Environmental Biology of Fishes 43: 109-119,1995. C’ 1995 Kluwrr Academic Publishers. Printed in the Netherlands.

Community ecology and conservation of the fishes of the Okavango Delta, Botswana

Glenn S. Merron & Michael N. Bruton J. L. B. Smith Institute o,f Ichthyology, Private Bag 1015, Grahamstown, 6140, South Africa

Received 28.10.1992 Accepted 14.10.1994

Key words: Diversity, Predictable, Unpredictable, Stability, Eurytopic, Stenotopic, Seasonality, Wetland, Africa

Synopsis

The Okavango Delta is a large inland swamp in northern Botswana which receives an annual flood from the highlands of southern Angola. There are distinct fish taxocenes in the Okavango which can be separated from each other by the physical characteristics of the different habitat types with which they co-evolved. An ac- count is given of the ecology and conservation of the fishes of the Okavango Delta. Their response to the annual flood regime, and the environmental factors that limit their distribution and abundance, are described. In the northern riverine floodplain and perennial swamp a higher species richness and ichthyomass was re- corded than in the seasonal swamp and drainage rivers. Suggestions are made on the conservation of Oka- vango fishes taking into account the ecological characteristics of the Delta.

Introduction

Floodplain ecosystems, such as the Okavango Del- ta in northern Botswana (Fig. l), are subject to regu- lar cycles of flooding and draining. The timing, mag- nitude and duration of these floods are not constant from year to year and in the Okavango these par- ameters are determined largely by the periodicity and amount of annual rainfall in the highlands of southern Angola (Wilson & Dincer’). Floodplain fish populations often depend on the annual flood cycle for their survival (Lowe-McConnell 197.5, Welcomme 1979). The floods periodically connect the water bodies on the floodplain to the river, and

‘Wilson, B.H. & T. Dincer. 1976. An introduction to the hydrol- ogy and hydrography of the Okavango Delta. pp. 33-47. In: A.C. Campbell, D.N. Nteta, J. Hermans & L.D. Ngcongco (ed.) Sym- posium on the Okavango Delta and Its Future Utilization. Bot- swana Society, Gaborone.

facilitate migrations and spawning of most fish spe- cies. By inundating low-lying regions, the flood wa- ters also entrain terrestrial plant and animal matter into the aquatic system where it forms the basis of the food chain. This allochthonous food input is uti- lized by the fish for gametogenesis and somatic growth (Lowe-McConnell 1975. Welcomme 1979, Bruton & Jackson 1983). The inundation of shallow African floodplains and adjacent terrestrial lands is also important for providing safe nursery sites for fish larvae and juveniles during their early stages of development (Bruton & Jackson 1983). Within the Okavango Delta a similar pattern of annually fluc- tuating water levels regulating the structure of the fish community is apparent, but the processes in- volved are not fully understood (Merron 1991).

The fishes of the Okavango Delta represent a val- uable natural resource for the people of Botswana. There are, however, a variety of man-induced and natural stresses on the Okavango Delta. These

110

111

stresses include: increased artisanal, commercial and recreational fishing; the spraying of insecticides to control the tsetse fly, Glossina morsituns, a carri- er of trypanosomiasis; and water abstraction schemes (Merron 1991).

These exploitation pressures are compounded by natural environmental perturbations. The extend- ed period of drought and low water levels through- out much of the 1980s markedly changed the char- acteristics of the Delta by decreasing the area of floodplain habitat.

The overall objective of this study was to contrib- ute to the understanding of the response of the fish- es to the annual flood cycle and to identify the key factor or factors of the flood which determine the nature of the fish communities. We hypothesize that the annual flood regime is important in main- taining fish populations and that the main factors determining the distribution and abundance of fish communities in the Delta are the permanence (re- tention time) of the water in particular habitats and whether or not the water is flowing.

This hypothesis is examined by comparing the re- sponse of the fishes in different areas of the Oka- vango to water level changes. Although all areas are subject to flooding, the enormous size of the Delta (15 000 km2 at high flood), causes the timing, magni- tude and duration of flooding to vary, with the greatest fluctuations occurring in the southern (downstream) end of the system.

Study area

The Okavango River rises in a series of headwater streams on the southern slopes of the Angolan high- lands and enters Botswana as a single broad river about 150 m wide and 4 m deep. The river meanders within a broad riverine floodplain (average width 15 km, approximate length 100 km) before branch- ing out to form the complex anastomoses of the Delta. The Delta fluctuates in area from 12 000-15 000 km* during the flood season and from 6 000-8 000 km* during the dry season (Campbell 1983).

The flood peak arrives in the northern riverine floodplain about March/April and reaches the southern drainage rivers in July/August (Wilson &

Dincer’). By the time the floodplains and rivers in the southern areas are full, the water level is low again in the northern regions. The timing, magni- tude and duration of the flood is not constant from year to year as each annual cycle depends on the rainfall history in Angola. An indication of the an- nual fluctuations in flood dynamics is shown by re- cords of the daily discharge of water in the Okavan- go River at Rundu in Namibia from 1983 to 1986 (Fig. 2). The highest discharge was recorded be- tween 1983-1984 whilst the 1984-1985 flood cycle was the lowest recorded. The 1985-1986 flood cycle, although higher than the 1984-1985 cycle, had a truncated duration when compared to former years.

The flow of water through the Delta can also be influenced by tectonic disturbances. sediment transport and channel blockages, which can pro- duce minor topographic changes that influence the distribution of water flow (Wilson & Dincerl. McCarthy et al. 1986). The Okavango is thus a dy- namic system that varies in space and time and in major physico-chemical factors, such as water avail- ability and quality, which may change each year in the different habitat types.

In terms of its hydrology, the Okavango is more stable (predictable) in the northern regions and less stable (unpredictable) in the southern regions. The magnitude of difference in the area inundated by water during different floods is in the order of 1 to 2

Fig. 2. Daily discharge of the Okavango River at Rundu for the period October 1983 to July 1986. Discharge data provided by the Department of Water Affairs. Namibia.

112

-c . A

;h,

In

,Ngaml

0 , 49 , , 8Q kilometers

Fig. 3. Map of the Okavango system in Botswana showing the location of the major sampling sites and habitats surveyed during this study: 1 = riverine floodplain, 2 = perennial swamp, 3 = seasonal swamp, 4 = drainage rivers, 5 = sump lake.

113

in the northern Delta and 1 to 10 in the southern seasonal areas (Fox2).

Five major habitat types are recognised in the Okavango Delta (Fig. 3). These are the riverine floodplain, perennial swamp, seasonal swamp, drainage rivers and sump lakes (Wilson & Dincer’). These five habitat types grade into one another. The slow pattern of inundation, which is due to the extremely low gradient (1:36000) results in about 95% of the water being lost to evaporation and evapotranspiration. Only about 2% of the inflow reaches the Thamalakane River in the south. Some important physical and chemical characteristics of the different habitat types are described by Thomp- son”, Wilson & Dincer’, Allanson et al. (1990) and Merron (1991). The limnology of this system has. however, not yet been fully studied and very little quantitative information exist.

Sampling sites, methods and general techniques

From November 1983 to December 1986 quarterly surveys were carried out in all habitats except the riverine floodplain. Quarterly surveys in the river- ine floodplain began in April 1985. The quarterly surveys comprised four distinct hydrological peri- ods including a filling, high, receding and low flood cycle. A comparative sampling regime using a varie- ty of fishing gear such as gillnets, seine nets and ro- tenone was performed during the quarterly surveys. In addition to the sites surveyed on a quarterly ba- sis, numerous other collections were made through- out the rivers and floodplains of the Okavango sys- tem during the 1980s (see Merron 1991).

Within the riverine floodplain sampling was con- ducted between the main river channel at Seronga and Dungu floodplain lagoon. Within the perennial

‘Fox. P.J. 1976. Preliminary observations on fish communities of the Okavango Delta. pp. Q-130. In:A.C. Campbell, D.N. Nteta. J. Hermans & L.D. Ngcongco (ed.) Symposium on the Okavango Delta and Its Future Utilization, Botswana Society, Gaborone. ‘Thompson, K. 1976. The primary productivity of African wet- lands with particular reference to the Okavango Delta. pp. 67- 79. In: A.C. Campbell, D.N. Nteta, J. Hermans & L.D. Ngcongco (ed.) Symposium on the Okavango Delta and Its Future Utihza- tion. Botswana Society, Gaborone.

swamp, sampling was conducted in the north-east- ern corner of Moremi Wildlife Reserve from Xaka- nixa Lagoon southwards to Magwexana Pools (Fig. 3). A rain pool habitat was also sampled in the pe- rennial swamp. Rain pools were only connected to the surrounding floodplains during the relatively high flood levels of 1984 and 1986 and were primar- ily maintained by local rainfall and underground seepage.

In the southern seasonal swamp, Nxaraga La- goon and the Boro River at the south-eastern end of Chief’s Island were surveyed as well as the sur- rounding floodplain and floodplain lagoons. Within the drainage rivers, the Thamalakane River at Mat- lapaneng and the Boteti River at Chanoga were sur- veyed. Lake Ngami, a sump lake (Fig. 3) was dry during most of the study period.

All fishes were sorted and weighed to the nearest gram. Although only 4 species of Synodontis were identified, Skelton & White (1991) recognize 2 addi- tional species.

A total of 32 datasets (i.e. 8 sampling sites multiplied by 4 respective flood levels) were ana- lysed using two-way indicator species analysis (TWINSPAN, Gauch 1982) to determine a commu- nity classification and to help explain the factors limiting the distribution and abundance of fishes. The dataset on fish collections for each quarterly survey was entered together with data on pertinent physical properties of each sampling site such as the depth of water, whether or not the site was peren- nially or seasonally flooded, flow rates, substrate type and amount of floating, emergent and submer- gent aquatic vegetation (see Merron 1991). The da- taset was then used to obtain an ordered dendro- gram that expressed the synecological relationships of fish species in different habitats.

Results

The Okavango Delta has a high species richness of fishes (65 species recorded during the present study period, Table 1). The number of species in different sampling sites varied markedly. Species richness and total catch, in mass, were highest in sampling sites within perennially flooded habitats (i.e. river-

114

Tablel. Checklist of fish species collected from respective sampling sites in the Okavango Delta, between November 1983 and December 1986.

Species Sampling site*

1 2 3 4 5 6 7 8

Mormyridae Hippopotamyrus ansorgii H. discorhynchus Marcusenius macrolepidotus Mormyrus lacerda Petrocephalus catostoma Pollimyrus castelnaui

Characidae Brycinus lateralis Hydrocynus vittatus Micralestes acutidens Rhabdalestes maunensis

Hepsetidae Hepsetus odoe

Distichodontidae Hemigrammocharax machadc H. multifasciatus Nannocharax macropterus

Cyprinidae Barbus afrovernayi B. barnardi B. bifrenatus B. eutaenia B. fasciolatus B. haasianus B. multilineatus B. paludinosus B. poechii B. radiatus B. thamalakenensis B. unitaeniatus Coptostomabarbus wittei Labeo cylindricus L. lunatus Opsaridium zambezensis

Bagridae Auchenoglanis ngamensis Zaireichthys chobensis

Schilbeidae Schilbe intermedius

Clariidae Clarias gariepinus C. ngamensis C. stappersi C. theodorae

Mochokidae Chiloglanis fasciatus Synodontis leopardinus

X

X

X

ri X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

11.5

Table 1. (Continued).

Species Sampling site*

1 2 3 4 5 6 7 8

S. macrostigma

S. rugromaculatus

S. woosnami

Poeciliidae Aplocheilichthys hutereaui

A. johnstoni

A. katangae

Cichlidae Hemichromis elongatus

Oreochromis andersonii

0. macrochir

Pharyngochromis darlingi

Pseudocrenilabrus philander

Sargochromis carlottae

S. codringtoni

5 giurdi

S. greenwoodi

Serranochromis angusticeps

S. longimanus

S. macrocephalus

S. robustus jallae

S thumbergi

Tilapia rendalli rendalli

7: sparrmanii

7: run,eti

Anabantidae Ctenopoma intermedium

C. multispinus

Athieomastacembelidae Athieomastacembelus frenatus

x )i x x

x

x

x

x

x

x

x

Y

X x x

x

x

x

X

x

x

X

x

X

x

x

X

x

x

x

X

X

X

x

X

X

X

X

x

X

X

X

X

X

x

x

X

x

x

x

X

X

x

x

x

x

x

x

X

X

x

x

x

X

X

X

X

x

x

x

x

x

X

x

X

X

X

x

X

X

X

X

x

X

X

x

X

x

x

x

X X

x x x

x

X x

X

x

X

X x

x

x

X

x

x

x

X X

x x X x x x

Total no. species 47 55 59 62 60 28 65 60 Total mass (kg) 581 489 481 599 394 102 649 605 No. of collections 13 13 13 13 13 13 8 8

* 1 = Boteti River (drainage), 2 = Thamalakane River (drainage), 3 = Nxaraga Lagoon (seasonal swamp), 4 = Xakanixa Lagoon (peren- nial swamp). 5 = Magwexana Lagoon (perennial swamp floodplain-connected lagoon), 6 = perennial swamp rain pools, 7 = mainstream channel (riverine floodplain), 8 = Dungu Lagoon (riverine floodplain-connected lagoon).

ine floodplain and perennial swamp), and lowest in sampling sites within seasonally flooded habitats (e.g. seasonal swamp and drainage rivers).

The use of the TWINSPAN multivariate commu- nity classification revealed a clear distinction, at the first level of division, between the species composi- tion (number and mass of fish) of sampling sites in perennially and seasonally-flooded habitats (Fig. 4). At the second level of division, a clear distinction

was again made between the species composition of perennially-flooded, fast-flowing (> 1 m sec.‘) and moderately-flowing (l-O.5 m set-‘) sampling sites. Within the seasonally-flooded habitats. the species composition of rain pools differed to slow flowing sampling sites (< 0.3 m sec.‘).

The third dichotomy grouped the species compo- sition of the perennial swamp at Xakanixa Lagoon (site 4) and the riverine Okavango mainstream

116

not l lOW flowing flowing

Fig. 4. TWINSPAN community classification based on the grouping of all fish collections and sampling sites in the Okavango Delta for the quarterly surveys between November 1983 and December 1986. Sampling site numbers are given in Table 1.

channel (site 7). The perennial swamp floodplain- connected lagoons (e.g. Magwexana Lagoons; site 5) and the riverine floodplain-connected lagoons (e.g. Dungu Lagoon; site 8) were grouped together.

Although sites 4 and 7 are in different areas of the Okavango, they are characterised by a continual flow of water. Sites 5 and 8 are connected to the larger river channels only during flooding.

Within the seasonally-flooded habitats, a distinc- tion was made between the species composition of Nxaraga Lagoon (site 3) and the drainage rivers (sites 1 and 2).

The fourth dichotomy further subdivides the drainage river sampling sites (sites 1 and 2). The species composition in all other sampling sites formed closely similar groups on their own (Fig. 4). No distinction between seasonal changes in the flood cycle within a sampling site and the respective species composition was evident at this level of analysis.

Fish species associated with the different habitats

Table 2 lists the 10 dominant species according to

mass for each habitat type. The piscivorous tigerfish Hydrocynus vittatus, was restricted to the riverine floodplain and perennial swamp. The absence of this species from the seasonal swamp and drainage rivers can be related to its preference for large, rela- tively clear and fast-flowing water bodies. The pisci- vorous African pike, Hepsetus odoe, was a dom- inant species in the slower-flowing, well-vegetated drainage rivers and seasonal swamp. Being an am- bush predator, H. odoe relies on dense vegetation for cover while waiting for prey, whereas H. vittatus is an actively-foraging predator in the perennially flooded habitats.

The ubiquitous silver catfish, Schilbe intermedi- US, was a dominant species in all habitats although its percentage mass contribution in the perennial swamp and riverine floodplain was considerably lower than in the seasonal swamp and drainage riv- ers (Table 2). Conversely, the relative mass abun- dance of the sharptooth catfish, Clarias gariepinus, and blunttooth catfish, C. ngamensis, was higher in the perennially-flooded habitat types. The spotted squeaker, Synodontis nigromaculatus, and three- spot tilapia, Oreochromis andersonii, were relative- ly abundant throughout all habitats. These species

117

Table 2. The percentage mass contribution of the 10 most common species occurring in four different habitats of the Okavango Delta, Botswana.

Species Habitats*

dr SS P” rp

Hydrocynus vittatus

Hepsetus odoe

Labeo lunatus

Schilbe intermedius

C/arias gariepinus

C. ngamensis

Synodontis leopardinus

S. nigromaculatus

Oreochromis andersonii

0. macrochir

Serranochromis angusticeps

S. robustus jallae

S. thumbergi

Tilapia rendalli

7: sparrmanii

Total mass (kg)

21.7 22.6

10.4 17.4

4.4

17.4 14.7 6.9 6.1

6.6 7.0 13.8 IS.8

3.4 s.3 5.1

3.4 4.2

3.0 5.8 3.6 3.9

14.7 9.x 9.9 10.7 6.3 2.9 4.6

7.9 9.x 9.3 3.5 2.8 2.3

2.6 3.x 2.2 6.3 4.7

1070 481 993 1254

* dr = drainage river, ss = seasonal swamp, ps = perennial swamp. rp = riverine floodplain

have sufficiently generalized habitat preferences to be tolerant of a wide range of biotic and abiotic con- ditions.

The leopard squeaker, Synodontis leopardinus, brownspot largemouth tilapia, Serranochromis thumbergi, and banded tilapia, Tilapia sparrmanii, were dominant species in the drainage rivers and seasonal swamp (Table 2). The Zambezi labeo, La- beo lunatus, was a dominant species in the perennial swamp floodplain-connected lagoon at Magwexa- na. Both the thinface largemouth tilapia, Serrano- chromis angusticeps, and nembwe, S. robustus jal- lae, were common in all habitats except the drain- age rivers.

Discussion

The TWINSPAN analysis on community similari- ties showed that marked differences in fish stocks existed between some areas of the Okavango Delta. The species assemblages and corresponding sam- pling sites formed a fairly well-defined sequence from those clustered in perennially and seasonally flooded habitats. The main determinants of fish tax-

ocenes in the Okavango appear to be a combination of the extent of time the water is present, and the nature of its flow. The above factors determine other physical features such as aquatic macrophyte communities, substrate type and oxygen values, as shown for other wetland systems (Lowe-McCon- nell 1975, Welcomme 1979, Bruton &Jackson 1983). These parameters also affect food quantity and quality (Bruton & Jackson 1983).

In the Okavango there is a longitudinal zonation from a relatively rheophilic state to a palustrine state. The seasonal swamp and drainage rivers were dominated by a low species richness of many small fish (< 300 mm SL), such as Schilbe intermedius, Synodontis leopardinus and Tilapia sparrmanii. In contrast, the riverine floodplain and perennial swamp was dominated by a high species richness of many large fish (> 300 mm SL), such as H. vittatus, C. gariepinus and L. lunatus. Merron (1991) also showed that a greater mass of fish in the riverine floodplain and perennial swamp was collected in large mesh nets (e.g. 96, 100 and 118 mm stretch- mesh) whereas small mesh nets (e.g. 24, 40 and 50 mm stretch-mesh) caught the greatest mass of fish in the seasonal swamp and drainage rivers.

118

In the Okavango Delta, the annual cycle of flood- ing plays an important role in determining the na- ture of the fish taxocene. Kushlan (1976, 1980) found that the seasonal fluctuation in water level in the Everglades Swamp in the U.S.A. was the most critical environmental factor affecting the demog- raphy of the fish taxocene. Welcomme (1979) showed that the higher the magnitude of the annual flood, the longer its duration on the floodplain and greater the overall production of fishes. The timing of the flood does not appear to be as decisive a fac- tor as the magnitude. This conclusion is based on the observation that the flood arrives mainly in the riverine floodplain and perennial swamp after the warmer summer months (March/April) and in the seasonal swamp and drainage rivers before the warmer summer months (July/August), which is typically the optimum time for fish spawning and growth. The critical parameter of the annual Oka- vango flood cycle is the magnitude of the flood, par- ticularly in the seasonal swamp and drainage rivers. If the magnitude of the flood is high, the water re- tention time will be longer and may lead to a longer spawning period with a greater percentage of the population spawning (Merron 1991). The duration and water retention time of flood waters in the Oka- vango Delta is characteristically long when com- pared to intense but short periods of water reten- tion as a consequence of flash flooding, as in a river- ine floodplain such as the Phongolo (Merron et al. 1993).

In a broad ecological context, the Okavango Del- ta can be considered to be a sub-climax ecosystem that is maintained by the annual flood regime. Al- though there are wide oscillations in the timing, magnitude and duration of the annual flood, the Okavango does receive a flood each year and a con- stant pattern over time is apparent. Some areas within the Okavango (e.g. riverine floodplain and perennial swamp) are relatively more stable than others (e.g. seasonal swamp and drainage rivers).

In the hydrologically stable riverine floodplain and perennial swamp, the community is more spe- ciose and interspecific connections assume increas- ing importance, such as seasonal movements and feeding relationships, like the annual catfish pack- hunting migrations (Merron 1993). It is postulated

that the reason why complex inter-relationships are only likely to develop in predictable habitats is that these interactions have co-evolved over time and are finely in tune with the relatively minor hydro- logical changes within these habitats. The widely fluctuating seasonal swamp and drainage rivers habitats do not permit the time necessary for com- plex interactions to develop.

The main flow of biotic and abiotic stimuli in the Okavango Delta appears to be from the riverine floodplain and perennial swamp to the seasonal swamp and drainage rivers (i.e. from a predictably perturbed system to a unpredictably perturbed one). A simple analogy may be made with an light- ning bolt. The energy of the bolt widens and dis- sipates in force as it moves further away from its ori- gin.

Since the annual flood regime is the basic inde- pendent abiotic variable influencing the distribu- tion and abundance of fishes in the Okavango, its relative constancy is an important determinant of how the fish taxocenes may respond to perturba- tions. Relatively unstable habitats such as the sea- sonal swamp and drainage rivers, which are subject to wide natural fluctuations, are predicted to be able to sustain a greater degree of human exploita- tion and perturbation than more stable riverine floodplain and perennial swamp. The fish fauna in- habiting predictably-perturbed habitats may be more vulnerable to the introduction by man of ma- jor, unnatural perturbations. Large scale fluctu- ations, such as a markedly different hydrological re- gime due to water abstraction schemes in the head- waters of Angola and/or Namibia, and indiscrimi- nant gillnetting and recreational fishing, may interfere with the integrity of the riverine flood- plain and perennial swamp by causing species that are finely in tune with their environment to become threatened. Many of these species (e.g. H vittutus, L. lunatus) may become dominated by species with wide habitat tolerances, which could result in the system becoming populated by more eurytopic spe- cies. In other wetlands such as the Niger, Chad, and Chiri systems, selective fishing pressure has result- ed in the decline in the stocks of larger species which are replaced by smaller and less desirable species (Welcomme 1979).

119

Man-made hydrological changes to wetland eco- systems have generally reduced the amplitude of oscillation in the annual water level fluctuations. On the Phongolo floodplain in South Africa the timing, magnitude and duration of floods released from the Pongolapoort Dam is now largely asyn- chronous with the normal flooding cycle (Merron et al. 1993). This unnatural change has had an adverse impact on the downstream fish fauna by reducing the relative abundance of stenotypic species and fa- vouring more eurytopic forms. Welcomme (1979) provides a detailed account of man-induced manip- ulations to floodplains and subsequent loss of spe- cies richness and reduced catches. In the Okavango Delta, conservation and resource management strategies must differ depending on whether the pe- rennially or seasonally flooded habitats of the Oka- vango are being considered for exploitation. It should also be stressed that additional research is vital in order to refine our knowledge of the ebb and flow of the fish communities in the Okavango.

Acknowledgements

We are grateful to the Office of the President, Re- public of Botswana and to the Ministry of Agricul- ture, Fisheries Unit, for permission to perform re- search in the Okavango Delta. This research has been supported by financial grants from the South- ern African Nature Foundation, the Kalahari Con- servation Society, the Okavango Wildlife Society and Nampak.

References cited

Allanson, B.R., R.C. Hart, J.H. O’Keefe & R.D. Robarts. 1990. Inland waters of southern Africa: an ecological perspective Monographiae Biologicae 64. Kluwer Academic Publishers, Dordrecht. 458 pp.

Bruton, M.N. & P.B.N. Jackson. 1983. Fish and fisheries of wet- lands. J. Limnol. Sot. South. Afr. 9: 123-133.

Campbell, A.C. 1983. The guide to Botswana. Winchester Press, Saxonwold. 670 pp.

Gauch, H.G. 1982. Multivariate analysis in community ecology. Cambridge University Press, Cambridge. 280 pp.

Kushlan, J.A. 1976. Environmental stability and fish community diversity. Ecology 57: 821-825.

Kushlan, J.A. 1980. Population fluctuations of Everglades fishes. Copeia 1980: 870-874.

Lowe-McConnell, R.H. 1975. Fish communities in tropical fresh- waters. Longman, London. 302 pp.

McCarthy, T.S.. W.N. Ellery, K.H. Rogers, B. Cairncross & K. Ellery. 1986. The roles of sedimentation and plant growth in changing flow patterns in the Okavango Delta, Botswana. S. Afr. J. Sci. 82: 579-584.

Merron, G.S. 1991. The ecology and management of the fishes of the Okavango Delta, Botswana, with particular reference to the role of the seasonal flood. Ph.D. Thesis, Rhodes Universi- ty, Grahamstown. 166 pp.

Merron, G.S. 1993. Pack-hunting in two species of catfish, Clarias gariepinus and C. ngamensis, in the Okavango Delta, Botswa- na. J. Fish Biol. 43: 575-584.

Merron, G.S., M.N. Bruton & P la Hausse de Lalouviere. 1993. Implications of water release from the Pongolapoort Dam for the fish and fishery of the Phongolo floodplain, Zululand. South. Afr. J. Aquat. Sci. 19: 3449.

Skelton, PH. & PN. White. 1990. Two new species of Synodontis (Pisces: Siluroidei: Mochokidae) from southern Africa. Ich- thy. Expl. of Freshwaters 1: 277-287.

Welcomme, R.L. 1979. Fisheries ecology of floodplain rivers. Longman. London. 317 pp.