The Ndumo area is situated in the north-eastern corner of South Africa's KwaZulu-Natal (KZN) province, south of Swaziland and southern Mozambique. The area is comprised of a small town (with limited facilities and infrastructure) and a number of rural villages that border the southern and western parts of the Ndumo Game Reserve. Ndumo is a provincial game reserve in South Africa that is managed by Ezemvelo KZN Wildlife. The whole area falls under the Jozini Local Municipality, which in turn forms part of the larger Umkhanyakude District Municipality in northern Maputaland (see Figure 1).

To the north of the Ndumo area, the Usutu River separates South Africa from Mozambique. In the east, the Pongola River runs through the Ndumo area from north to south and into the larger Phongolo River and its floodplain. The area lies within a summer rainfall area and has a subtropical climate with an average annual rainfall of between 670 mm and 1000 mm, with temperatures ranging from 13 °C to 40 °C (Morgenthal et al. 2006).

A cross-sectional quantitative survey design (Sarantakos 2005) was adopted as a basis for this study. In total, 521 participants participated in the survey. Their average age was 37.28 years (s.d. = 13.12) and participants had been living in the Ndumo area for a mean of 28.18 years (s.d. = 16.24). Table 1 depicts additional demographic characteristics of the sample.

Entry into Ndumo (i.e. various villages) was negotiated on behalf of the research team by an experienced research assistant. The research assistant also used the opportunity to gather basic data that was used as a basis for the development of a questionnaire.

Further data was collected by means of a structured questionnaire (Creswell 2007) during a second visit to the area. The questionnaire consisted of 40 questions that were divided into 4 sections: demographic characteristics; questions relating to the consumption of fish and fish products; questions relating to local fishing habits; and, finally, questions relating to local trading in fish.

A group of 16 fieldworkers were sourced from local villages and trained to recruit participants, to obtain informed consent, and to administer the questionnaire by means of structured face-to-face interviews.

All data was collected over a 5-day period in August 2013, with fieldworkers receiving a set of questionnaires in the morning, which they administered during the course of the day. Questionnaires were collected at the end of each day and scrutinised for completeness.

During the entire process of data collection, care was taken to do so in as ethical a manner as possible. Full, informed and signed consent was obtained from all participants. Participants were advised that their responses would be treated confidentially and that they had the right to withdraw from the research at any time without penalty.

Data was analysed using the Statistical Package for the Social Sciences, version 21 (SPSS 21). Basic descriptive statistics (means and measures of variance) were calculated for all items in the questionnaire. Wherever relevant, findings were compared across demographic groups using chi-square tests, with cross tabulation and independent t-tests to verify the statistical significance of any differences that emerged, while bivariate correlations were used to examine relationships between variables (Field 2005).

As reflected in Table 2, fish was found to be the third most frequently consumed protein per week (after chicken and beef) among all participants. This contradicts earlier findings (Ross & Ross 2012) that fish was the main protein source in the floodplain area. Among those who consumed fish, which included 88.9% of the participant group, fish was consumed on average twice a week. On those days that fish was consumed, participants did so 1.4 times per day on average (s.d. = 0.59).

Additionally, participants were asked to specify which species of fish they ate and how frequently they did so. Table 3 provides a summary of these findings.

Overall, it was found that all species of fish listed in Table 3 were consumed in fairly high numbers by a substantial portion of the participant group, with the exception of the recently introduced redclaw crayfish (Cherax quadricarinatus) (see Du Preez & Smit 2013). Reasons for the low levels of utilisation of this (non-native invasive) species are still unclear and should be examined more closely in future studies. The two tilapia species that occur in the region were the most preferred and frequently consumed species of fish, with red-breasted tilapia (Tilapia rendalli) being much more popular than Mozambique tilapia (Oreochromis mossambicus). However, while the African sharptooth catfish (Clarias gariepinus) was consumed by only 54.8% of the participants, it was the third most frequently consumed fish of all the species.

Bivariate correlations were used to examine the relationships between age, the number of years that participants resided in the Ndumo area and their consumption of fish, both generally and in relation to specific species. Age was uncorrelated to general fish consumption, but it was found that the longer participants lived in the Ndumo area, the more likely they were to eat red-breasted tilapia (T. rendalli) (r = 0.12, p = 0.02), brown squeaker (Synodontis zambezensis) (r = 0.15, p = 0.02), Mozambique tilapia (O. mossambicus) (r = 0.13, p = 0.02) and, in particular, tigerfish (Hydrocynus vittatus) (r = 0.24, p = 0.00), suggesting that those living in this region develop an acquired taste for these species.

Asked how they obtained the fish they consumed, 46.1% of participants (54.4% of females and 37.3% of males) indicated that they bought the fish from a vendor in the area; 37.8% (similar for both genders) obtained the fish from family or friends, whereas 16.1% (25.5% of males and 7.2% of females) caught the fish themselves. A chi-square test with cross tabulation confirmed that the association between gender and where participants obtained the fish was significant [χ²(2, N = 457) = 31.18, p = 0.00] and of moderate strength (ɸ = 0.26).

Participants indicated that they commonly fished in an average of 4.32 (s.d. = 5.07) different locations. However, the majority of participants (70.48%) only fished in one to three locations. Among these sites, 81.02% of fish was caught in rivers and streams, 13.60% in pans and 1.9% in other (unspecified) locations, suggesting that fishing was mostly focused around the river system and that pans were not highly exploited for fishing. The main reason for this seems to be that the majority of the pans in the Ndumo region falls within the confines of the Ndumo Game Reserve and, as such, are protected.

Participants indicated that the bulk of fish caught was intended to be sold (54.73%) and that 44.28% of fish caught was for personal consumption.

The most commonly employed fishing techniques were handlines (26.77%), traditional baskets (25.81%) and drag nets (22.26%), followed by gill nets (17.10%) and, to a much lesser extent, fishing rods (8.19%). Thus, non-commercial techniques were used and no evidence of large-scale fishing was uncovered. This indicates that local communities in the area fish primarily for personal subsistence use rather than for large-scale commercial reasons.

The species of fish that participants usually caught and the average number of fish per species is summarised in Table 4.

Red-breasted tilapia and, to a lesser degree, Mozambique tilapia emerged as the species that were most frequently caught in the region. Whilst the vast majority of participants indicated that they actively sought to catch African sharptooth catfish, a much lower number of these fish was caught. However, as can be seen from the large standard deviations associated with the number of fish caught, great variability exists in this regard, with some fishermen and -women catching 1 or 2 fish per week and 1 individual reporting catching as many as 200 tilapia species per week.

These differences could possibly be attributed to the fact that whereas most fishermen and -women fished for personal consumption and as such only needed a relatively small number of fish, a few residents did so for small-scale commercial purposes and therefore made far larger catches. No statistically significant gender differences were found in relation to the types or numbers of fish species caught.

In order to determine whether any changes or trends pertaining to fish populations had been observed by local fishermen and -women, participants were asked whether they had noticed any changes in the number of fish over the past 10 years. Responses were measured on a five-point scale ranging from ‘decreased a lot’ (1) to ‘increased a lot’ (5). A score of 3 indicated that there was no observed change. Participants’ mean score was 2.91 (s.d. = 1.22), suggesting that they observed very little change. The same was true for the average size of the fish, which was viewed as decreasing only fractionally (M = 2.89; s.d. = 1.25).

Lastly, data were also collected from participants who sold fish, but did not catch it themselves. The average trader indicated that he or she had been selling fish in the area for 7.50 years (s.d. = 5.87) and that they obtained 85% of the fish from local fishermen and -women and 15% from other undisclosed suppliers.

Traders indicated that they mostly sold their fish in nearby towns (40.7%) (mainly Ndumo, but occasionally Jozini, approximately 75 km farther south) and in their own villages (30.7%). To a lesser extent, fish was also sold in other villages where they did not reside (15.9%) or ‘next to the road’ (referring to areas outside towns or villages) (11.1%). This suggests that the local fishing industry, though limited in opportunities, contributes directly and meaningfully to the economy of the Ndumo area, in which income-generating opportunities are very scarce (Ogilvie pers. comm. 2014).

Table 4 indicates the number of fish that were sold by traders during the course of a typical week. It also reflects the mean and modal prices (in South African rand [ZAR]) that is charged for an average-sized fish of a given species.

The findings reported in Table 5 suggest that the two species of tilapia and African sharptooth catfish were the two species most frequently sold and also the most expensive of all the fish species that were exploited in the area. However, when the modal prices for all the fish species were considered, it would appear that fish of all species were most commonly sold for ZAR 10 (with the exception of African sharptooth catfish, which sold for a modal amount of ZAR 20; and the redclaw crayfish, which sold for a modal amount of ZAR 15). As such, these species likely play an important role in the context of the local economy within the Ndumo and the greater Phongolo River floodplain region.

None of these species are regarded as threatened or endangered, with the exception of the tigerfish, which is a protected species in South Africa. However, even though residents tend to increase their consumption of this species the longer they reside in the Ndumo area, this species is consumed fairly infrequently and does not appear to be specifically targeted as much as being incidentally caught.

Assessing the impact of human activities on fish populations in the Ndumo area is hampered by the fact that very little research has been conducted to determine the size and reproductive rate of fish populations in this ecologically sensitive region (Smit pers. comm. 2014). As such, it is hoped that the findings of the present study would provide the necessary baseline data on fish consumption in the region so that the ecological impact on local fish populations can be assessed once future studies on fish populations and reproductive rates in the area have been completed.