SASS-biotope" basis in instances where one or other SASS-biotope is absent from a monitoring or reference site. By sampling SASS- biotopes separately, differences in the availability of SASS-biotopes between reference and monitoring sites may be taken into account, and subsequent results will thus reflect conditions other than those resulting from habitat differences. Flow conditions and season are important additional factors that need to be taken into consideration when doing SASS, defining reference conditions and interpreting SASS data.

E4.4 Temporal variability in macro invertebrate assemblages

Generally, seasonal differences were less pronounced than biotope-related differences and were more prevalent in the Western Cape compared to Mpumalanga. SASS Scores, specifically the number of taxa and ASPT, were significantly different among seasons in the Western Cape, with fewer taxa recorded in winter compared to summer and significantly higher ASPT values recorded in winter and spring in comparison to summer and autumn. Whilst more taxa were recorded in autumn than in spring, a higher proportion of sensitive and high-scoring taxa were recorded in spring. Temporal variability did not, however, curtail the detection of disturbance at monitoring sites.

In terms of defining reference conditions cognizance should be taken of the sampling season, particularly in regions that exhibit a relatively high degree of seasonal variability such as the Western Cape. When identifying expected or reference taxa for a seasonally variable region, details pertaining to the seasonal trends in individual taxa should be provided, since seasonal absences of certain taxa may affect the bioassessment results. Initial classification of reference sites based on seasonally-composite data provides a more robust classification of reference sites and is to be recommended.

E4.5 Environmental variables

Environmental variables at all scales were identified as potential predictor variables and were thus considered important in grouping sites with similar macroinvertebrate assemblages. In Mpumalanga, catchment-level variables included altitude and longitude, lending support to the observed distinction in macroinvertebrate assemblages between upland and lowland sites. Temperature, a correlate of altitude, was important, as was the depth of the shallow-water habitat (e.g. cobble rime, bedrock rapid). Biotope-group predictor variables varied to some degree with aspects such as geological-type, canopy cover and the percentage of mud identified as important in the stony-habitat classification, in comparison to the depth of the deep-water habitat and the percentage of gravel/sand and mud in the vegetation classification.

The utility of a spatial framework within which reference sites are selected and bioassessment is undertaken is confirmed by these results. The importance of additional factors such as substratum that influence macroinvertebrate assemblages, is highlighted by the number of river type variables, at the scale of site and habitat, that were identified as important discriminators of macro invertebrate assemblages in both the composite classification and biotope-specific classifications.

E4.6 Variability in macro invertebrate assemblages within a region

The final chapter (Chapter 7) draws together aspects from all preceding ones, by examining spatial and temporal variability of macroinvertebrate assemblages within the most spatially and temporally variable group of sites identified in Chapter 3, namely upland sites of the Fynbos bioregion of the Western Cape. The degree of dissimilarity was a minimum of 47%, even when differences in the availability of biotopes, i.e. separating sites with- and without-vegetation, were included. Results confirmed that differences between sites in the two subregions, namely mountain streams and foothill-cobble beds, were not significant, although upland sites did form distinct Groups, particularly when mountain stream sites were considered in isolation.

Of importance from a bioassessment perspective, SASS Scores calculated for these upland sites were less variable than the macroinvertebrate assemblages and did not preclude the detection of disturbance at monitoring sites. Biological bands derived for data interpretation that utilised the relationship between ASPT and SASS4 Score provided a means whereby variability resulting from differences in the availability of biotopes and seasonal differences could be taken into account. Examination of the relative frequency of occurrence of taxa within each biological band revealed three different trends in response to increased disturbance. One group of taxa, many of which were high-scoring, sensitive taxa characteristic of minimally-disturbed upland sites, and many of which showed a preference for the stones-in-current biotope, decreased as disturbance increased. A second group of taxa, including several tolerant and low-scoring taxa such as Muscidae and Oligochaetes, increased in response to disturbance. A third group of taxa remained relatively unaffected by increased disturbance and included several hemipterans, dragonflies and damselflies.

Development of biocriteria is an important process in the effective protection of aquatic ecosystems and the confidence with which a judgement of biological condition is made depends on the soundness and scientific validity of the bioassessment tool (e.g. the biotic index) and the reference condition defined.

E4.7 Incorporation of abundance into SASS

SASS is a qualitative index and does not include abundance as part of the index. Rather, abundance is used as a descriptive aid for data interpretation. Examination of data in this study showed that there is a highly significant linear correlation between unweighted and weighted SASS Scores. This indicates that the inclusion of rank abundances did not alter the assessment of disturbance appreciably. The key difference was a broadening of the SASS4 Score range, particularly of the upper limit, suggesting that greater resolution may be attained between minimally disturbed sites and mildly disturbed sites, i.e. biological bands A and B. The adherence to the current practice of using the rank abundance estimates as additional descriptive and interpretive tools of the macroinvertebrate assemblage at a site is probably sufficient for interpreting bioassessment data.


In conclusion, this study has shown that spatial and temporal heterogeneity are features of South African river systems. For effective management of these lotic systems it seems clear that intrinsic spatial and temporal heterogeneity and variability need to be understood and incorporated within the context of bioassessment. On the basis of the results of this study, it is possible to partition spatial variability such that defining reference conditions based on several similar reference sites is feasible. Adopting a regional framework, within which reference sites are selected and reference conditions defined, facilitates initial partitioning of variability resulting from differences at the regional and subregional levels.

Further spatial partitioning is necessary at the habitat level, specifically separation of SASS-biotopes during the bioassessment and analysis phase. In this way, differences in the availability of SASS-biotopes between reference and monitoring sites may be taken into account, and subsequent results will thus reflect conditions other than those resulting from habitat differences. Temporal variability, whilst not as obvious as biotope differences, needs to be considered when defining reference conditions, with certain taxa more common in one or other season. The importance of seasonal differences was shown to vary between geographic regions. Temporal variability did not, however, curtail the detection of disturbance at monitoring sites.

Notwithstanding the spatial and temporal variability, and the identification of environmental variables at all scales acting on and influencing macroinvertebrate distributions, it is possible to define a reference condition for macroinvertebrates. This study has shown that a reference condition comprised of biocriteria in the form of SASS scores and expected SASS-taxa allows the identification of disturbed sites.

Recommendations for future research and management aspects are provided below.