The determination of flow and habitat requirements for selected riverine macroinvertebrates

Abstract

The focus of the Department of Water and Sanitation (DWS) has changed from one which addresses the quality and quantity of water resources in isolation to one which integrates these attributes with that of aquatic ecosystem integrity. The right to water for basic human needs as well as to ensure a functioning ecosystem is entrenched in the National Water Act (Act 36 of 1998) through the setting of the Reserve (Rowlston 2011). The Ecological Reserve is defined as the quality and quantity of water required for protecting aquatic ecosystems in order to secure ecologically sustainable development and use of the relevant water resource. Although macroinvertebrates are used to set environmental flows in South Africa and abroad, very limited information is available about their flow requirements (Brunke et al. 2001, Schael 2002, Jowett 2002a, Cassin et al. 2004, Clifford et al. 2004, Hanquet et al. 2004, Kleynhans and Louw 2007). In southern Africa some information is available on certain Ephemeroptera in the Inkomati System (Matthew 1968) and some species occurring in the Lesotho Highlands (Skoroszewski and de Moor 1999). A structured approach is required to determine macroinvertebrate environmental requirements taking into account the different life stages, ecoregions, seasonality and substratum The purpose of this study was to determine the habitat requirements of selected riverine macroinvertebrate taxa. The main aim was to determine the preferred ranges of water depth, velocity and temperature, as well as the substratum types required by Ephemeroptera, Trichoptera, Coleoptera and Diptera. In order to determine the habitat requirements for the selected riverine macroinvertebrates, 266 quantitative samples were collected at 52 sites between July 2005 and February 2009. Samples were taken with a Surber sampler and hand net at a number of localities at each site to cover all substratum types, velocity and depth groupings. Basic in situ water measurements (temperature, dissolved oxygen, pH and Electric Conductivity) were also collected at each site. The macroinvertebrates were preserved in 80% ethanol and identified to family. The water velocity was also measured at 5 – 10 cm intervals at each locality from as close to the bottom as possible to the water surface. No comprehensive study has been done on the distribution of aquatic macroinvertebrates in South Africa. The only distribution maps available are those of selected insect families drawn mostly from existing museum and literature records (Picker et al. 2003, Griffiths et al. 2015). The geographical distribution of 10 Ephemeropteran, 16 Trichopteran, 10 Coleopteran and 14 Dipteran families were determined using data collected from this study as well as archived datafrom the National Rivers Database, the Biobase and the National Freshwater Invertebrate Collection housed at the Albany Museum in Grahamstown. The distribution ranges for each of the insect families were then compared to distribution ranges in the literature. The distribution of each of the families was also associated with Level I Ecoregion, geomorphological zone and altitude range. The range extension in a number of taxa such as the Calamoceratidae (Trichoptera) and the Ptilodactylidae (Coleoptera) as well as the identification of questionable distribution records for a number of mostly south-western Cape endemics such as Barbarochthonidae, Sericostomatidae and Glossosomatidae was highlighted. The need to archive voucher specimens, not only for new or unidentified taxa, but also to validate the range distributions of known taxa is emphasised. The scarcity of distribution records for a number of families, most notably that of the Hydrosalpingidae (Trichoptera), Ptilodactylidae, Limnichidae (Coleoptera) and Ephemeridae (Ephemeroptera) is also noted.

Redundancy analysis (RDA) was done using Canoco 5.04 (ter Braak and Šmilauer 2012) to determine the environmental factors contributing most to the distribution of the different invertebrate families. Results from the RDA were then used to draw response curves for the selected families using the Species Response Curve function in Canoco 5.04 (ter Braak and Šmilauer 2012). A Generalised Additive Model (GAM) with Poisson distribution and log link function (family as the response and depth, velocity at 60% of depth and substratum category as predictors) were used with stepwise selection using the Akaike Information Criterion (AIC) and two degrees of freedom (2 DF) factor to smooth the curves. Significance of relationships was regarded as p < 0.05.

The concept of Habitat Suitability Curves (HSCs) was developed as part of the Instream Flow Incremental Methodology (IFIM) and the Physical Habitat Simulation System (PHABSIM) in the 1980s by researchers at the United States Fish and Wildlife Service (Bovee 1982, Bovee 1986). Habitat Suitability Curves (HSCs) were determined for the selected taxa using the methods described in Bovee (1986) and Jowett et al. (2008). Separate univariate curves were developed for frequency and abundance and the average of the two curves was then used to derive the final HSC. A second order polynomial regression was performed for the depth and velocity curves using Excel 2010. No regression was done on the substratum curves as they represent discrete categories rather than a range of values. Based on the reasoning of Jowett et al. (2008), only families with at least twenty individuals and that were present in at least ten samples were selected for the HSCs.

Environmental factors such as velocity, pH, temperature, latitude and longitude, as well as Ecoregion, geomorphological zone and substratum type were determining factors in the distribution patterns of the insect families under consideration. Not all of the factors were important for all of the families. While certain common families (e.g. Baetidae, Chironomidae) showed no preference for any of the environmental factors under consideration, others (e.g. Simuliidae, Blephariceridae) are associated with very fast flowing water over cobbles, the Caenidae with the GSM biotope and Dytiscidae with vegetation. The distribution of taxa with a more limited geographical range such as the more subtropical Calamoceratidae and the burrowing mayflies (Ephemeridae and Polymitarcyidae) are associated with Ecoregion as well as latitude and longitude while the distribution of the southwestern Cape endemic mayflies (Teloganodidae) and cased caddisflies (Sericostomatidae, Glossosomatidae) are also associated with low pH values. The importance of noting the developmental stage of insects such as larva, pupa and adult is highlighted most notably by the different environmental requirements of the beetles where the larval and adult stages sometimes have different requirements.

These results provide a first step in setting habitat requirements for selected families of Ephemeroptera, Trichoptera, Coleoptera and Diptera, and the need for more data on certain families (Prosopistomatidae, Sericostomatidae, Glossosomatidae, Haliplidae, Ephydridae, and Syrphidae) is pointed out. Although depth does not appear to be a determining factor in the occurrence of the macroinvertebrate families investigated here, there is still a requirement for investigating the effect of particularly shallower depths as there might be a threshold value below which the macroinvertebrates could potentially be affected. There is a real danger of damaging the riverine macroinvertebrate communities if depth is ignored and the focus is solely placed on substratum and velocity as there can still be water of a suitable velocity but the depth might be too shallow to enable long-term survival of the resident macroinvertebrates. The Macroinvertebrate Response Assessment Index (MIRAI) was developed as part of a suite of EcoStatus indices to be used in the Ecological Classification Process (Thirion 2007). The MIRAI is based on the principle that macroinvertebrates integrate the effect of the modification of drivers (hydrology, geomorphology and physico-chemical conditions). The degree of change from natural is rated on a scale of 0 (no change) to 5 (maximum change) for a variety of metrics. Each metric is weighted in terms of its importance to determining the Ecological Category under natural conditions for the specific locality. The main aim of the Ecological Classification process is to acquire a better understanding of the reasons for the system’s deviation from the natural or reference condition. The distribution of an aquatic macroinvertebrate assemblage is determined by the tolerance of the individuals in the population to an array of environmental factors. It is therefore essential that all habitat features are considered when evaluating the suitability of habitat for aquatic macroinvertebrates. The approach followed in assessing macroinvertebrate response to driver characteristics is based on a qualitative combination of information gained by field surveys, the available habitat as a result of driver conditions, and the traits of the macroinvertebrates present (Lamaroux et al. 2004, Horta et al. 2009). The Habitat Suitability Curves (HSCs) were converted to values out of 5 and rounded to the nearest 0.5 to fit in with the system used in the suite of EcoStatus models. Where no or not enough data were available, information from the literature was used to assign the preference values to each taxon. Teresa and Casatti (2013) suggested that values greater than 0.7 can be regarded as preferred conditions. It was therefore decided to use preference values greater than 3.5 to indicate a strong preference for a certain habitat feature. No changes were made to the physico-chemical (water quality) metric group as these ratings are based on the sensitivity values (QVs) used in the South African Scoring System (SASS) version 5 (Dickens and Graham 2002). The preference values in MIRAI v1 were then updated to reflect these new values. This new version (MIRAI v2) was then tested by running it for 44 sites covering a large geographical range and ecological conditions. The results from the two different versions of MIRAI were compared using a one-way analysis of variance (ANOVA) and a linear regression analysis done using Excel 2010.

The changes to the velocity and substratum preference ratings in the MIRAI model resulted in only small changes to the total MIRAI score. The MIRAI category generally remained the same or at most changed by half a category. The largest percentage change was 5.7% at the Sterkstroom (Site A2STER-MAMOG) that remained in a D Category, and 5.3% at the Jukskei River (Site A2JUKS-DIENR) that changed from a DE category to an E category for MIRAI v2. The relatively small changes in MIRAI, together with the high correlation values (>0.9) means that the results from the two versions should be compatible and no adjustments will be needed to the results obtained from the original MIRAIs. The impact of the larger changes to the Flow modification and Habitat modification metric group will however need to be investigated. The following three hypotheses were tested in this study;  The macroinvertebrate assemblage structure can be differentiated based on Ecoregion delineation and geomorphological zonation. The results indicated that this is true for certain taxa while other taxa have a countrywide distribution and have been recorded from most geomorphological zones. However, the macroinvertebrate assemblage structure as a whole can be differentiated based on Ecoregion and geomorphological zone. This hypothesis is thus accepted.  The macroinvertebrate assemblage structure can be differentiated based on environmental factors such as substratum, depth and velocity as well as physico-chemical parameters. The results indicated that this is true for certain taxa but not for others. It is also clear that some of the environmental factors play a role in the distribution of certain taxa but not others (e.g. temperature is a determining factor for Blephariceridae, but not for Simuliidae). Depth was not a significant factor in determining the distribution of the insect families under consideration. However, the macroinvertebrate assemblage structure as a whole can be differentiated based on a combination of environmental factors. This hypothesis is therefore accepted.  The different habitat requirements of the macroinvertebrate taxa in terms of velocity and substratum type can be used to refine the macroinvertebrate taxa’s preference values in the Macroinvertebrate Response Assessment Index (MIRAI), to assess the ecological condition of the macroinvertebrate assemblage. The preference ratings, based on the HSCs as well as information from the literature and personal experience, were successfully used in MIRAI v2 to determine the ecological condition of the macroinvertebrate assemblage at 44 sites spanning a range of conditions. The high correlation values (>0.9) for the different MIRAI metrics tested clearly indicates that the hypothesis can be accepted. The following recommendations were made:  Information collected post January 2014 should be used to update the distribution ranges. The updated distribution maps and associated KML files should be placed on the RQIS website.  Gaps identified should be filled by actively targeting areas with limited or no data and a concerted effort made to collect information on the distribution of taxa with limited records.  A sampling programme should be designed and implemented to include the water surface as a possible habitat and the study area expanded to include polluted sites as well.  HSCs and preference ratings should be developed for taxa not included in this study.  Specimens should be identified to genus or species level and HSCs and preference ratings for these genera or species determined where possible.  A wider range of depths should be included in order to determine if there is a minimum depth requirement that should be used when determining Environmental Water Requirements.  Information obtained should be included in the RHAMM and FIFHA models.  A list of possible reference taxa per Level II Ecoregion, geomorphological zone and altitude range should be compiled based on information obtained during this study. These lists should be included in MIRAI v2 as well as the RHAMM model. This will enable a user to compile a reasonable reference condition for a site, based not only on the list but also on the natural characteristics of the site. Ideally these reference conditions should be placed in a central location such as the RQIS websites where other researches can access it.  The effects of the changes in the flow modification and habitat modification metric group results between the two MIRAI versions should be investigated and the following questions answered: o Does it have an impact on the explanation for the Ecological condition/ impacts at the site? o Does it explain the impacts more realistically than the information obtained using MIRAI v1?