Agricultural land-use is a leading cause of water quality deterioration, biodiversity loss and impairment of stream functionality. Understanding the mechanisms by which agricultural land-use impair stream ecosystems is important for their effective management, especially in Africa. In this study, a combination of analytical tools, including macroinvertebrate taxonomic- and trait-based community analysis, functional indices, functional feeding groups and stable isotopes were used to investigate the effects of an increasing gradient of agricultural disturbance on the community composition, functional diversity, and food web of aquatic macroinvertebrates in the Kat River. Eight sites grouped into four site categories that represent a decreasing gradient of agricultural pollution (LUC 1< LUC 2 < LUC 3 and LUC 4) were selected. Macroinvertebrates and physiochemical variables and aquatic and terrestrial basal food sources were sampled from the eight sites over four sampling occasions; dry (winter and spring) and wet (summer and autumn) periods using the SASS 5 protocols.
The taxonomy-based analysis showed different responses of macroinvertebrates to agricultural disturbance, with taxa such as Lymnaea spp., L. columella, Appasus spp. Biomphalaria spp., Trithemis spp. and Oligochaeta identified as potentially tolerant indicators of agricultural pollution. These taxa were positively correlated with the highly disturbed LUC 1 sites, and increasing levels of NH4-N, NO2-N, temperature and TDS. Conversely, Caenis spp., Afroptilum spp., Pseudocloeon piscis, Pseudocloeon spp., Baeti harrisoni, and Potamonautes spp. were sensitive to agricultural pollution, indicating strong negative associations with LUC 1 sites and NH4-N, NO2-N, salinity, temperature and TDS. Further, a multimetric index (MMI) was developed, validated and applied to assess agricultural disturbance in the Kat catchment. Of the 29 metrics that satisfactorily discriminated the LUC 4 site from the LUC 1, 2 and 3 sites, only eight metrics were non-redundant and integrated into Kat River MMI. The metrics integrated into the final MMI were Decapoda abundance, EPT/Chironomidae abundance, %EPT abundance, %Ephemeroptera abundance, %Caenidae abundance, %Hydropsychidae abundance, %Oligochaeta+chironomidae abundance and Shannon index. The developed MMI proved effective as a biomonitoring tool for assessing the ecological health of agricultural pollution in the Kat River.
The trait-based analysis showed that traits such as haemoglobin, spiracle, adult aquatic life stage, active swimming and predatory lifestyle were positively correlated with LUC 1 sites, and were deemed tolerant-trait indicators of agricultural pollution. Shredding, medium body size (>10–20 mm), crawling and a preference for macrophytes were negatively correlated with LUC 1 sites, and were deemed sensitive-trait indicators of agricultural pollution in the Kat River. Functional diversity responded predictably to agricultural pollution, as functional indices such as functional richness, significantly declining along disturbance gradient during the dry and wet periods. The functional feeding group results revealed that gatherers and scrapers dominated in the Kat River, and together represented 0.27–0.43 of the invertebrate composition. Shredders were the lowest represented in the Kat River, with a relative abundance of 0.18. The FFG results showed that filter-feeders and predators increased in abundance along increasing environmental stress gradient, whereas shredders’ abundance decreased along the environmental stress gradient.
Analysis of stable carbon (δ13C) and nitrogen (δ15N) isotopes were used to estimate the contributions of aquatic and terrestrial resources to consumers across the four LUC and periods. Carbon contributions, determined using mixing models (Stable Isotope Analysis in R), revealed that consumers assimilated mainly aquatic sources (filamentous algae, macrophytes and biofilms), and this assimilation increased as agricultural disturbance increased across the two seasons. Terrestrial-derived food sources did not show evident variations among the LUCs, but C4 grasses changed along an increasing gradient of agricultural pollution during the two seasons. Further, there was enriched 15N of consumers, especially scrapers, predators and filter-feeders, along the disturbance gradient, whereas that of shredders declined along an agricultural disturbance gradient. NH4-N was the variable that affected consumers δ15N values, indicating a significant positive correlation with δ15N values for the majority of the consumers, especially gatherers, shredders and scrapers.
The results of the study highlight the strength of a complementary approach to biomonitoring agricultural pollution in riverine systems. For example, the taxonomic analysis indicated changes in community composition, and the trait-based approach provided insights into the key stressors associated with agricultural pollution as a cause of water quality deterioration. The study contributes significantly to our understanding of riverine ecology in South Africa and, in particular the Kat River, in the context of agricultural pollution, which remains one of the leading causes of pollution of riverine ecosystems.