A preliminary fish survey of the estuaries on the south-east coast of South Africa, Kei Estuary to Mdumbi: A comparative study

COPYRIGHT © The Author(s) Published under a Creative Commons Attribution 4.0 International Licence (CC BY 4.0) A preliminary ichthyofaunal and physico-chemical survey of estuaries on the south-east coast of South Africa from the Kei Estuary to the Mdumbi Estuary was undertaken between October and November 1997. Twentyseven (27) estuaries were surveyed along this stretch of coastline and these were grouped into three estuary types: small (< 10 ha) predominantly closed estuaries, moderate to large (> 10 ha) predominantly closed estuaries and predominantly open estuaries. Multivariate analyses revealed significant differences between predominantly closed estuaries and predominantly open estuaries in terms of both their physico-chemical characteristics and their fish communities. There was no difference between small and moderate to large predominantly closed estuaries. A significant relationship was also observed between the physico-chemical characteristics of the estuaries and their fish communities. The estuaries in the study area fall within the warmtemperate biogeographic region; temperate species dominated the fish communities of all the estuaries. This survey represents one of the few fish surveys undertaken along this little-studied section of the coastline. A preliminary fish survey of the estuaries on the south-east coast of South Africa, Kei Estuary to Mdumbi: A comparative study


INTRODUCTION
Research into fish communities in the Eastern Cape Province of South Africa has excluded to a large extent the former Transkei region of the province (Mbande et al., 2005), such that information on most systems along this coastline is classified as poor or non-existent (Whitfield and Baliwe, 2013). This paper focuses on the southern Transkei, which is regarded as falling within the warm-temperate biogeographic region; the boundary between the warm-temperate and subtropical regions is situated at the Mdumbi Estuary (Harrison, 2002). The fish species diversity in South African estuaries increases from west to east (Harrison, 2002) and, as such, the southern Transkei estuaries are expected to have more species than the estuaries further south/southwest (e.g. James and Harrison 2010a;2010b;2011;2016). As the southern Transkei estuaries are situated south of the biogeographic boundary, estuaries in this region are likely dominated by temperate species with moderate numbers of tropical species.
The overall ecology, including the fish assemblage, of the large predominantly open Kei (Plumstead, 1984;Plumstead et al., 1985), Mbhashe (Plumstead, 1984b;Plumstead, 1990;Plumstead et al., 1989) and Mthatha (Plumstead, 1984;Plumstead et al., 1989) estuaries was studied in the 1970s and 1980s. Limited information has been published on the fish fauna of the temporarily open/closed Nqabara (Marais and Prinsloo, 1980), Ngoma/Kobule ( Van der Elst, 1978) and Ngqusi/Inxaxo (Wasserman et al., 2010) estuaries. As part of a national assessment of South African estuaries, a fish survey was undertaken along the south-east coastline between the Kei Estuary and the Mdumbi Estuary; basic physico-chemical variables, fish community data and a comparative analysis are provided. Although this survey was conducted more than 20 years ago, this data provides useful baseline information on the fish fauna of this poorly studied region, particularly in the light of climate change related distribution shifts.

Physico-chemical
During each survey, selected physico-chemical parameters were measured at various sites within each system, ranging from the mouth area (Site 1) upstream; the number of sites varied depending on the size of each system. Water depth and transparency were measured using a 20 cm diameter Secchi disc attached to a weighted shot line graduated at 10 cm intervals. Temperature (°C), salinity (psu), pH, dissolved oxygen (mg•L -1 ), and turbidity (NTU) were measured using a Horiba U-10 Water Quality Checker. Where water depth permitted (usually >0.5 m), both surface and bottom waters were measured. The mouth state of each system at the time of sampling was also noted.

Ichthyofauna
The ichthyofauna of each estuary was sampled using a 30 m long x 1.7 m deep x 15 mm bar mesh seine net fitted with a 5 mm bar mesh purse, and a fleet of multi-mesh gill nets. The gill nets were either 10 m or 20 m in length and 1.7 m in depth and consisted of three equal sections of 45 mm, 75 mm and 100 mm stretch meshes. Seine netting was carried out during daylight hours in shallow (< 1.5 m deep), unobstructed areas with gently sloping banks. Fish caught were identified and measured to the nearest millimetre standard length (SL) before being released. Where large catches of a species were made, a sub-sample was kept and returned to the laboratory where the fish were identified, measured and weighed to the nearest 1.0 g; specimens that could not be identified in the field were also kept and processed in the laboratory. All fishes were identified by reference to Smith and Heemstra (1991) and Skelton (1993); taxonomic identities of certain species were adjusted using information provided in Whitfield (2019). The total fish species composition, by number and mass, was calculated for each system. The relative biomass contribution of each species was calculated using actual recorded masses as well as masses derived from length-mass relationships provided in Harrison (2001). Fishes were also classified according to their biogeographic affinity (endemic/temperate, temperate, tropical, widespread) and the contribution of each group calculated for each estuary based on the number of species, abundance, and biomass.

Estuary classification
Estuaries were divided into two main groups on the basis of predominant mouth condition, according to the classification given in Harrison and Whitfield (2006a). The two main groups were predominantly open estuaries and predominantly closed estuaries. Predominantly closed estuaries were further subdivided into two groups based on surface area: small closed estuaries with a surface area below 10 ha and moderate to large closed estuaries with a surface area above 10 ha.

Multivariate analyses
Data were analysed using the Plymouth Routines in Multivariate Ecological Research (PRIMER) package (version 6.0) with PERMANOVA+ add-on (PRIMER-E, Plymouth Marine Laboratory, UK). A principal component analysis (PCA) was undertaken on the overall mean (surface and bottom) values of the physico-chemical variables recorded in each system. Each parameter was first examined for normality; turbidity, depth and dissolved oxygen required log-transformation (ln [1 + x]). The data were also examined for any inter-correlations (Pearson r); pH exhibited significant correlations with both dissolved oxygen and salinity and was omitted from the analysis. Temperature and depth also showed a significant correlation; however, these parameters were retained in the analysis. A PCA was performed based on the following normalised parameters: depth, temperature, salinity, dissolved oxygen, and turbidity. An analysis of similarities (ANOSIM) was also undertaken (using the normalised Euclidean distance similarity measure) to test for significant differences between estuarine types. Specimens not identified to species level (e.g. Mugilidae) as well as exotic species (e.g. Micropterus spp.) were excluded from the analysis. Abundance and biomass data were first standardised and then square-root transformed before calculating a Bray-Curtis similarity matrix. Standardisation removed the effect of variable sampling while transformation scales down the importance of dominant species (Field et al., 1982;Clarke and Warwick, 2001). A non-parametric multivariate analysis of variance (PERMANOVA) (Anderson, 2001) was applied to both the abundance and biomass data to examine differences in fish communities between estuary types. A similarity percentages analyses (SIMPER) was also undertaken to identify species that characterise estuary types as well as those that discriminate between estuary types. Relationships between physico-chemical and biotic resemblance matrices were also investigated using the RELATE routine; the measure of agreement is the Spearman rank correlation coefficient (Rho) between the corresponding elements of the two similarity matrices.

RESULTS
A total of 26 systems were sampled between Kei Estuary and the Mdumbi Estuary. Two systems, (Sundwana and Thsani) comprised small coastal streams and were not considered further. Of the remaining systems, 8 were predominantly open estuaries and 16 were predominantly closed estuaries. Of the predominantly closed estuaries, 12 were moderate to large (>10 ha) systems and 4 were small (<10 ha) systems.

Small predominantly closed estuaries
The four small predominantly closed systems were all closed at the time of sampling. All estuaries were relatively shallow, with average water depths generally not exceeding 1.4 m ( Table 1). Mean water temperatures ranged between 20.2°C (Jujurha) and 22.7°C (Ncizele and Mbhanyana). Mean salinities were almost fresh (0.5) in the Mbhanyana and averaged from 16.3 (Kwa-Suka) to 27.2 in the Ncizele. Salinities were fairly uniform throughout most of the systems with no clear horizontal or vertical gradients. Only the Jujurha exhibited a horizontal decrease in salinity from 27.9 in the lower reaches to 16.6 in the upper reaches. A vertical salinity and temperature gradient was also recorded in the Jujurha in a 3.5 m deep site in the otherwise very shallow system (Table 2). Average dissolved oxygen values ranged between 4.0 mg•L -1 (Kwa-Suka) and 7.7 mg•L -1 (Jujurha). Mean turbidity values were highly variable and ranged from 15.3 NTU (Jujurha) to 85 NTU (Mbhanyana). Average pH values were between 7.4 and 7.9 (Table 1).

Moderate to large predominantly closed estuaries
All 12 of the moderate to large predominantly closed estuaries were closed to the sea at the time of this survey. Mean water depths ranged from 0.5 m (Nenga) to 3.4 m (Qolorha) ( Table 1). Water temperatures averaged between 20.2 (Kumpenzu) and 24.2°C (Cebe). Water temperatures increased from the lower to the upper reaches of the estuaries. Vertical temperature stratification, with a 1°C or more decrease in temperature from the surface to the bottom, was only evident in the Ngadla, Ntlonyane and Nkanya estuaries (Table 3). Mean salinities ranged from 16.5 (Qolorha) to 31.6 (Nkanya) ( Table 1). A pronounced horizontal salinity gradient was present in the Qolorha, Ngqwara, Ngadla, Ntlonyane and Nenga estuaries, with salinities decreasing upstream from the mouth. These estuaries received marine water from overwash events. Pronounced vertical salinity stratification was observed in the Gxara, Qolorha, Ntlonyane, Nenga and Maphuzi estuaries (Table 3). Mean dissolved oxygen values ranged from 4.5 mg•L -1 (Qolorha) to 10.7 mg•L -1 (Mapuzi), with most values exceeding 5.0 mg•L -1 . The water column in these estuaries was clear (<10 NTU) to moderately turbid (<50 NTU). Mean pH values ranged from 7.6 (Ngogwane) to 8.1 (Maphuzi) ( Table 1).

Predominantly open estuaries
The predominantly open estuaries were all open at the time of sampling. Mean water depths recorded in the eight predominantly open estuaries ranged from 1 m (Qhorha and Shixini) to 3.1 m (Mtata) ( Table 1). Water temperatures averaged between 17.0°C (Qhorha) and 22.4°C (Xhorha). Water temperatures in most systems increased upstream of the mouth, except for the Inxaxo arm of the Ngqusi/Inxaxo Estuary, where water temperatures decreased upstream (Table 4). The Mbhashe and Mtata estuaries were freshwater-dominated with mean salinities of 14.6 and 12.6 recorded in these systems, respectively (Table 1). In these systems the surface water was predominantly fresh, with pronounced vertical salinity stratification (Table 4). In contrast, mean salinities in the other six predominantly open estuaries were all above 21 (Table 1) and very little vertical salinity stratification was evident (Table 4). Mean dissolved oxygen values ranged between 6.6 and 7.9 mg•L -1 . The Qhorha, Mbhashe and Mtata estuaries were very turbid (> 80 NTU), whereas the other estuaries were fairly clear (<23 NTU). The mean pH in all estuaries was similar to seawater (7.9-8.1) ( Table 1).

Multivariate analysis
The PCA classification ( Fig. 2) divided the estuaries based on salinity (Axis 1) and depth, temperature and turbidity (Axis 2). The first two axes accounted for approximately 64% of the variation between the samples. Large predominantly open systems (Mbhashe, Mtata, Qhorha) were situated towards the upper right section of the plot associated with high turbidities, depth and dissolved oxygen (Fig. 2). The predominantly open Qhorha, Ngqusi/Inxaxo, Shixini, Khobonqaba, Xhorha and Mdumbi were situated towards the middle right associated with high salinities and low turbidities. The predominantly closed estuaries showed a gradation from estuaries with low salinities to estuaries with high salinities situated towards the right of the plot (Fig. 2). Although there was overlap between estuary types, the marine-dominated predominantly open estuaries were all situated towards the right of the plot. The ANOSIM test revealed a weak but significant difference between estuary types (Global R: 0.25; p < 0.05). Pairwise tests showed that there was no significant difference between small predominantly closed estuaries and moderate to large predominantly closed estuaries (R: 0.26; p > 0.05), however, significant differences were observed between predominantly open estuaries and both small and moderate to large predominantly closed estuaries (R: 0.24-0.35; p > 0.05).

Multivariate analyses
The nMDS plot based on abundance produced a pattern where predominantly open estuaries clustered together and separated from predominantly closed estuaries, which were situated to the left of the plot. There was no separation between small and moderate to large predominantly closed estuaries (Fig. 3a). In terms of biomass, the separation between predominantly closed and open systems was less distinct, with one medium to large predominantly open estuary (Ntlonyane) clustering together with the predominantly open estuaries (Fig. 3b). The PERMANOVA test based on abundance data revealed significant differences between predominantly closed and open estuaries (Pseudo F = 3.1048, p = 0.006). Biomass yielded similar results, with the two estuary types being significantly different (Pseudo F = 4.0978, p = 0.003).

DISCUSSION
This survey provides baseline information on the estuaries and fish assemblages found along a poorly studied section of the South African coastline.  (James et al., 2007;Vorwerk et al., 2003;James and Harrison, 2016). The lower numbers of estuarine species recorded during this survey may be due to prolonged closed conditions. During extended closed periods fish populations in predominantly closed estuaries can decrease considerably due to predation (James et al., 2007).
This study found a significant link between estuary typology (and physico-chemical characteristics) and the fish communities present. Predominantly open estuaries have a near-permanent connection with the sea and are characterised by moderate to high salinities and high species richness. Predominantly closed systems have an intermittent connection with the sea and are characterised by shallow, warmer waters. Species richness in these systems is typically lower than predominantly open estuaries, although marine species may be introduced into these systems via barrier overwash. This study represents a unique survey of multiple estuaries along a little-studied section of the South African coastline.

ACKNOWLEDGEMENTS
This project was undertaken by the CSIR (Durban) and funded by the Department of Environmental Affairs. The authors would also like to thank the Walter Sisulu University for logistical and operational support during field sampling operations.

AUTHOR CONTRIBUTIONS
Conceptualisation, field work and sample analysis was done by TH. Interpretation of the results and writing of the first draft was done mainly by NJ.  (2016)