AN ECOLOGICAL STUDY ON THE VITTATUS

IN THE OLIFANTS AND LETABA RIVERS WITH SPECIAL

REFERENCE TO ARTIFICIAL REPRODUCTION

By

Christopher Lodewyk Gagiano

THESIS

submitted in fulfilment of the requirements for the degree

MASTER IN SCIENCE

in

ZOOLOGY

in the

FACULTY OF SCIENCE

at the

RAND AFRIKAANS UNIVERSITY

SUPERVISOR: Dr. G.J. STEYN CO-SUPERVISOR: Prof. H.H. Du PREEZ

MAY 1997

This thesis is dedicated to my Parents, Koos and Letta.

The Heavens gave to thee the mind, the soul oh so purely free but never shall we know where anger deepest flow

Where shall it all go the creatures that we know in time and space will man ever after see what Heavens gave to thee

With mind to rule the life of creatures all belove the breath off those depend on us to spare their end ACKNOWLEDGEMENTS

To God, who blessed me to deeply love and work with His creatures, and who looked safely after me, and gave me the strength to do this project.

My supervisor and mentor, Dr. Gert Steyn, who gave me this project to do. Without his help, motivation, advice, critique, and patience, this piece of work would not have seen the light.

My co-supervisor Prof. Hein du Preez, who also made this possible through his patience, advice and support.

My beloved wife Tanya, who's love and persistent encouragement certainly inspired me a lot to finish this piece of work.

My dearest Parents Koos and Letta. What more can I say other than to thank you for to what you have done for me. Your love, inspiration and assistance throughout my long years of study certainly taught me more than books ever could.

Dr. Andrew Deacon, friend and mentor who helped me in so many ways that can impossibly be mentioned. Thank you for teaching me so many things during our field trips.

To Gerhard Strydom, colleague and friend, who also helped me in so many ways. Your organising of the logistics certainly contributed to the success of this project.

My faithful field trip companion Charlie Nkuna, who accompanied me on many a hot day in the Lowveld sun. The evenings spent with you alone in the bush will always be with me.

Naledi Mare, who spent many hours in the computer room to help me with all sorts of data processing.

The National Parks Board who allowed me to do this project in the . I also wish to thank the rangers on who's sections I have worked, Johan Steyn, Leighton Hare and Arrie Schreiber for their assistance.

This project was supported by the Kruger National Park Rivers Research Programme and financed by the Sanlam Research Unit for Environmental Conservation. TABLE OF CONTENTS Pg

Chapter 1. Introduction. 1

Chapter 2. Literature Review. 5 2.1. Representatives of the Characidae in Southern Africa 5

2.2. Conservation Status of 7

2.3. Migration 10 2.4. Ecological position 13 2.5. Feeding and growth 13

2.6. Reproduction 18

2.6.1. Spawning 18 2.6.2. Sex ratio 20 2.6.3. Gamete production 21 2.7. Artificial breeding 22

Chapter 3. Aspects of the population dynamics of the tigerfish Hydrocynus vittatus from the Olifants and Letaba Rivers in the Kruger National Park. 27

3.1. Introduction 27 3.2. Materials and Methods 29 3.2.1. Choice and description of sampling localities 29

3.2.2. Sampling procedure and data collection 31

3.2.3. Data analysis 31

3.3. Results and discussion 33

3.3.1. Distribution and abundance 33

3.3.2. Age and growth 34 3.3.3. Mortality 38

3.3.4. Sex ratio 39

3.4. Conclusion 40 Chapter 4. Feeding habits of the tigerfish .(Hydrocynus vittatus) in the lower Olifants and Letaba Rivers, Kruger National Park. 41 4.1. Introduction 41 4.2. Materials and Methods 42 4.3. Results 42 4.4. Discussion 48 4.5. Conclusion 51

Chapter 5. Gonad development and fecundity of the tigerfish (Hydrocynus vittatus) in the lower Olifants and Letaba Rivers, Kruger National Park. 52 5.1. Introduction 52 5.2. Materials and Methods 53 5.3. Results 54 5.4. Discussion 58 5.5. Conclusion 61

Chapter 6. Notes on the induced reproduction and development of the tigerfish Hydrocynus vittatus (Characidae) embryos and larvae. 63 6.1. Introduction 63 6.2. Material and methods 66 6.2.1. Selection and handling of broodstock 66 6.2.2. Artificial stimulation 66 6.2.3. Artificial insemination and hatching 67 6.2.4. Development of embryos and larvae 68 6.2.5. Rearing of larvae and juveniles 68 6.3. Results and discussion 68 6.3.1. Selection and handling of broodstock 68 6.3.2. Induced spawning 69 6.3.3. Artificial insemination and incubation 70 6.3.4. Development of embryos and larvae 71 6.3.5. Free embryo and behaviour of larvae 72 6.3.6. Juveniles 73 6.4. Conclusion 74 Chapter 7. Tooth replacement of tigerfish Hydrocynus vinatus from the Kruger National Park 75 7.1. Introduction 75 7.2. Materials and Methods 76 7.2.1. Field observation 76 7.2.2. Laboratory observation 77 7.3. Results 77 7.3.1. Field observation 77 7.3.2. Laboratory observation 79 7.4. Discussion 79

Chapter 8. General discussion and conclusion. 81

Chapter 9. References cited. 86 ABSTRACT

Hydrocynus vittatus, commonly known as the tigerfish, plays an important role in riverine ecology. It is a top predator which roams the open waters of most larger river systems in southern Africa. Their presence in a freshwater ecosystem has a dramatic impact on the fish community structure. It is known that dams and weirs have a negative effect on the migration of the tigerfish. It is also evident that tigerfish do not occur in certain areas in some of the rivers where they have been present historically. The Olifants and Letaba Rivers in the Kruger National Park (KNP) are two of a few rivers within where tigerfish do occur. The KNP represents the edge of the most southern distribution of tigerfish in southern Africa. It was therefore expected that the tigerfish do not function optimal in the Olifants and Letaba Rivers as they are subjected to waters with high concentrations of silt and low flow which influences migration and successful breeding. Breeding migrations does however take place during the summer months after which the tigerfish returns to the Massingire Dam in Mozambique to avoid the colder winter temperatures in the rivers. Gonad development coincide with the yearly summer rainfall patterns. A deviation of the expected 1:1 male:female sex ratio to favour the males was experienced in both rivers, which may be the result of over population. Females were found to grow to a larger size than the males and were extremely fecund. Although H. vittatus is believed to be mainly piscivorous, other food items such as invertebrates, played an important role in the diet of small and large tigerfish in both the Olifants and Letaba Rivers. Invertebrates were mostly preyed upon which implies that optimal feeding conditions for the tigerfish does not prevail in these systems and that they have to adapt to satisfy their feeding requirements. Tigerfish is more abundant in the Olifants than in the Letaba River. The overall growth performance or phi prime (4)) values for H. vittatus in the was determined and compares well to the overall growth performance of tigerfish in the Okavango River and Lake Kariba. However the maximum length calculated for tigerfish in the Olifants River (Lco = 52.40 cm ) is smaller than the Lco values (56.06 cm) for the Okavango River. The mortality rate of tigerfish in the Olifants River exceeds those in the Letaba River which means that the life expectancy is longer in the Letaba as opposed to the Olifants River. Successful artificial spawning revealed some of the secrets of the reproduction strategy of this species. Tigerfish has semi pelagic eggs, are very small (0.65 mm), negatively buoyant and slightly adhesive for bentic and epibiotic incubation, and it is expected that tigerfish would spawn in open water, on a sandy substrate in the vicinity of aquatic vegetation. First hatching took place at 22 h 30 min after fertilization. Vertical movement of the larvae lasts for two days, which allows for downstream movement and dispersement of the larvae. It was found that tigerfish replace their teeth on a regular basis as they grow larger. Transition from conical to functional dentition takes place 45 days after hatching. Replacement of sets of teeth occurs during all phases of its lifespan. It is a quick proses of three to six days during which all teeth are replaced in both the upper and lower jaws. OPSOMMING

Die tiervis, Hydrocynus villains, speel 'n belangrike rol in rivier ekologie. Tiervis is 'n roofvis wat in die meeste groot riviersisteme van Suidelike Afrika voorkom en die teenwoordigheid daarvan in 'n ekosisteem 'n groot potensiele impak op die visgemeenskapsstruktuur van daardie sisteem het. Dit is bekend dat damme en keerwalle 'n negatiewe invloed op die migrasie van tiervis het en gevolglik is tiervis the meer teenwoordig in sekere gedeeltes van riviere waar hulle histories wel gevind is nie. Die Olifants- en Letabarivier in die Nasionale Krugerwildtuin (NKW) is twee van 'n paar riviere in Suid Afrika waar die tiervis wel voorkom. Die NKW verteenwoordig feitlik die mees suidelikste verspreiding van tiervis in suidelike Afrika en daar kan verwag word dat tiervis the ekologies optimaal funksioneer in hierdie riviere nie. Tiervis is onderhewig aan hoe slik konsentrasies gedurende vloede in die Olifantsriver en Iaagvloei toestande in die Letabarivier beinvloed die migrasie en broeisukses negatief. Broeimigrasies word gedurende die somermaande deur die tiervis onderneem, waarna hulk na die Massingiredam in Mosambiek gedurende die wintermaande terugbeweeg om die koue watertemperature van die riviere te vermy. Gonade ontwikkeling wind saam met die jaarlikse somerreenval plaas. Daar is 'n afwyking van die verwagte 1:1 mannetjie:wyfie geslagsverhouding wat die mannetjies bevoordeel in beide die Olifants- en Letabariviere. Hierdie afwyking kan moontlik te wyte wees aan oorbevolking in die betrokke sisteem. Wyfies behaal 'n groter liggaamsmassa as die mannetjies en is besonder vrugbaar in vergelyking met ander visspesies. Alhoewel H. villains hoofsaaklik piscivories is, speel invertebrate 'n groot rol in die dieet van beide groot en klein tiervisse in die Olifants- en Letabariviere. Die dieet van die tiervisse in die Olifants- en Letabariviere bestaan hoofsaaklik uit insekte en dui daarop dat optimal voedingstoestande the in hierdie riviere geld nie. Daar is gevind dat tiervis meer volop in die Olifantsrivier is in teenstelling met die Letabarivier. Die algehele groeiprestasie oftewel die phi prime (4)) van tiervis in die Olifantsrivier is bepaal en vergelyk goed met die phi prime waardes in die Okavangoriver en die Karibameer, alhoewel die maksimum lengte vir tiervis in die Olifantsrivier (Lao = 52.40 cm) kleiner is as die Loo waardes (56.06 cm) in die Okavangorivier. Die Die mortaliteitskoers van tiervis is hoer in die Olifantsrivier in vergelyking met die Letabarivier en bring gevolglik mee dat die verwagte lewensduur Langer is in die Letaba as in die Olifantsrivier. Die suksesvolle kunsmatige teel van die tiervis het sekere belangrike aspekte van die voortplantingstrategie bekend gemaak. Tierviseiers is baie klein (0.65 mm), semipelagies met dryfvermoe en is effens klewerig vir bentiese of epibiotiese inkubasie. Die vermoede bestaan dus dat tiervis op 'n sanderige bodem in die omgewing van akwatiese plantegroei broei. Eiers begin 22 uur 30 minute na bevrugting uitbroei waarna 'n aanhoudende vertikale migrasie vir twee dae gehandhaaf word. Tandwisseling geskied op 'n gereelde basis soos wat die tiervis groei. Koniese tande word 45 dae na bevrugting deur funksionele tande vervang. Tandwisseling geskied gedurende alle fases van sy lewensduur. Dit is 'n spoedige proses van drie tot ses dae en geskied gelyktydig in beide die onder- en bokaak. CHAPTER ONE

INTRODUCTION

The tigerfish, H. vittatus (Castelnau, 1861), is probably one of the most strikingly beautiful freshwater fishes that inhabits the warm waters of the east flowing rivers of Southern Africa. The colours of the tigerfish immediately draws the attention as the silvery flank and bluish sheen on the back are reflected in the sun. Parallel to the lateral line are a series of dark lines which nuns across the body. The tail fin is blood-red with a thin pitch-black edge, immediately catching the attention while the other fins are tinged with orange or red. The scales are large, 44-48 in the lateral line, and 15-16 scales around the caudal pendule. The dog like teeth are large, very sharp and set in hard bony upper and lower jaws (Jubb, 1967).

The tigerfish has earned itself a reputation as one of the most sought after freshwater sport fish in the world. This reputation is well earned as the tigerfish is a fierce fighter once hooked. With speed and agility it tests the competence of the angler. The most spectacular moment once the fish is hooked is the way it shatters the surface of the water when it jumps into the air in an attempt to free itself. The head of the fish is then shaken vigorously in an attempt to shake the hook from the mouth. This normally happens a few times before the fish is lured in. The Upper and Middle River with Lake Kariba is probably the areas where most anglers go to catch tigerfish. Numerous fishing Lodges are situated on the banks of this river. An international championship in Zimbabwe on Lake Kariba is held each year whereby hundreds of anglers compete to catch the largest tigerfish during the International Tigerfish Fishing Tournament (I.T.F.T.).

The Hydrocynus occurs widely in Africa but is absent from Lake Malawi and the rivers of Kenya and Somalia, the Kunene and Kafue Rivers (Bell-Cross, 1965;1966), and coastal rivers of Angola (Poll, 1967a), but occurs in the Okavango River and upper swamp (Skelton et al, 1985) and throughout the Zaire basin to and the West African rivers as well as in Lake Chad and in the (Daget & Durant, 1981). The Chobe River (locally called the Kwando) is the

1 largest of the west bank tributaries of the Zambezi, where Hydrocynus is present. The Lungwebungu is the next largest tributary, where Hydrocynus occurs well over 200 miles into Angola throughout its course (Bell-Cross, 1965;1966).

Hydrocynus villains is the only representative of this genus in southern Africa. There are, however, five species of Hydrocynus that occurs practically throughout the range of the family Characidae in Africa. Hydrocynus tanzaniae occurs in the eastward flowing rivers of Tanzania, the Wami Ruaha and Rufiji river systems. The lateral stripes of H. tanzaniae are distinct and differs from all other Hydrocynus species in the presence of elongated 3rd and 4th dorsal and anal fin rays (Brewster 1986). has a deeper body (mean =24.4% of the standard length) than the other species. They occur in Nilo-, Upper Guinea, Cameroun, Togo, Ghana and Ivory Coast (Brewster 1986). is the largest of the genus and can attain a mass of 50 kg! Hydrocynus goliath is restricted to the Oubangui River, and the central and upper Zaire (Congo) basin (Brewster, 1986).

In her review of the tigerfish genus of southern Africa (H. villains) and those in central Africa (H. forskahlii), Brewster (1986) concluded that they are the same and that the name H. forskahlii should prevail for both of them. A significant difference in their review of the genus was later found by Paugy & Guegan (1989) and confirmed that the species H. forskahlii to be the central African tigerfish, and H. villains the southern African tigerfish.

The Zambezi rises in the Mwinilunga District of Zambia and flow about 55 miles before entering Angola. Several fish collections in that area failed to produce Hydrocyrtus, as were confirmed by the local fisherman, who state that it is caught a little south in Angola. Water falls in the vicinity of Cazombo might restrict its upstream distribution. Tigerfish are plentiful where the Zambezi re- enters Zambia at Lingelengende, and occurs continuously southward to the Victoria Falls (Bell- Cross, 1965;1966). Hydrocynus vittatus occurs in extremely small channels in the main river of the Upper Zambezi River but apparently seldom penetrates tributaries of equivalent size. This apparent behavioral inhibition may be correlated with the danger of becoming cut of when waters recede after the rainy season. Hydrocynus vittatus is also abundant in Lake Kariba and southward below the dam in the Lower Zambezi. The most southern distribution of H. vittatus is the Pongola

2 System and in South Africa they are mainly found in the Kruger National Park. As of present, all the major rivers and some of the larger seasonal rivers in the Kruger National Park are inhabited by the tigerfish although the distribution of the tigerfish has changed. Earlier on, tigerfish were found as far as Mica in the Olifants river and in the Greater Letaba River at Tzaneen. Large numbers of tigerfish were also found in the Greater Letaba River at Letaba Ranch (Hugo, pers comm). Tigerfish were also found west of Hazyview in the Sabie river (Pienaar, pers comm).

Together with many other fish species, the tigerfish had been inhabiting the waters of the Lowveld of South-Africa for thousands of years. The identification of fish remains from a Late Iron Age site on the Letaba River certainly provides some information as to the fish community of the Letaba River more than a thousand years ago. Amongst some of the remains found were a number of identified skeletal parts of the tigerfish (Hydrocynus spp), which were included in the diet of the communities living there during those years (Plug & Skelton, 1991).

The distribution ofHydrocyma clearly indicates that a linkage between the Zambezi River System and River System of some sort must have existed thousands of years ago. The sparse distribution of H. vittatus in the south and the presence of other species such as lateral's in Kwazulu-Natal confirm that these river systems were linked to the Zambezi River Systems. This was indeed the case as described by Balon & Coche (1974). The Victoria Falls were believed to become the isolating barrier of the previously separate rivers i.e. the pre-Upper and pre-Middle- Lower Zambezi. The Victoria Falls were formed 99 km downstream from its present position, approximately 500 000 years ago. In the Pliocene the Upper Zambezi River and its probable tributaries, the Upper Kafue-Kalomo systems, flowed south together with the present Chobe- Okavango into the Ngami-Makarikari Lake and Limpopo valley. The Middle Zambezi was then a separate river with headwaters somewhere in the region of the present Matetsi River. A tectonic upward in the Caprivi area broke the connection of the Upper Zambezi with the Ngami- Makarikari drainage and the water of the former escaped eastward along a depression and spilled over the sheer drop at the old Middle Zambezi headwaters. At this point of origin of the Batoka Gorge, the existence of the united Upper-Middle Zambezi River and Victoria Falls, began (Balon & Coche, 1974). The monolithic surface of the basalt rocks along the entire length of the gorges has a system of fissures probably formed when the lava cooled. These were later enlarged by

3 upward tectonic movements. The fissures formed distinct fault lines. The gorges are eroded along such fault lines were a protective barrier of solid lava had been dissipated. When the river was spilling over an edge of solid basalt the erosion was slowed down until the water found the weakest point in the edge. When the river succeeded in eroding its way into another fault line, the process of upstream cutting was greatly speeded up (Balon & Coche, 1974).

Although the Upper Zambezi River and the Limpopo System was linked earlier on, the habitat of these two systems differ considerably today. The Upper Zambezi System has many flood plains with clear swift flowing waters and a sandy bottom and banks. The Limpopo System in contrast has irregular flow and carries large volumes of silt with virtually no flood plains. Hydrocynus villains seems however to survive in both these systems in spite of these differences. It would however seem that the Zambezi River is the typical habitat for H. villains.

Many studies on the ecology of the tigerfish had been done in the Zambezi River (Jackson, 1961a; 1961b; Winemiller, and Kelso-Winemiller 1994) and in Lake Kariba, (Badenhuizen, 1967; Matthes, 1968; Balon, 1971; Bowmaker, 1973; Kenmuir, 1973, 1975; Langermati, 1980, 1984; Marshall, 1985;). Few work has however been done on the Limpopo System where the tigerfish still do occur (Gaigher, 1970; 1973). Surveys over the last decade have shown that the most successful tigerfish population in South Africa occurs in the Olifants River. In spite of the totally different character of the Olifants River, as compared to the Zambezi System this species is extremely successful in the Olifants River. This leads to the key question which motivated this study: How does the ecology of the tigerfish compare between these systems. Furthermore was the artificial reproduction studied in order to learn more about the unknown reproduction strategy of this species.

4 CHAPTER TWO

LITERATURE REVIEW

2.1 Representatives of the Characidae in southern Africa

The species Hydrocynus vittatus is a member of the Characidae (Order ), and is one of the largest families of freshwater fishes found in both Africa and the Neotropics (Brewster, 1986).

Throughout the world, freshwater fish faunas are dominated by Ostariophysan fishes, and the riverine faunas of Africa is no exception to this. Africa is the only major continental area where the three groups of Ostariophysan fishes namely: the Cypriniformes, Characiformes and Siluriformes all occur, and it seems clear that the Ostariophysi must have originated in Gondwanaland prior to the separation of Africa and South America (Lowe-McConnel, 1988). The widespread distribution of certain families and sub-familial taxa is one of the outstanding features of the African fish fauna. The , Characidae, Bagridae, Schilbeidae, Amphiliidae, Clariidae, Mochokidae, Cyprinodontidae, Cichlidae and are found in most east and west Afro-tropical river systems (Greenwood, 1983).

The characins form a large and diverse group of fishes which are widespread in tropical waters of Africa (Skelton, 1988). The Characidae family consists of 18 genera and some W9 species (Lowe-McConnel, 1988). The Characidae family has big eyes, bony cheeks, sharp teeth, big silvery scales, and a short dorsal fin. They feed normally on a wide variety of food types and occurs in shoals (Skelton, 1993). There are approximately five genera which are represented by six species in southern Africa. Hydrocynus vittatus is the only representative of the genus Hydrocynus that occurs in this region. The genus Brycinus is represented by two species i.e. B. imberi and B. lateralis. Brycinus imberi (Peters, 1852) keeps to a wide variety of habitats including big rivers and marshes and pans of flood plains. They occur in the east coast rivers of the Phongolo northwards to the Rufigi in Tanzania. They are absent in the Upper Zambezi system

5 (Skelton, 1993). Brycinus lateralis (Boulenger, 1900) shoals in clear slow flowing or standing water with an abundance of vegetation. They occur in the Zambezi System, Okavango-, Kunene-, Busi River and in the St Lucia catchment in Natal (Skelton, 1993). Micralestes acutidens (Peters, 1852) shoals in clear lentic or lotic open waters. These small fishes occurs in the Kunene, Okavango, Zambezi and east coast Rivers southwards to the Phongolo (Skelton, 1993). The remaining two genera are absent in South Africa. Rhabdalestes maunensis (Fowler, 1935) keeps to flood plains with shallow waters and lots of vegetation. They are found in the Kunene, Okavango, Upper Zambezi and Kafue River Systems (Skelton, 1993). Hemigrammopetersius barnardi (Herre, 1936) are found amongst vegetation of rivers and lakes. They are confined to the Lower Zambezi, Pungwe and Busi Systems (Skelton, 1993).

The wide distribution of the Characidae family in Africa ensures that many different systems are inhabited. This includes the huge man made lakes in Africa such as Lake Kariba and Cabora Bassa. River systems had been changed to lentic habitats which have an enormous impact on the ichthyuofauna and their composition in relation to one another. The ecology of these species must subsequently differ largely from one genera to the other. Not only does differences occur between genera, but also between different species. This is evident in Brycinus imberi and Brycinus lateralis. Some riverine fishes disappear out of the new lake faunas in an regular sequence, generally because they cannot adapt to the changed breeding conditions. Lakes challenge species to exploit new sources of food and breeding conditions. Once the lake is stabilized, species new to the area appear to become abundant (Lowe-McConnel, 1988). The phenomenon of the sudden replacement of an original species by a new species in Lake Kariba is most probably explained by the breeding biology of the two characin species B. lateralis and B. imberi. Brycinus lateralis occurs mainly in the Upper Zambezi system and Kafue River, but has never been found in the Middle Zambezi. Brycinus imberi inhabited the Middle and Lower Zambezi. These two species has never been found together. After the creation of the lake, B. lateralis was found in Lake Kariba (Middle Zambezi) in 1963 for the first time. A few years later B. lateralis proved to be exclusive in Lake Kariba and has replaced B. imberi (Balon 1971). The creation of Lake Kariba proved to be an ideal habitat for B. lateralis which in turn affected the whole ecology of B. imberi negatively.

6 Brycinus imberi spawn during the raising of the water level (rainy season) on freshly inundated grassy shores. Contrary to B. imberi, B. lateralis spawn on submerged flora, roots of Salvinia etc. irrespective of the variation of the water level. B. lateralis therefore suddenly found not only free space but exceptionally good spawning conditions. Individual specimens driven over the edge of the Victoria Falls therefore created in a short time a prosperous population (Salon, 1971). Not only did the creation of Lake Kariba on itself changed the ecology of the above mentioned species, but also that of the tigerfish. Shortly after the introduction of the freshwater sardine Limnothrissa miodon, and B. lateralis were the major food item of the tigerfish. This situation soon changed where L. miodon became the major food item of the tigerfish. There has thus been a major shift from B. lateralis and the cichlids to L.miodon in the diet of H. vittatus (Kenmuir 1973).

It is clear that the ecology of any given species is not a set and final matter, but that changing environments necessitate changes in their behaviour to survive and sometimes to thrive.

2.2 Conservation status of Hydrocynus vittatus

Tigerfish H. villains in Southern Africa is not listed in the Red Data Book of Fishes (Skelton, 1987), and is thus currently of no concern. It is however important to note that the status may differ from one river system to another due to various factors including water quantity and quality, fishing pressure and the loss off habitat.

Apart from the limited distribution of H. vittatus in South Africa the tigerfish is rare in South Africa and not readily found in the rivers of the Kruger National Park. Due to aggregation at natural migration obstructions and isolation in clear deep pools, tigerfish are more frequently found in the Olifants and Letaba Rivers. The tigerfish is thus relatively protected in South Africa as they are isolated in the Kruger National Park, but extremely vulnerable to anthropogenic activities which can influence the water quantity and quality in their catchment. Gaigher (1973) and Steyn (1987) mentioned that the reproductive potential of H. vittatus is negatively influenced

7 due to migration problems and limited spawning localities due to the erection of weirs and the shortage of water. Similar problems were experienced by many other African freshwater fishes (Bell-Cross, 1965a).

The fact that the Letaba River had been extensively used for irrigation purposes, contributed to its present poor status, providing little or no suitable habitat for the tigerfish to survive, both in the higher reaches of the river and within the boundaries of the Kruger National Park.

The water quality of the Olifants River is also of concern. Although the water of the Olifants River entering the Park is highly mineralized, bioaccumulation of selected metals seems to be low. Concentrations of selected metals in certain tissues and organs of the tigerfish were studied by Du Preez and Steyn (1992). The selected metals were detected in all the tissue examined, but in variable concentrations indicating differential bioconcentration of the metals. The bioconcentration factors were generally low suggesting low bioavailability of the metals.

Extremely high silt deposits, especially during floods, in the Olifants River below the Phalaborwa Barrage certainly has a negative effect on the numbers of tigerfish close to the western boundary of the KNP, as they prefer clear clean water (author, personal observation). Silt are deposited along the river as it moves downstream, resulting in somewhat cleaner water further down. Van Loggerenberg (1981) also mentioned that the numbers of tigerfish higher up in the Olifants River was so few that it could not be netted with seine nets. Surveys conducted with seine nets and gill nets during 1990 until 1992 also produced very few tigerfish. The previous drought of 1991/92 certainly also must have had its negative effects on the river with regards to habitat loss as the water flow were drastically reduced. Large surviving pools also reduced significantly in size as the days got hotter and the water evaporated in the blazing sun.

There is an abundance of tigerfish in the Zambezi System, but is currently under heavy fishing pressure. As early as 1965-6 Bell-Cross reported that man is undoubtedly the heaviest predator on adult tigerfish, because 4 and 5 inch gill nets are commonly used by fisherman in Lake Kariba. However he did not consider the effects of angling significant. Langerman (1984) reported that the impact of the pelagic fishery on tigerfish stocks in Lake Kariba is potentially the most damaging of all three fisheries and that the gillnet fishery is likely to have the largest biological impact on tigerfish. The use of smaller gillnets to achieve greater catches in the short term, certainly contributes to negative long term effects (especially recruitment) of the population (Langerman, 1984). Kenmuir (1972) also mentioned that if gill netting were to have continued for a few years at Lakeside where they have worked, it might have eradicated the tigerfish population. The most serious problem relating to tigerfish in Lake Kariba has been and remains the illegal netting in river systems. The problem is not only that of cropping the adult stock but also of destroying a large portion of the egg pool (Kenmuir, 1972). Although fears that the annual I.T.F.T. (International Tigerfish Fishing Tournament) is already affecting the tigerfish resource, since catches may consist of up to 1000 large females all full of eggs, Langerman (1984) concluded that the recreational fishery is unlikely to have any adverse biological impact on the tigerfish population. Tigerfishing is a very popular ecotourism attraction along the Zambezi River and Lake Kariba and tigerfish is heavily exploited by anglers visiting the numerous Fishing Lodges and then kept as trophies. Angling safaris on the Zambezi is common and the larger trophy fishes are removed from the system. These larger fishes are most probably the females as they grow to a much larger size than the males. This simply means that the prime breeding material for the genus is removed which will undoubtedly have negative results for the species in the long term. This especially give reason for concern with respect to the future of the Upper Zambezi population as the tigerfish in the Upper Zambezi represent an unique colour variation for the species H. vittatus (Chapter 6).

The tigerfish in the Okavango River is currently under ecological pressure, most likely due to subsistence fishery as mentioned by Van Zyl (1992), although he states that the survival of H. vittatus is at this moment not endangered. Hay (1995) reported that only a few large individual specimens of H. vittatus were collected during a study in the Okavango River in an area where no fishing is allowed and concluded that large individuals were basically non-existent in the rest of the river. Nikolskii (1969) noted that a decline in a predator population may be directly correlated to fish exploitation.

9 Electrophoretic analysis of tigerfish liver, white muscle, heart and testis samples revealed genetic variation at 20% (Upper Zambezi River, Namibia) and 63% (Olifants River, South Africa) of the protein loci studied (Kotze et al, 1996). Average heterozygosity values ranged from 1.9% in the Upper Zambezi River to 6.4% in the Olifants River, with a genetic distance value of 0.005 between these populations. These results clearly shows the relative genetic vulnerability of the tigerfish in the Upper Zambezi River in comparison to the tigerfish in the Olifants River, should there be overexploitation, any sudden environmental changes, virus infections or diseases. It is however interesting to note that the genetic variation of the tigerfish in the ()Hants River is much higher than the Upper Zambezi River although the Olifants River between the Phalaborwa Barrage and the Massingire Dam in Mozambique is a considerable smaller biotope as compared to the Upper Zambezi System.

The low genetic variation in the Upper Zambezi River could be due to genetic isolation by the Victoria Falls which acts as a migration obstruction prohibiting exchange of genetic material. The population in the Olifants River has a far greater genetic variation altough the Massingire Dam also acts as a migrational obstruction. This obstruction however is very recent (28 years) as compared to the 500 000 years of the Upper Zambezi River.

The limited genetic variation clearly shows that inbreeding is already taking place in the Upper Zambezi River. Of great concern is the disappearance of genetic material from trophy fishes in a population where inbreeding is already happening. Conservation strategies is thus important to prevent the loss of genetic variation and subsequently a unique species.

2.3 Migration

The tigerfish is potamodromous which means that they do migrate to areas in freshwater to spawn otherwise than normally during the year (Bowmaker, 1973). The "triggering" mechanism for upstream migration in fish, as generally believed, can not be attributed to any single factor but results from the single combination of many physical and chemical factors associated with flooding (Bowmaker, 1973). A single substance "Petrichor" may provide the final stimulus for

10 spawning Lake (1967). "Petrichor" is an oil consisting of a complex mixture of organic compounds containing acidic, basic and neutral fractions which is released from the dry ground when it is dampened.

The two factors which is the main cause for the annual spawning migrations are believed to be adequacy of cover and sufficiency of food. The most important one seemed to be safe refuge from adult predation (Jackson, 1961b). Feeding activity of tigerfish however decreased during the spawning migrations in the Sanyati Gorge while it was not the case in Lake Kariba (Kenmuir 1973). Tigerfish were found not to feed at all during the annual spawning migrations, after which feeding activities increased once spawning had taken place (Matthes, 1968).

Spawning migrations of the tigerfish do occur each year during the height of the rainy season when the river is in flood (Soulsby, 1960; Jackson, 1961a; 1961b; Matthes, 1968; Kenmuir, 1973). Migrations does not always start during the height of the rainy season as a large congregation of Hydrocynus were witnessed at Parrot Bent some 1.5 km from the entry point of the Mwenda River, Lake Kariba, prior to the first rains in October and November (Bowmaker, 1973). It could have been that the tigerfish anticipated the floods by the change in water temperature. Jackson (1961a) however stated that temperature which normally plays a role in the onset of the reproductive activity, may almost be discounted as the variation between the summer and winter water temperature in the Middle Zambezi is almost the same or sometimes equal.

It was observed that Hydrocyon villains' migrate upstream in the out of during the height of the rainy season and that migrational runs of tigerfish occurs in waves as if shoals had been formed and were moving together (Soulsby, 1960). Large migrational runs were observed to take place on overcast days when the river was flowing sufficiently (Bowmaker, 1973) and shoals of migrating tigerfish had been witnessed by Kenmuir (1973) in the Sanyati Gorge. During these migrational runs, tigerfish up to 4 kg or more were leaping up to heights of 1.8 metres or more. Contrary to the above mentioned upstream migrations of the tigerfish Gaigher (1967, 1970) has found that the tigerfish migrate downstream to spawn in Lake Chualo, Mozambique in the Incomati River system. Gaigher (1967; 1970) witnessed that an upstream

.1 Previously synonymous with Hydrocynus vittatus

11 migration of the tigerfish take place after spawning and concluded that migration of the tigerfish is thought to be a feeding migration and not a spawning migration as were found to be the situation in other systems as mentioned previously.

No migration of H. villains from the Middle Zambezi River to the Upper Zambezi River can take place because of the obstruction caused by the Victoria Falls. This means that 11. villains in the Upper Zambezi River is an isolated population due a natural obstruction.

Natural migration of tigerfish is restricted in South Africa. This has been the reason for the disappearance of the tigerfish in certain stretches of the Lowveld rivers. Obstruction of migration routes have been caused by a shortage of water and the erection of dams and weirs, restricting the movement of tigerfish and other fish species (Gaigher, 1970, 1973; Kruger, 1972; Clay, 1976; Pienaar, 1978). Tigerfish spent only a part of the year in Swaziland and that the building of dams and weirs have greatly changed the distribution of a lot of fish species in those catchments (Clay, 1976). It had been found that tigerfish do make use of fish ladders to migrate upstream in the River in the Kruger National Park (Olivier, 1994). Three tigerfish were collected in the Kanniedood fish ladder during surveys when the dam wall was overflowing. An interesting statement was made that there is indeed no fundamental distinction between physical barriers and ecological factors which limits dispersion of fish (Crass, 1962). Such ecological factors include sudden temperature variations of the water. Tigerfish is known to inhabit the warmer waters of Africa, and a sudden dramatic drop in temperature could kill them. Dead tigerfish had subsequently been observed in the Crocodile River after a hailstorm. It seemed that the sudden decline of the water temperature caused the death of the tigerfish. The Zambezi River is much larger in volume than the rivers of the Limpopo System. It is also situated more towards the equator which means that the water temperature is slightly warmer and more stable, with fewer dramatic temperature variations taking place.

12 2.4 Ecological position

Hydrocynus vittatus predominantly frequents the 2m and 5m depth zonations, with fewer being caught at a depth of 10m in Lake Kariba (Mitchell, 1978). were found to be restricted to the surface waters in the deeper parts of Lake Kainji, and that no specimen of H. forskahlii2 was captured below a depth of 10m (Lewis, 1974). Winemiller and Kelso-Winemiller (1994) reported that Hydrocynus forskahlii occupies almost exclusively the open waters of the main river channel in the Zambezi River. The fact that they are open water loving species was also observed by the author in the Olifants, Letaba, Sabie and Crocodile Rivers within the Kruger National Park. Tigerfish favour upper reaches of rapids for they would almost swim into the rapid before darting back into the more quiet waters.

2.5 Feeding and growth

Hydrocynus vittatus is an open water loving species and preference would naturally be given to prey species which is most abundant in the river or lake system occupying the same pelagic habitat (Jackson, 1961b; Kenmuir, 1973; Lewis, 1974; Lauzanne, 1975; Winemiller and Kelso- Winemiller, 1994). The availability of cover for prey species plays an important role for them not to be preyed upon. The advantage of availability of cover is two-fold; there is an abundance of , crustaceans and other food, and there is complete protection from large predators, especially H vittatus Jackson (1961b). As the level of Lake Kariba dropped with a subsequent decrease in cover the degree of fullness of tigerfish stomachs increased (Kenmuir, 1973). The importance of cover for prey species can clearly be seen in the findings of Kenmuir (1973) who found tigerfish in very poor condition in a pool in the Gache River. Only 14 out of a total of 139 tigerfish had stomach contents of some sort. A few boulders and a shallow sloping beach provided sufficient cover for the prey species despite the lack of any vegetal cover.

. 2 Previously synonymous with Hydrocynus vittatus

13 Bell-Cross (1965;1966) describe Hydrocynus as: a - partly insectivorous/partly piscivorous for young of the year while they do take adult Barbus and other "small" fish; b - piscivorous on fish which are less than 10 cm when adult, and c - piscivorous on fish which grows greater . than 10 cm when adult. Predation would be on adults of small species and young of larger species (Lauzanne, 1975), which is related to the size of the predator and the season of the year. The adult of small species are mainly being preyed upon by tigerfish with a length between 100 and 200 mm throughout the year while the young of large species are mainly being preyed upon during the flood season. In Lake Kariba the impact of predation by tigerfish is mainly restricted to the small and young fish of the families Characidae and Cichlidae (Matthes, 1968). The first year class is restricted to very shallow water in the main river and is the only year class subject to predation by all other classes ofHyd,rocynus (Bell-Cross, 1965;1966). Tigerfish, and other fish species has a critical period to which they are more susceptible to predation once they have exceeded that critical length.

When tigerfish reaches a length of approximately 50 mm, they prefer fish in their diet as was found in Lake Kariba and Lake Chad (Jackson, 1961b; Holden, 1970; Lauzanne, 1975). Tigerfish from the Incomati River and larger than 55 mm (FL) preferred fish in their diet (Gaigher, 1970). Insects, predominates as food item during seasons with low water levels while the dominant food item during the high water levels were found to be fish with insects virtually absent. Prey fishes within the suitable size-range for the tigerfish are not present in sufficient quantity during seasons of low flow prior to the breeding season (Jackson, 1961a). Insects as food item played a major role (86.8%, 63.6% and 60%) in the diet of tigerfish between 180-300 nun (FL) in Lake Kariba (Kenmuir, 1973). In Lake Albert, a small prawn, Cardinia nilotica, was present as a food item in 96% of tigerfish specimens caught at offshore stations (Holden, 1970).

Juvenile tigerfish has three major components in their diet namely zooplankton, insects and fish. Zooplankton were preyed upon after five days when the larvae were 5 mm in length. The change- over in diet from zooplankton to insects is approximately at 35 mm (FL) (Kenmuir, 1975). In the Chari River tigerfish up to 30 mm (SL), were almost exclusively zooplanktophytes, and from 30-45 mm (SL) they followed a mixed diet in which insects dominate while tigerfish larger than 50 mm (SL) were almost exclusively ichthyophagous (Lauzanne, 1975). Kenmuir (1973)

14 mentioned that tigerfish appears to become piscivorous at a length of 28 mm (FL). This is in line with the findings of Gaigher (1970) that tigerfish less than 25 mm (FL) fed almost exclusively on Entomostraca.

The change-over to a predominantly fish diet, occurs at a length of more or less 80 mm (TL) while the smallest specimen recorded feeding on fish was 52 mm (TL) (Jackson, 1961a). The change-over in diet to an almost ichthyophagous one occurs when the young tigerfish are between 40 and 50 mm (TL) with no intermediate stage of eating (Matthes, 1968). This was supported by Kenmuir (1975) who later found that fish-feeding starts at 40 mm (FL). The diet of tigerfish larger than 100 mm (SL) and longer consisted mainly of insects, shrimps and fish during the fall in the south-eastern Archipelago of Lake Chad. The importance of insects were evident in the 100-200 mm classes after which it became insignificant. Shrimps and fish were the main diet of fish with lengths between 200-300 mm. while fish only became the main diet for Hydrocyon3 larger than 300 mm (Lauzanne, 1975).

The diet of the tigerfish in the tributaries and the south-eastern Archipelago is totally different to that of the fish in the open waters of Lake Chad. Hydrocyon are exclusively ichthyophagous throughout the year in the open waters of the lake. Insects and shrimps that constituted an important part of the diet in the Archipelago were absent from the stomach contents, presumably because of their extreme rarity in this biotope. The same diet for the fish in the river system was observed as in the open waters (Lauzanne, 1975).

The impact of predation by tigerfish were restricted to fish generally not more than 40 percent of the predator's length, although this was not always the case (Jackson, 1961b; Matthes, 1968; Kenmuir, 1973). This ratio would not exceed 50% of the predator's length as was observed by Jackson (1961b) and Gaigher (1967). On one occasion a prey species of 64% of the length of a tigerfish was found by Matthes (1968). Winemiller and Kelso-Winemiller (1994) found the predator prey ratio between 7% and 62% with a mean of 27%. No significant relationship between the body lengths and prey could be derived for the tigerfish in the Upper Ogon River , although maximum prey size increased with body length (Adebisi, 1981).

. 3 Hydrocyon Genus previously used synonymous to Hydrocynus

15 The effect of predation on prey fish by tigerfish is not to be taken lightly, and it seems probable that many of the spawning migrations of many African fishes are directly influenced by the impact of predation by this fish Jackson (1961b). The greatest predation pressure is caused by tigerfish with a length of 550 mm or 3 kg (Jackson, 1961b).

Although several species of Hydrocynus occurs in Africa, there seems to be a big difference in the normal feeding behaviour between some of the species. Hydrocynus brews has a specialind mode of feeding whereby they are biting pieces from, or mutilating large fish. In most instances the posterior or ventral part of the prey species had been bitten away, though occasionally the body of the prey, although badly lacerated, was still intact. Once maimed the prey fish usually floated to the surface to have pieces repeatedly bitten away by the predator (Lewis, 1974). The method described above is however not the only one by which H. brevis feeds. Typically of all the other Hydrocynus species they also catch smaller prey and swallow them head first, whole in, as described by many authors before.

A total of nine occasions had been witnessed during which H. vittatus attacked struggling fish on the end of a line (Bell-Cross, 1965;1966). Jackson (1961b) reported of a crocodile being shot in the head and emitting great quantities of blood, after which a number of large Hydrocyon were instantly on the spot and biting at the wound. This specific incident proves Lauzanne (1975) partially wrong when she stated that Hydrocyon forslcalii of all sizes and in all biotopes attack only living prey which they swallow whole. In this case, the tigerfish reacted purely to a blood stimulus in the water from the bullet wound in the head of the crocodile. Although they were biting at the wound they must have known that the blood was coming from something much bigger than them. Their reaction simply emphasises their whole nature as being that of ferocious predators roaming the waters in search for any given prey.

It was also witnessed by (Jackson, 1961b) that H. vittatus had been biting Tilapia macrochir indescriminantly as they passed by. On another occasion a 210 mm Labeo altivelis was removed from a tigerfish of 455 mm which was severed in two. The head and tail sections were lying side by side in the stomach of the tigerfish. The Labeo had a neat cut at right angles to the body. This however is not common feeding behaviour of H. vittatus. Tigerfish usually attacks from behind

16 and biting at their tails. The damaged fish are then finally eaten (Matches, 1968). It was observed by the author that prey species were bitten in the middle of the body before turned and swallowed, head first.

The effect of the area of free movement of fish in a body of water does play a role in their growth rate (Chen and Prowse, 1964). In other words, the larger the volume of water, the larger the fish will become in that body of water, in comparison to fish in a small volume of water. Growth of tigerfish is rapid when food supply is good and shelter from predators is available. It is also necessary to grow fast to exceed a critical length in order to avoid predation (Jackson, 1961a). It was estimated that the growth of tigerfish was between 140-180 nun in the first year and up to 250 mm in the second year in riverine conditions (Jackson, 1961a). A growth rate of 160-200 mm per year for the first year and 310-320 mm for the second year in Lake Kariba (Badenhuizen, 1967). A similar growth rate with the first year between 170-210 mm, the second year at 260-320 mm and the third year at 380 mm for the tigerfish in Lake Kariba was found by Kenmuir (1973). A much slower growth rate was documented for tigerfish in the Incomati System (Gaigher, 1967). A growth rate of 90 mm in the first year, about 170 mm for the second year and 250 mm for the third year was found by the latter author.

The growth rate found by Salon (1971) is very similar to that of all the above mentioned authors with the exception of Gaigher (1967). The growth rate was found to be between 140-180 mm for the first year and 250 mm for the second year. The second year class of these fish had the most active gonads, of which 93% were males. These males had a faster growth rate than the females, especially in the older age groups (Salon, 1971). Winemiller and Kelso-Winemiller (1994) found that the growth rate of the tigerfish in the Zambezi River floodplain was 170 mm and 230 mm for the first and second year classes respectively. Drought periods play an important role in the growth rate of tigerfish (Dansoko el al, 1976). It was pointed out that the young of the year suffered most from the drought. Balon (1971) concluded that tigerfish with a faster growth rate, has a better chance of survival. This observation was supported by Kenmuir (1973) who found that the larger sizes had a more rapid growth rate, probably due to food availability and space. The very high mortality rates (84%) of tigerfish in their first season of growth emphasises the faster growth rate and critical length of the fish to avoid predation (Baton, 1971).

17 There seems to be a difference in the relative maximum size of H. villa:us throughout Africa (Gaigher 1970, Kenmuir 1973, Van Zyl 1992). Tigerfish with a mass of 7.4 kg and 8.8 kg were sampled in the Upper Zambezi River System (Bell-Cross, 1965;1966) which is much smaller than the 17.7 kg tigerfish which was caught in Lake Kariba (Jubb, 1967). Tigerfish with a mass of 5.4 - 5.9 kg were caught during earlier years in the Sabie, Crocodile and Letaba rivers (Pienaar, 1978). Tigerfish oft 7 kg were also regularly caught during the earlier years in the Olifants River (South Africa) below the Olifants Wilderness Trails Camp in the Kruger National Park (Bryden, Kloppers, and Van Niekerk, pers comm). It would therefore appear that the tigerfish grow to larger sizes in the Zambezi River System than in the most river systems in South Africa, and the Zambezi River therefore seems to be the ideal habitat for H. vittalus.

2.6 Reproduction

2.6.1 Spawning Although the spawning period for all the fishes of the order Ostariophysi (including H. vitlatus) is of extremely short duration (Jackson, 1961a), the duration of breeding is closely linked with the duration of river flow and can be as long as five months (Kenmuir, 1973). No reference to time was given by Bowmaker (1973) when he stated that Hydrocynus breeds "more quickly" than Mormyrus, once the river had been entered. The actual spawning period of tigerfish coincides with the rainy season when the rivers are in flood (Jackson, 1961a;1961b; Gaigher, 1970; Bowmaker, 1973; Kenmuir, 1973; Lauzanne, 1975; Badenhuizen, 1976; Van Zyl et al, 1992 and Hay, 1995). It was however found that the tigerfish spawn prior to the floods in the Okavango Swamps (Merron and Bruton, 1989). Tigerfish experienced two major spawning cycles in the Okavango River (Van Zyl, 1992). A possible second breeding cycle was mentioned by Kenmuir (1973) in the Sanyati Gorge, although he could not confirm that due to a lack of data. On one occasion Jackson (1976) recorded a two month old juvenile tigerfish in early November in the Middle Zambezi River, indicating an abnormal spawning time. This is an indication that the gonads of the females had matured during abnormal seasons or conditions.

18 Surveys conducted by Van Loggerenberg (1982) indicated that the tigerfish have spawned at the confluence of the Olifants and Letaba Rivers in the Kruger National Park during January 1982. The Olifants and Letaba Rivers were in flood just after new year although the Kruger National Park itself has received little rains during that period. Water were released by the Phalaborwa Barrage in the Olifants River while the Fanie Botha Dam was overflowing in the Letaba River creating flood conditions in these two rivers. The manipulation of the water in the rivers have stimulated the tigerfish to spawn during that specific period. Surveys conducted revealed that only spent females were found (Van Loggerenberg, 1982). Steyn (1987) reported tigerfish spawning in a deep isolated pool in the Letaba River during February 1986. Females were holding back when the Letaba River was not flowing. When the river started flowing and the pool received some water it was enough to stimulate the fish to spawn. Spawning will however not always take place as the normal cycles is not always maintained. This could lead to the atrophy of eggs.

Bowmaker (1973) observed atrophy of Hydrocynus in the Mwenda area and speculates that this may result from too long a waiting period prior to the sufficient flooding of the river, but could also result from disease or some breakdown in the ethological breeding sequence. The presence of atrophied gonads in an area may be a reflection of the nature of river flow in that area, and atrophied gonads may occur more frequently in the vicinity of the smaller rivers (Kenmuir, 1973). Tigerfish with atrophied eggs as a result of no river flow was witnessed by Langerman (1984) in Lake Kariba.

The question as to the exact habitat the tigerfish prefers to spawn, has not yet been answered. Different habitats as possible ideal spawning, or definite spawning grounds have been given by various researchers. Hydrocynus brevis breeds in the flooded plains, while H. fors/obi is confined to smaller river beds (Daget, 1954). This view was also shared by Badenhuizen (1967) who found that tigerfish migrates to breeding grounds in shallows upstream in rivers. Flood plains are described as ideal spawning conditions for the tigerfish which is created with the influx of rising water in Lake Chualo in Mozambique (Gaigher, 1967; 1970). Tigerfish were found to reproduce both in the south-eastern Archipelago and river systems of Lake Chad (Blache, 1964), while the exact locality of spawning by tigerfish going up the Sanyati River during river flow is not known

19 (Kenmuir, 1973). With the rivers in flood, the need for vegetal cover always seems to be important. This would ensure a safe refuge for the larvae and abundance of food amongst the submerged vegetation while the larvae are transported downstream.

The behaviour of the tigerfish embryos and larvae are adapted specifically for downstream dispersion along the river course by the current (Bowmaker, 1969b; 1973; Kenmuir, 1973; Chapter 6). This rather strongly suggest that the tigerfish would select a spawning site upstream from the nursery area.

2.6.2 Sex ratio The expected sex ratio of a fish population is a 1:1 male:female ratio (Nikolskii, 1969). Certain factors however normally causes a deviation of this ratio. A shorter life span and higher mortality rate of males shifts the ratio towards the females (Kenmuir, 1973; 1984). Disproportions of the number of mature males and females, shortened breeding periods and unsynchronised maturation of the male and female tigerfish are associated with low recruitment in the waters of the Kruger National Park (Steyn, 1987).

The expected 1:1 sex ratio was found in the Okavango River except for the 1992 winter survey which favoured the females (Hay, 1995). This was a shift towards the 1:1 ratio since Van Zyl (1992) found the females to be favoured in the Okavango River. The deviation of the 1:1 sex ratio during different seasons was also evident in Lake Kariba (Kenmuir, 1973). The sex ratio favoured the females (1.35:1) during the non-breeding months, whereas the males were favoured (4:1) during the peak breeding season. The sex ratio favoured the males at 1.8:1. from Mwenda Bay in Lake Kariba (Langerman, 1984). This male:female ratio would however ensure fertilization of the eggs during the spawning periods.

20 2.6.3 Gamete production Sperm counts of fish is typically high which is needed to fertilize normally hundreds or thousands of eggs. Fish which spawn in running water tend to have high sperm counts with short motility duration when compared with still-water spawners (Ginsburg, 1972). The semen of tigerfish is white and has a water-like viscosity. The white colour of tigerfish semen reflects a high sperm count, while a low sperm count is reflected by a transparent appearance (Kruger, Smit, Van Vuren & Ferreira, 1984). The sperm count of the tigerfish (11.90 ± 4.81 x 10 6/nun3) is much higher than the sperm count (0.04 x 10 6/mm3 of O. mossambicus (Kruger et al, 1984; Steyn, 1993). The high sperm count of tigerfish may therefore either reflect a stream spawning activity or a high fecundity (Steyr!, 1993).

Fishes spawning in cold water tend to have a shorter duration of sperm motility than warm water species (Steyn, 1993). The duration of motility of the tigerfish spermatozoa (86 sec) is relatively short compared with that of Clarias gariepinus (120 sec) (Steyn and Van Vuren, 1987). Steyn (1993) concluded that the relatively short motility duration and high sperm count of the warm water tigerfish suggests that this species is a stream spawner.

The males in Lake Bengeullu, Zambia have a length of 390 nun (Griffith, 1975), and 300 mm (FL) in Lake Kariba when they become sexually mature (Kenmuir, 1972). According to Langerman (1984), forty percent of the males became sexually active at a length of 240 mm (FL). There seems to be a big difference in the lengths of males reaching sexual maturity as obtained by Kenmuir (1972) and Langerman (1984) respectively in Lake Kariba. The findings of Langerman (1984) is similar to that of Gaigher (1975) for the male tigerfish (200 mm FL) in the Incomati River system. Male tigerfish reaches sexual maturity at a minimum of 170 mm FL and 180 mm FL respectively in the Okavango River (Van Zyl, 1992; Hay, 1995). Although the average length when 50% of the males reaches sexual maturity was not given by Van Zyl (1992), one can assume that the length would be similar to the findings of Hay (1995), who found the length to be 260 nun (FL).

21 Females seems to reach sexual maturity later in life, compared to the males (Kenmuir, 1973; Langerman, 1984). Female tigerfish reaches lengths of up to 700 mm (FL) and more, while males hardly grows to lengths beyond 500 mm (FL) (Kenmuir, 1973). Females reaches sexual maturity at a length of 360 mm (FL) in the Incomati River system (Gaigher, 1970), but in Lake Kariba, 50 percent reaches sexual maturity at 260 mm (Langerman, 1984). The minimum length at which females reaches sexual maturity in the Okavango River is 420 mm (FL) (Van Zyl, 1992) and 275 mm (FL) (Hay, 1995) respectively. The length given by Van Zyl (1992) can certainly not be used as the minimum average length when females reaches sexual maturity in the Okavango River, as that was the only active female sampled. The findings of Hay (1995) for the minimum length at which females becomes sexually active in the Okavango River is more acceptable. The fact that female tigerfish grows larger than males suggests that egg counts would be high in order to ensure fertilization and the survival. Female tigerfish grows larger than the males (Jackson, 1961a; Kenmuir, 1973; Van Zyl, 1992).

Tigerfish were found to be very fecund, with large numbers of eggs (Pott, 1969; Kenmuir, 1973; Hay, 1995; Steyn, pers comm). Although egg counts varied in length groups, egg counts increased with length. A large specimen can produce up to a million eggs (Langerman, 1984). This would ensure the survival rate of the species as mortalities of the eggs and larvae are expected to be extremely high.

2.7 Artificial breeding

More than two decades ago the need to restock some of the Lowveld waters in South Africa was mentioned by Gaigher (1968), since the distribution of tigerfish was negatively influenced by the erection of dams and weirs. There are two possible methods of restocking the Lowveld waters with tigerfish. The first is to capture and translocate the tigerfish; and the second one is to propagate the species through induced reproduction. The first option would not be possible since there is no locality in the Transvaal were enough large and small tigerfish can be collected. The Incomati River System was known to have large numbers of tigerfish (Gaigher, 1968). This situation has changed dramatically, mainly due to weirs being built in Mozambique and the

22 absence of suitable breeding grounds in South African waters (Gaigher, 1968). The outflow of capital to the neighbouring countries to catch tigerfish can be counteracted if South Africa can create its own "Zambezi River" (Kruger, 1972). The need for propagation of tigerfish is thus very important, and can be truly justified.

There are, however, certain factors to take into account when dealing with tigerfish, as it is a very sensitive species. Water temperature fluctuations have to be kept to the minimum when transporting the fish. Water temperature where tigerfish are kept should be constant at all times and should not decrease below 15°C, although tigerfish do survive at that specific temperature. Kruger (1973) reported mortalities due to temperature shock after successfully transporting the fish to the Marble Hall fisheries research station. Mortalities of tigerfish occurred in the Piet Grobelaar dam in the Kruger National Park when the water temperature decreased to 14.5°C (Van der Merwe, pers comm). Dead tigerfish were also observed in a laboratory after the water temperature decreased to 14.5°C (Steyn, pers comm).

Probably the most important aspect of transporting tigerfish is to ensure high oxygen levels. The fish should be handled and transported with great care. An anaesthetic should also be applied to calm the fish when placed in a container and to prevent injuries against the sides of the container (Kruger, 1972).

The first breeding programme of H.vittatus was initiated in 1972. The first female was injected with a 2 mg carp (Cyprinus carpio) pituitary extract. The specimen died after 48 hours, but the gonads were very well developed with eggs which were loose from one another. Three females were injected a year later with pituitary extract but unfortunately died 3-5 days later. The gonads of these females did not show any signs of development and were green in colour. The constant loss of females when injected with carp pituitary could have been due to poisoning as the fish started losing scales and haemorrhage between the scales was visible.

23 Van Loggerenberg (1979) initiated a project to translocate tigerfish from the Kruger National Park to certain dams in the Lowveld. The main objective of the project was to determine where the tigerfish would spawn in the eight selected dams. The Klaserie Dam was first chosen because it is a relative small dam (123 ha) and situated close to the Olifants River from where the tigerfish would be collected. Tigerfish were collected close to the Gorge Picnic spot in the Olifants River and translocated to the Mala-Mala Dam of the Phalaborwa Mining Company. A minimum of mortalities were reported in the Mala-Mala dam after transportation (Van Loggerenberg, 1979). The tigerfish which were translocated to farm dams after being collected in the Komati River were also reported to be in good condition. Unfortunately, Van Loggerenberg (1980) did not report on the growth of the tigerfish in the Klaserie Dam but the growth of tigerfish in the dams close to Komatipoort were very satisfying (Van Loggerenberg, 1981).

A breeding experiment was conducted on tigerfish in the Transvaal Sugar Board (TSB) Dam at Ten Bosh (Van Loggerenberg, 1981). Five females were injected with a combination of (C. carpio) pituitary and Pregnyl for the first session. A second injection after nine hours of only (C. carpio) pituitary was then given to the females. One female died on the second day of treatment and two other females did not respond to the hormones. One other female did produce a portion of her eggs but after a dissection showed that the ovaries were only partially developed. The eggs of this female were fertilised in the same manner as it was done with the eggs of the Letaba River which is described in the next paragraph. The fifth female was in poor condition and also dissected, the eggs were removed, fertilised, and incubated. The eggs were inspected under a microscope on a hourly basis to see if any development had taken place. After one hour of fertilisation cleavage of one of the eggs was observed. No further development of any eggs were witnessed during the following inspections (Van Loggerenberg, 1981). According to Van Loggerenberg (1981) the fact that egg development was poor, could well have been because of a high concentration of DDE-insecticide as were found by Van Rensburg (1981) in the tissues of tigerfish and Clarias gariepinus at Komatipoort.

During their investigation of the tigerfish in the breeding season of 1979/80 in the Mala-Mala Dam at Phalaborwa no ripe running females were found while all the males were found to be ripe running (Van Loggerenberg, 1980). Due to the lack of ripe-running females in the previous year

24 it was thought that it would be better to obtain the breeding stock just below the Engelhardt Dam in the Letaba River. It would have been easier to keep the tigerfish there in 500 litre plastic tanks and to treat the fish with hormones on site. The plan was then to transport the fertilised eggs in plastic bags with water to Phalaborwa where they would be placed in a breeding apparatus. On 21 November 1980 two females were treated with a combination of Estrumate and Pregnyl in a ratio of 1:3. The dosage of hormones were calculated as for every one kilogram of biomass to

1000 International Units (I.U.) hormones in total for a single injection. Another female was in a poor physical state after transport and the eggs were removed. Most of the eggs were in a ripe condition. Only one male was treated with 150 I.U. Estrumate. The semen of a male was then mixed with the eggs of that female. The eggs which were removed from the female were rinsed with oxygen rich water to get rid of the unwanted semen and ovarian tissue, five minutes after fertilization. The eggs were put in plastic bags and oxygen was supplied to the eggs in such a way that a small turbulence was created. These eggs were kept for about thirty hours in the plastic bags. During this period the water was changed twice in the plastic bags with river water, and the dead egg material removed. One of the two females which were treated for a second time on 22

November with hormones were also dissected and the eggs removed and fertilised. The fertilised eggs were inspected on the 23 November and were suspect. Upon further inspection it was found that Paramecium Protozoa were feeding on empty eggs shells. A few larvae which were apparently dead were suspended in the water. The cause of the death of the eggs during that night could not be explained, although some speculations were made. The remaining female was treated again on 23 November, but did not produce any eggs when stripping was attempted (Van

Loggerenberg, 1981).

On 10 December 1981 two females were three times, with a nine hour interval, treated with hormones. Three more females were collected later and released in a pool in the river. A net was placed in the pool to prevent the fish from escaping into the rest of the pool. A hippopotamus walked through the net resulting in all the tigerfish to escape. No more females could be collected later during a follow up survey (Van Loggerenberg, 1982).

25 A few years later (1986) Steyn and Deacon had limited success with the breeding of tigerfish at the rangers post at Crocodile Bridge. The embryos developed up to 12 hours after which no further development took place. The same problems with development were also experienced during December 1991 at Charara next to Lake Kariba in Zimbabwe (Steyn, pers comm). These experiments were however extremely important as this had lead to the successful artificial spawning of the tigerfish which will be dealt with in Chapter Four.

26 CHAPTER THREE

ASPECTS OF THE POPULATION DYNAMICS AND DISTRIBUTION OF THE TIGERFISH HYDROCYNUS VITIATUS IN THE LOWER OLIFANTS AND LETABA RIVERS, KRUGER NATIONAL PARK.

3.1 Introduction

The study of population dynamics may be defined as the attempt to measure rates of birth, growth and death, and their inter responses in a precisely defined population. It may be thought of as including the quantitative aspects of the ecology of a population (Regier, 1966).

Tigerfish may attain a length of 140 mm to 180 mm in the first year and 250 nun in the second year in Lake Kariba (Balon, 1971). Tigerfish is a long lived species which may attain a length of 590 mm according to the Loo values calculated by Van Zyl (1992). The maximum size of Hydrocynus villains differ considerable from one system to another. Bell-Cross (1965;1966) reported tigerfish with a total length of 705 mm (7.4 kg) from the Upper Zambezi, while the Zambian Fish Records Association recorded a specimen of 8.8 kg from the Zambezi River. The maximum sizes reported from the Okavango and Lake Kariba were 620 mm and 800 mm (15.5 kg) respectively (Kenmuir, 1983; Van Zyl, 1992 & Hay, 1995). Tigerfish of 5.4 - 5.9 kg were regularly caught in the Sabie, Crocodile and Letaba Rivers until 1978 (Pienaar, 1978). Maximum sizes of 3.8 kg and 3.5 kg tigerfish has been recorded by (Van Loggerenberg, 1982; Steyn, 1987) respectively in the Crocodile, Letaba and Olifants Rivers. Consequently it appears as if there is a progressive decline in the maximum sizes of tigerfish in the KNP.

Tigerfish individuals within a group which has a faster growth rate, has a better chance of survival (Balon, 1971). The very high mortality rate (84%; Z= 0.851) of tigerfish in their first season of growth, emphasises the benefit of a faster growth rate and critical length to avoid predation (Balon, 1971; Van Zyl 1992). Females had a faster growth rate than the males, especially in the

27 older age groups, in Lake Kariba (Balon, 1971). These findings are also supported by the phi prime values calculated for males and females in the Okavango River (Van Zyl, 1995; Hay, 1995).

Due to the difference in growth rate between the sexes of the same age group, males and females separate naturally and arrange themselves in schools of similar length groups in order to avoid cannibalism within the same age group. This hypothesis is based on behaviourial observations within an aquarium supported by several years of field observations (Steyn, pers comm). A discrepancy between males and females often occurs in nature and a specific sex tend to aggregate at certain localities, possibly to a uniformity in size (Steyn, 1987). For example, the larger females within the same brood tend to stick together and predate on their smaller male counterparts which will shoal separately.

In the Incomati River System, Lake Kariba and the Okavango River, males reach sexual maturity at a minimum length between 170 and 200 mm (Gaigher, 1975; Langerman, 1984; Van Zyl, 1992; Hay, 1995). The average length when 50% of the males reaches sexual maturity in the Okavango River is 260 mm with the minimum length of the females slightly larger at 275 mm (Hay, 1995). Fifty percent of the females reaches sexual maturity at a length of 260 mm in Lake Kariba (Langerman, 1984), which is similar to the males in the Okavango River (Hay, 1995). A total of 84% of the females caught in Lake Kariba were larger than 400 mm, indicating that the older or larger tigerfish contribute to the breeding potential of the population (Kenmuir, 1973). Female tigerfish grows larger than males (Jackson, 1961a; Kenmuir, 1973; & Van Zyl, 1992), and also reaches sexual maturity later in life (Kenmuir, 1973; Langerman, 1984). Tigerfish has a high fecundity in comparison to other fish species (Pott, 1969; Kenmuir, 1973; Hay, 1995; Chapter 6). This greater fecundity is an evolutionary response to ensure the survival of the species as mortalities of the eggs, larvae and yearlings are very high which is demonstrated by the high mortality rate calculated (Balon, 1971; Van Zyl, 1992).

28 The purpose of this section of the study was to compare some of these aspects with other tigerfish populations, as the KNP tigerfish population is represents the edge of distribution of the species and it can therefore be expected that the values for population parameters will not be optimal. Furthermore it is evident that the distribution range of the tigerfish in the Northern Province and Mpumalanga has changed.

3.2 Materials and methods

3.2.1 Choice and description of sampling localities Tigerfish populations were studied over a period of thirteen months (October 1991 - September 1993) in the Olifants and Letaba Rivers in the Kruger National Park. Eight localities were selected along the Olifants and Letaba Rivers. These localities does not necessary represent different reaches (based on geology, geomorphology and rainfall) as identified by Venter (1991) in these rivers but was selected on the basis of obstacles which causes aggregation during migration and sometimes isolation at low flow and consequently serve as ideal sampling localities. Six surveys were done during the autumn and winter months (March; April; May; June; July; August) and seven during the spring and summer months (September; October; November; December; January; February) which will be referred to in this study as winter and summer. Four localities (localities 1-4) were selected along the Olifants River from Mamba weir on the western border of the KNP to the confluence of the Letaba River ± 5 km from the border with Mozambique and four localities (localities 5 - 8) were selected in the Letaba River between Engelhardt Dam and the confluence with the Olifants River (Fig 3.1).

Locality one is situated directly below Mamba weir within reach one according to the classification of Venter (1991). This reach has a single channel with mostly a flat river bed and shallow stream. Short rapids occur over firm or rounded rock, with deep pools only occurring occasionally. The river bed consists of sand and gravel, alternating with small rocky pools. Riparian vegetation is moderately dense with trees such as Fiats sycomorus, Trichilia emetica, Lonchocarpus capassa, Acacia robusta and Diospyros mespiliformes. Hanging reeds (Phragmytes spp.) are limited to isolated small patches.

29 localities in the Olifants an Locality two is downstream from the Balule low water bridge and above the potholes at the Olifants Wilderness Trail Base Camp. A natural obstruction to fish during periods of low flow occurs at the potholes. This locality falls within reach two. The channel of reach two is mostly irregular and branches off to form small (5 - 10 m) and sometimes deep channels between the islands. The river bed consists of irregular deposits of silt or sand on rock or islands. Dense reeds occur on islands and sometimes also trees such as F. sycomorus and Breonadia saligna. The riparian vegetation is scattered to moderately dense with trees such as F. sycomorus, T. emetica, L. capassa, A. robusta, D. mespiliformes and Colophospermum mopane. Hanging reeds (Phragmytes spp.) are very dense in some localities (Venter, 1991).

The third locality is below the potholes in the river stretch below the Wilderness Trail Camp. Reach three has a V - to U - shaped single channel with deep pools and short rapids. At the Olifants camp and Trails camp, the channel is deeply cut into the rock to form a series of low waterfalls and deep little ravines. The river bed consists of rock with relatively thick depositions of red silt in deep pools and round loose cobble-stones in the rapids. Virtually no vegetation occurs on the river bed and reeds are limited to rocky places or alluvial islets. The river banks are open with trees such as F. sycomorus, T. emetica, L capassa, A. robusta, and A. xanthophloea. Hanging reeds (Phragmytes spp.) occurs alternatively (Venter, 1991).

Locality four is at the Olifants/Letaba confluence downstream into the Olifants Gorge. This locality lies within reach four. This valley consists of a V- to U shaped, narrow and deep gorge in the Lebombo Mountains. The banks are steep (often vertical) and high (40 - 60 m). It has a single U- shaped channel with many deep pools and little rapids. There are no vegetation on the river bank and bed (Venter, 1991).

Locality five is directly downstream from the Engelhardt Dam with locality six five kilometres further down. The river bank is slightly steep and low with the channel as an irregular river bed that is highly divided. Dense reed beds occur with a larger tree composition of F. sycomorus, T. emetica, L capassa, C. mopane and Combretum imberbe. Dense reed beds over the whole width of the river bed and trees like F. sycomorus and Breonadia saligna are sometimes found in the middle of the river bed (Venter, 1991).

30 Figure 3.1 Locality one within reach one in the Olifants River.

Figure 3.2 Locality two within reach two in the Olifants Figure 3.3 Locality three within reach three in the Olifants River.

Figure 3.4 Locality four within reach four in the Olifants River. Figure 3.5 Locality five within reach four in the Letaba River.

Figure 3.6 Locality six within reach four in the Letaba River. Figure 3.7 Locality seven within reach five in the Letaba River.

Figure 3.8 Locality eight within reach five in the Letaba River.

c Localities seven and eight are situated within reach five, approximately two km upstream from the confluence with the Olifants River. It has a channel width of 30 - 50 metres with a gradient of 10.0 m/km. The riverbank and bed is solid rock with fallen rocky boulders. Little vegetation occurs along the river (Venter, 1991).

3.2.2 Sampling procedure and data collection Tigerfish were caught with rod and reel in the Olifants and Letaba Rivers (Fig 3.1). Brass lures of different sizes were used as artificial bait to prevent sample bias. The sampling method was however selective for specimens larger than 80 mm as this was the smallest size which could be obtained with lures.

Electro-narcosis as method was not used for most of the data collection took place in deep pools with great numbers of crocodiles. Since sampling took place in a National Park, and only tigerfish were sampled for, the use of rotenone was not deemed necessary. After specimen collection the standard length (SL) was measured (mm), and the mass (g) was determined with a Kern electronic balance (model 440/53). Scales were collected dorsolateral, between the head and'dorsal fin, for age determination. They were then placed in a small envelope and marked with all the relevant information of the specimen, date and locality of collection. The sex was determined by dissection.

3.2.3 Data analysis In order to analyze the data with the Length Frequency Stock Assessment (LFSA) software, the data sampled over a period of thirteen months was categorised under two sections namely September - February representing summer and March - August representing winter. Length data was sorted into 10 mm length groups to obtain length frequencies for the two seasons in the Letaba and Olifants Rivers respectively (Fig 3.2). These length frequencies were separated into cohorts to determine the age groups and population structure (Table 3.1).

Age determination was done after the scales were first soaked in water for 24 hours and then washed to remove excessive mucus and dead tissue. Age determination based on scale readings from tigerfish in the Olifants River were done while the scales were still wet and correlated with length groups (Table 3.2). The length frequencies of H. vittatus and scale readings were used to

31 determine the Von Bertalanffy growth parameters (Loo & k) with Bhattacharya's method, link of means and the programme VONBER in the Length Frequency Stock Assessment (LFSA) software.

The Von Bertalanffy growth curve was determined with the following formula: 1-t) = Lop (1 - exp 441-w)) where k = curvature constant, Dr = L - infinity t = time (years), to = t - zero

The overall growth performance as reflected by Munro's phi prime (4) was determined by using Loo and k. 4) (phi prime) = In k + 2 * In Loo

Total mortality (Z) was determined with the catch curve and Pauly's empirical formula (Sparre et al, 1989): Z = M + F where M = natural mortality coefficient & F = fishing mortality coefficient

The percentage survivors from the annual recruitment after one year was determined with the Exponential decay model (Sparre et al, 1989): -(Z*(t-Tr)) N(t) = N(Tr)exp where N = number of survivors, t = time (years), Tr = recruitment age, Z = total mortality coefficient

32 3.3 Results and discussion

3.3.1 Distribution and abundance The abundance and distribution of tigerfish in the Olifants and Letaba Rivers varies considerably in relation to the localities and the different seasons. Tigerfish are more abundant in the Olifants River in comparison with the Letaba River in spite of the intensive survey conducted during October 1991 in the Letaba River.

Although tigerfish are found along the entire Olifants River in the KNP they are very unevenly distributed in numbers but rare towards the western border near the Phalaborwa Barrage. Van Loggerenberg (1981) also mentioned that the numbers of tigerfish higher up in the Olifants River in the KNP was so few that it could not be netted with seine nets. It is however not possible to determine to what extent the abundance of the tigerfish below the Mamba weir has changed, for it was never recorded in detail. During the summer months tigerfish was more abundant (n=1249) than the winter months (n=209) in both the rivers. A grand total of 1100 tigerfish were collected in the Olifants River of which 932 and 168 tigerfish were collected during the summer and winter months respectively. Tigerfish is more abundant towards the eastern border in the Letaba River below the Engelhardt Dam. A total of 358 tigerfish with 317 in the summer and 41 in the winter months were collected.

Tigerfish seems to aggregate in large numbers below the Olifants Wilderness Trail Camp during certain periods of the year. It is evident that aggregation of tigerfish takes place before the onset of the summer at the natural migration obstruction below the Wilderness Trail Camp. The numbers of tigerfish decline after the floods due to distribution, indicating that the obstructions had been overcome. As the winter months approach, the numbers of tigerfish decline drastically, since only 168 tigerfish had been collected during the winter. This implies that tigerfish migrates downstream in order to tolerate winter temperatures. This assumption is supported by the observation of Van der Merwe (pers comm) whom observed tigerfish dying in the KNP when water temperature decreased to 14.5 °C at night during the winter of 1994. Tigerfish in the Letaba River were more readily caught during the summer months than the winter months in spite of the fact that the pools were also isolated during the summer and not only during the winter months.

33 A decline during the winter months could also be due to a reduced metabolism, urge to feed and activity associated with the lower winter temperatures. During this study it was found that the best locality for collecting tigerfish would be below the Olifants Wilderness Trail Camp during the months October until February before the water level rises to overcome migration obstacles.

Although the tigerfish is relatively protected in South Africa as this species is largely isolated in the Kruger National Park, it is vulnerable to anthropogenic activities which can influence the water quantity and quality in this seemingly haven. The reproductive potential of H. vittatus is negatively influenced due to migration problems and limited spawning localities due to the erection of weirs and the shortage of water (Gaigher, 1973; Steyn, 1987) and that the distribution of tigerfish in South Africa has diminished due to anthropogenic activities (Chapter 6).

3.3.2 Age and growth Although the sampling method used during this study was selective, a good representation of the population larger than 8.5 cm (SL) was obtained. A larger variety of length groups were represented in the length frequencies of the Letaba population and varied between 85 and 405 mm (Fig 3.2). Length frequencies displayed a maximum of three cohorts in the Olifants River with a minimum length group of 85 mm and a maximum length group of 375 mm. Although the first two cohorts for the winter season in the Olifants River are relatively close to each other, and can probably be seen as one, the separation index (Si.) of 2.078 is an indication that the two cohorts can statistically be separated (Table 3.1). Only two cohorts could be identified for the Letaba population sampled during summer. Inadequate data was obtained to separate more than one cohort from the winter data of the Letaba River. In both the rivers fish larger than 295 mm (SL) were scarce however a good representation of fish smaller than 295 mm (SL) was evident during the sampling periods.

The average lengths of cohorts separated in Table 3.1 demonstrate that the tigerfish in the Letaba

River are larger than the same age groups in the Olifants River. Tigerfish in the Letaba River were larger than those in the Olifants River during surveys in 1986 (Steyn, pers comm). This might be due to a smaller population size tigerfish in the Letaba as compared to the Olifants River.

Competition for food and living space is thus not as fierce as in the Olifants river. Furthermore

34

Olifants - Winter 0 0 i tisti JolaciumN R I 0 — I oLL —E —E

Letaba - Summer Letaba - Winter ID 0 0 S ID 11 qsy JolaqtatiN tisg PJaquint4 a n m1 I 71.p. N R ; 0 T -in 1 .5 a • O o-1 CU t.0 er; 0) CU Ctl col 0.4 Et 2 0 C N C a) C tin 0 -4 Cr 0 >-. 0 N OD .) far less crocodiles are also found in the Letaba River, which subject them to less predation by these efficient predators. The absence of large tigerfish from these rivers as documented by Pienaar (1978) is possibly a result of the massive Massingire Dam inside Mozambique. It could well be that the larger tigerfish prefer to stay within the larger volume of water in the dam. Consequently younger fish are then forced out of the dam to escape predation pressure and to expand their living space. This could also serve to explain the large numbers of the same age group of tigerfish that aggregate below obstacles in these rivers.

Table 3.1 Average length (mm) of the different cohorts obtained by the Bhattacharya method for H. villains in the Olifants and Letaba Rivers during two different seasons.

Season Cohort No. Average Standard deviation Cohort size Separation Index

length (nun) (s.d.) 00 (SS)

Olifants River

Summer 1 769 16 761 R96 44 --- 2 328 9.000 30.400 4.619 3 366 10.313 3.160 3.933

Winter 1 236 13.043 20.000 --- 2 270 18.835 137.93 2.078 3 327 7.238 9.070 4.379 Letaba River

Summer 1 270 17 56 275 73 --- 2 330 13.03 31.92 3.907

Winter 1 262 22.20 29.44 ---

The calculated maximum length (Lco = 52.40 cm) according to the Wetherall method, was much larger than the largest specimen sampled in the Olifants (375 mm) and Letaba (405 mm) Rivers. The L4 values calculated for the Olifants and Letaba Rivers is smaller than the L4 value (Lco = 56 06) calculated for the Okavango River (Van Zyl, 1992). The maximum sizes sampled from the

35 Olifants and Letaba Rivers respectively were also substantially smaller than the maximum lengths recorded in the Okavango River (620 mm) and Lake Kariba (570 mm) (Balon, 1971; Van Zyl, 1992). Table 3.2 Length frequencies of age groups as determined from scale readings of H. villains from the Olifants River.

Length group Age (years) (mm) 1+ 2+ 3+ 0+ 5+

121 - 130 1 151 - 160 1 181 - 190 1 201 - 210 1 211 - 220 1 221 - 230 1 231 - 240 4 7 241 - 250 26 3 251 - 260 46 9 261 - 270 49 16 271 - 280 12 19 281 - 290 3 12 1 291 - 300 3 6 3 301 - 310 6 311 - 320 4 1

321 - 330 2 2 331 - 340 1 361 - 370 1 1 371 - 380 1 391 - 400 441 - 450 1

36 Two methods, having a set asymptotic length of 52.40 cm, was used to calculate the curve (k) values: where Gullant and Holt Plot, k = 0.387; Munro's method, k = 0.40. These k values are significantly. The overall growth performance or phi prime (4)), as calculated with the different curvature parameters are: 4) = 6.97; k = 0.387 and 4) = 7.00; k = 0.40.

Annuli counted on the scales under a microscope showed that the tigerfish can reach an age of five + years in the Olifants River (Table 3.2), altough they are very scarce. The mean age group of all the specimens sampled from the Olifants River are those of two and three years old. Data from scales of H. vittatus in the Olifants River were also taken to determine the growth parameters, where Lao = 35.193 and k = 0.449 with 4) = 6.32 (scale readings) and 4) = 6.97 & 7.00 (length frequency).

The phi prime values as obtained by length frequencies and scales are supposed to be equal (Sparre et a1,1989). The values of the phi prime from the scale readings and that of the length frequencies are similar due to the fact that we are dealing with the same population, in spite of the different values obtained for the asymptotic lengths.

The overall growth performance of the H. vittatus in the Olifants River compare well to those found by Van Zyl (1992) and Hay (1995) in the Okavango River although the two systems can not be compared with each other. Similar growth rates were found by Beattie (1982) in Lake Kariba. It would therefore appear that tigerfish have the same growth performance irrespective of the system they occupy.

According to the catch curve the tigerfish population in the Olifants and Letaba Rivers can reach an age of six and seven years respectively. The age given by the catch curve compares well with the age determined by scale readings for the tigerfish in the Olifants River. Tigerfish could reach an age of 8 to 10 years in the Okavango River (Hay, 1995; Van Zyl, 1992) and seven years in Lake Kariba (Langerman, 1984). Annuli counted has indicated that H. vittatus can reach an age of five years in the Olifants River, which closely corresponds with the year class calculated by using the catch curve method. The fact that H. vittatus can reach an age of five years in the Olifants River, might also be a greater reflection of mortalities in that river.

37 3.3.3 Mortality Length frequencies are used when determining the catch curve values to determine the mortality of a population. The small sample size in the length frequencies in the Letaba River may therefore influence the values of the catch curve resulting in misinterpreted data.

The total mortality (Z) for the tigerfish population in the Olifants and Letaba Rivers is Z = 1.488, and Z = 1.067 respectively. The Z- values of a light, medium and heavy exploitation are: Z = 0.6, Z = 0.9 and Z = 1.2 respectively (Sparre et al, 1989). According to these values, there is a medium and high exploitation rate of tigerfish with 34.4% and 22.6% tigerfish recruits to survive after one year in the Letaba and Olifants River respectively. Although the tigerfish population in the Letaba River has a higher survival rate as opposed to those in the Olifants River, it is still lower than the survival of tigerfish in the Okavango River. The total mortality of the tigerfish in both the Olifants and Letaba Rivers is higher than the mortalities recorded in the Okavango River (Z = 0.851) and in Lake Kariba (Z = 0.557) by Van Zyl (1992) and Balon (1971) respectively, but still much lower than the total mortality in the Niger (Z = 2.58) as recorded by Dansoko (1976).

It is unknown in this stage why there is such a high mortality rate of tigerfish in the two rivers, but possible causes are: a - heavy predation by the large number of crocodiles and other predators in the river and area present; b - heavy fishing mortalities by the Mozambique residents in the Massingire Dam; c - unsuitable habitat for efficient recruitment; d - high silt loads during floods.

The (Crocodilus niloticus) were reported to be a predator of the tigerfish, although some contradiction as to that exists (Balon, 1971). If the nile crocodile does happen to prey on tigerfish in the Letaba and Olifants Rivers, they could have a big impact on the tigerfish population. The Olifants Gorge and River up to the Wilderness Trails Camp is probably one of the areas with the highest concentration of crocodiles per kilometre in a river system known in Africa (Viljoen, pers comm). In comparison the Letaba River has less crocodiles. Counts of up to 155 crocodiles 500 meters downstream from the Olifants Wilderness Trail Camp in a 150 meter river stretch had been made by the author. This could well explain the high exploitation rate of

38 the tigerfish in the Olifants River and not in the Letaba River. Other known predators of the tigerfish is the Cape Clawless Otter (Aonyx capensis), Fish Eagle (Habanns vocifer) and the Pel's Fishing Owl (Scotopelia peli) which occurs along the rivers.

The very high concentrations of silt in the Olifants River during floods (Buermann, Du Preez, Steyn, Harmse, and Deacon, 1995) could have a negative effect on the distribution and abundance of tigerfish towards the western border of the Kruger National Park. Recruitment and larval survival rate might also be negatively influenced by the high concentrations of silt.

3.3.4 Sex ratio The male:female (M:F) ratio differs somewhat in the Letaba and Olifants Rivers. In both rivers the males seems to be more abundant than the females during the summer and winter season. The sex ratio of H. villains in the Letaba River is 5.1:1 and 8:1 in the Olifants River respectively. During the non-breeding season the M:F ratio (7.1:1) favoured the males, although the ratio was much higher (9.9:1) during the breeding season (September - February) in the Olifants and Letaba Rivers combined. The very high sex ratio of males to females which was found in especially the Olifants River can not clearly be explained. More males than females are produced when a population is under stress (Holtzhauzen, 1989). These stress conditions seems to be a result of negative environmental conditions such as water temperature, photoperiod and over population on the parental fish. External factors such as photoperiod and especially stress as a result of over population was reported to be the main cause for the shift towards males in the sex ratio of a population (Tave, 1986), rather than fluctuations in water temperature (Mires, 1974). As a result of the high population of tigerfish in the Letaba and especially the Olifants River, and the stress associated with that, the high male:female ratio could therefore be easily explained.

The high male:female ratio in the Olifants and Letaba Rivers could also be as a result of the larger females preferring to stay within the larger volume of water in the Massingire Dam in Mozambique.

39 3.4 Conclusion

The phi prime values from length frequencies and scale readings were found to be similar due to the fact that we are dealing with the same population. These values compare well with the phi prime values for tigerfish in the Okavango River (Van Zyl, 1992) and Lake Kariba (Balon, 1971). The mortality rate of tigerfish in the Olifants and Letaba Rivers is larger than those in the Okavango River and Lake Kariba possibly indicating that reproduction requirements are not optimal. Tigerfish in the Olifants and Letaba Rivers have a life expectancy of six and seven years respectively as opposed to the ten to twelve years in the Okavango River (Van Zyl, 1992). A deviation from the expected 1:1 M:F sex ratio was experienced in both the Olifants and Letaba Rivers which could well be explained by stress as a result off over population. The population dynamic values obtained, indicates that conditions are not optimal for tigerfish in the KNP except for the growth parameters.

40 CHAPTER FOUR

FEEDING HABITS OF THE TIGERFISH (HYDROCYNUS VM'ATUS) IN THE LOWER OLIFANTS AND LETABA RIVERS, KRUGER NATIONAL PARK.

4.1 Introduction

There can be no doubt concerning the predatory impact of tigerfish on any given population of fish in a system. Their role in an ecosystem certainly has been the subject of many studies, and rightfully so too, not only because of their predatory impact, but also of their economical value. Most work has however been done in large rivers and lakes in Africa and include studies in Lake Kariba (Jackson, 1960; Matthes, 1968; Kenmuir, 1973; 1975), Lake Cabora Bassa (Jackson, 1976), Lake Albert (Holden, 1970), Lake Kainji (Lewis, 1974), Lake Chad (Lauzanne, 1975), large rivers such as the Zambezi river (Jackson, 1961a; 1961b; Winemiller and Kelso-Winemiller, 1994), the Okavango Swamps in Botswana (Van Zyl, 1992; Hay, 1995) and the central Delta of the Niger (Dansoko et al, 1976). Few work has however been done on smaller river systems towards the southern range of their distribution where they are relatively scarce (Gaigher, 1967; 1968; 1969; 1970).

Tigerfish is known to be a successful occupant of clear well oxygenated water if one consider for instance the Zambezi River. One should expect that the feeding behaviour of a predator, especially a piscivore, would be influenced in a system with a high turbidity or low visibility. Apart from the physical/ chemical environment, the population size of an organism is mainly determined by recruitment success and availability of food. During this investigation the feeding habits were compared with other systems.

41 4.2 Material and methods

Tigerfish were caught with a selection of artificial lures and angling gear in the Letaba and Olifants Rivers within the Kruger National Park. The size range of artificial lures used provided an equal catchability of all possible sizes larger than 80 mm (SL). Samples were collected on a monthly basis during the period October 1991 - September 1993. Specimens were gutted and the stomachs removed for laboratory analyses.

The stomachs, including the contents, were preserved and stored in plastic bottles with 10% formaldehyde. The fixed stomach contents were analyzed under a stereo microscope in the laboratory. Tigerfish sampled, were placed in different feeding categories due to different composition of their stomach contents. The categories were defined as: Invertivore - invertebrates as food item Piscivore - fish as food item Carnivore - fish and invertebrates as food item Opportunistic - Any other material such as plant material etc.

A linear regression analysis was used to determine if there was any correlation between the length of the tigerfish and the fullness of the stomach.

Four seasons were identified which the stomach contents were divided in: these included spring (September; October; November), summer (December; January; February), autumn (March; April; May) and winter (June; July; August).

4.3 Results

A total of 84% of all the stomachs examined with contents had invertebrates as food item. In both the Olifants and Letaba Rivers invertebrates were the most important food item (54% and 32%) followed by fish (7% and 11%) (Fig 4.1). Four specimens examined (3%), had more than one food item in their stomachs however invertebrates were present in all four stomachs. Only one

42 tigerfish stomach (1%) contained both invertebrates and fish. Other food items included plant material and invertebrate eggs. The seed of an exotic invader plant, Xanthium Strumarium which grows abundantly on the banks of the Olifants River were also found in the stomach of a tigerfish. Three teeth of a tigerfish were found in a its own stomach, which must have been swallowed after replacing a set of teeth (Chapter 7). A large percentage of the stomachs in the Olifants and Letaba Rivers (37% and 55%) were empty (Table 4.1).

The stomach contents of only thee tigerfish had fish remains that could be identified during this study. Prey species that were identified, were Synodontis zambezensis and Brycinus imberi. A 74 mm (TL) Synodontis zambezensis was obtained from a 230 mm (SL) tigerfish, which comprises 32% of the predator's length. The other specimens were in an advance stage of digestion and could not be measured. Consequently, no correlations between predator length and prey could be made.

The percentage of invertebrates taken by tigerfish in the Olifants River, decreased progressively from spring to winter. However, the relative percentages of fish that could be sampled also decreased in the same order as described above (Table 4.2). The presence of invertebrates in stomach contents did not show the same tendency in the Letaba River. In both the rivers a definitive trend could be observed as to the feeding dependence on invertebrates during the different seasons of the year (Table 4.2). With a correlation coefficient of 0.0386 and 0.0523, no correlation could be found between length classes and feeding preference for tigerfish in the Olifants and Letaba Rivers respectively, although the peak of number of fish with full stomachs were found to be between length groups of 250 and 280 mm (Table 4.3).

Identification of the invertebrates preyed upon, were not always possible, as they were in an advanced state of digestion. Sixteen percent of the invertebrates could thus be not identified. A total of 59% of all the stomachs examined had contents, and the remaining 41% stomachs were found to be empty.

Nine invertebrate Orders were present of which the Ephemeroptera (30%), and Hemiptera (29%) were the most abundant (Figure 4.1). Twelve families are represented in the nine orders of

43 •▪

invertebrates on which the tigerfish preyed. The invertebrate families notonectidae, corixidae and chironomidae represented the most food important items for the tigerfish (Table 4.4).

Table 4.1. Food preference of H. vittatus in the Letaba and Olifants Rivers. Percentage and number of fish in each category Composition of diet category Letaba River Olifants River

n= % n=

Stomach empty 31 55 66 37

Invertivore 18 32 97 54

Piscivore 6 11 12 7

Carnivore 0 0 1 1

Opportunistic 1 2 2 1

Olifants River Letaba River

Invertivore Piscivore IM Carnivore Opportunistic

Figure 4.1. Food preference of H. vittatus in the Olifants and Letaba Rivers.

44 Table 4.2. Food preference of H. viltatus in the Letaba and Olifants Rivers for the different seasons. Percentage and number in each category Season Composition of diet category Letaba River Olifants River n= % n= %

Spring Invertivore 7 28 49 44 Piscivore 2 8 3 3 Carnivore 0 0 0 0 Opportunistic 0 0 1 1 Summer Invertivore 3 12 33 29 Piscivore 2 8 3 3 Carnivore 0 0 1 1 Opportunistic 0 0 1 1

Autumn Invertivore 0 0 1 1 Piscivore 0 0 2 2 Carnivore 0 0 0 0 Opportunistic 0 0 0 0

Winter Invertivore 8 32 14 12 Piscivore 2 8 4 4 Carnivore 0 0 0 0 Opportunistic I 4 0 0

45 Table 4.3. Number of H. vittatus stomachs with and without contents against length frequency intervals in the two rivers.

River Stomach status in Letaba River Stomach status in Olifants River

Length Empty Full % full Empty Full % full 90 1 130 1 150 1 4 190 1 4 210 I 220 I 1 230 1 4 2 2 240 3 1 4 8 7 250 5 4 16 3 13 11 260 5 4 16 11 31 28 270 7 5 20 19 31 28 280 1 4 16 14 13 11 290 2 6 7 6 300 1 1 4 5 4 4 310 2 1 4 2 1 1 320 1 1 2 2 330 370 2 8 380 1 450 1

46 Table 4.4. The different invertebrate orders and families, and the percentage composition of the invertebrates diet present.

Orders n = % composition of Families which could n = invertebrate diet be identified

Coleoptera 1 1 Megaloptera 1 1 Isoptera 4 3 Hymenoptera 6 4 Formicoidea 1 Odonata 11 8 Trichoptera 13 9 Diptera 22 15 Chironomidae 11 Culicidae 7 Simuliidae 2

Hemiptera 42 29 Aphididae 1 Naucoridae 3 Gerridae 4 Corixidae 12 Notonectidae 16

Ephemeroptera 43 30 Beatidae 1 Caenidae 2

n = 143

47 Figure 4.2. Percentage of invertebrate orders in the diet of H. vittatus

4.4 Discussion

There can be no doubt as to the enormous impact a predatory fish such as the tigerfish has on any given system. Their whole existence being that of inquisitive, hungry, aggressive, fearful roamers of the open waters they occupy, making them formidable predators. The speed they can attain in the water also attributes towards their role as super predators, and being highly feared by prey species.

Although tigerfish are top predators, a total of 41% of the stomachs examined during this study were found to be empty. This is much lower than the tigerfish with empty stomachs (89.9%) in a pool in the Gache River, Zimbabwe (Kenmuir, 1973) and (57.3%) in the Upper Zambezi drainage of Zambia (Winemiller and Kelso- Winemiller, 1994). Lewis (1974) stated that the percentage of fish with empty stomachs varied considerably from month to month in Lake Kainji, Nigeria. These variations could not be correlated with breeding seasons, lake levels or water turbidity.

48 It was found that tigerfish feeds almost .exclusively (84%) on invertebrates in the Letaba and Olifants Rivers. An analysis of the stomach contents of H. forskali indicated a very high percentage (96%) of invertebrates taken at some of the offshore stations in Lake Albert (Holden, 1973). During April 1969 to March 1970 and April 1970 to March 1971 only 5.3% and 18.1% of tigerfish stomachs examined had invertebrates as contents in Lake Kariba (Kenmuir, 1973). In Lake Kainji, Lewis (1974) found 1.7% of H. forskahhi to have taken invertebrates and H. brevis none.

Not all the invertebrates in the stomach contents could be identified, as they were in an advanced state of digestion. The orders Ephemeroptera and Hemiptera represented the main invertebrate component in the diet of the tigerfish which is in contrast to tigerfish in the Upper Ogon River, Nigeria, that all the tigerfish examined which had food in their stomachs, contained fish, and that Ephemeropteran and Tricopteran nymphs were sparingly included in their diet (Adebesi, 1981). The families Notonectidae and Corixidae (order: Hemiptera) inhabits clear slow flowing water, where they congregate, and for this reason seems to be favoured by the tigerfish. The Chironomidae family of the order Diptera were mostly preyed upon by tigerfish. They consisted of 55% of all the Diptera taken and seemed to be favoured by fish (Scholtz and Holm, 1985).

The presence of invertebrates in large (320 mm SL) tigerfish clearly indicates that there is indeed no clear cut change to an exclusively ichthyophagous diet. The change-over from plankton and other invertebrates to an almost exclusively ichthyophagous diet occurs when the young tigerfish are between 40 and 50 mm (IT). This is a very clear cut change with no intermediate stage of, for instance, insect-eating (Matthes, 1968). Lauzanne (1975) recorded a change over in diet at 50 mm, but noted that invertebrates are still taken by larger specimens although they form a minor part of the diet. Large tigerfish (300+ mm) were reported to feed on invertebrates during the hot dry months of the year (Jackson, 1961a). Deep pools with a lack of cover from plants and rocks for small fish certainly contributes to the fact that more invertebrates than fish are preyed upon. Small prey are quickly reduced in numbers once they are forced to enter the open pools. Tigerfish is then forced to prey on any available prey. Being predators, tigerfish is also opportunistic feeders and will sometimes prey on plant material which appears alive in currents or on the surface driven by wind.

49 The availability of prey in any given biotope determines to which extent it is preyed upon. Seasonal changes certainly played a role in the diet of the tigerfish in the Olifants River. Invertebrates as food item decreased from spring to winter which is in contrast to the findings of Jackson (1961a; 1961b) that the insect diet of the tigerfish increases as the volume of flowing water was enormously reduced in comparison to the flow during the rainy season. Fish were the most important food item from spring to late summer and invertebrates in the winter months in the Inkomati River when the water level has dropped and the small fish were consequently decimated (Gaigher, 1970). Small seasonal shifts in the diet of H. forskahlii were experienced in the Zambezi River floodplain (Winemiller and Kelso- Winemiller, 1994).

Specimens larger than 350 mm in the northern part of Lake Chad were entirely ichthyophagous (Tobor, 1972). Tigerfish in the Okavango and Kunene Rivers were exclusively piscivorous as fish was the only prey species preyed upon (Van Zyl, 1992; Hay, 1995). These findings indicates that fish as prey species is a preferred food item, if abundant and readily available, to the tigerfish. This is in contrast to the 14% of tigerfish which had fish as food item in the Olifants and Letaba Rivers in their diet. Factors such as limited prey size and availability contributes to the small number of fish being preyed upon by the tigerfish in the Olifants and Letaba Rivers.

The maximum prey size increases with the body length of the predator (Adebisi, 1981), which means that the larger the prey, the larger the predator has to be in order to feed in it. The impact of predation on fish were generally not more than 40 percent of the predator's length, although this was not always the case (Jackson, 1961b; Matthes, 1968; Kenmuir, 1973). This ratio would not exceed 50% of the predator's length (Jackson, 1961b; Gaigher, 1967). This was proven wrong when a prey species of 64% of the predator's length was reported in Lake Kariba (Matthes, 1968). Winemiller and Kelso-Winemiller (1994) found the predator prey ratio between 7% and 62% with a mean of 27% in the Zambezi River floodplain. Only one tigerfish in the Olifants River had a fish as stomach contents that could be measured. This fish was approximately 32% of the length of the tigerfish. Due to the lack of sufficient data, it was not possible to determine the exact relationship between the predator and prey lengths in the Olifants and Letaba Rivers.

50 Predation pressure were mainly caused by tigerfish with a length of 250-280 mm (SL) in the Olifants River. This is however much smaller than the 550 mm found in Lake Kariba (Jackson, 1961b). The majority of the tigerfish population in Lake Kariba is up to 580 mm after which the numbers decline, although the majority were in size range 150-350 mm (Kenmuir, 1973). One could therefore assume that the greatest predation pressure is caused by tigerfish in that size range, as is the case in the Olifants River. This would lead to fierce competition for food, especially on prey more or less 40% of the length of the tigerfish. The narrow pools and absence of vegetal cover for prey species in certain parts of the Olifants and Letaba Rivers increases the risk of prey species to be preyed upon by the tigerfish. All these factors leads to the fact that so many tigerfish which had been collected, had no stomach contents as all.

4.5 Conclusion

Tigerfish of all sizes in the Olifants and Letaba Rivers were found to feed almost exclusively on invertebrates. This is in contrast to'figerfish from other systems where fish plays a major role in their diet. The percentage of invertebrates as food item decreased from spring to winter in both the rivers. Nine orders of invertebrates were included in the diet of the tigerfish indicating that every available food source is exploited. Competition for food is fierce, especially in the length groups up to 280 nun .

51 CHAPTER FIVE

GONAD DEVELOPMENT AND FECUNDITY OF THE TIGERFISH (HYDROCYNUS VIITATUS) IN THE LOWER OLIFANTS AND LETABA RIVERS, KRUGER NATIONAL PARK.

5.1 Introduction

In South Africa the tigerfish is mainly restricted to the Pongola River and the rivers in the Kruger National Park. In the KNP it seems if the most successful populations is confined to the Olifants River in the area t . 2 km upstream from the confluence with the Letaba River downstream towards the Massingire Dam (Chapter 3). During spring when water temperature becomes more favourable, potamodromic behaviour of tigerfish can be witnessed when thousands of tigerfish aggregate at natural migration obstructions under low flow conditions below the Olifants Wilderness Trail Camp. Although successful in this section of the river including the Letaba River below the Engelhardt Dam, little is known about the ecology of tigerfish in these systems. Surveys done by Pienaar (1978), and a study on the male reproductive physiology of tigerfish, juveniles were never found in these systems (Steyn, 1987). Although juveniles were never found, Steyn (1987) made the conclusion that tigerfish has spawned in a specific pool in the Letaba River approximately a kilometre upstream from the confluence with the Olifants River. This assumption was based on the fact that eight spent female tigerfish were surveyed exactly two weeks after a total of eighteen ripe females were sampled in that very same isolated pool. Ripe - running males were also sampled in that pool during these surveys. Spawning has thus occurred in the pool which is densely covered on the one side with Phragmites reeds. It would however appear that the circumstances for spawning and rearing of the larvae is not ideal which has a negative influence the survival rate of the larvae. This could also explain why Gaigher (1967) never found juvenile tigerfish in the Incomati River system.

52 The question now arises why tigerfish is so successful in the Olifants River in comparison to other rivers in the KNP, taking into account that the habitat of the Olifants River is much more degraded than rivers such as the Sabie River due to high concentrations of silt. Recruitment must take place for the tigerfish to be successful in the Olifants and Letaba Rivers. Gonad development and fecundity were therefore compared with other systems.

5.2 Material and methods

Tigerfish were caught in the Olifants and Letaba Rivers (Chapter 4). The gonads were visually inspected to determine the sex and stage of maturation. Gonad maturity was evaluated and then classified according to the gonosomatic index (GSI) of Balon (1971). The stages of gonad development and abbreviations are given below: I - Immature RR M - Ripe-running male A M - Active male RR F - Ripe-running females A F - Active female S M - Spent males R M - Ripe male S F - Spent females R F - Ripe female

Only the eggs of ripe-running females were prepared to determine the fecundity of the Letaba River tigerfish. The ovaries were opened and the eggs were fixed in Gilson's fluid and stored in separate plastic containers at 25°C. The standard length, gonad mass and state of development were recorded. After a minimum fixation period of two months the eggs were shaken to remove excessive ovarian tissue. The eggs were then washed with running water and dried in a oven for three hours at 60 degrees Celcius. Five hundred eggs were then counted and then weighed to an accuracy of four decimals. The remainder of eggs were also weighed and by extrapolation the total egg count was calculated for each of the relevant specimens.

53 5.3 Results

The sex ratio for all seasons favoured the males (7.1:1), although the ratio is much higher (9.9:1) during the breeding season (September - February). A breakdown of the different gonad maturation classes shows that a close similarity exists between the Olifants and Letaba Rivers, in both the development of the gonads throughout the different months of the year, and especially the number of ripe-running males, although much more fish were collected in the Olifants River (Tables 5.1 & 5.2). Gonad development is progressively evident from July until February, which coincides with the breeding activity of the tigerfish (Table 5.3). The M:F ratio differs somewhat in the Letaba and Olifants Rivers. In both rivers the males are much more abundant than the females. The M:F sex ratio of tigerfish in the Letaba River is 5.1:1 and 8:1 in the Olifants River respectively.

Sexually active fish were found throughout the year although the greatest number of ripe and ripe- running fish were found during the months September to February, indicating the breeding season (Table 5.3). The majority of the males (86%) were ripe-running when captured, especially from October to November, indicating sexual maturity before the rains started. During this study, only five ripe-running females were caught below the Engelhardt Dam in the Letaba River in January 1992 with a mean standard length of 297 mm. No ripe-running females were caught in the Olifants River. Very small quantities of eggs was released from the urogenital pore when gentle pressure was applied to the abdomen. It is evident that the number of eggs increased with length (correlation coefficient 0.918) and fecundity also varied amongst fish of the same length (Table 5.4). The minimum lengths of maturity was 251 mm for the females and 122 mm for the males.

54 Table 5.1. Gonosomatic Index (GS!) of H. vittatus in the Olifants River.

Month Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May in

CSI

I 1 1

IAM 16 2 1 19

IAF 15 10 6 1 32

AM 5 8 13

AF 11 1 12

RM , 6 1 1 8

RF 1 1 1 3

ARM 2 2 191 64 12 26 33 330

RRF 0

SM 1 9 1 1 2 14

SF 1 1

ir 03 41 9 26 200 65 12 28 37 0 0 2 433

Table 5.2. Gonosomatic Index (OSI) of H. vittatus in the Letaba River.

Month Jun Jul Aug Sep Oct Nov Dcc Jan Feb Mar Apr May o GM

I 2 2

IA M 5 4 9

IA F 2 7 2 1 12

AM 5 5

AF 3 2 5

RM 1 1

RF 1 1 2

RR NI 41 10 68 1 120

RRF 5 5

SM 1 1

SF 1 1

n •• 0 4 17 3 49 11 70 6 1 2 0 0 163

55 Table 5.3. Gonosomatic Index (GSI) of H. villains in the Olifants and Letaba Rivers.

Month Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May a

CS1

I 1 2 3

IA M .16 7 5 28

IA F 17 7 10 8 2 44

AM 10 8 18

AF 14 1 2 17

R M 6 2 1 9

R F 1 2 1 1 5

RR M 2 2 232 74 80 27 33 450

RR F 5 5

S M 1 10 1 1 2 15

SF 2

n= 3 45 26 39 249 76 82 34 38 2 0 2 596

56 Table 5.4. The fecundity of H. villains in the Letaba River, Lake Kariba, Pongola and Okavango River.

Length (mm) Mass (g) Eggs (n=)

Letaba River (Standard length)

251 293 85400

255 280 63400

292 437 330300

319 615 476000

370 827 485100

Lake Kariba (Fork length)

311 325 78750

470 1700 593400

540 2722 672400

550 3100 827650

562 3100 742800

566 3000 625000

625 4725 580500

629 4425 724500

685 6123 779590

Pongola River (Fork length

510 306100

525 477500

630 873500

Okavango River (Fork len h)

400 74612

520 35735

57 5.4 Discussion

There seems to be a major influx of tigerfish into the Olifants River during the breeding season from the Massingire Dam situated in Mozambique. A dramatic deviation from an expected 1:1 male to female ratio was evident, which favoured the males in the Olifants (8:1) and Letaba Rivers (5.4:1). The high sex ratio in which the males are favoured is as a result of stress. More males than females are produced when a population is under stress (Holtzhauzen, 1989). Various stress related factors which pushes the sex ratio towards the males are dealt with in Chapter 3. Although males were favoured (1.8:1) in Lake Kariba, is it closer to the expected 1:1 ratio and can be expected as normal (Langerman, 1984). The sex ratio favoured the females (8:1) in the Incomati River system (Gaigher, 1970). This is in direct contrast to the sex ratio of the tigerfish in the Olifants and Letaba Rivers since the Incomati River system is also a system that flows into Mozambique.

The sex ratio increased in favour of the males during the breeding season in both the Olifants (9.1:1) and Letaba (7.1:1) Rivers. An influx of males prior to females into the rivers from the Massingire Dam in Mozambique could also be a factor to explain the high male:female ratio. Males were more abundant (4:1) than females during the peak breeding season which changed to females (1:1.35) during the non-breeding months in Lake Kariba (Kenmuir, 1973). The high male sex ratio would however ensure fertilization of the eggs during the spawning periods, especially in the Olifants and Letaba Rivers. The expected 1:1 sex ratio was found in the Okavango River (Hay, 1995), although the females were slightly favoured during the winter of 1992 which is similar to the findings of Kenmuir (1973) in Lake Kariba. Females were however favoured during the summer and winter in the Okavango River (Van Zyl, 1992). A shorter life span and higher mortality rate of males pushes the ratio towards the females (Kenmuir, 1973; 1984).

A definite progression of tigerfish gonad maturity is evident towards the summer rainy season in the Olifants and Letaba Rivers. Gonad development of males and females were found to be very similar except for the ripe-running males. The amount of ripe-running males which were collected in the Olifants River, emphases again the influx of pre-breeding tigerfish prior to the first rains.

58 The majority (94.3%) of the males were ripe-running and were emitting semen, when handled. A small number of tigerfish were caught during August 1992 and were already ripe-running indicating sexual maturity prior to the first floods. Bowmaker (1973) also found that the male tigerfish were sexually precocious and were emitting milt while migrating in the company of pre- breeding females.

Males were found to reach sexual maturity prior to females (Kenmuir, 1973; Langerman 1984), and that a proportion of the population ripens a considerable time before the breeding season which commences with the spawning migrations of the tigerfish. Spawning migrations commences prior the first floods in the Olifants and Letaba Rivers. The spawning period seems to be of short duration (Chapter 6) This would ensure the tigerfish to spawn immediately after a suitable habitat and partner had been found. Jackson (1961a) also mentioned the spawning period of the tigerfish to be of extremely short duration.

Only after the first floods are the tigerfish able to migrate into the Letaba River. This does not however mean that the tigerfish do not spawn in the Letaba River, as small tigeifish had been sampled in that river. Tigerfish of 85 mm (SL) were collected at the confluence of the Olifants and Letaba Rivers and above the natural obstruction in the Olifants River, indicating that spawning migrations had taken place. Gaigher (1970) however reported that tigerfish migrates downstream to spawn in the flooded plains of the lakes in Mozambique in the Incomati River system after the first floods. Hydrocynus brevis were also reported to spawn in flooded plains (Daget, 1954).

The mere presence of a great number of ripe-running tigerfish prior to the first floods in the Olifants River, indicates that tigerfish migrates upstream to spawn. A large concentration of tigerfish was noticed two months prior to the fist rains. The presence of tigerfish concentrating in certain areas prior to the first rains for spawning migrations was recorded by Jackson (1961a) and Bowmaker (1973). Van der Waal (pers comet) found a small tigerfish (20 mm) in a rock pool next to the main stream in the Kavango River close to Bogani, during October 1977. On January 7, 1974 he also found tigerfish of 30-50 mm in length 1.6 km North of Katima Mulilo in a small stream from a "mulapo" (flooded vlei) in the Zambezi River, indicating that the tigerfish had waited for the first rains to spawn. Small tigerfish between 120-160 nun TL were also caught in

59 June 1992 by the author in an isolated pool measuring 20m x 10m in the Luvuvhu River in the KNP, indicating that breeding took place the previous season, i.e 1990/91. An abnormal spawning time for the tigerfish was reported by Jackson and Rogers (1976) when seven two month old tigerfish were collected in the a shallow weedy bay adjoining and connected to the Zambezi river. A spent female was collected in the Letaba and Olifants river respectively during July, which could also indicate an abnormal spawning time for the tigerfish.

Badenhuizen (1967) found that 93% and 99% of the active and spent fish respectively were males, which is similar to the 94.6% active, ripe and ripe-running, and 88.2% spent males respectively in both the Letaba and Olifants rivers. Forty percent of the males were found to be sexually active at a length of 240 mm in Lake Kariba (Langerman, 1984) which is slightly larger than the 200 mm (FL) of the males in the Incomati River system, (Gaigher, 1970). This compare fairly well with the respective findings of Van Zyl (1992) and Hay (1995) who reported minimum lengths of 170 mm (FL) and 180 mm for male tigerfish in the Okavango River. A sexually mature male was collected at a minimum standard length of 122 mm and only 54 g in mass, which is much smaller than the minimum lengths recorded in the Okavango River. However males reaches sexual maturity at an average standard length of (265 mm). This is smaller than the findings of Kenmuir (1973) and Griffith (1975) who found the males to be sexually active at a length of 300 mm FL and 390 mm respectively in Lake Kariba and Lake Bengweulu respectively.

Females seems to reach sexual maturity later in life, as was confirmed by Kenmuir (1973) and Langerman (1984). This means that they would be larger than the males when reaching sexual maturity. A number of tigerfish were collected by the author in the Crocodile River during December 1992, and it was noted that the sexually active females were larger than the males. The average standard length (284.6 mm) of females, were found to be slightly larger than the males (265.7 mm) when reaching sexual maturity in both the Olifants and Letaba Rivers. Kenmuir (1973) noted females to reach lengths of up to 700 nun (FL) and more, while the males hardly grows to lengths beyond 500 mm (FL).

The minimum length at sexual maturity for females in the Olifants and Letaba Rivers is 251 mm (SL) although the average standard length is 284 mm. Females reaches sexual maturity at 360 mm

60 (FL) in the Incomati River system (Gaigher, 1970) which is higher than the minimum lengths in the Olifants and Letaba River. A very large minimum length (420 mm FL) at which females reaches sexual maturity in the Okavango River was reported by Van Zyl (1992) which differs considerably from the 275 mm (FL) (Hay, 1995). The length given by Van Zyl (1992) can certainly not be used as the minimum average length of females reaching sexual maturity in the Okavango River, as that was the only active female sampled. The findings of Hay (1995) is therefore more acceptable. Fifty percent of the females in Lake Kariba reach sexual maturity at 260 mm (Langerman, 1984), which is a bit smaller than the above findings. Steyn (pers comm) found the tigerfish to be very fecund, with great numbers of eggs. Tigerfish collected in the Letaba River produced a great number of eggs. An average amount of 588 eggs per body mass(g) was calculated for the tigerfish collected in the Letaba River, which is higher than the 194 eggs/body mass (g) in Lake Kariba (Langerman, 1984). This indicates that tigerfish is slightly more fecund in the Letaba River than Lake Kariba. Great fecundity was also reported by Pott (1969), Kenmuir (1973) and Hay (1995). A large specimen which was examined by Langerman (1984) proved to have almost a million eggs. This would ensure the survival rate of the species as mortalities of the eggs and larvae are expected to be extremely high (Chapter 6).

On one occasion a large ripe-running female was sampled by the author in the Luvuvhu River, which is situated north of the Letaba River. This specimen was collected during the drought period of June 1992 and the ovaries were filled with atrophied eggs. It could be an indication that breeding did not take place that season (1991/92) as a result of the drought, and no river flow. Tigerfish with atrophied eggs was also sampled by Bowmaker (1973) and Langerman (1984) which attributed it to no river flow.

5.5 Conclusion

Gonad development of tigerfish in the Olifants and Letaba Rivers showed a clear trend throughout the different seasons. The development of gonads were similar for tigerfish in the Olifants and Letaba Rivers. Males were favoured throughout the year, resulting in a deviation from the expected M:F sex ratio. The deviation of the sex ratio is a result of stress in the population due

61 to various factors. The minimum length at which females reach sexual maturity is larger than the males which would ensure greater fecundity. The large number of ripe-running males would in turn ensure fertilization and thus the survival of the tigerfish.

62 [CHAPTER SIX

INDUCED REPRODUCTION AND DEVELOPMENT OF THE TIGERFISH HYDROCYNUS VITTATUS (CHARACIDAE) EMBRYOS AND LARVAE

6.1 Introduction

The tigerfish, Hydrocynus vittatus is a ferocious predator, almost entirely piscivoric and has a well- deserved reputation as one of the world's most spectacular freshwater game fish species. This species is a member of the Characidae (Order Characiformes) one of the largest families of freshwater fishes found in both Africa and the Neotropics (Brewster, 1986). The family includes many well known species such as the attractive neon tetras (Skelton, 1993). The genus Hydrocynus is represented by six species, all endemic to Africa. They are pikelike predators, commonly termed `tigerfishes' for their prominent dentition and dark lateral stripes. Most of the species reach 40 or 50 cm, or even 68 cm for Hydrocynus brevis. The record size is held by Hydrocynus goliath, which can attain a length of 150 cm and a weight of 50 kg (Gery, 1977). In Southern Africa Hydrocynus vittatus occurs in the Okavango River, Zambezi River and lowveld reaches of coastal systems south to the Pongola River (Fig. 6.1). It is also present in the Zaire, , Rufigi and large Nilo-Sudanian rivers in North and (Skelton, 1993). The tigerfish is a lowland species that is not tolerant of cold water and migrates downstream to lower lying reaches of river systems during winter where water temperatures are higher and more stable (Van Loggerenberg, 1983). As a result of the increasing construction of dams, weirs and reduced flow conditions in many rivers, the fish migrating upstream during the summer months are effectively prevented by these man-made obstructions from reaching their original summer distribution areas. In winter large scale fish mortalities above weirs and dam walls in the Incomati and Pongola Rivers have been observed due to the inflow of cold winter water. On account of the low level of the rivers and the absence of fish ladders, the fish could not negotiate the weirs to migrate downstream to avoid the cold water and consequently died of cold (Van Loggerenberg, 1983). As a result of this, populations above migration barriers could not be

63

Loggerenberg, 1983). As a result of this, populations above migration barriers could not be repopulated and tigerfish south of the Limpopo are now confined to lower reaches, occasionally higher than 300 m above sea level (Gaigher, 1969).

Early reports on the distribution of tigerfish in South Africa mentioned their presence at altitudes greater than 300 m in the Transvaal, Swaziland and KwaZulu-Natal. They have been found in the Pongola River up to the confluence with the Pivaan (Fig. 6.1) at an altitude of 457 m (Crass, 1964). This, however, was before the erection of the weir at Grootdraai approximately 32 kms above Pongola Town (Pott, 1969). During the years 1955-1961, Harry Wolhuter, who was game ranger at Pretoriuskop in the Kruger National Park, at that time used to catch tigerfish in the Sabie River to the west of Hazyview near to what is presently known as Sabie River Bungalows (Fig. 6.1). Tigerfish were also frequently found at Mica in the Olifants River outside the Kruger National Park. In 1943, U. de V. Pienaar (former head of the Kruger National Park and National Parks Board) witnessed an angler catching a tigerfish in the Greater Letaba River at Tzaneen (pers comm). The upstream distribution in the Limpopo River has always been limited by waterfalls to the east of Messina (Gaigher, 1969). The construction of large irrigation dams such as the Fanie Botha Dam in the Letaba River (Fig. 6.1) near the border of the original summer distribution range, rendered the opportunity of establishing tigerfish populations which will be able to survive the winter outside the borders of the Kruger National Park. The large volume of water in such a dam provides a more temperature stable environment during winter as opposed to the original river where only river pools would have occurred (Van Loggerenberg, 1983). Several smaller irrigation dams on private game farms adjacent to the Kruger National Park also provide suitable habitat for the survival of tigerfish in winter. The Transvaal Directorate, Nature and Environmental Conservation has for many years envisaged a project to restock traditional tigerfish waters and establish populations within suitable irrigation dams (Gaigher, 1969; Van Loggerenberg, 1983). Such a project never materialized because an abundant source of small tigerfish was not available in South Africa from where restocking could be done. Due to the sensitive nature of this species to handling, transporting tigerfish over long distances from neighbouring countries was not attempted. The fish would have to be restocked every year due to its inability to breed naturally in such waters. According to Gaigher (1967) who did an

64 ecological survey in the Incomati River System, tigerfish did not breed in South Africa and had to migrate to the lowlands of Mozambique were they presumably spawn during December and January along the shallow, grass covered fringes of lakes and small streams (Pienaar, 1978).

The possibility of supplying tigerfish for restocking from an artificial breeding programme was investigated by Gaigher (1969) and Van Loggerenberg (1983). Several attempts to spawn tigerfish artificially were unsuccessful due to a lack of knowledge on the reproductive biology of this species, as well as many logistical problems. During a study on the male reproductive biology of tigerfish (Steyn, 1987), the major factors associated with its artificial breeding were identified. It was found that the successful artificial spawning of the tigerfish was being hampered by factors such as unsynchronized maturation of both sexes, discrepancy between the number of mature males and females; shortened breeding seasons and restricted access to breeding stock in general (Steyn & Van Vuren, 1991). It is possible to overcome some of these problems by gamete preservation or administration of a slow release pellet containing gonadotropic releasing hormone (Crim & Glebe, 1984). The latter option did not seem a viable proposition due to the sensitive nature of wild tigerfish which is highly stressed during handling in captivity, especially over prolonged periods (personal observations). To facilitate spawning synchrony and gamete availability, a technique was developed for the cryopreservation of tigerfish spermatozoa (Steyn & Van Vuren, 1991). Spermatozoa can now be thawed when it is required for artificial insemination during an artificial breeding programme.

The objectives of this section of the study were to find a successful technique to artificially reproduce this species and to document embryonic and larval development. Artificial breeding will lead to the establishment of a domesticated broodstock which in turn will resolve the problems associated with artificial reproduction of wild tigerfish. Furthermore, knowledge of artificial fertilization, embryonic and larval development and behaviour will reveal some aspects of the reproductive strategy of this species in its natural environment.

65 6.2 Material and Methods

6.2.1 Selection and handling of broodstock In order to distinguish between males and females to facilitate broodstock management during an artificial breeding programme, tigerfish from the upper Zambezi River and rivers of the Kruger National Park were inspected for external morphological differences.

Sixteen brood fish were collected with angling equipment during December 1992 in the Crocodile River, Kruger National Park. The fish were transported to Skukuza in a 10001 container filled with river water to which an anaesthetic, 2 phenoxy ethanol was added. The water was oxygenated with medical oxygen. At Skukuza the fish were transferred to a 2 m x 1.5 m Hydrex portable pool, filled with river water. The water was recirculated continuously with a submersible pump and the surface of the dam was covered with shade net to stabilize water temperature at 28°C and to prevent fish from jumping out. An acclimatization period of two weeks was allowed prior to breeding experiments. A third of the water was replaced every consecutive day to maintain water quality. Live tilapia was used as food.

6.2.2 Artificial stimulation Five mature females were selected from the acclimatized fish and transferred to separate holding nets (1 m x 0.5 m x 0.8 m) suspended in another 2 m diameter Hydrex portable pool. Only one mature male was available and was isolated accordingly. Holding nets were also covered using shade net to prevent fish from jumping out.

Fish to be treated were anaesthetized with 2 phenoxy ethanol (0.3 ml r' H20) prior to hormonal administration. Each of the selected broodstock was injected intramuscularly, anterior to the dorsal fin. One of the females was treated with a combination of human chorionic gonadotropin (HCG) and catfish pituitary extract (CPE). Each of the remaining females were treated with a cocktail of gonadotropin-releasing hormone (GnRH) and a dopamine receptor antagonist. Pimozide (Sigma P1793) and GnRHa, Des-Gly l°, [D-Ala6] Gn Rh-ethylamide (Sigma L4513) were suspended and

66 dissolved in a 0,8% saline vehicle containing 0,1% sodium metabisulphite. An analogue of salmon gonadotropin-releasing hormone (SGn Rita) with a sequence of Gly 6 Trp7 Leui (Sigma L4897) was prepared in the same manner. Mother product, Aquaspawn (Spawnrite Ltd) contains D-Lys 6-Trp7- Tyra-Gil RH and dopamine receptor antagonist, domperidone. A relatively unknown product, Lymnospawn (All Seasons Aquatics) which contains D-SerBut 6 Gn RH ethylamide, reserpine, deoxycorticosterone and prostaglandin 2, was used to treat the fifth female. The single male was treated with CPE to increase semen volume. The details of treatment are summarized in Table 6.1. Experimental design was not planned to evaluate each of the above products against each other but merely to establish a suitable technique from where further experimentation can be planned.

6.2.3 Artificial insemination and hatching Prior to stripping the broodstock, fish were anaesthetized with 2 phenoxy-ethanol (0.3ml t' H20) and dried with a wet cloth to prevent contamination of the gametes with water or faeces. Eggs were stripped into a round bottomed plastic bowl and divided into two separate batches awaiting insemination. Semen from the mature running male was stripped into a 50 ml glass beaker when needed. Artificial insemination was performed with each of the two batches of eggs according to the wet and dry fertilization techniques respectively (Steyn el aL 1989). River water (28±1°C) was used as activating medium and two minutes was allowed for fertilization to occur as tigerfish spermatozoa maintains motility for only 86 seconds (Steyn, 1993).

Inseminated batches of eggs were rinsed in river water, transferred into separate 350 ml glass incubation funnels and delicately kept in suspension by adjusting the gentle flow of water through the incubation funnel. Water to the incubation funnels (28±1°C) was gravity fed and drained from the funnels into a 2 m x 1 m portable pool (Hydrex). Water was recirculated with a submersible pump to a header dam (2 m x 1.5 m) from where water was supplied to the incubation funnels.

67 6.2.4 Development of embryos and larvae Embryo development was observed and photographed in hourly intervals by means of a Nikon

Stereoscopic Zoom microscope (SMZ-10) fitted with a Nikon 8301 camera. After hatching, free embryos were transferred to glass aquaria for observation and rearing. Development of larvae was photographed daily for a period of six days.

6.2.5 Rearing of larvae and juveniles Larvae were fed on a liquid suspension of cooked hens egg albumen, supplemented occasionally with infusoria. As the larvae became larger, the diet was switched to Anemia nauplii and Daphnia. Juveniles were fed on mosquito larvae until they were large enough to consume shredded ox liver. When the fish attained a size of approximately 6-8 cm fork length, they were transferred to 1000 litre aquaria for grow-out as part of a domestication program.

6.3 Results and discussion

6.3.1 Selection and handling of broodstock Tigerfish displayed distinct colouration differences between mature males and females in the upper Zambezi River. Mature males from the upper Zambezi River have a very prominent yellow colouration in all of the ventral fins and dorsal two thirds of the caudal fin. The ventral section of the caudal fin has a bright red colouration. A faint intermittent black line stretches across the fin rays of each of the ventral fins. Mature females from the upper Zambezi River have orange fins and resemble the colouration of both sexes from the Kruger National Park. It is not possible to distinguish sexes on an external morphological basis other than the unique and most spectacular colouration which is present within the upper Zambezi variation. From the studies by Balon (1971) and Kenmuir (1972) in Lake Kariba as well as personal observations, it is concluded that this unique colouration is not present in tigerfish from Lake Kariba which might indicate that tigerfish cannot survive the fall over of the Victoria Falls and possibly gave rise to a distinct gene pool above the falls. In the Pliocene the

Upper Zambezi River and its probable tributaries, the Upper Kafue-Kalomo systems flowed south

68 together with the present Chobe-Okavango into the Ngami-Makarikari Lake and Limpopo Valley. The middle Zambezi was then a separate river with headwaters somewhere in the region of the present Matetsi River. A tectonic upwarp in the Caprivi area broke the connection of the Upper Zambezi with the Ngami-Makarikari drainage and the waters of the former escaped eastward along a depression and spilled over the sheer drop at the old Zambezi headwaters (Salon & Coche, 1974). The absence of the unique morphological variation in the Limpopo system and Middle Zambezi indicate that these characteristics have developed after the Upper Zambezi River was disconnected from the Ngami-Makarikari drainage and after formation of the Victoria Falls. A genetic investigation revealed a low genetic variability in the tigerfish population from the Upper Zambezi River. The low amount of genetic variability in the former population compared to that of other fish species from the same geographical area, and to that of H. vittatus from the Limpopo system, can be attributed to restricted gene flow due to isolation (Kotze et al, 1996).

Successful artificial insemination of fish depend to a large extent on the selection of suitable broodstock, each sex to be treated according to its own protocol and requirements. This is especially true of tigerfish where broodstock from the wild has to be transported over long distances, often inaccessible terrain, thus limiting the transport capacity. Under these conditions, it is advisable to restrict the number of males to the minimum to provide adequate space in order to ensure successful transport and keeping of a fair number of egg donors. It must be kept in mind that only a minute volume of semen is needed to achieve sufficient fertilization as tigerfish has a mean number of 11.9 x 106 spermatozoa for each micro litre of semen (Steyn, 1993).

6.3.2 Induced spawning During this investigation viable eggs could not be stripped from any fish treated according to GnRH and dopamine antagonist protocol. Furthermore, ovulation was partial with at least 28 hours reaction time after hormonal induction (Table 6.1). The slow response after treatment was probably not due to insufficient hormonal treatment as the concentration of GnRH and dopamine receptor antagonist was relatively high compared to results obtained with other species. Fertile eggs could only be stripped from one female which was treated according to the CPE and HCG protocol. This female

69 Table 6. 1. Hormone application schedule of H. vittatus broodstock

Broo dstock Ma le 1 • Fema le 1 Female 2 Fema le 3 Female 4 Female 5 -- cr c .-. -Tr tr. :Tr tn a, VI" c.0 , ,r VI' ■ Tr Le. '.... VI VI Length of Broo dstock 37/43 ■ 01/4 FUT!. (cm) ,— N c: — — c, , .--. VD en , CO CI T:r We ight o f Broo dstock (g eV VD CV

) :7- 0 a 0 0 cc 0 C., c eu -.- .7, + C a 6 7,, 0 1:4 .c t te. a 1 o ,,, >, - 17 I c a - ce .c T c" a. >. c ,

. - 3 CPE + F1CG . ,

6 7 Gly Trp Leu'-Gn Rh D-Ser- But6-Gn Rh

(salmon ) Deoxy corticosterone

Prostaglandin 2 Z ._ a. g a) 3 ' :2 C. E . E . Dopamine recep tor None c Domperiodone Reserp ine. an tagon ist < -7 ill r c. c c r;" Name of Product/s Catfish P itu itary Extrac t Catfish Pituitary Ex tract Sigma E re tn. 1.4513 Sigma L4897 " 0, Human Chorionic human Chorionic 6

& Sigma Gonadotrophin (Pregnyl) Gonadotrophin (Pregnyl) Sigma ..n ci ! =

Dosage o to E f first trea tment 250 iu.11CG fish 250 iu Pregnyl/kg 100 ug/kg Gn Rh 100 ug/kg Gn Rh I ml/kg 44 n

(zero hours) 4 p c g itu itary/fish 1/4 pitu itary/kg 10 mg/kg Pim 10 mg/kg Pim Unknown

Dosage and time of secon d None 1 pituitary/kg at 9: 00 Same as above Same as above ' Same as above Not trea ted trea tment (time after zero at 28 hours at 28 hours a t 28 hours hours)

Time to stripping a fter 18 hours 1 8 hours No spawn 28 hours 28 hours 28 hours 28 hours first trea tmen t 40 hours 40 hours 40 hours 40 hou rs

Extent of spawn 2 ml Complete No spawn Partial (2m1) Partial ( 15m1) Partia l (2m1) 4 ml ( 180 ml) Partial ( 15m1) Partial (20m1) Complete ( 120m1) 2 ml 0 .— — —:. 41 E Approximate number ?.. E c ± 342 000 eggs No spawn 13800 egg s 12 8500 eggs a) 13800 eggs of gametes 128500 eggs 138000 6) 1228000 eggs

> 80% g rade 5 No spawn Not fertile, Not fertile, Not fertile, motility con taminated contaminated with con taminated with with blood blood blood responded in a relatively short time (18 hours) after treatment (Table 6.1). It must, however, be emphasised that these results are preliminary and should be investigated further once a domesticated stock has been established.

Investigations into the structure and synthesis of gonadotropin releasing hormones (GnRH) of vertebrates (King & Millar, 1980) and the mechanism by which their action is potentiated (Peter et a!, 1987) has lead to exciting prospects in the culture of fish species. Although several fish species have been spawned successfully with this technique, further research is needed, in particular, optimal application protocols for each species (Carolsfeld et a!, 1988). However, before the combination of dopamine antagonist and GnRH can be accepted as useful in applied practice for induced spawning of cultured fish, the new procedure must meet certain criteria: a high rate of ovulation occurs consistently from one group to another within each species, ovulations are complete rather than partial, the time to ovulation following injection is short and predictable, the ovulated eggs are fertile and viable, and induction of ovulation by this technique does not affect subsequent reproductive cycles of the same brood fish (Peter et al, 1987).

6.3.3 Artificial insemination and incubation

The eggs from all the females had a soft turquoise blue colour. Counts revealed that tigerfish have a mean number of 1900 eggs ml'' sample. The single female which responded by complete egg release, delivered 180 ml which by extrapolation contained 342 000 eggs. Based on the number of eggs obtained from the latter female, the fecundity was calculated as 201 eggs body weight. These results confirm the very high fecundity also reported by Pott (1969) and Kenmuir (1972). The mean number of eggs calculated by utilizing Kenmuir's data reveals a mean number of 219 eggs g"' body weight (S13±73). This is however much smaller than the 588 eggs g-' for tigerfish in the Letaba River as a result of hydration of the eggs when ripe-running.

Tigerfish eggs are small with a mean diameter of 0.65 mm (n=11) after water hardening. The eggs are slightly adhesive and negatively buoyant. Hatching success was approximately 40% but a large number of embryo's were trapped and died as a result of coagulation caused by the slight adhesiveness

70 Figure 6.2 Embryonic development of H. vittatus : PV = perivitelline space, BD = blastoderm, BO = cytoplasmic blowout, EA = embryonic axis, OG= oil globule, BP = blastopore, HD = head, TL = tail, OV = optic vesicle, SM =somite, TB = tail Figure 6.2. Embryonic development of H. vittatus (continued): AN = anus, GL = globule, AV = auditory vesicle. solution (Schoonbee & Prinsloo, 1984) in order to remove egg adhesiveness prior to funnel incubation. Alternatively, a nylon screen with appropriate mesh size can be used as an incubation substrate without removing egg adhesiveness.

6.3.4 Development of embryos and larvae Development are presented in figure 6.2 a-n. The initial stages of cleavage were not photographed due to care that had to be taken to ensure faultless operation at the onset of incubation. The very small (0.65 mm) egg has an average sized perivitelline space (yolk diameter is 75% of total water hardened egg diameter). A small perivitelline space would indicate that the eggs are not supposed to be bounced along a rocky river bed (Cambray & Meyer, 1988). The morula stages were attained ±3 h after fertilization and developed into the blastula at approximately 4 h (Fig. 6.2a). The blastoderm occupied nearly half of the yolk surface at this stage. At 8 h after fertilization, gastrulation has resulted into formation of the embryonic shield and successive stages followed to establish the embryonic axis after 10 h (Fig. 6.2c). Before the embryonic axis is formed, a cytoplasmic blow out at the blastopore into the perivitelline space indicated the cessation of embryonic development of unfertilized eggs at 9 h (Fig. 6.2b). Until this point, spawning success cannot be predetermined as it is known that fish eggs can develop spontaneously irrespective of fertilization. Parthenogenetic development proceeds to the morula or gastrula stages after which decomposition follows (Withler, 1980).

From 11 h onwards an embryo body becomes discernable which encircles approximately 3/4 of the egg yolk circumference (Fig. 6.2d-g). Nine somites and the optic vesicles were visible at 14-15 h (Fig. 6.2e). The auditory vesicle is situated anterior of the somites and gradually shifts towards the head as embryonic development progresses. The tail bud dislodged at 15-17 h and the first body movement was visible as irregular twirling twitches (Fig. 6.2g). The onset of cardiac pulse (62 min -I) was observed at 20 hours. First hatching took place at 22 h 30 min after insemination (07 h 30 min 10th December 1992) and continued for ±3 h (Fig. 6.2h). Newly hatched embryos were 2.9 mm (TL) (n=2), the yolk was now ellipsoid in form and occupied ± 43% of the body length (Fig. 6.2i). An oil

71 globule became visible within the yolk as embryos grew longer and the yolk diminished in size (Fig. 6.2k). The oil globule remained and was still visible at the change to exogenous feeding (Fig. 6.21-n).

6.3.5 Free embryo and behaviour of larvae As soon as tigerfish embryos hatch, they start with a continuous vertical migration which last for two days. Free embryos actively swim to the surface, break the surface, inactively drop to the bottom and after benthic collision, the upwards migration immediately starts again. This monotonous motion is continuous, displaying no diel photokinetic periodicity. The vertical migration probably serve the purpose to place the free embryo in suspension within the water column to accomplish transport by drift to reach a planktonic soup with appropriate food particles. Vertical migration of free-embryos was also observed in the walleye, Stizostedion vitreum but they were able to retain their positive phototaxis and surface-suspension-ability until the water was agitated. Turbulence would cause the free embryos to loose its surface suspension, subsequently falling to a lower stratum and requiring vertical swimming to regain its surface position (Mc Elman & Balon, 1985). According to Kenmuir (1972), tigerfish young displayed photokinetic periodicity in Lake Kariba. The young were found to inhabit the surface waters during the day and to descend to greater depth during the night. This was established by towing a circular plankton net over a period of 66 h. During our investigation, photokinetic periodicity was observed from 4 days after hatching and seems to be related to buoyancy control. Larvae became inactive at night, sinking lower in the water column.

At three.days after hatching, free embryos are able to regulate buoyancy and to maintain position within the water column. The ability to maintain buoyancy within the water column seems to be associated with the emergence of an oil globule (Fig 6.2k) at 72 h after hatching. In Stizostedion vitreum the ability of free embryos to remain in motionless suspension at the surface is also associated with an oil globule. Interestingly the embryos initial inability to overcome the buoyancy of the large oil globule results in a resting position with the ventral side up (Mc Elman & Balon, 1985). The yolk of tigerfish embryos are almost completely absorbed after four days (96 h after hatching) and change to exogenous feeding occurs between four to five days (4-5 mm it) (Fig. 6.2n). Ten days after hatching, larvae were large enough to consume Anemia nauplii non-selectively. From the change to

72 exogenous feeding up to 12 days after hatching, larvae feed opportunistic by swimming vigorously in the area where the food is concentrated. Selective feeding becomes evident from thirteen days after hatching at a size of approximately 8 mm (TL).

From the feeding behaviour, it is evident that tigerfish larvae are dependent on a plankton rich nursery area which must contain a substantial component of the smaller planktonic organisms. The mode by which tigerfish larvae prey on plankters is accurately described by Bowmaker (1973) and was also confirmed during our study - 'on approaching a small plankter a larva would stop one to two millimetres away, deliberately coil its body into an S-shape on the horizontal plane, at all times keeping the plankter directly in front of the head, and then lunge at the plankton in a single spasmodic movement. In many instances the lunge was unsuccessful, due either to its misdirection, rapid avoidance by the plankter or the plankter being too large for the mouth of the larva.'

6.3.6 Juveniles Metamorphosis from larva to juvenile form was visible after 17 days after hatching at a total length of ±10 mm. The characteristic tigerfish shape and fin configuration was visible at this stage. Upon feeding with Anemia nauplii and plankton, juveniles went into a feeding frenzy, feeding ferocious and selectively, expelling suspensoids which were not edible. The food and feeding habits of juvenile tigerfish are well documented (Gaigher, 1967; Matthes, 1968; Kenmuir, 1975). The sides of the fish are silver-gold in appearance and it has adult-like eyes by nineteen days after hatching. A faint red dot is visible at the apex of the mandible and transition from conical to functional dentition has already taken place by 45 days after hatching at a length of ±40 mm (TL). However, it should be noticed our results are based on macroscopic observation in an aquarium, the transition of juvenile to functional and replacement dentition is described in detail by Roberts (1967) and Brewster (1986). The parallel stripes along the sides which is characteristic of H. vittatus gradually becomes visible from 72 days post-hatch at a size of approximately 50 mm (TL).

73 6.4 Conclusion Results showed that it is possible to induce release of gametes in H. vittatus by administration of HCG and pituitary extract. Gonadotropin releasing hormone revealed promising results but require further investigation to overcome the problems of non-viable eggs. Although the tigerfish has been spawned artificially, natural spawning has not been witnessed yet. According to Baton (1984; 1990), most fishes belong to the reproductive guild of non-guarding, egg scattering pelagic spawners which is characterised by small nutrient poor ova produced in high numbers, delayed embryonic differentiation and a long larval period terminated by metamorphosis. The walleye Stizostedion vitreum is an example of this reproductive style and can be classified as a non-guarding, egg- scattering lithopelagophil (Baton, 1990). The tigerfish conform to these characteristics but additionally has negatively buoyant eggs which are slightly adhesive for benthic or epibiotic incubation. From these results as well as field observations (Gaigher, 1967; Kenmuir, 1972; Steyn, 1987), it seems if tigerfish spawns on a sand substrate in the vicinity of aquatic vegetation.

74 [CHAPTER SEVEN'

TOOTH REPLACEMENT OF TIGERFISH HYDROCYNUS VI77'ATUS FROM THE KRUGER NATIONAL PARK

7.1 Introduction

Tooth replacement in characins is not an uncommon phenomenon, as it has been studied by Monod (1950), Roberts (1967), Kenmuir (1972), Gaigher (1975), Tweddle (1982) and Brewster (1986). It has been suggested by Roberts (1967) and confirmed by Kenmuir (1972) that the dentition of tigerfish is entirely conical. Conical teeth appear generally at a length of 14 mm and change to tricuspid dentition at a length of approximately 22 mm, followed by adult conical dentition. Fry are also fully scaled by 25 mm suggesting that development of tricuspid dentition begins when scale development has finished (Roberts, 1967; Kenmuir, 1972). Change to functional conical teeth occurs at about 32 mm standard length. Brewster (1986) also reported on transition from conical to tricuspid dentition at a length of 16-25 mm (SL). A specimen of 40 mm (SL) had conical teeth anteriorly in both jaws but each tooth had lateral cusps in the form of very small, nipple-like protuberances. The posterior teeth of this specimen were all tricuspid with well developed_ lateral cusps. In specimens larger than 50 mm SL, the lateral cusps on the anterior teeth are difficult to detect but the teeth at the posterior margin of each jaw are always tricuspid (Brewster,1986). Each tricuspident replacement tooth can be separated into three conical elements, thus suggesting that such teeth are formed by fusion of three "juvenile" conical elements. Five tooth replacement cavities are present on each side of both jaws. The anterior four each contain a single large tooth, whereas the fifth may contain one or two small, usually tricuspid teeth (Brewster, 1986).

Kenmuir (1973) reported anglers who had caught toothless tigerfish, and he himself had also caught a 30 cm toothless tigerfish in a gill net in Lake Kariba. This author also mentioned a tagged toothless tigerfish which had been recaptured several months later, still toothless. Tweddle (1982) observed a tigerfish in which the protruding teeth were not firmly imbedded in their

75 sockets, and could easily be moved by light finger pressure. It has been suggested that teeth replacement is quick and that teeth might be shed simultaneously (Kenmuir, 1973). The latter author speculated further that if replacement occurs on one side of the jaw only, followed by the other side, the teeth should show different degrees of wear on each side of the jaw. Gaigher (1975) reported a captured tigerfish with small teeth and concluded from X-ray evidence that the teeth were newly formed replacement teeth at the time of capture. All the teeth were in the same state of development, indicating that tooth replacement in both the upper and lower jaws occurs simultaneously. A tigerfish with small teeth protruding 3 mm from the gums had been collected by Tweddle (1982). This specimen was about 40 cm long and weighed 740 grams.

Analysis of stomach contents of a number of H. vittatus revealed teeth in some of the samples and many teeth were found on the bottom of an aquarium in which tigerfish were raised (Begg, 1973). How quick tooth replacement takes place was not known until a complete set of teeth was observed being replaced over a period of one month in an aquarium (Begg, 1973). However, frequency of tooth replacement is unknown.

During this section of the study, the frequency of tooth replacement was studied in tigerfish from the Olifants and Letaba Rivers and the process was observed in laboratory held specimens.

7.2 Materials and methods

7.2.1 Field observation Tigerfish were caught with artificial lures from October 1991 until September 1993 in the Olifants and Letaba rivers in the Kruger National Park. Fish were sampled at regular intervals each month to investigate the frequency of tooth replacement. The mass, standard length and gonosomatic index (GSI) were recorded and the age was determined according to the procedure of Balon (1971). The jaws of each fish were examined for any signs of tooth replacement.

76 7.2.2 Laboratory observation.

Twelve adult tigerfish with a mean fork length of 287 mm ± 13 mm and with a mean mass of

180.85 g were kept in three 1000 litre aquaria at a temperature of 27 ± 1°C for a period of seven

months. They were fed on a mixed diet of live fish, beef liver and hake fillets. The sequence of events during tooth replacement and frequency thereof was recorded.

Tigerfish were artificially bred (Chapter 6), and the larvae were kept and raised in a glass

aquarium. Thirty of these fish were observed in a 1000 litre aquarium to determine first

replacement of adult conical teeth. After the fish has attained an average length (FL) of

approximately 50 mm, the bottom of the aquarium was siphoned clean on a daily basis and filtered to detect any teeth. Some of the teeth were then prepared for scanning electron microscopic

observation.

7.3 Results

7.3.1 Field observation Fourteen tigerfish were captured whilst replacing teeth in the Olifants and Letaba rivers. Seven

of these fish were caught in each of above rivers (Table 7.1).

Most tooth replacing fish had teeth in their sockets, but all the teeth were very loosely imbedded

and could easily be removed by hand. One tigerfish visually had no teeth, while other specimens

had small teeth which were hardly protruding from the gums. All the teeth were always in the

same stage of development, indicating that tooth replacement was synchronised in both the upper

and lower jaws. The gums of fish with loosely imbedded teeth appeared swollen. Tooth replacement is apparently not restricted to a certain age and may occur at any time during the

year. The frequency of replacement was however slightly higher during spring and summer.

Stomach analysis revealed digested fish and invertebrates in three of these specimens. Three tigerfish teeth were also found in one specimen which had an empty stomach (Table 7.1).

77 Although nine of the fourteen specimens were male, a higher frequency of tooth replacement can not be associated with a given sex due to an unequal sex ratio in both rivers (Chapter 5).

Table 7.1. Length, age mass, stomach contents and GS! of tooth replacing tigerfish from the Olifants and Letaba Rivers.

Fish No Length Age Mass Stomach GSI Date River (FL) (y) (g) contents 1 256 2+ 317 none RR M 10/91 Letaba 2 291 2+ 405 none 10/91 Letaba 3 260 2+ 193 none RR M 10/91 Letaba 4 240 2+ 177 none 10/91 Letaba 5 304 3+ 551 none 06/92 Letaba 6 220 1+ 150 empty IA M 08/92 Letaba 7 261 2+ 268 empty A M 08/92 Letaba 13 262 2+ 296 teeth RR M 01/92 Olifants 9* 270 2+ 320 fish RR M 02/92 Olifants 10 265 2+ 287 empty RR M 02/92 Olifants 11 252 2+ 231 Inver RR M 02/92 Olifants 12 254 2+ 252 none 06/92 Olifants 13 263 2+ 298 none 06/92 Olifants 14 252 2+ 288 Inver ARM 09/93 Olifants

GSI: Gonosomatic Index: IA M = Inactive avtive male; A M = Active male; RR M = Ripe-running male Stomach contents: inver = invertebrates 'Tigerfish with no teeth protruding from the gums

78 7.3.2 Laboratory observation The twelve tigerfish in captivity progressively killed each other during fights until only three remained, one in each aquarium. During the six months observation period, two of the fish replaced their teeth twice, independently on different occasions. The third tigerfish replaced its teeth three times. During all these occasions some of the teeth were retrieved from the aquarium bottom. Tooth replacement was complete and rapid and the whole process was completed in three to five days. Initially the teeth become irregular arranged, pointing dorsolatterally into different directions, resembling that of a ragged tooth shark. The mouth becomes swollen internally, appears sensitive and the fish refuses to accept food. The teeth are shed in a disorderly fashion and teeth of both jaws are lost showing no specific pattern. Teeth shedding is completed in one to two days and sensitive gums persist for approximately one day after which a complete set of teeth is visible in both jaws. Initially the teeth are small, but are restored to normal size within three days after protruding from the gums. Normal feeding behaviour returns gradually as soon as the new set of teeth is visible.

Tigerfish which were artificially bred and raised in the aquarium shed their first set of adult conical teeth after six to seven months at an average length of approximately 100 mm (FL). A total of 106 teeth were retrieved from the aquarium bottom within above period. Both tricuspid and adult conical teeth were found. The base of each discarded tooth has a conical hole in which the tip of the replacement tooth is probably imbedded prior to tooth replacement (Figures 7.1-6).

7.4. Discussion

Tooth replacement was observed in only a small percentage (1.2%) of tigerfish caught and examined in the Kruger National Park. Although tooth replacement was only observed in tigerfish in the age classes 1+ and 2+ years, Roberts (1967) stated that a 20-30 mm standard length tigerfish with tricuspid teeth has similar replacement trenches to those in Alestes. With growth rate increments of 193-196 mm per year and 110-140 mm per year for the first and second years respectively (Balon, 1971; Kenmuir, 1973), one should expect tooth replacement to take place at least a couple of times during the first two years of development.This is confirmed by the fact

79 Figures 7.1-3 Conical teeth displaying different angles of the hole in the centre of teeth. Figure 7.4. A tricuspid tooth. Note the difference in size as opposed to conical teeth. Figure 7.5. A hooked conical tooth situated anteriorly in the jaw. Figure 7.6. Enlargement of the centre of a tooth fragment. that the third set of teeth is replaced after 6-7 months after hatching (100 mm FL). According to Kenmuir (1973), a toothless tigerfish was tagged and recaptured several months later, still toothless. This may be explained by the frequency of replacement as was observed in the aquarium. This specimen was probably recaptured while it was replacing a second or third set of teeth. Furthermore, it is hard to imagine that such a ferocious predator like tigerfish would be without any teeth for such a long period of time. During this study it was observed that 57.1% of the tooth replacement takes place as from October until February which is the warmer months of the year. No conclusive evidence can thus be made regarding tooth replacement and the different seasons.

Although reluctancy to feed while replacing teeth was observed in the laboratory, it is not sure to what extent foraging behaviour is affected in nature. The presence of only 1.2% of tooth replacing specimens in catches done with angling gear supports the observation on suppressed foraging behaviour, especially if the frequency of tooth replacement is considered. The presence of fish in the stomach of one individual is probably the result of food scarcity prior to tooth replacement. The presence of invertebrates in the stomach of one individual is understandable, due to the small size and soft texture, the sensitive gums of tooth replacing tiger fish should not be affected. The three teeth found in one tigerfish stomach were probably its own teeth which were accidentally ingested. Roberts (1967) and Begg (1973) also found worn teeth in the stomach of some characins and made the same assumption. It appears that some teeth are accidentally swallowed while teeth are shed.

In conclusion tooth replacement is relatively quick, as was observed several times in the aquarium. Since the teeth of the tigerfish plays an important role in the feeding habits of the fish, one can expect tooth replacement to be a rapid process. Begg (1973) on the other hand said that tooth replacement took a month to complete. However, definitive detail about replacement was not given by Begg.

80 [CHAPTER EIGHT

GENERAL DISCUSSION AND CONCLUSION

The tigerfish is and will always be important in the environment in which it occurs as it plays an important role in the trophic structure as a top predator in riverine ecology. Five species off tigerfish occurs in Africa of which only Hydrocynus vittatus is found in Southern Africa. Their distribution is however limited to a few relative small rivers in South Africa. Tigerfish only occurs in the lower reaches of the rivers that is situated in the Eastern Lowveld and Northern Kwazulu Natal. Water shortage and the erection of dams and weirs restricts the distribution of tigerfish to the warmer waters of the lower reaches of these rivers. Tigerfish occurs in all the major rivers of the KNP. There is a major difference in the numbers and distribution of tigerfish within the rivers of the KNP. Large populations of tigerfish are restricted to only a few localities in specific rivers in the KNP, such as the Olifants and Letaba Rivers. The vulnerability of tigerfish in South Africa and the KNP is therefore partially due to the limited distribution of tigerfish in the rivers in which it occurs.

Tigerfish is a sensitive species which simply means that any man induced changes to their environment could have a negative effect on their survival. Not only could a small number of fish in a system, but a whole population be affected by any such changes. Anthropogenic activities which affects the water quantity and quality of the rivers in which tigerfish occurs is important. High silt deposits in the Olifants River (Buermann et al, 1995) could already have a negative influence on the numbers and distribution of tigerfish close to the western border of the KNP, as they prefer to live in clear clean water. Fishing pressure by pelagic fishery, illegal netting, subsistence fishery and the I.T.F.T. are mentioned as factors which have a negative influence on tigerfish populations in Lake Kariba and the Okavango River. Fishing pressure, especially during the breeding season, is also a big threat to the tigerfish population since not only are individuals removed from the system, but also the egg pool which is suppose to ensure recruitment. Tigerfish concentrate prior to and during the breeding season at natural obstructions which renders them to much danger of being caught by fisherman and gill nets.

81 Although tigerfish seems to be more abundant in the Olifants than the Letaba River, they seem to be very unevenly distributed along the course of these rivers with a high concentration towards the lower reaches of the rivers. Seasonal changes also seems to play a role in the abundance of the tigerfish in these rivers as they migrate downstream to find better winter temperatures. It is then important for the tigerfish to overcome any man made or natural obstruction during the yearly summer floods when migrating upstream in order to spawn. It is also important to note that the yearly flood should be big enough for the tigerfish to overcome these obstructions when migrating upstream. In river systems where there are some existing dams, enough water should be allowed to pass through to fulfil these flood requirements and to simulate a natural flood. This would ensure that the distribution of the tigerfish is not negatively affected and that spawning migrations can take place. Breeding migrations takes place annually prior to the first floods. It had been reported that tigerfish migrate downstream to spawn in the Incomati River (Gaigher, 1967; 1970), but is in contrast to the findings during this study and in Lake Kariba (Bowmaker, 1973), Lake Mweru (Soulsby, 1960) and the Zambezi river (Jackson, 1961a). This was confirmed when Olivier (1994) found tigerfish migrating upstream in the Kanniedood fish ladder in the Shingwedzi River of the KNP to spawn.

A large variety of length groups was observed in both the Olifants and Letaba Rivers, indicating different year classes. This was supported when three cohorts were obtained by length frequencies in the Olifants and two in the Letaba River. Tigerfish in the Letaba River seems to be larger than the same age groups in the Olifants River. This can possibly be explained by the quality of the water. With far less silt in the Letaba River during the summer months visibility would be better for feeding purposes as they are known to be fierce predators. There are also less tigerfish in the Letaba River, which means less competition for food and living space.

The overall growth performance based on length frequencies could only be calculated for tigerfish in the Olifants River as the small sample size from the Letaba River made this calculation unreliable. The data showed that the overall growth performance for tigerfish in the Olifants River compared well to that of tigerfish in the Okavango River (Van Zyl, 1992; Hay, 1995) and Lake

82 Kariba (Beatie, 1982). Scales from tigerfish in the Olifants River were also used as a method to determine the overall growth performance and were found to be similar, which proves the effective use off length frequencies to determine the overall growth performance for the species.

The catch curve values indicated that tigerfish in the Olifants and Letaba Rivers can reach an age of six and seven years respectively which is less than the eight to ten years for the same species in the Okavango River. The catch curve values obtained, compares well with annuli counted (five + years) for tigerfish in the Olifants River. The higher total mortality of tigerfish in the Olifants River in comparison to the Letaba River is a good indication of the catch curve values which indicates that tigerfish in the Letaba River has a longer expected life expectancy than tigerfish in the Olifants River. The better water quality and smaller population size in the Letaba River might contribute to the longer live expectancy of the tigerfish in the Letaba River.

Fundamental differences between the feeding habits of tigerfish in the Olifants and Letaba Rivers have been observed in comparison to tigerfish in other systems. Large tigerfish preyed heavily on invertebrates, indicating that prey species is either absent or too large for the tigerfish to prey on. Fish formed a small component of the diet of the tigerfish in the Olifants and Letaba Rivers which is a clear indication of the fish species composition and prey availability in these rivers. The abundance and distribution of tigerfish in the Olifants and Letaba Rivers varies considerably according to the different seasons (Chapter 3) This however had an influence on the diet of the tigerfish since a decrease of invertebrates were witnessed from spring to summer in the Olifants River. A total of nine invertebrate orders were identified in the diet of the tigerfish of which the Ephemeroptera and Hemiptera were the most abundant.

Tigerfish in the Olifants River has also a higher M:F sex ratio than in the Letaba River. This could be explained by stress as a result of over population in the river as mentioned by Holtzhauzen (1989). A deviation from the 1:1 M:F sex ratio was experienced throughout the year to favour the males (7.1:1) in both the Olifants and Letaba Rivers. This ratio increased in favour of the males during (9.9:1) during the breeding season. Holtzhauzen (1989) reported that stress such as over population contributes towards a deviation of the M:F sex ratio in favour of the males. A close similarity existed in the gonad development of tigerfish in The Olifants and Letaba Rivers.

83 Large numbers of ripe-running males were collected in both rivers prior to the first rains. The mere presence of the large concentration of ripe-running males clearly indicates that the tigerfish do undertake breeding migrations. Females reach sexual maturity later in live and were found to be very fecund.The presence of so many ripe-running males would therefore ensures fertilization, since mortality amongst eggs and larvae is extremely high.

It is possible to induce release of gametes in H. vittatus by administration of HCG and pituitary extract. Gonadotropin releasing hormone revealed promising results but require further investigation to overcome the problems of non-viable eggs. Tigerfish eggs are negatively buoyant and slightly adhesive for benthic or epibiotic incubation. From these results as well as field observations it seems if tigerfish spawns on a sandy substrate in the vicinity of aquatic vegetation. The vertical movement of newly hatched larvae indicates passive transportation downstream which in turn indicates that slow flowing water is important when selecting a spawning area.

Tooth replacement was observed both in the field in the laboratory for tigerfish. Tooth replacement is a relative quick process which lasts between three and five days. The gums of the tigerfish replacing teeth becomes swollen and the teeth loosely imbedded. Replacement of teeth occurs simultaneously in the upper and lower jaws. A set of conical teeth is replaced by tricuspid teeth at a length of 22 mm which is in turn replaced by functional conical teeth at a length of 32 mm (Brewster, 1986). The first set of adult conical teeth is replaced after six to seven months at an average length of 100 mm.

It could therefore be concluded that the ecological requirements for tigerfish in the Olifants and Letaba Rivers are not optimal, since the tigerfish is at the edge of its distribution range of southern Africa in these rivers. The only ecological requirement that was relatively similar to tigerfish in other systems was the growth rate of tigerfish in the Olifants River. Differences between maximum age and size, recruitment and mortality were evident from tigerfish populations in the Olifants River and in systems such as the Zambezi and Okavango River and Lake Kariba. Tigerfish appears to be smaller in river systems south of the Zambezi River, indicating that optimal growth conditions does not prevail for the tigerfish in the relatively small rivers in South Africa. The fecundity of tigerfish in the Letaba River were found to be slightly higher than those in Lake

84 Kariba, although recruitment seems to be low with high mortalities in the Olifants and Letaba Rivers. The high male:female sex ratio in both these rivers as a result of stress, is also an indication that optimum ecological conditions are not met for the tigerfish in the Olifants and Letaba Rivers. Tigerfish is a ferocious predator which preys on fish and is therefore regarded as a piscivour. However, invertebrates are the main food source for tigerfish in the Olifants and Letaba Rivers, indicating that feeding conditions are not optimal for tigerfish and are they therefore regarded as invertivours in the Olifants and Letaba Rivers.

Although the distribution of tigerfish is limited in the KNP and ecological conditions are not optimal, the genetic variation of tigerfish in the Olifants River is much higher than the tigerfish found in the Zambezi River. It would however appear that the Zambezi and Okavango River is the typical habitat for the tigerfish Hydrocynus vinatus in southern Africa.

85 [CHAPTER NINE

REFERENCES CITED

Adibisi, A.A. 1981. Analysis of the stomach contents of the piscivorous fishes of the Upper Ogon River in Nigeria. Hydrobiologia.79:167-177. Badenhuizen, T.R. 1967. Some notes on the population dynamics of Hydrocynus vittatus Castelnau in Lake Kariba. Hydrobiologia. 30:527. *Bailey, R.G. 1969. The non- fishes of the eastward flowing rivers of Tanzania, East Africa. Rev. Zool. Bot. afr. 80 (1-2):17-199. Balon, E.K. 1971. Age and growth of Hydrocynus vittatus (Castelnau, 1861) in Lake Kariba, Sinazongwe area. Fish. Res. Bull. Zambia. 5:89-118. Balon, E.K. 1971. Replacement of Alestes imberi Peters, 1852 by A. lateralis Boulenger, 1900 in Lake Kariba, with ecological notes. Fish. Res. Bull. Zambia. 5:119-162. *Balon, E.K. 1974. Fish production of a tropical ecosystem. IN: E.K. Balon & A.G. Coche Eds.Lake Kariba: A man made tropical ecosystem in Central Africa. The Hague Publishers. 484-497. Balon. E.K. 1984. Patterns in the evolution of reproductive styles in fishes.IN: G.W. Potts & .J. Wootton (ed.). Fish Reproduction strategies and tactics, Academic Press, New York. Balon, E.K 1990. Epigenesis of an epigeneticist: the development of some alternative concepts on the early ontogeny and evolution of fishes. Guelph Ichthyol. Rev. 1: 1-48. *Beattie, I.H. 1982. A re-examination of tigerfish growth rates in Lake Kariba. L.K.F.R.I. Project Report (45):I-4 (Cyclostyled). 'Begg, G.W. 1972. Some abnormalities and an interesting anatomical feature occurringin the • tiger-fish Hydrocynus vittatus at Lake Kariba. Piscator. 86:110-112. 'Bell-Cross, G. 1965a. Observations of the movement of fishes across the Congo/Zambezi watershed in the Mwinilunga District of N. Rhodesia. Paper in Proc. Med. and Sci. Cong. Lusaka. 1963. *Bell-Cross, G. 1965;1966. The distribution of fishes in Central Africa. Fish Res. Bull. Zambia 4:3-20.

86 'Bell-Cross, G. & Minshull, J.L. 1988. The fishes of Zimbabwe. National Museums and Monuments, Harare, Zimbabwe. Blache, J. 1964. Les poissons du bassin du Tchad et du bassin adjacent du Mayo-Kebbi. MOmoire O.R.S.T.O.M. No. 4 Paris. 438. Bowmaker, A.P. 1969 b. Fish population fluctuations associated with anadromesis at the Mwenda River mouth, Lake Kariba. News Lett. Limnol Soc. South Afr. 13, Oct.1969. Bowmaker, A.P. 1973. Potamodromesis in the Mwenda River, Lake Kariba. Geophys. monogr.17: 159-164. Bowmaker, A.P.M. 1973. An hydrobiological study of the Mwenda River and its mouth, Lake Kariba. Ph.D. Thesis, University of the Witwatersrand, Johannesburg. Brewster, B. 1986. A review of the genus Hydrocynus Cuvier 1819 (Teleostei: Charachiformes).Bull. Br. Mus. Nat. Hist. (Zool). 50(3):163-206. *Bruton, M.N. & Kok, H.M. 1980. The freshwater fishes of Maputaland. IN: Studies on the ecology of Maputaland. (Eds M.N. Bruton & K.H. Cooper): 210-244. Rhodes University. Buerman, Y., H.H. Du Preez, G.J. Steyn, T.J. Harmse, and A.R. Deacon. 1995.. Suspended silt concentrations in the lower Olifants River (Mpumalanga) and impacts of silt releases from the Phalaborwa Barrage on water quality and fish survival. Koedoe. 38(2):11-34. Cambray, J.A & K. Meyer. 1988. Early ontogeny of an endangered, relict cold-water cyprinid from Lesotho, Oreodaimon quathlambae (Barnard, 1938). Rev. Hydrobiol. trop. 21:309- 333. Carolsfeld, J., S.M. Ramos, R. Ormanezi, J.H. Gomes, J.. Barbosa, & B. Harvey. 1988. Analysis of protocols for application of an LHRH analog for induced final maturation and ovulation of female Pacu (Piaractus mesopotarnicus Holmberg 1887). Aquaculture. 74:49-55. Chen, F.Y., Prowse, G.A. 1964. The effect of living space on the growth rate of fish. Ichthyologica. 3(1-2):11-20. Clay, D. 1976. An investigation into the distribution of fish in Swaziland. Rev. Zool. afr. 90(3). Crass, R.S. 1962. Physical barriers and the dispersion of freshwater fish with particular reference to Natal. Ann. Cape Prov. Mus. 11. South Africa. Crass, R.S. 1964. Freshwater fishes of Natal. Shuter and Shooter, Pietermaritzburg. Crim, L.W. & B.D. Glebe.1984. Advancement and synchrony of ovulation in Atlantic salmon with pelleted LHRH analog. Aquaculture. 43:47-56. 87 •Daget, J. 1954. La Oche dans le fleuve Niger. Afr. J. Trop. Hydrobiol. Fish. Special issue II:107-114. *Daget, J. & Durand J.R. 1981. Poissons. In Fiore et faune aquatiques de l'Afrique Sahelo- soudanienne. (Ed J.R. Durand & C. Levague). ORSTOM, Collection Initiations, Documentations Techniques. 45:687-771. Dansoko, F.D. 1975. 'Contribution a l'etudede la biologie des Hydrocyon dans le delta central du Niger'. IN: Welcomme, R.L. 1979. Fisheries ecology of floodplain rivers. Longman, London. Dansoko, D., Breman, H. and Daget, J. 1976. Influence de la secheresse sur les populations D'Hydrocynus dans le Delta Central du Niger. Cal O.R.S.T.O.M., ser. Hydrobiol.. vol X(2): 71-76. Du Preez, H.H. & Steyn, G.J. 1992. A preliminary investigation of the concentration of selected metals in the tissues and organs of the tigerfish (Hydrocynus vittatus) from the Olifants River, Kruger National Park, South Africa. Water SA. 18:2. Gable, V. 1965. Problems associated with the distribution of freshwater fishes in Southern Africa. South African Journal of Science. 61:11. Gaigher, I.G. 1967. Some aspects of the ecology of the tigerfish, Hydrocynus vittatus Castelnau in the Incomati River System. Provincial Fisheries Institute, Lydenburg. Progress Report No.2. cyclostyled. Gaigher, I.G. 1968. 'n Opname van die visserybronne van die Incomati-sisteem met spesiale verwysing na die ekologie van die tiervis. Projek no 7. Finale verslag. Provinsiale Visseryinstituut, Lydenburg. Gaigher, I.G. 1969. Aspekte met betrekking tot die ekologie, geografie en taksonomie van varswatervisse in die Limpopo- en Incomatiriviersisteem. Ph.D. Thesis, Rand Afrikaans University, Auckland Park, Johannesburg. Gaigher, I.G. 1970. Ecology of the tigerfish (Hydrocynus vittatus) in the Incomati River System, South Africa. Zoologica Africana. 5(2):211-227. Gaigher, I.G. 1973. The habitat preferences of fishes from the Limpopo River System, Transvaal and Mosambique. Koedoe. 16: 103-116. Gaigher, I.G. 1975. Evidence for tooth replacement in the tiger-fish Hydrocynus vittatus. Amoldia Rhod.7 (10):1-4. Gery, J. 1977. Characoids of the world. T.F.H. Publications, New York.

88 Ginsburg, A.S. 1972. Fertilazation of fishes and the problem of polyspermy, (ed.) T.A. Detlaf, translated from Russian. Jerusalem, Israel Program for Scientific Translations. 'Greenwood, P.H. 1983. The zoogeography of African Freshwater fishes: Bioaccountancy or Biogeography? In Systematics Association Special Volume No 23. Evolution, time and space: the emerge of the biosphere. (Eds R.W. Sims, J.H. Price, and P.E.S.Whalley):179-199. Academic press, London and New York. *Griffith, J.S. 1975. Annulus formation and growth of tiger fish Hydrocynus vittatus in Lake Bengweulu, Zambia. Trans. Am. Fish Soc., 104(3):499-505. Hay, C.J. 1995. The development of a data base for the assessment of the biotic integrity and sustainable utilisation of the Okavango River, Namibia. Ph.D. thesis, Rand Afrikaans University. Holden, M.J. 1970. The feeding habits of Alestes baremose and Hydrocynus forskali (Pisces) in Lake Albert, East Africa. J. Zool., Lond. 161:137-144. Holtzhauzen, J.A. 1989. Die voortplantingsbiologie, teelt en hormoongeinduseerde geslagsverandering van vier tilapiasoorte. M.Sc. thesis, Rand Afrikaans University. Jackson, P.B.N. 1960. Ecological effects of flooding by the Kariba Dam on the Middle Zambezi fishes. Proc. Fed. Sci. Cong. Rhod. Sci. Soc. Salisbury, Rhodesia. Jackson, P.B.N. 1961 a. Kariba Studies. Icthyology. The fish of the Midile Zambezi. Manchester Univ. Press. Jackson, P.B.N. 1961 b. The impact of predation, especially by Tigerfish, Hydrocynus vittatus (Cast.) on African freshwater fishes. Proc. Zool. Soc. Lond. 136(4):603-622. Jackson, P.B.N., Rogers, K.H. 1976. Cabora Bassa fish populations before and during the first filling phase. Zoologica Africana. 11(2):373-397. *Jubb, R.A. 1967. Freshwater fishes of Southern Africa. Balkema, Kaapstad. Kenmuir, D.H. 1972. Report on a study of the ecology of the tigerfish Hydrocynus vittatus Castelnau in Lake Kariba. L.K.F.R.I. Project Report No. 6. Kenmuir, D.H. 1973. The ecology of the tigerfish Hydrocynus vittatus Castelnau, in Lake Kariba. Occ. Pap. Natn. Mus. Rhod. B5(3):115-170. Kenmuir, D.H. 1975. The diet of fingerling tigerfish, Hydrocynus vittatus Cast., in Lake Kariba, Rhodesia. Amoldia Rhod. 7(9):1-8. 'Kenmuir, D.H. 1983. Fishes of Kariba. Wilderness Pupblications, Harare. *Kenmuir, D.H.S. 1984. Fish population changes in the Sanyati basin, Lake Kariba, Zimbabwe.S. Afr. J. Zool. 19(3): 194-209. 89 King, J.A. & R.P. Millar. 1980. Comparative aspects of luteinizing hormone-releasing hormone structure and function in vertebrate phylogeny. Endocrinology 106:707-716. Kotze, A., Van der Bank, F.H. & Steyn, G.J. 1996. Allozyme variation in two populations of Hydrocynus vittatus (Pisces, Charachidae). Genetics (in press). Kruger, E.J. 1972. Die broei en propagering van tiervis Hydrocynus vittatus Castelnau, in die Transvaal. Eerste vorderingsverslag, Projek TN 6/4/2/1/8. Laeveldse Visserynavorsingstasie. Marble Hall. Kruger, E.J. 1973. Die broei en propagering van tiervis Hydrocynus vittatus Castelnau, in die Transvaal. Tweede vorderingsverslag, Projek TN 6/4/2/1/8. Laeveldse Visserynavorsingstasie. Marble Hall. Kruger, J.C. De W., Smit, G.L., Van Vuren, J.H.J. & Ferreira, J.T. 1984. Some physical and chemical characteristics of the semen of Cyprinus carpio L. and Oreochromis mossambicus (Peters). J. Fish Biol. 24:263-272. *Lake, S.J. 1967. Rearing experiments with five species of Australian freshwater fishes. Aust. J.Mar. Freshwater. Res. 18:137. Lauzanne, L. 1975. Regimes alimentaires D'Hydrocyon forskalli (Pisces Charachidae) dans le Lac Tchad et ses tributaires. Cah. O.R.S.T.O.M., sot Hydrobiol. 9(2):105-121. Langerman, J.D. 1984. Optimum harvest strategies for tigerfish in Lake Kariba, Zimbabwe. Ms.C. dissertation, University of the Witwatersrand. Lewis, D.S.C. 1974. The food and feeding habits of Hydrocynus forskahlii Cuter and Hydrecynus brevis Gunther in Lake Kainji, Nigeria. J. Fish. Biol. 6:349-363. Lowe-McConnel, R.H. 1988. Broad characteristics of the ichthyofauna. In Biology and ecchEy of African freshwater fishes. (Eds C. Leveque, M.N. Bruton & G.W. Ssentongo): 93-110. Editions de I'ORSTOM. Marshall, B.E. 1985. Changes in the abundance and mortality rate of tigerfish in the Eastern Basin of Lake Kariba. J. Liam!. Soc. south. Afr. 11(2):46-50 Matthes, H. 1968. The food and feeding habits of the tigerfish, Hydrocyon vittatus (Cast.,1861) in Lake Kariba. Beaufortia. 15(201):143-153. Mc Elman, J.F. & E.K. Balon. 1985. Early ontogeny of walleye, Stizostedion vitreum, with steps of saltatory development. IN: E.K. Balon (ed.) Early life histories of fishes. Dr. W. Junk Publishers, Dordrecht, The Netherlands. *Merron, G.S. & Bruton, M.N. 1989. Recent fisheries research in the . South African Journal of Science. 85:416- 417. 90 *Merron, G.S. 1991. The ecology and management of the fishes ofthe Okavango Delta, with particular reference to the role of seasonal floods. Ph.D. dissertation. Rhodes University. Mires, D. 1974. On the high percentage of tilapia males encountered in captive spawnings and the effect of temperature on the phenomenon. Bamidgeh. 26(1):3-11. Mitchell, S.A. 1978. The diurnal activity patterns and depth zonation of Marginal fish as shown by gillnetting in Lake Kariba. Kariba studies. 9.Trustees of the National Museums and Monuments of Rhodesia. 'Mond, T. 1950. Notes d'ichtyologie ouest-africaine. Bull. Inst. Franc. Afrique. Noire. 12: (Cited by Roberts, 1967). *Nikolslcii, G.V. 1969. Theory of fish population dynamics as the biological background for rational exploitation and management of fishery resources. Oliver and boyd, Edinburgh. Olivier, L.P. 1994. 'n Studie om die effektiwiteit van die Kanniedooddam visleer in die Shingwedzirivier, Nasionale Krugerwildtuin, to bepaal. Nasionale Hoer Diploma, Technikon Pretoria, RSA. *Paungy, D. & Guegan, J.F. 1989. Note a props de trois especes d'Hydrocynus (Pisces, Characidea) du bassin du Niger suiver de la rehabilitation de l'espece Hydrocynus vittatus (Castelnau, 1861). Revue Hydrobiol. trop. 22(91):63-69. Peter, R.E., H. Lin & G. Van der Kraak. 1987. Drug/hormone induced breeding of Chinese telecAts. IN: Idler, D.R., L.W. Crim, & J.M. Walsh (ed.). Proceedings of the third International Symposium on Reproductive Physiology of fish. St. Johns, Newfoundland, . Canada. Pienaar, U de V. 1978. The freshwater fishes of the Kruger National Park. Pretoria: National Parks Board. Plug, I., Skelton, P. 1991. Fish and other faunal remains from a Late Iron Age Site one The Letaba River, Kruger National Park. Koedoe 43(1). *Poll, M. 1967a. Contribution a la faune ichthyologique de l'Angola. Companhia de diamantes de Angola (Museo de Dundo). Publicacoes cultarais. 75:1-381. *Pott, R. McC. 1969. The fish life of the Pongola River and the effects of the erection of a dam on the fish populations. Progress Report Sport Fisheries, Project No. II. Provincial Fisheries Institute, Lydenburg, Tvl.

91 •Regier, H.A. 1966. A perspective on research on the dynamics of fish populations in the great lakes. Prog. Fish. cult. 28(1):3-18. *Roberts, T.R. 1967. Tooth formation and replacement in Characoid fishes. Stanford. Ichyol. Bull. 8(4):231-247. Scholtz, C.H., Holm, E. 1985. Insects of Southern Africa. Butterworth Publishers, Mayville, Durban. Schoonbee, H.J. & J.F. Prinsloo. 1984. Techniques and hatchery procedures in induced spawning of the European carp, Cyprinus carpio and the Chinese carps, Ctenopharyngodon idella, Hypopthalmichthys molitrix and Aristichthys nobilis in Transkei. Water SA. 10:36-39. •Skelton, P.H., Bruton, M.N., Merton, G.S., Van der Waal, B.C.W. 1985. The fishes of the Okavango drainage system in Angola, South West Africa and Botswana : and distribution. Ichthyol. Bull. J.L.B. Smith Inst. of Ichthyol. 50:1-21. *Skelton, 1987. South African Red Data Book - fishes. South African National Scientific Programmes Report, No 137. CSIR, Pretoria. Skelton, P.H. 1988. The disribution of African freshwater fishes. IN: Biology and ecology of African freshwater fishes. (Eds C. Lev'eque, M.N. Bruton & G.W. Ssentongo): 93-110. Editions de l'ORSTOM. Skelton, P.H. 1993. n' Volledige gids tot die varswatervisse van Suider Afrika. Southern Boekuitgewers. *Soulsby, J.J. 1960. Some Congo Basin fishes of Nortern Rhodesia. N.Rhodesia J. 4:231. Sparre, P., Uisin, E. & Venna, S.C. 1989. Introduction to tropical fish stock assessment. FAO Fisheries Technical Paper. No 306.1. Rome. Steyn, G.J. 1987. Aspekte van die spermatologie van die tiervis (Hydrocynus vittatus) en die kriobewaring van semen van geselekteerde varswatervissoorte. PhD. proefskrif, Randse Afrikaanse Universiteit, Johannesburg. Steyn, G.J. & Van Vuren, J.H.J. 1987. Some physical properties of the semen from artificially induced sharptooth catfish (Clarias gariepinus). Comp. Biochem. Physiol. 86A:315-317. Steyn, G.J., J.H.J. Van Vuren & E. Grobler. 1989. A new sperm diluent for the artificial insemination of rainbow trout (Salmo gairdneri). Aquaculture 83:367-374. Steyn, G.J. and J.H.J. Van Vuren. 1991. Cryopreservation of the spermatozoa of two African freshwater fishes (Characidae). S. Afr. I Wildl. Res. 21:76-81.

92 Steyn, G.J. 1993. Physico-chemical characteristics of tigerfish semen. S. Afr. J. Wildl. Res. 23:44-47. 'Tave, D. 1986. Genetics for fish hatcherymanagers. AVI Publishing Company, Inc. Westport, Conneticut. Tobor, J.G. 1972. The food and feeding habits of some Lake Chad commercial fishes. Bul. I.F.A.N., 34, ser. A, 1:179-211. Tweddle, D. 1982. Tooth replacement in the tiger-fish Hydrocynus vittatus Castelnau. Luso: J. Sci. Tech. (Malawi). 3(1):33-35. 'Van der Waal, B.C.W. 1976. n' Visekologiese studie van die Liambezimeer in die Oos-Caprivi met verwysing na visontginning deur die Bantoebevolking. Ph.D. proefskrif, Randse Afrikaanse Universiteit, Johannesburg, South Africa. Van Loggerenberg, N.P. 1979. Die teelt en verspreiding van die tiervis Hydrocynus vittatus Castelnau, en die vestiging van ander Afrika visspesies in die Transvaal. Projek TN 6/4/2/2/1/18. Provinsiale Visseryinstituut. Lydenburg. Van Loggerenberg, N.P. 1980. Die kunsmatige teelt van die tiervis Hydrocynus vittatus (Castelnau), en die biologiese aspekte wat daarmee verband hou. Projek TN 6/4/2/2/1/18. Provinsiale Visseryinstituut. Lydenburg. Van Loggerenberg, N.P. 1981. Die kunsmatige teelt van die tiervis (Hydrocynus vittatus Castelnau, 1861), in Transvaal Eerste verslag aan die Nasionale Parkeraad. Projek TN 6/4/2/2/1/18. Provinsiale Visseryinstituut. Lydenburg. Van Loggerenberg, N.P. 1982. Die kunsmatige teelt van die tiervis (Hydrocynus vittatus Castelnau, 1861), in Transvaal. Vierde Projekverslag. Projek TN 6/4/2/2/1/18. Provinsiale Visseryinstituut. Lydenburg. Van Loggerenberg, N. 1983. Conservation of tigerfish and fish farming techniques. Fauna and Flora 40:30-31. Van Rensburg, N.O.S. 1981. Die monitor van gechlonineerde koolwaterstowwe in die Trartsvaalse watersisteme en die omvang van skade deur plaagdoders op die waterbronne. Eerste ongepubliseerde verslag. Jaarlikse Vakkundige Vergadering. Afdeling Natuurbewaring van die Tansvaal. Van Zyl, B.J. 1992. 'n Visekologiese ondersoek van die Okavango- en Kuneneriviere met spesiale verwysing na visontginning. Phd proefskrif, Randse Afrikaanse Universiteit, Johannesburg.

93 Venter, F.J. 1991. Fisiese kenmerke van bereike van standhoudende riviere in die Nasionale Krugerwildtuin. Kruger National Park Rivers Research Programme: First annual research meeting 18-20 March 1991. Winemiller, K.O. and Kelso-Winemiller, L.C. 1994. Comparitive ecology of the African pike, , and tigerfish, Hydrocynus forskahlii, in the Zambezi Rriver floodplain. Journal of Fish Biololy. 45:221-225. Withler, F.C. 1980. Chilled and cryogenic storage of gametes of Thai carps and catfishes. Can. Tech. Rep. Fish. Aquat. Sci. No. 948. * References not cited

PERSONAL COMMUNICATIONS.

Bryden, B.R1992. Chief Ranger: Kruger National Park, 1983 - 1997. Hugo, C. 1992. Manager: Letaba Ranch. 1969 - 1973. Kloppers, J.J. 1992. Manager: Nature Coservation: Kruger National Park, untill 1992. Pienaar, U. de V. 1992. Executive Director: National Parks Board. 1987 - 1990.

Van der Merwe, J. 1994. Chief Ranger; Kruger National Park 1997 - present Van Niekerk, H.H. 1994. Chief Pilot: Kruger National Park. 1974 - 1997 Viljoen, P.C. 1996. Senior Research Officer: Kruger National Park, untill 1596.

94