Western Cape River and Catchment Signatures
REPORT NO 1303/1/05
October 2005
Executive Summary: 

Introduction

In a previous Water Research Commission Project (final report 754/1/01 “Assessing the ecological relevance of a spatially-nested geomorphological hierarchy for river management”: King & Schael 2001) invertebrate data were collected from headwater reaches (mountain streams and foothills) of 17 relatively undisturbed (here called Least Disturbed or LD) sites and 11 disturbed (D) sites in the Western Cape, all sites being located within the Fynbos bioregion.  In each site up to 12 invertebrate samples were taken from the full available range of hydraulic conditions (substratum-flow combinations), with ancillary physico-chemical data taken at the level of sampling area or site.  A species list was compiled for each of the LD sites, from its set of invertebrate samples.  These lists were used in multivariate similarity analyses to search for sites that were similar in terms of their invertebrate assemblages.  The hypothesis put forward was that the sites would cluster by geomorphological/biological zones, that is, into a mountain-stream group of invertebrates and a foothill group.  Instead, the sites clustered by catchment, with a second division by channel bed form (bedrock or alluvial).  River zone only appeared at the third level of division.  Within any one catchment, with the invertebrate samples dealt with individually instead of pooled, each sample clustered with others from its site and separate from those of other sites.  These distinctions have been termed catchment and river signatures.

The possible implications of such signatures are wide: traditional management of South African rivers recognises all headwater systems within any one biome as being essentially the same.  With this reasoning, some rivers could potentially be sacrificed to land and water developments with the knowledge that other similar rivers remained.  The data from King & Schael (2001), however, suggested that this assumption may not be true and that all rivers may be different, in ways as yet not understood.  The small follow-up project reported on here was thus designed to investigate the nature and causes of the river/catchment signatures, and to assess if there were management implications.

The study was a desktop one, confined to additional analyses of the database compiled in the original project (King & Schael 2001).  A number of univariate and multivariate analyses were done to address the project objectives as listed below, investigating the nature of the species assemblages, their secondary environmental causes, and the effect of disturbance on catchment signatures as noted in King and Schael (2001).

Project Objectives

The project objectives, as agreed in the original contract between the University of Cape Town and the Water Research Commission are summarised below.
  1. Explain the nature and proximal causes of catchment and river signatures.
  2. Identify, if possible, the underlying causes of the catchment and river signatures.
  3. Describe the link between various kinds of disturbance and the progressive loss of catchment and river signatures.
  4. Reach consensus with other researchers on the management implications of the finding of Objectives 1 – 3, and transfer the conclusions to the management arena.
  5. Assess the influence (if any) of the selection of sampling points within a site on SASS scores.  Validate the present biomonitoring techniques, or suggest modifications if necessary to enhance standardised sampling.
  6. Refine the database framework (MS Access program interface) for wide user-ship in scientific and technical management arenas; complete user interface for easy data access and additions.
These objectives have been addressed in full and are reported on in this document.  Chapter 1 provides the background and impetus for the research.  Objectives 1 – 3 are addressed in Chapters 2 – 5, and objective 5 in Chapter 6.  Improvements to the database are reported in Chapter 7.  Objective 4 was addressed through a workshop at the University of Cape Town attended by a select group of research scientists, which is reported in Chapter 8.  Conclusions and recommendations appear in Chapter 9.

Are catchment signatures real or an artefact of the research? (Chapter 2)

The main objective of this chapter was to investigate whether the catchment signatures described by King & Schael (2001) were an artefact created either by the sampling regime or by the process used for identifying the invertebrates to species level.  It was deemed possible that a bias in the results could have been brought about by the combinations of habitats sampled in each site: sites within any one catchment could have had a similar suite of available habitats and this could have resulted in their invertebrate samples grouping together.  Alternatively, sites/catchments could have appeared more similar or dissimilar than they really were because their invertebrates had been identified by different staff.  Several permutations of the data set were, therefore, analysed in an attempt to detect any possible bias.  First, the habitat types were standardised for all LD sites by selecting only those habitats that were common to all sites.  The invertebrate assemblages of these habitat types were re-analysed using hierarchical clustering, multi-dimensional scaling (MDS) plots and analysis of similarity (ANOSIM).  Second, the same set of habitat-types was used, but their invertebrate samples were analysed using different combinations of taxonomic levels (species to family) and groups (e.g. only the mayflies), to detect any possible bias in the data.  Through all these permutations of the data, the catchment signatures persisted.  Only the Chironomidae (midges) displayed a weak catchment signature, reflecting their good dispersal abilities and their ubiquitous presence across the landscape.  In summary, the catchment was shown to be the landscape feature that best explained invertebrate species distributions.

This conclusion supports the findings of Wishart et al. (2002), who reported that the genetic structure and flow of selected invertebrates and fish revealed that catchments in the Fynbos bioregions of the Western Cape were unique entities.  
That study and this one identify catchment as an important large-scale unit with biological and ecological significance and, in fact, Wishart et al. (2002) regarded catchments as the best functional unit for conservation of instream biota.

What characteristics of the data set cause the catchment signatures (Chapter 3)

Investigation of the characteristics of the invertebrate data that produced catchment signatures revealed that there was no one over-riding cause.  It was concluded that the signatures were not caused by unique species within each catchment, nor by a unique mix of taxa in each catchment, nor by unique proportions of the same set of taxa within each catchment.  Instead, the signatures were caused by subtle changes of species within each major taxon group from catchment to catchment.  As an example, of the 11 stonefly species collected in the study, three occurred only in the Eerste-Molenaars catchment group, one only in the Olifants-Berg group, and the Breede had a mixture of these species.  Those stonefly species in the Eerste-Molenaars group were shredders, whilst those in the other two catchment groups were deposit feeders, suggesting fundamental differences in the ecological functioning of the sites.  Of interest was the maintenance of the basic proportional presence of each major taxon group through all the species changes: in comparisons of the major catchment groups, for instance, the percentage of the taxa that were true flies held at 29-44%, whilst stoneflies always contributed 4-9%.

Of all the analyses, the only one that eradicated the catchment signatures was that using functional feeding groups (FFGs).  In this analysis, species and morph species of each invertebrate taxon is replaced by the manner of feeding it employs.  This resulted in the samples re-grouping in a way that coarsely reflected a split between mountain and foothill sites and thus reflected the more commonly held view that the main division of invertebrate assemblages is between the various longitudinal zones of the river.  The two sets of results described above suggest that sites in each of the river zones (mountain or foothill) were functioning in a similar way, with the same proportions of the major invertebrate groups, even though the actual mix of species differed.

What effect does disturbance have on catchment signatures? (Chapter 4)

Although the original similarity analyses (King & Schael 2001) revealed that most disturbed (D) sites were distinctly different from the least-disturbed (LD) sites, the subsequent analyses in this project did not reveal a clear picture of what caused this.  Most D sites had similar invertebrate densities, levels of diversity and species richness to the LD sites.  Standard measures used to detect differences between reference (LD) sites and D sites, such as % ephemeropterans, plecopterans and trichopterans (EPT), number of families, % dipterans and % non-insects, also did not demonstrate statistical differences, although box plots did show small differences between the two.

Once again, subtle changes of species, rather than shifts in major taxonomic groups, distinguished D from LD sites, and the overall impression was that even though the D sites were showing visual signs of disturbance, this was not sufficiently intense, except for one or two sites, to cause major changes in invertebrate assemblages.  Thus, an array of disturbances, such as a picnic area, a dam with downstream water loss, agricultural land and a bull-dozed riverbed (Table 4.1), did not appear to be having a major affect on ecosystem functioning (as judged by invertebrate assemblages), whilst a dam with a bottom release of very cold water (Holsloot), and an infestation of grey poplar (Cecilia) appeared to be more intense disturbances causing the early stages of more substantial ecosystem change.

What are the underlying environmental factors causing the catchment signatures? (Chapter 5)

The shifts in species in LD sites from catchment to catchment must be caused by underlying environmental conditions.  Possible environmental candidates are water chemistry, the presence or absence of fish, biogeographical influences such as temperature, topography and rainfall, or paleogeographical influences such as past drainage networks.  These were assessed in terms of their possible contributions to catchment signatures.  The main catchment signatures were caused by: 1) the Olifants and Berg River sites, which clustered together; 2) the Eerste and Molenaars Rivers sites, which clustered together; 3) the Breede River sites; 4) the Palmiet River site; and 5) the Table Mountain sites, which were in different small catchments but clustered together.

Water chemistry data from DWAF gauging weirs in all the studied catchments for which it was available revealed that regional water chemistry differences did not appear to be a possible cause of the catchment signatures.

Fish can have significant effects on invertebrate population structure and so data from CapeNature were used to search for possible correlations between the presence of native and alien fish and the catchment signatures.  The tentative conclusions, based on few data, suggest that the presence of both native and alien fish plays some role in the catchment signatures, but other environmental variables are also implicated.

One such variable could be the geographical location of sites.  The catchments stretch from the Olifants in the north, to the Palmiet in the south, and to the Table Mountain group in the south-west (Figure 5.3).  The MDS plot of invertebrate assemblages from LD sites can be re-orientated to coincide with this geographical layout (Figure 5.4), suggesting that the catchment signatures may be at least partly attributable to biogeographical distribution patterns.  Knowing that the signatures are caused by species replacing species within each major taxonomic group (Chapter 3), these replacements could be a result of some species reaching the edge of their distributional ranges and being replaced by others.  On assessing individual taxa, however, some were shown to be confined to the north or south, or absent from Table Mountain, but the majority were either wide-spread across the region or scattered, and so again there was no clear picture of species distributions causing the catchment signatures.

Paleogeographically, the Olifants and Berg systems are thought to have once had a common estuary, and this may have caused their grouping together to provide one catchment signature.  Table Mountain was once an isolated island off mainland Africa, which may account for its studied rivers – though in different small catchments - sharing a common catchment signature.  At present no suggestion can be made as to why the Molenaars sites clustered with the Eerste sites rather than with those from the Breede, which is its parent river.  Much remains unknown and unexplained about the causes of the signatures and an exploration of “river capture” and past drainage networks is just one avenue of research that could be followed to shed further light on them.

In conclusion, despite some positive correlations and avenues revealed for further research, none of the analyses reported here provide proof of a single environmental driver for the catchment signatures.  Rather, the signatures appear to be the result of complex interactions of many variables over long geological time.

The effect of sampling point selection and identification on SASS scores (Chapter 6)

The data used in this study were not collected using SASS techniques.  They were used, however, to assess if the choice of sampling points for a SASS collection could affect the resulting SASS score.  Different permutations of sampling points had little impact on the outcome in terms of computed scores, producing essentially the same results as those of Dallas (2001) and fairly similar to Dickens & Graham (2002).

A further outcome of this analysis was the demonstration that a SASS score does not detect physical degradation of a river, and indeed was not designed to do so, being designed to assess pollution levels.  Researchers should therefore avoid using SASS as an index of river health in terms of physical disturbance.

Catchment and river signatures workshop (Chapter 8)

A workshop was held at the University of Cape Town with the delegates being a mix of scientists and managers.  The presenters represented different disciplines in river science: invertebrates, fish, riparian vegetation and conservation planning.  Each came with their own data sets and suggestions on the nature of catchment and/or river signatures, resulting in a range of ideas on what caused them.  The invertebrate and fish specialists found signatures in biotic data, the vegetation specialists in a combination of biotic and physical data (species, geology, soils) and the conservation specialist in purely physical data.  It was concluded that biotic river and catchment signatures do exist and reflect biodiversity at the landscape level, but there was no clarity on how they form and what they indicate about the functioning of individual catchments.

It was agreed that catchment/river signatures do have management implications, especially in the Reserve Determination process within South Africa’s current water law.  Managers need further clarity, however, on the validity of these signatures across disciplines and for the rest of South Africa.  Although it is recognised that further research is needed to confirm the nature, validity and importance of signatures, it was recommended that catchments be better integrated into classifications of rivers at the second hierarchical level, after ecoregions but before longitudinal zones.

In summary, the main conclusions and recommendations from the workshop are as follows.
General conclusions (Chapter 9)

River and catchment signatures are real.  As shown in this project, they are biotic fingerprints of upper rivers and catchments in the Western Cape, distinguishing each from the others.  There is no reason to believe they do not exist in other parts of South Africa and, indeed, elsewhere although this remains to be shown.  Lower rivers may also have biotic signatures, but as a majority of lower rivers are considerably degraded the data of natural conditions needed to detect the signatures may be largely missing.

We now know that the signatures are due to unique mixes of species per river and catchment from a regional pool of mostly common species, but still have very little understanding of what caused the specific mixes and what their significance is.  The signatures do not appear to be strongly correlated with water chemistry, the presence or absence of native or alien fish, systematic biogeographic species changes across the region or paleogeographic influences, although all of these seem to play some role.  Rather the signatures are probably the result of complex interactions of the above and other variables over many millions of years.

Many questions thus remain unanswered to a large degree because we do not have sufficient understanding of the biology and habitat requirements of riverine species to attempt further interpretation of the signatures data set.  Some additional insights have emerged however.
The main management implications of river and catchment signatures are as follows.
Recommendations (Chapter 9)

Recommendations for future work and consideration are as follows.