The Okavango Delta is a vast mosaic of various habitats consisting of swamps, islands and river channels whose aquatic, semi-aquatic and terrestrial phases change constantly at different temporal scales, driven by the flood regime (McCarthy et al., 2003; Ramberg and Wolski, 2008). It is located in a dry sub-tropical area with a mean annual rainfall of 475 mm and experiences large annual variations in temperature where October is the hottest month while July is the coldest (Milzow et al., 2009). Rain normally falls in the period November—March while annual flooding from the Angolan highlands occurs in the period April - September (Ramberg and Wolski, 2008). Annual precipitation, which is out of phase with seasonal flooding (Porter and Muzila, 1989; Ramberg et al., 2006a), contributes approximately between 5% (Andersson et al., 2003) and 42% of the total water input into the delta, while the rest comes as discharge from the Angolan highlands (Ramberg and Wolski, 2008). Total water storage in the delta is about 10 Km³ (about a year’s inflow of water) which supports diverse vegetation (Porter and Muzila, 1989) “aquatic” and wildlife species (Ramberg, et al., 2006b). The delta’s hydrology is dynamic (i.e. changes in flow patterns from one part of the delta to the other), that varies in response to changes in seismic activity, vegetation dynamics, animal activity (such as hippos) and human intervention (Wilson, 1973; Porter and Muzila, 1989; Wolski and Murray-Hudson, 2006; Milzow et al., 2009), causing the flow in the anastomosis of channels to change at any given time due to variations in these factors.

Peak discharge in the delta’s panhandle occurs in March/April (Wolski, et al., 2005) and the flood pulse travels progressively down the delta, taking a maximum of 6 months to reach the distal ends of the system (Andersson et al., 2003). The sinusoidal flooding cycle (Figure 2) in the delta results in a period of minimum inundation (November - March) to a period of maximum inundation (May - September) (Andersson et al., 2003; McCarthy et al., 2003; Wolski et al., 2005). Water depth variations in the permanently flooded areas are usually very small, while normally in the order of 1–2 m in the seasonally inundated parts of the Delta (Ramberg et al. (2006b).

Also inter-annually the flows in the delta have a cyclical behaviour with a 17.5 year periodicity in the annual average and maximum flows (Mazvimavi and Wolski, 2006). However, there is high inter-annual variability in flooding patterns where good flood years may be followed by poor flood years and the extent of flooding in the previous year and local rainfall also affect the extent of flooding in any 1 year (Milzow et al., 2009; Mendelsohn et al., 2010). While inter-annual variations in rainfall cause variability (lows and highs) in its flooding regime (Wolski and Murray-Hudson, 2006), seismic Earth movements also cause different parts of the delta to periodically undergo drying episodes (Milzow et al., 2009). Thus, flooding dynamics in the delta are highly dynamic and critical towards a comprehensive understanding of ecological processes in the delta.

Catch statistics are a major challenge in most fisheries around the world (FAO, 2020) particularly in inland fisheries (Welcomme, 2011) and Botswana is not an exception. Generally, data on fish catches are fragmented (Arntzen, 2005) and irregular. Therefore, there is a lack of accurate catch statistics from the Okavango Delta, and most of the initial records are best estimates (Mosepele, 2003). Subsequently, the period between 1970 and 1987 is characterized as a time of poor/uncertain data, while fish yield data from 1996 are relatively better (Mosepele, 2003). Table 1 summarizes fish production data from the Okavango Delta. Turpie et al. (2006) attributes these discrepancies to the presence of different actors in the delta’s fishery. However, inter-annual variability in flooding patterns can also cause discrepancies in fish production estimates. Fishing effort in the delta is driven by the seasonal flood regime (Mosepele et al., 2018), where most traditional/artisanal fishers are only active during good flood years (Mosepele, 2001). Furthermore, Mosepele (2000) and Mosepele (2001) highlights that catch data is only collected from gill net fishers, which would underestimate the total annual production from the fishery. The extent of flooding in the delta also opens up new fishing grounds which would increase the total annual fish production from the system. According to Mosepele (2019), fishery yield is driven by the delta’s flood regime at a 2-year time lag. This fishery-flood relationship is an illustration of Junk et al.’s (1989) flood pulse concept.

Turpie et al. (2006) estimated the total annual private use value of the delta’s fishery at approximately US$ 490,000 (based on 2005 exchange rates), which was approximately 9% of the total value from the delta’s wetland resources. Despite this relatively low use value, the delta’s fishery is a source of livelihoods and food security for subsistence fishers (Mosepele et al., 2006) who increase fish catches as the first and second coping strategy during periods of food shortage (Mosepele et al., 2006; Ngwenya and Mosepele, 2008; Mmopelwa et al., 2009). Subsistence fishers either sell their surplus fish for cash or barter them for grain (Ministry of Agriculture, 1997; Mosepele, 2001; Mmopelwa et al., 2009). Therefore, subsistence fishing is a source of income for 40% of subsistence fisher households in the delta, where fishing income is used on food, toiletry, and clothing (Ngwenya and Mosepele, 2008). This is in accordance with Mmopelwa et al. (2009) who observed that subsistence fishing has socio-cultural, socio-economic and food security value to the delta’s subsistence fishers.

Furthermore, the presence of a small-scale and yet profitable commercial fishery (Mmopelwa et al., 2005) makes it an integral part of the delta’s rural livelihoods strategy (Mosepele and Ngwenya, 2010). The small-scale commercial fishery is not only a source of rural employment (Mosepele, 2001; Mmopelwa et al., 2005; Mosepele and Ngwenya, 2010), but fishery revenue is also invested into either agriculture (Mendelson et al., 2010; Mosepele and Ngwenya, 2010) or other economic activities (Mosepele, 2001). Kgathi et al. (2018) observed that the small-scale commercial fishery also contributes to fish regional trade, which may contribute to a reduction of Botswana’s food import bill. Generally, most of the fish catch goes towards livelihood support (Arntzen, 2005) which makes fish not only a valuable safety-net (Mendelson et al., 2010) but also a resource (Mosepele, 2001). Traditional (artisanal) fishing is also identified as a natural safety net against HIV/AIDS comorbidities in the delta (Ngwenya and Mosepele, 2008). Based on income generated, fishing was the second most important economic activity in the region after cattle farming in the 1990s (Mosepele, 2003).

Newly inundated floodplains are an important nursery habitat for fish recruitment (Junk et al., 1989; de Oliveira et al., 2020). In the Okavango Delta the inundated areas are dominated by juvenile cichlids (e.g. Oreochromis andersonii, Tilapia sparrmanii and Coptodon rendalli), catfish (Clarias gariepinus), and cyprinids (e.g. Barbus bifrenatus and B. barnardi) during the first month of flooding. Fish fry and juveniles were observed at increasing frequency starting from the second month of flooding (Høberg et al., 2002). The boom of primary producers and zooplankton initiated by the seasonal flooding, serves as abundant food sources for the juvenile fish and small fishes (Siziba et al., 2013) and also some adult fish (Mosepele et al., 2012). The subsequent decrease in zooplankton biomass corresponding with an increased frequency of juvenile fish over the flooding season is largely due to predation and decreased primary productivity (Høberg et al., 2002; Siziba et al., 2013). This suggests that failed or poor floods cause a bottle neck in fish production due to failed zooplankton production (Siziba et al., 2012).

Juvenile fish growth on the inundated floodplains is rapid within the first year of life (Dudley, 1974). Rapid growth ensures that juvenile fish are large enough to (i) avoid being stranded in the floodplains at receding floods, and (ii) avoid heavy predation when migrating into the permanent channels at draw-down (Booth and Merron, 1996). Foraging by juvenile fish in the inundated areas is an adaptation for taking advantage of high zooplankton biomass (Lindholm et al., 2007), and these shallow areas also act as a predator refuge due to highly fluctuating DO concentrations (Kolding 1993; Mosepele et al., 2017). Less frequently flooded areas (those only flooded occasionally at very high flows) show exceptional “booms” in zooplankton biomass and juvenile fish (Siziba et al., 2011b), especially after a low flood year (Siziba et al., 2013). During poor flood (i.e. low flood) years, the zooplankton biomass is less exposed to fish grazing, while predation appears to be a strong regulator of zooplankton biomass during good flood years (Lindholm et al., 2007). Large flood years result in extensive flooded areas which appear to particularly facilitate fish breeding, growth and survival and ultimately increased fish production (Lowe-McConnell, 1987; de Graff, 2003). Thus, the flood volume in the Okavango Delta is a major driver of fish production, where relative fish biomass during a high flood year can be double that of a low flood year (Lindholm et al., 2007).

Alternating wetting and drying processes are necessary in floodplains to increase nutrient turnover, maintain primary production dynamics (Junk et al., 1989) and hence fish production. However, the pattern of rise and fall of the hydrograph is influencing floodplain fish production. According to King et al. (2003), a “relatively slow rate of rise and fall” of the seasonal hydrograph creates optimum conditions for fish species to utilize the floodplain for recruitment. Conversely, a rapid rise and fall in the hydrograph may offset the balanced time lag between primary production and fish production (Tockner et al., 2000), which may result in less successful fish production. However, short lived hardy species in floodplain systems can adjust quickly to extreme hydrological events (Junk et al., 1989; Junk, 2002).

The preceding overview has highlighted the dynamic interactions and processes between floodplain fish communities and the highly dynamic environment in the delta. Floodplains are unstable, seasonally fluctuating ecosystems characterized by strong intra and inter annual variability, where the flood pulse is a key driver of practically all processes (Junk et al., 1989; Schongart and Junk, 2007). Inland fisheries in Africa are generally small-scale and labour intensive (Welcomme, 2011). They are characterized by multi-species assemblages, of different sizes exploited by diverse fishing gears and methods (van Zwieten et al., 2003; Welcomme, 2011; Kolding and van Zwieten 2014; Mosepele, 2019). In the Okavango Delta, the hydrological regime is a major driver of change in the biology and ecology of the fish community (Lindholm et al., 2007; Mosepele et al., 2009; Linhoss et al., 2012; Mosepele et al., 2012; Bokhutlo et al., 2015; Mosepele et al., 2017). Like the habitat, the fisheries are dynamic, fluctuating, and constantly changing and are never in stable equilibrium and the environmental drivers are in general much more important in regulating productivity than the fishing effort (Jul-Larsen et al., 2003, Kolding and van Zwieten 2006; Kolding and van Zwieten 2012). This makes conventional management approaches based on steady state assumptions inconsistent and difficult (Staples et al., 2004; Mosepele, 2008; Welcomme et al., 2010; Mosepele, 2014).

Except for a few highly commercialized fisheries in freshwater systems like the Amazon and Mekong (Welcomme et al., 2014), most tropical floodplain fisheries are a major source of localized food and nutrition and mostly serving as subsistence for riparian households (Junk, 2002; Mosepele et al., 2006; Welcomme, 2011). Their primary value to local communities is their contribution towards household income and food security (Mosepele et al., 2006), though some African inland fisheries are slowly morphing towards commercial or recreational fishing as well (Kolding and van Zwieten, 2014). Fishers in floodplain fisheries systems use various traditional techniques (Cerdeira et al., 2000; Kolding et al., 2003; van Zwieten et al., 2003) to adapt and optimize utilization of the ever changing fish assemblages, and the same is observed in the Okavango Delta (Mosepele et al., 2007; Mmopelwa et al., 2009; Mosepele, 2019). Floodplain fisheries are thus also a major source of traditional ecological knowledge (Mosepele, 2008) and cultural heritage (Junk, 2002). Therefore, floodplain fisheries management plans should incorporate these characteristics (i.e. cultural values and traditional knowledge) into their management objectives.

Gear restrictions and mesh regulations are fixed constant attributes and remain some of the easiest and cheapest regulations to implement in fisheries management regimes (Misund et al., 2002), and these have been widely implemented in floodplain fisheries. The fundamental question in fisheries management is how to regulate the fishing mortality, which is a combination of how to catch the fish (regulated by gear and mesh restrictions) and how much fish to catch (which is based on effort regulation). The key approach to regulate the ‘how’ question is to control gear selectivity (see next section), while effort on the other hand is sometimes regulated to maintain the aggregate fishing effort to obtain a “maximum sustainable yield” (MSY). An efficient economic exploitation of the fishery is assumed to save fish stocks from over-exploitation/collapse (Bene et al., 2010; Kolding and van Zwieten, 2014). Arguments such as these are attractive to policy makers and introduce policies aimed at effort reduction. The classical argument is that fishing effort is the main factor influencing fish stock dynamics, which is otherwise assumed in “steady state” and since catch is a function of effort, it needs to be managed. The alternative assumption would be that effort is controlled by the current production (Kolding and van Zwieten 2011; Kolding and van Zwieten 2014), and therefore largely self-regulated as in natural predator-prey relationships. According to Mosepele and Kolawole (2017), law enforcement in fisheries is prioritised over rural people’s livelihoods. This is a consequence of implementing classical management approaches in fisheries management. Subsequently, anecdotal evidence indicates that people’s livelihoods were curtailed through the implementation of these management approaches in the delta (Daily Maverick, 2017).

A key theoretical argument for regulating the gear selectivity is to protect the young fish and target the big fish in order to prevent so-called growth overfishing (Kolding and van Zwieten, 2011). Most fishing gears are selective regarding species, sizes and habitats fished (Kolding and van Zwieten, 2014, Figure 3 and Table 2) but regulating selectivity on certain sizes or species will invariably change the natural composition of the various components in the ecosystem (Garcia et al., 2012). For example, males of O. andersonii, O. macrochir and C. rendalli (these are the three most important commercial fish species in the Okavango Delta), grow larger than females (Dudley, 1974). Hence, selective harvesting with large mesh sizes would tend to select the males from the populations of these three species resulting in unbalanced sex ratios. Such scenario can alter the breeding sex ratio of an exploited population and ultimately reduce its reproductive potential (Fenberg and Roy 2008). Focusing exploitation exclusively on the mature part of the population will also alter the demographic composition and potential recruitment. It therefore makes ecological sense to also target younger and more productive age classes than only old big fish, the so-called (BOFFFs, Big Old Fat Fecund Females, Hixon et al., 2014), which are the engines of new recruitment by being more fecund and having better egg quality than smaller/younger fish (Trippel, 1995; Walsh et al., 2006; Kolding et al., 2015a). Smaller/younger fish are also relatively more productive than bigger/older fish (Law et al., 2012). Therefore, in order to maintain the natural structure and composition of fish communities it has been suggested to exploit populations in proportion to their natural productivity, the so-called ‘Balanced harvest’ concept (Garcia et al., 2012; Law et al., 2012; Zhou et al., 2019).

Similar species from different habitats in the delta display different phenotypic life history strategies (Merron and Bruton, 1988; Mosepele, 2000; Mosepele and Mosepele 2005; Bokhutlo et al., 2015) where lower delta species are generally smaller, upper delta species are generally larger (Merron and Bruton, 1988; Mosepele et al., 2011). While O. andersonii from the lower delta has slower growth than those from upper Delta, O. macrochir and C. rendalli from the lower Delta grow faster than their upper delta conspecifics (Mosepele, 2000). Moreover, lower delta populations of these three cichlids were found to mature earlier than those from the upper delta (Mosepele and Mosepele 2005). A similar observation was made for C. gariepinus (Bokhutlo et al., 2015). Wild tilapias are frequently observed to mature early and breed prolifically in small shallow water bodies, but not in larger, deeper environments (Lowe-McConnell 1982; Kolding 1993), and this ‘stunting’ is apparently driven by the fluctuating oxygen conditions found in shallow environments (Kolding et al., 2008).

From a multispecies point of view, the smallest fish species (Total Length) in the delta is approximately 3 cm while the largest species is over 1 m with a graduation of sizes in between them (Mosepele, 2019, Figures 3, 4). Implementing conventional mesh (or gear) regulations will skew fishing mortality towards larger sizes of the community size spectrum (Figure 4), causing a structural and demographic change of the fish community, and possibly also effecting functional changes. Selective fishing, can in the long run also cause evolutionary change in exploited populations (Rochet, 1998; Law, 2000). As a consequence, exploited stocks undergo changes in growth and maturation (Rochet, 1998; Law, 2000), and selective fishing essentially causes ecosystem imbalances (Schindler et al., 1998; Law, 2000; Kolding and van Zwieten, 2011). According to the Convention on Biological Diversity (CBD), a major component of the Ecosystem Approach to Fisheries (EAF) is to maintain the structure and function of the natural communities as close as possible to the natural stages.

Classical single-species assessment models are incompatible with multi-species, multi-gear fisheries (Mosepele, 2008; Welcomme, et al., 2010; Welcomme, 2011). A more balanced exploitation pattern, harvesting species of all sizes and all trophic levels in proportion to their individual productivity, is likely the best management approach for floodplain fisheries in terms of both yield and maintaining the fish community structure (Kolding et al., 2003; Mosepele 2014; Kolding et al., 2015b; Mosepele 2019). There are at least nine different fishing gears/methods observed in the Okavango Delta, with specific catch signatures (Figure 3), but which collectively harvest the fish community across different age classes, species and trophic levels (Figures 4, 5) (Mosepele et al., 2003; Mmopelwa et al., 2009). A cumulated log-converted multispecies and multi-gear catch curve (Figure 4; Table 3) shows that the fishery is approximately “balanced” on all sizes above 28 cm TL (being not significantly different from a theoretical slope of −6), but smaller sizes and species are still under exploited compared to larger sizes. Such multi-species harvesting pattern, by the diversified gear assemblage (Figure 3) is a common attribute of floodplain fisheries (Kolding et al., 2003; Kolding and van Zwieten, 2014). The only fish stock assessment of the delta so far (Mosepele and Kolding, 2003), showed that i) the fish stocks were generally under-exploited and ii) that the fish community was being rationally exploited by using several different fishing gears and methods to harvest the delta’s diverse species assemblage (Mosepele, 2019, Figure 5).

Currently, some of these gears (e.g. mosquito nets) and fishing methods (e.g. drive fishing) are prohibited in the delta (Botswana Government, 2008; Mosepele, 2008; Mosepele, 2014). However, there is no empirical evidence to justify these regulations. About 70% of the species exploited by mosquito nets are generally very small species (e.g. Barbus radiatus, Aplocheilichthys johnstoni, etc.), which are not caught by other methods (Mosepele et al., 2003; Mosepele, 2019, Figure 3). Restricting this gear will result in decreased catches of these small sized underutilized species, which are primarily harvested by women for subsistence household consumption. Moreover, small fish are usually eaten whole, with heads, bones, skin and viscera, and represent a concentrated source of multiple essential nutrients compared to only eating the flesh of larger species (Longley et al., 2014). Drive fishing is a traditional and efficient method for exploiting O. andersonii, O. macrochir and C. rendalli (Mosepele et al., 2007; Mosepele, 2019). However, prohibiting drive fishing (which would only legalise stationary gill net setting) will skew gill net fishing mortality towards O. andersonii (Mosepele, 2000; Mosepele, 2019), while Coptodon rendalli, well known for escaping stationary gillnets (Mosepele, 2000; Kolding et al., 2003; Mosepele et al., 2007), will remain relatively unexploited. In addition, prohibiting drive fishing will result in reduced revenue for the delta’s commercial fishers which are primarily targeting cichlids. A blanket prohibition of some fishing methods and gears, without informed justification may not only cause ecosystem imbalances, but may also reduce the food security aspect and socio-economic value of the fishery to riparian communities. The principle of Balanced Harvest (BH) has been strongly criticized by Froese et al. (2015), because they argue it does not conform to ‘basic population dynamics’ as developed by Beverton and Holt (1957). However, the basic population dynamics were single-species, steady state models where fish grow without eating (Tilley et al., 2020), whereas BH is a multi-species concept with a concrete proposal for implementing the Ecosystem Approach to Fisheries (EAF) (Kolding et al., 2015b), which does not only make ecological and biological sense in floodplain fisheries (Kolding et al., 2003; Mosepele, 2014), but it is also sensitive to the traditional fishing patterns and cultural value of floodplain fisheries (Mosepele, 2008).

Diversified fishing techniques, as traditionally practiced in the Okavango Delta and many other African inland fisheries (Kolding and van Zwieten, 2014), ensure that most species across various sizes and habitats-in the fish community are exploited. It also allows impoverished households (especially those headed by women), to have access to high quality protein and nutrients, which again ensures that young children from these fishing households have a relatively good nutritional status (Nnyepi et al., 2007; Longley et al., 2014; Kolding et al., 2019). BH was intended to reduce adverse ecological impacts of fishing while also supporting sustainable fisheries (Garcia et al., 2012). Fisheries management should also preserve cultural and heritage practices of fishing communities when these are not proven destructive, because, “culture is a fundamental human right” (Junk, 2002). Therefore, we advocate for balanced fishing that allows the utilisation of diverse fishing gears and methods to exploit the delta’s fish community.