Trajectories of Change in South Africa's Freshwater Fish Fauna

Kajee, Mohammed

Trajectories of Change in South Africa's Freshwater Fish Fauna

Thesis / Dissertation

2025

Publisher

University of Cape Town

Department

Department of Biological Sciences

Faculty

Faculty of Science

Abstract

In South Africa, freshwaters are the most threatened ecosystems in the country, facing pressure from water abstraction, land-use change, pollution, invasive species and climate change. These anthropogenic pressures are directly impacting the diversity, abundance, and survival of freshwater fauna and flora across the country. Freshwater fishes are recognised as South Africa's most threatened species group, with one third of species classified as threatened (Vulnerable, Endangered, Critically Endangered) by the International Union for the Conservation of Nature (IUCN). Understanding patterns of diversity, threat, invasion, and protection status are vital for management. However, access to reliable and comprehensive long-term data on fish distributions, are limited. This thesis aims to collect, collate, and analyse historic and current freshwater fish records for all fish species (native and non-native) known to occur in South Africa, to better understand distribution, diversity, and threat patterns in space and time. A comprehensive freshwater fish occurrence dataset for South Africa was compiled and uploaded to the Freshwater Biodiversity Information System (FBIS, freshwaterbiodiversity.org). An 18-month historic-data collation effort resulted in the accrual of 35 955 records of freshwater fish from South Africa spanning 194 years (1828–2022), that have since also been uploaded to the Global Biodiversity Information Facility (GBIF). Together with pre-existing GBIF records (24 861), a total of 60 837 freshwater fish records are thus now available for South Africa. The data show a marked decline in the number of native fish occurrence records over the last decade relative to the previous five decades. Conversely, the number of occurrences for non-native fishes increased over the past three decades. A breakdown of the data is provided for all native (n = 105) and non-native (n = 24) species as well as for all endemic (n = 40) and threatened (n = 27) species. Using occurrence records collected for this thesis, spatial patterns in distribution, diversity, threat, invasion, and protection status of freshwater fishes in South Africa were analysed. Few studies have undertaken such analyses at ecologically and politically appropriate spatial scales, largely because of limited access to comprehensive biodiversity data sets. Results show that record density varies spatially, at both primary catchment and provincial scales. The diversity of freshwater fishes also varied spatially: native species hotspots were identified at provincial level in the Limpopo, Mpumalanga and KwaZulu-Natal provinces, endemic species hotspots were identified in the Western Cape, and threatened species hotspots in the Western Cape, Mpumalanga, Eastern Cape, and KwaZulu-Natal. Non-native species distributions mirrored threatened species hotspots in the Western Cape, Mpumalanga, Eastern Cape, and KwaZulu-Natal. Some 47% of threatened species records fell outside of protected areas, and 38% of non-native species records fell within protected areas. Concerningly, 58% of the distribution ranges of threatened species were invaded by non-native species. A data breakdown is provided for each of South Africa's nine provinces including total number of records, and the numbers of native, non-native, endemic and threatened species. These data provide a much-needed update of the known status and distribution of freshwater fishes in the country. To support future spatial modelling of South Africa's freshwater fish species distributions, a robust and repeatable method for applying Species Distribution Models (SDMs) to freshwater fishes was developed. Specifically, this thesis sought to investigate whether SDMs could be applied to a wide range of freshwater fishes in South Africa and whether the inclusion of invasive species spatial data would result in improved model accuracy and performance. Using a Bayesian Additive Regression Trees (BART) algorithm, the distributions of three black bass (Micropterus salmoides, M. dolomieu, and M. punctulatus) species and two native species (Clarias gariepinus and Pseudobarbus burgi) were modelled. A total of 148 hydrological and 16 biological predictor variables were specifically developed for this chapter, along with the collation of 75 environmental predictor variables. The results demonstrated that, with appropriate data cleaning and using a statistically robust method for selecting predictor variables, model outputs performed well, with the resulting modelled distributions congruent with known distributions and historic descriptions of the species' presence. All models for invasive, native, and threatened species performed well, with AUC values ranging from 0.923–0.994 and TSS ranging from 0.701–0.940. Furthermore, model performance for P. burgi was not improved by the inclusion of biological predictors (invasive species). The AUC for the original P. burgi model (including only climatic, environmental, and hydrological predictors) was 0.994, whereas the adapted model (including both climatic, environmental, and hydrological predictors as well as biological predictors) was marginally lower, with an AUC value of 0.993. This indicates that the inclusion of invasive species as biological predictors may not be necessary to produce accurate predicted species distributions. This thesis also presents a case study demonstrating how the FBIS system was used to provide freshwater fish spatial data products to a national conservation decision-support tool – The Department of Forestry, Fisheries, and the Environment (DFFE) National Environmental Screening Tool. The Tool uses empirical and modelled biodiversity data to guide Environmental Impact Assessments of proposed developments. Occurrence records for 34 threatened freshwater fishes occurring in South Africa were extracted from the FBIS and verified by taxon specialists, resulting in 6 660 records being used to generate modelled and empirical national distribution layers. This represents the first inclusion of freshwater biodiversity data in the Screening Tool, and future iterations of the Tool will incorporate additional freshwater taxa. This case study demonstrates how systems like the FBIS can play a pivotal role in the data-to-decision pipeline through supporting data-driven conservation and management decisions at a national level. The collective findings of the thesis are concluded in Chapter 6, which highlight the importance of collecting, collating, and consolidating biodiversity data at a national level. The dataset presents the largest (by number of records) collection of freshwater fish records in Africa to date, with data spanning 194 years. This thesis also presents an updated version of South Africa's Species List for Freshwater Fishes, which includes all known native (n = 107) and non-native (n = 27) species occurring in the country, and key species attributes like species origin, endemism, and threat status. By consolidating historical data for all known freshwater fishes, this thesis presents the first comprehensive assessment of the status of South Africa's freshwater fish fauna, providing valuable insights into spatial patterns of richness, endemism, threat status and non-native species, as well as assessing how effective the country's protected area network is for protecting freshwater fishes. This thesis addressed the critical need for a deeper, data-driven understanding of spatial and temporal variation in South Africa's freshwater fishes – an essential foundation for sound management of these species and their habitats into the future.