Fish are well established around the world as indicators of ecological integrity (Harrison & Whitfield 2006; Mims & Olden 2013; Sosa-López et al. 2005). Similar to other aquatic organisms, fish species face several challenges from the environment they occur in and are subject to a plethora of stressors, both natural and anthropogenic (Du Plessis 2019; O’Brien et al. 2018). The ever-increasing anthropogenic pressures have generally led to a steady decline in aquatic organisms, including fish species (Dudgeon 2014; Rodell et al. 2018). Fish kills and population declines are often reported worldwide as a result of deteriorating aquatic ecosystems (Ferreira & Pienaar 2011; Thronson & Quigg 2008). This deterioration has been attributed to several factors, including river fragmentation, flow modification, pollution, the introduction of alien species and global climate change (De Moor 1996; Ellender & Weyl 2014; O’Brien et al. 2019). Fish make use of the available physical habitat as refugia, for feeding and spawning, and as nurseries for their offspring (Godin 1997; O’Brien et al. 2013). Maintaining these environments is essential, but they are often inaccessible and difficult to sample, making observations hard to determine (Hussey et al. 2015; Lennox et al. 2017). Field-based studies are becoming increasingly important because of their holistic approach to measure animal behaviour and movements attributed to the environment they inhabit (Cooke & Shcreer 2003; Cooke et al. 2017). Techniques such as capture-mark-re-capture (CMR) and environmental monitoring generally lack the representation of aquatic organisms’ ecological interactions and fine-scale movements within their environments (Cooke et al. 2017). Increasingly telemetry techniques are being used to determine various stressors of aquatic organisms in the wild and their interactions with other organisms (Burnett et al. 2020; Cooke et al. 2004; Thorstad et al. 2013). Telemetry techniques are employed to monitor fish within their natural environment and can provide the required information needed to understand fish behaviour and fine-scale movements (Burnett et al. 2020; Wisniewska et al. 2016). Consequently, this technique has become the preferred method to study fish behaviour worldwide (Cooke et al. 2013; Hussey et al. 2015; Lennox et al. 2017). Recent developments in telemetry studies have presented opportunities for better management actions to safeguard the sustainability of natural resources (Abecasis et al. 2018; Block et al. 2016; Cooke et al. 2017).

Fish telemetry studies involve the attachment of electronic tags, and it has been shown that these affect the initial behaviour of tagged fish following tagging procedures (Økland et al. 2005). This has revealed that the predation of tagged fish is more likely to occur directly after the tagging procedure (Beland, Kocik & Sheehan 2001; Halfyard et al. 2017; Schultz et al. 2017; Thorstad et al. 2003). In southern Africa, there have been reports of African fish-eagles, Haliaeetus vocifer, preying on telemeter tagged fish shortly after release (Burnett et al. 2020; Thorstad et al. 2003). Furthermore, the movement of telemeter tagged individuals does not always reflect normal behaviour until several days after tagging procedure, possibly the reason for the predation of individuals shortly after release (Thorstad et al. 2003). The predation effect on adult fish is generally unknown and can be exacerbated when using telemetry methods (Bolland et al. 2019; Moxham et al. 2019; Thorstad et al. 2003). Despite the above, the effects of telemeter tagging of fish are regarded as relatively minimal compared with the amount of information that is obtained through telemetry studies (Jepsen et al. 2002, 2004). The effect of predation, and other mortality accounts, after recovery from the telemeter tagging procedure, has largely been neglected, with mortality increasingly becoming an important consideration in aquatic telemetry studies (Klinard & Mately 2020). These effects have been known primarily during the recovery period post-telemeter tagging procedure; however, predation thereafter could be higher than previously considered because of natural causes and environmental stressors (Jacobs 2017; Thorstad et al. 2003).

Anthropogenic changes to the environment take precedence in threatening local populations of fish within South Africa and elsewhere (Dudgeon et al. 2006; Rodell et al. 2018; Woodborne et al. 2012). In recent years, there were increased studies within the Kruger National Park, particularly in the Olifants River, to understand the biological and ecological effects of anthropogenic stressors on various fish species and the Nile crocodile Crocodylus niloticus (Ashton 2010; Bouwman et al. 2014; De Villiers & Mkwelo 2009; Ferreira & Pienaar 2011; Smit et al. 2013). Moreover, telemetry has successfully been conducted on tigerfish Hydrocynus vittatus within southern Africa to determine their biology and ecology (Jacobs et al. 2019; O’Brien et al. 2012; Roux et al. 2018). These studies noted some interference of aerial predation on H. vittatus; however, none have specifically addressed the predation issue on H. vittatus post-telemeter tagging nor during recovery.

The tigerfish, Hydrocynus vittatus, is a relatively large, migratory Characin that is known to make extensive use of river reaches (Jacobs et al. 2019; Roux et al. 2018). Their higher trophic level and sensitivity to water temperature and oxygen levels make them good ecological indicators of environmental change in the riverine ecosystems they occur (Roux et al. 2018; Smit et al. 2013). They are listed as protected in the South African National Environmental Management of Biodiversity Act for Critically endangered, endangered, vulnerable and protected species (DEAT 2004). These characteristics make H. vittatus an ideal candidate for telemetry studies (Burnett et al. 2020; Jacobs et al. 2019; Jepsen et al. 2015). In the present study, we used the activity rates of H. vittatus as a line of evidence to assess (1) the effects of telemeter tagging on their activity rates in three potential post-tagging recovery periods and (2) their vulnerability to predation and the causality related to environmental stressors placed on the river catchment in the Kruger National Park, South Africa. We predicted that (1) H. vittatus are affected by the tagging procedure for only a few days and (2) the predation pressure on H. vittatus populations here would be higher than previously perceived. We then highlight the potential causalities of the increase in predation of H. vittatus based on environmental variables measured in the study area and similar environmental impacts elsewhere. The knowledge of natural stressors on H. vittatus populations can contribute to sound decision-making for their conservation and management and their river ecosystems. Understanding the impact of predation of fish species may explain the effect these predators have on the taxa and the aquatic environment.

The Letaba River, which originates in the Limpopo province of South Africa, joins the larger Olifants River in the Kruger National Park and then flows into Mozambique (Figure 1). Our study was conducted around the confluence of the two rivers. The Olifants River flows through the Gauteng, Mpumalanga and Limpopo Provinces. It is socio-economically important and subjected to anthropogenic factors (e.g. intensive mining, industrialisation, urbanisation and agriculture) in most of its secondary catchments before it enters the Kruger National Park (De Villiers & Mkwelo 2009; Gerber et al. 2016; Rogers & O’Keeffe 2003). As a result, a wide range of environmental stressors impact the water quality and quantity entering the Kruger National Park, ultimately affecting the aquatic biota here (Du Preez & Steyn 1992; Gerber et al. 2016; Roux et al. 2008). The Letaba River has anthropogenic impacts similar to the Olifants River that affect the water quality and quantity downstream (Rogers & O’Keeffe 2003; Woodborne et al. 2012).

To track tagged H. vittatus individuals remotely and in real time, we established a remote network of receivers (base [BS] and relay [RS] stations) in the Olifants and Letaba Rivers (Figure 1) as per the Fish tracking (FISHTRAC) methodology described by Burnett et al. (2020).

We captured and tagged a total of nine adult H. vittatus individuals, with a mean (± standard deviation [SD]) standard length of 452 mm ± 35 mm. Of these, we successfully tracked and monitored seven (Table 1). Two tags (HVIT 6 and HVIT 9) went out of range immediately after release and did not contribute to the data collected (Table 1). Telemeter tagged H. vittatus were monitored for a minimum of 22 days, with the longest tracked for 156 days (mean ± SD: 66 ± 52 days). The variability in tracked days was because of the mortality of tagged individuals or fish moving out of the network range. The sample size for field studies met the criteria recommended by Sequeira et al. (2019), stating that individuals of ≤ 10 should be used for studies that are still at an exploration phase, such as the present study.

A typical activity pattern, as displayed by an H. vittatus individual (Figure 2), showed diurnal behaviour, with regular higher activity rates during the day when temperatures were warmer. This individual was tagged on 10 May and is indicative of the behaviour after recovery. The individual showed daily diel activity rates as they fluctuated throughout the study showing periods of lower activity rates interspersed with periods of higher activity rates, still maintaining a diurnal diel behavioural pattern.

In our study, we successfully showed the effects of telemeter tagging on the post-tagging activity rates of H. vittatus. Typically, the impact of the tagging procedure on tagged fish species in their natural environment is difficult to measure because of the limited in situ measurements and the inability to provide a control experiment (Frank et al. 2009, Jacobs et al. 2019). Despite this, we showed that the fine-scale variability in the movement was evident between different days in the study. This indicated that the response of H. vittatus was unique per day, and effects from the tagging procedure were no longer reflective from as early as 3 days post-tagging procedure, as shown in our linear model and the PCA analyses. This supports Jepsen et al.’s (2004) finding that the minimal disturbance in fish behaviour is seen during recovery, and the tagging procedure may only affect behaviour within 7 days post-tagging. Knowing that the tagging procedure is regarded as a stressor on the behaviour, the fish will navigate its recovery whilst maintaining daily circadian rhythms whilst facing associated environmental conditions. Although none of the individuals were predated with 14 days after the tagging procedure, this period of recovery suggests that telemeter tagged H. vittatus may be susceptible to predation as in other studies (Thorstad et al. 2003).

The minimum length of the recovery period for fish can be as short as 8 h to a day depending on the species, tagging methodology and the fish species’ environment (Jepsen et al. 2002; Metcalfe et al. 2016). Økland et al. (2005) suggested 7 days, specifically for H. vittatus. Our study suggests that recovery can extend up to 14 days, however, but can be seen from as early as 3 days post-tagging procedure. On the contrary, the diel behaviour of tigerfish can return to normalcy within the 24-h period and may be crucial to the recovery of the fish. More recently, Jacobs et al. (2016) used upstream and downstream movement of H. vittatus to assess the immediate and long-term effects of telemeter tags and found that the movement of the individuals was predominantly downstream (60%). This was associated with reduced spatial movement within the first 3 days of tagging. As such, tagging procedures have immediate effects (< 3 days after tagging) and minimal in the long term (> 7 days) (Jacobs et al. 2016; Jepsen et al. 2004). In our study, 2 weeks was shown to fall within the periods of an exponential increase in activity rates; however, fluctuations after this period fell as low as activity rates seen after 7 days. Paukert et al. (2001) found similar findings and concluded that activity patterns of individuals could be abnormal to up to 2 weeks following release after tagging. Although these studies (Jacobs et al. 2016; Økland et al. 2005; Paukert et al. 2001) used manual monitoring techniques, in our study, the activity rates showed that daily circadian rhythms were reflective of normal behaviour from within 3 days post-tagging. The fine-scale observations using activity rates in our study were conclusive. The activity rates of H. vittatus were not only reflective of their full recovery within 2 weeks of tagging but could also be indicative of their need to maintain biological processes (feeding and resting) as seen in the diel behavioural activity patterns. This supports the use of H. vittatus as suitable candidates for telemetry studies (Økland et al. 2005) and can be used to understand the response of the fish to environmental variables (Burnett et al. 2020).

The present study also used the activity rates of H. vittatus as a line of evidence to assess their vulnerability to predation and the causality related to flows and predatory events in inland aquatic ecosystems. Of the seven H. vittatus that were successfully monitored, three were removed by predators after the tagging procedure recovery periods tested. Through manual tracking and recovery of radio tags, the African fish eagle (H. vocifer) appeared to be a key predator of adult H. vittatus. This was supported by the activity rates for individuals HVIT2 and HVIT5 and their change in environments from the freshwater to terrestrial environments. During the tag recovery exercise, HVIT2 and HVIT5 tags were found below a feeding roost and surrounded by fish scales and bones, confirming that H. vocifer were the likely predators of H. vittatus. Real-time data showed that these events occurred during the day when H. vocifer and other piscivorous birds such as P. haliaetus were active, of which both are known to be diurnal species, the latter being migratory (Østnes et al. 2019; Stewart et al. 1997).

Notwithstanding H. vocifer being a common natural predator of H. vittatus, otters (Lutrinae family) and Nile crocodiles cannot be ruled out as potential predators as they are known to feed on fish (Games & Moreau 1997; Rowe-Rowe 1977). As shown by HVIT8, a predation event took place at night, resulting in the tag being removed from the water and later returned to below detectable limits of the telemeter tag, given adequate network coverage in the area. The occurrence of this event at night suggests that a mesocarnivore like the Cape clawless otter (Aonyx capensis) were responsible as they are nocturnal predators, known to be common in the Olifants and Letaba Rivers (Pienaar 1964; personal observation). The high presence of piscivorous predators in our study area suggests the high impact of these predators on fish communities. Our study suggests that the impact of predation is considerably higher than what was initially thought and should be considered as a factor in disturbances to fish communities. Similar studies have shown this using an avian approach in estuaries (Cowley, Terörde & Whitfield 2017), yet little is known about riverine ecosystems where predators are linearly spaced. This study suggests that predation on fish species is not solely avian related and that mammalian and reptilian predation could also be significant and should be considered especially for adult fish.

We acknowledge that one of the limitations in our study was the relatively small sample size, which was partly because of the cost of telemeter tags and accessibility to the field sites that limited the ability to increase the sample size. As such, larger sample sizes are recommended in further studies to improve confidence in using activity rates as a line of evidence and understanding the results shown here. In the present study, we were also limited to the use of adult fish because of the mass of the telemeter tag (20 grams [g]); thus, only individuals above 1000 g could be tagged as determined by the 2% tag to body weight ratio (Jepsen et al. 2002). Despite these restrictions, our findings provide inferred population responses, strengthening the cause and effect data associated with H. vittatus. Considering this, the results of our study are useful in the management of water resources, especially for downstream users such as the Kruger National Park, and can lay a foundation for future use of this technique.

The low flow observed in both rivers during the survey is also a matter of concern. One of the immediate results of these low flows, potentially resulting from water abstraction, dams and drought, is the formation of isolated pools with little to no flows (Pearce 2018). Stears and McCauley (2018) reported that during these low flow and dry periods, the lack of sufficient oxygenated water and removal of nutrients could cause oxygen levels to decrease and nutrients to reach toxic levels. Although not measured in the current study, Stears and McCauley (2018) reported that in hypoxic conditions oxygen-dependent fish species such as yellowfishes (Labeobarbus spp.) tend to swim close to the water surface where oxygen levels are comparatively higher, making them vulnerable to predators. Hydrocynus vittatus is known to use the entire water column to adapt to changes in food resources and water quality (O’Brien et al. 2014; Smit et al. 2013; Steyn et al. 1996). These adaptations to altering environmental conditions could increase their exposure to predators as they swim closer to the surface in hypoxic conditions (Stears & McCauley 2018). Other opportunistic predators such as baboons Papio ursinus have been observed predating on stressed fish in these circumstances (Stears & McCauley 2018).

Hydrocynus vittatus are oxygen-dependent species (Vignaud et al. 2002) and are likely to be under more pressure during periods of low flow when the river becomes fragmented into isolated pools than other species. They are also known to generally stay within their home ranges (Jacobs et al. 2019; Økland et al. 2005), with their movement often restricted by in-stream barriers (Roux et al. 2018). As such, they are more likely to be affected by low oxygen levels when trapped in isolated pools. The reduction in flows and in-stream barriers (dams and natural waterfalls) on both the Olifants and Letaba rivers resulted in relatively few opportunities for H. vittatus to move to reaches where flows were favourable. As observed in this study, this traps them in isolated pools until conditions improve or mortality ensures. This was supported by the remote network established for this study, not detecting reach scale movement of tagged fish. The reduction in water flows from both river systems coincided with the predation events, showing that flow stressors increase the risk to predation and the ecological degradation of the ecosystem for H. vittatus. Two main in-stream barriers (Engelhard and Mingerhout Weirs) upstream on the Letaba River form artificial barriers obstructing already low flow conditions and have no economic or biodiversity value for the lower Letaba River system (O’Brien et al. 2019). Flows beyond these gauging weirs typically go unmonitored, and our study detected no flow in the Letaba River whilst Engelhard Dam recorded minimal flows at < 0.314 m³s⁻¹. The activity rates and predation events coincided with this reduction in the flow and susceptibility of H. vittatus to hypoxic conditions (Smit et al. 2013). This highlights the importance of maintaining environmental flows into the Olifants Gorge. The discrepancy in the flows between Engelhard Weir and the Olifants Gorge suggests that this weir cannot meet environmental flows downstream and questions the ability of the current infrastructure to adequately monitor and meet the environmental flow requirements for the Olifants and Letaba Rivers. With this, the Engelhardt Wier makes for a good case to be removed as it jeopardises the biodiversity in the Letaba River as with other such in-stream barriers (O’Brien et al. 2019). All-inclusive real-time and remote monitoring techniques have been developed to minimise the loss in river connectivity and monitor rivers, and these can continue to be implemented in the Olifants-Letaba Catchment (Burnett et al. 2020; Pollard et al. 2012).

Telemetry methods should, where possible, have larger sample sizes and make use of smaller and lighter tags (< 20 g air mass) with proportionally less body to tag mass ratios for adult fish. These recommendations will improve the confidence in results and the recovery process as well as increase the representable size classes in telemetry studies of this nature. This will further ensure the inclusion of smaller fish species as well as earlier life stages in telemetry studies. Our study also demonstrated that external telemeter tags affected the activity patterns of individuals in the short term (< 3 days) but minimal in the long term. Despite the stress caused by tagging, inadequate flows can contribute to the vulnerability of H. vittatus to predation, especially when environmental flows are not adequately maintained. The reduction in flows may play a role in increasing the risk of predation of H. vittatus and their overall condition in the freshwater environment. Similarly, additional environmental stressors may be compounded by reduced flows, and if not addressed, it would further jeopardise the population of H. vittatus in the Kruger National Park.

The excessive use of water resources by users upstream of the Kruger National Park (Dabrowski, Oberholster & Dabrowski 2014; De Villiers & Mkwelo 2009; Fouché 2008) is typically the main drivers of reduced flows in the downstream reaches of the Letaba and Olifants rivers. Although compounded by other environmental stressors, altered flows expose H. vittatus to predation, increasing the likelihood of such events during the low flow season when exacerbated by abstraction upstream. In our study, the extremely low flows observed in the Letaba River were further exacerbated by the presence of unused dams in the Kruger National Park, mainly Engelhardt Wier, that limited the release of water downstream. Its removal should be considered to contribute to maintaining low flows and the biodiversity in the Letaba River to ensure that environmental flow conditions for Kruger National Park and into Mozambique are met.

Setting environmental flows and implementing them could be the solution to the conservation and sustainability of H. vittatus in the Kruger National Park. Suitable allocations of environmental flows to the Letaba and Olifants Rivers will have a direct influence on this species by reducing environmental stress, avoiding predators and improving their general fitness. The fine-scale detection of activity rates recorded using the telemeter tagged fish could further contribute to an understanding of the response of fish to environmental conditions enabling long-term monitoring of anthropogenic stressors placed on these and other river systems.

The value of telemetry studies such as ours is that they can significantly assist in understanding the behavioural responses of fish species to environmental stressors, such as flows, and also other water quality parameters. Monitoring the behavioural responses in real time and remotely can further assist the implementation of environmental flows in ensuring optimal functioning of the ecosystems and increasing the output of ecosystem services provided by freshwater ecosystems. The conservation of H. vittatus is essential in South Africa as they serve as keystone species and meeting their requirements will assist in the conservation of other aquatic species maintaining the integrity of those freshwater ecosystems.