University of Groningen The biology and impacts of Oreochromis niloticus and Limnothrissa miodon introduced in Lake Kariba Chifamba, Chiyedza Portia

(1)

The biology and impacts of Oreochromis niloticus and Limnothrissa miodon introduced in

Lake Kariba

Chifamba, Chiyedza Portia

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.

Document Version

Publisher's PDF, also known as Version of record

Publication date: 2019

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

Chifamba, C. P. (2019). The biology and impacts of Oreochromis niloticus and Limnothrissa miodon introduced in Lake Kariba. Rijksuniversiteit Groningen.

Copyright

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

(2)

528794-L-bw-Chifamba 528794-L-bw-Chifamba 528794-L-bw-Chifamba 528794-L-bw-Chifamba Processed on: 6-2-2019 Processed on: 6-2-2019 Processed on: 6-2-2019

Processed on: 6-2-2019 PDF page: 1PDF page: 1PDF page: 1PDF page: 1

The biology and impacts of Oreochromis niloticus and

Limnothrissa miodon

introduced in Lake Kariba

(3)

The research presented in this thesis was conducted at the University Lake Kariba Research Station and the Department of Biological Sciences of the University of Zimbabwe and at the Lake Kariba Fisheries Research Institute, Zimbabwe, according to the requirements of the Graduate School of Science, Faculty of Science and Engineering, Institute of Evolutionary Life Sciences (GELIFES), University of Groningen.

This research was funded by Nuffic grant (grant number NFP-PhD.11/ 858) awarded to Portia Chifamba, International Foundation for Science (IFS) Grant awarded to Portia Chifamba (grant number A/3159-1, Tonolli Memorial Fund Fellowship of the International Society of Limnology (SIL), and University of Zimbabwe Research Grant. The printing was supported by the University of Groningen (RUG).

The preferred citation for this thesis is:

Chifamba PC (2017) The biology and impacts of Oreochromis niloticus and

Limnothrissa miodon introduced in Lake Kariba. PhD thesis, University of

Groningen, Groningen, The Netherlands. Cover design: Portia C. Chifamba & Jan H. Wanink Lay‐out: Jan H. Wanink & Portia C. Chifamba Figures: Portia C. Chifamba Pictures including cover: Portia C. Chifamba, Jan H. Wanink, social media Printed by: Ipskamp Printing, Enschede, The Netherlands ISBN: 978‐94‐034‐1472‐0 ISBN: 978‐94‐034‐1471‐3 (electronic version) © 2019 Portia C. Chifamba (pcchifamba@yahoo.co.uk)

(4)

The biology and impacts of

Oreochromis niloticus and

Limnothrissa miodon introduced

in Lake Kariba

Proefschrift

ter verkrijging van de graad van doctor aan de Rijksuniversiteit Groningen

op gezag van de

rector magnificus prof. dr. E. Sterken en volgens besluit van het College voor Promoties.

De openbare verdediging zal plaatsvinden op maandag 4 maart 2019 om 16.15 uur

door

Chiyedza Portia Chifamba

geboren op 2 augustus 1963

(5)

Prof. dr. H. Olff

Prof. dr. B.D.H.K. Eriksson

Beoordelingscommissie

Prof. dr. ir. C. Both

Prof. dr. E. van Donk

(6)

This thesis is dedicated to my parents who made it all possible.

Oh, ik ben zo blij

Urombo

by Jane Eugenia Chifamba

Urombo Urombo Urombo Urombo

Mwanasikana muka utarire

Nyika yaugere igungwa rebvura

Nyatotarira rinoda kukunyudza

Vakomana venyika vanokunyengedza

Vanokufurira unyangadze vakuseke

Uri chirombe, chifuza, chinzenza chamakoko

Urombo Urombo Urombo Urombo

Ambuya vangu vakafa, havakasiya mari

Nambuya vako vakafa, havakasiya mari

Taivakirwe chikoro chevanasikana

Vaidzidzira kuchengeta nhaka dzavapwere vavo

Nzvimbo dzacho dzotinetsza

Hwangova urombo

O urombo. O urombo. O urombo. O urombo

(7)

(8)

Chapter 1 Introduction 9

PART I BIOLOGY AND IMPACTS OF OREOCHROMIS NILOTICUS

Chapter 2 Replacement of the indigenous Oreochromis mortimeri by the invader Oreochromis niloticus in the Southern-African Lake Kariba: in relation to differences in their reproductive potential

25

Chapter 3 Growth rates of alien Oreochromis niloticus and indigenous

Oreochromis mortimeri in Lake Kariba, Zimbabwe

41 Chapter 4 Diet overlap between a native and an invasive tilapia species in

the Southern-African Lake Kariba 57

Chapter 5 Comparative aggression and dominance of Oreochromis

niloticus (Linnaeus, 1758) and Oreochromis mortimeri

(Trewavas, 1966) from paired contest in aquaria

77

PART II BIOLOGY AND IMPACTS OF LIMNOTHRISSA MIODON

Chapter 6 Developing a sustainable pelagic fishery in an African reservoir: trends in the catches of the introduced freshwater sardine

Limnothrissa miodon and associated species in Lake Kariba,

Zimbabwe

93

Chapter 7 Growth of the freshwater sardine, Limnothrissa miodon (Boulenger 1906) estimated from diurnal increments in otoliths

117

Chapter 8 Synthesis 139

References 165

Authors affiliations and addresses 193

List of publications 197

Summary 201

Samenvatting (Dutch summary) 207

Curriculum Vitae 213

Acknowledgements 217

(9)

(10)

Chapter

1

Introduction

(11)

(12)

11

1

World-wide introduction of exotic fish species in waterbodies has had both beneficial

and adverse ecological, social and economic outcomes of varying magnitude (Reynolds & Greboval 1988; Welcomme 1988, Balirwa 1992; Witte et al. 1995). Thus, informa-tion on impacts of introduced species and how well they have adapted to their new conditions, is essential when considering management options both to protect indigenous species and to enhance fisheries. In addition, information on the ecological impacts can enhance our understanding of ecological processes caused by introductions.

Differences in the modes of introduction might have some bearing on the type of impacts arising from the introduction. Fish introductions are either deliberate or planned, in order to fill a vacant niche or accidental, as a result of escapees invading an already occupied niche (De Silva & Sirisena 1987; Welcomme 1988; Balirwa 1992; van Zwieten et al. 2011). The deliberate introduction of a large predator, Nile perch (Lates niloticus), has well established negative ecological impacts, having caused the extinction of many haplochromine species from Lake Victoria. At the same time, Nile perch brings about large economic benefits from harvesting (Reynolds & Greboval 1988; Witte et al. 1995; Twongo 1995). Nile tilapia, Oreochromis niloticus (Linnaeus 1758), was also deliberately introduced into Lake Victoria. Oreochromis niloticus is today one of the three commercially important species caught in the lake but has caused the disappearance of some native tilapias (Goudswaard et al. 2002; Njiru et

al. 2005). The introduction of O. mossambicus increased fish catches in reservoirs

in Sri Lanka (De Silva & Sirisena 1987), just as the deliberate introduction of

Limno-thrissa miodon in Lake Kivu (de Iongh et al. 1995).

Lake Kariba provides a typical case study both for planned introductions of fish into supposedly open ecological niches as well as accidental, unplanned introduc-tions into occupied niches. This was possible as the creation of Lake Kariba, a reservoir in the Zambezi River, created a completely new lacustrine ecosystem whose physical attributes such as oxygen concentration, water depth and distance from the lake margin differed profoundly from the lotic system to which the native riverine species were adapted (Jackson et al. 1988). This new, complex matrix not only presented economic opportunities for fisheries development and fish farming but also raised ecological challenges.

First, the newly created and therefore vacant pelagic niche was filled by a freshwater sardine, Limnothrissa miodon (Boulenger 1906). Limnothrissa miodon is native to Lake Tanganyika and was introduced to improve fish production (Bell-Cross & Bell-(Bell-Cross 1971). Physical and environmental conditions in Lake Kariba were unlike those in Lake Tanganyika, and therefore changes in L. miodon’s bio-logical characteristics were anticipated. For example, L. miodon in Lake Kariba was considered stunted (Marshall 1987a) until studies on growth rate using otoliths showed that the fish in Lake Kariba follows the same growth trajectory as in the

(13)

12

native Lake Tanganyika (Chifamba 1992). However, patterns appear to suggest that

L. miodon may have inhibited the expansion of a small fish (Brycinus lateralis)

native to the Upper Zambezi and affected the habits of the predatory tigerfish

Hydro-cynus vittatus (Woodward 1974). Since its introduction, the L. miodon population has

been subjected to considerable fishing pressure which may have induced evolution of its life history parameters. Therefore, further research on L. miodon is essential to determine its adaptation to its new environment, its interaction with native fish species and factors affecting the biology of this fish.

Secondly, ponds on the Lake Kariba shore provided opportunities for the farming of O. niloticus which resulted in escapees entering the lake and invading the niche already occupied by the congeneric native O. mortimeri (Chifamba 1998, 2006). This negative interaction of the exotic and native species thus presented an opportunity to study factors that confer competitive advantage to the invader by comparing growth, diet and aggression in these two species. Research on the biology of O. niloticus and

L. miodon in Lake Kariba would determine adaptations of these introduced species

to their new environment.

Both introduced fish species are important in the Lake Kariba fisheries, L. miodon as the main catch of the pelagic fishery and O. niloticus as one of the important species in the artisanal fishery. Hence, the objective of introducing L. miodon into Lake Kariba to improve fish production was realised. However, catches have declined since the beginning of the fishery, resulting in economic losses and a need to improve the management of the fishery. Informed management of the fish resource is needed to ensure sustainable fisheries. Such knowledge is currently needed for explaining changes in the productivity of the sardine industry that is thought to have crashed as a result of overexploitation or environmental changes.

The environment

Lake Kariba physical characteristics

Lake Kariba was formed on the middle Zambezi River at 485 m altitude in 1958 and was then the world’s largest reservoir (5820 km2_{at maximum storage) (Coche 1974).}

The reservoir became full in 1963 and the weight of water was so large that it increased The aim of this thesis is to establish the degree of suitability of the ecosystem created by the Lake Kariba dam to the introduced species Oreochromis niloticus and Limnothrissa miodon, to identify factors that may have caused O. niloticus to displace the native species Oreochromis mortimeri, and to investigate potential factors causing the decline of the L. miodon fishery.

(14)

13

1

Figure 1.1 Map of Lake Kariba showing the Lake basins.

the seismic activity in the area. In 1963 alone, five tremors above 5.0 on the Richter scale were experienced (Tumbare & Sakala 2000). Lake Kariba has a length of 277 km, width of 32 km at its widest point, mean depth of 29 m, and a maximum depth of 120 m. The Zambezi River contributes most (77%) to the water volume in the lake, whilst other rivers contribute 16% and rainfall 7% (Balon & Coche 1974). The dam was constructed primarily for hydro-electricity generation. Hence, the bulk of the water is lost through hydro-electricity turbines and 14% by evaporation. Retention time of the water is about 3 - 4 years and the lake level experiences an annual change of 1 - 5 m, resulting from inflowing floods and drawdowns through turbines and spillage through the sluice gates (Karenge & Kolding 1995).

The lake is separated into five basins, marked by chains of islands and narrows (Figure 1.1). The uppermost two lake basins, Mlibizi and Binga, are riverine due to the influence of the Zambezi River. They are flushed out in May by the Zambezi River floods and thereby assume turnover characteristics earlier than the other three basins, which are truly lacustrine and have temperature-induced turnover. This river-lake environment gradient has a profound effect on the fish species composition (Begg 1974).

(15)

14

Rainfall and temperature

The Kariba area has one rainy season, with most rain falling in December and January. Annual rainfall is between 400 and 800 mm depending on location (Begg 1970). The amount of rainfall affects lake level and drawdown. Lake Kariba is monomictic and has a thermal stratification from September to early June. Mixing or turnover usually takes place in July each year. Stratification begins around September, with a thermo-cline at around 15 m depth, which gradually moves down to around 35 m at the time of turnover. Stratification prevents deep water in the hypolimnion from mixing with the epilimnion (Begg 1970).

Lake Kariba is a warm lake with a surface water temperature of 28 to 30 0_{C and a}

hypolimnion of about 22 0_{C when stratified. When mixed, the whole water column is}

22 0_{C (Marshall 2012a). The mean maximum air temperature in Kariba has steadily}

risen from about 33.1 0_{C in 1968, to 35.5}0_{C in 1998 (Magadza 2010). As a result of}

the temperature increase, the depth of the thermocline decreased from 10 to 15 m in 1986 – 1987, to a consistent 5-m depth in 2007 – 2008 (Ndebele-Murisa et al. 2014). A shallow thermocline likely means a reduction in the optimal habitat for phyto-plankton and fish. Cochrane (1978) found a correlation between the catches of L.

miodon and water volume above the thermocline, therefore the recent change of

thermocline depth may have affected L. miodon abundance and catches negatively. Temperature and hydrological factors (rainfall, riverflow and lake level) are correlated with L. miodon catches through nutrients brought in by the rivers and the effect of temperature on phytoplankton and zooplankton on production (Chifamba 2000).

Phyto- and zooplankton productivity

During the stratification period, the hypolimnion becomes depleted of oxygen and the epilimnion of nutrients due to photosynthesis. Turnover increases the availability of nutrients in the epilimnion and the euphotic zone, increasing phytoplankton pro-duction (Ramberg, 1987; Masundire, 1989). Hence, turnover, rainfall and river mouths are associated with increased plankton production (Magadza, 1980; Ramberg, 1987; Masundire, 1992, 1994; Cronberg, 1997). Temperature therefore mediates nutrient cycling in the lake and is an important driver of productivity in the lake. Temperature also drives seasonal and annual variation in phytoplankton and zooplankton com-munities. Cyanobacteria (blue-green algae) dominated from December to May and diatoms from June to September, whilst 60% of the annual biomass consisted of cyanobacteria in 1982 – 1983 (Ramberg 1987). Chlorophyceae dominated during periods of relatively low temperature compared to Cyanophyceae.

Comparable results were found in a laboratory study by Sibanda (2003). The growth rates (% increase in number of cells per day) of Chlorophyceae in the labora-tory declined at water temperatures above 25 0_{C, becoming negative above 28}0_C.

At the same time, the growth rate of Cyanophyceae increased almost exponentially up to 34 0_{C, resulting in a transition temperature from Chlorophycea to}

(16)

Cyanophy-528794-L-bw-Chifamba 528794-L-bw-Chifamba 528794-L-bw-Chifamba 528794-L-bw-Chifamba Processed on: 6-2-2019 Processed on: 6-2-2019 Processed on: 6-2-2019

15

1

ceae domination of about 28 0_{C (Magadza 2011). In Lake Kariba, that has warmed by}

a mean of 1.54 0_{C between 1965 and 1990, the mean epilimnion summer temperature}

reached this transition limit in 1987 (Magadza 2011).

The rise in water temperature in the lake has indeed caused a change in the phyto-plankton towards a community dominated by cyanobacteria (Magadza 2011). Due to toxicity and morphology, cyanobacteria are poor food for zooplankton relative to small chlorophytes and flagellates (Wilson et al. 2006). This is reflected in the shift in the zooplankton from large- (Calanoida; Daphnidae) to small-bodied (Bosmina; Cyclopoida) species, and a strong reduction in abundance, that accompanied the transition to cyanobacteria in the lake. Being a zooplankton feeder, L. miodon was expected to be negatively affected by these changes. Though a causal relation could not be established, it is remarkable that a decline in the catches of L. miodon started shortly after the mean epilimnion summer temperature exceeded 28 0_{C (Magadza}

2011). This thesis evaluates changes in individual growth rate of L. miodon as a possible mechanism by which temperature can affect L. miodon production.

The shift in the phytoplankton composition is also of interest in relation to competition in the inshore planktivorous fish, the endemic Oreochromis mortimeri and introduced O. niloticus. The question is whether such a shift in the plankton community would then favour O. niloticus, which is known to feed and digest cyanobacteria (Moriarty 1973). Thus, to understand the recent changes in the fish populations in Lake Kariba, it is important to study the diet composition of both introduced (e.g. O. niloticus) and native species (e.g. the potentially competing indigenous planktivorous species, O. mortimeri).

The Fish

Fish species composition changes

At the future Kariba Dam site in the Zambezi River, which Jackson (1960) described as a sandbank river with little vegetation, Hydrocynus vittatus, Distichodus sp.,

Barbus sp. and Labeo sp. were dominant, while cichlids such as O. mortimeri and

some small fish species were rare. The fishes were subjected to a seasonal period of flooding, when food and shelter were plenty, and a dry season when flow was low and remaining pools in the river small, thus providing little food and shelter (Jackson 1960). It could be anticipated that the transformation from a riverine to a lacustrine ecosystem would create new conditions that would alter fish distribution patterns. Initially, nutrients were high after impoundment from the decomposing submerged vegetation and leaching from the soil. The new lake was characterised by a high productivity of algae, the invasive water fern, Salvinia molesta and fish. Nutrients decreased with time as the lake matured. Macrophyte beds developed in the inshore area, increasing habitat diversity, shelter and food, supporting an increase in fish species such as Tilapia sparrmanii.

(17)

16

Surveys done before the creation of Lake Kariba found between 28 and 31 species of fish in the Middle Zambezi River (Jackson 1961; Harding 1964; Bell-Cross1965). In the late 1990s, when the reservoir was filled, 45 species were reported by Marshall (2006) and 50 by Songore & Kolding (2003). Species such as Brycinus lateralis,

Barbus poechii, Labeo lunatus, Oreochromis andersonii, Sargochromis giardi, Sargochromis carlottae, Serranochromis macrocephalus and Serranochromis robustus which before the creation of the dam were restricted to the Upper Zambezi

River, were captured in the lake (Balon 1974). However, of these, only B. lateralis and S. macrocephalus were caught in a lake-wide survey in 2006 (Zengeya & Marshall 2008).

Fish species in the new water body revealed preference for diverse conditions. Many riverine species, especially tilapias, prefer still-water pools and marshes in a river (Jackson 1966). These fishes found the lake’s stable and stagnant environment favourable and thrived particularly in the most lacustrine Sengwa, Bumi and Sanyati basins (Figure 1.1) (Begg 1974). In these basins, the cichlid fishes, mainly

Oreo-chromis mortimeri, SargoOreo-chromis condringtoni and Tilapia rendalli, made up

between 64.1 and 96.2% of the catch in 1968 to 1970. In an unpublished report from 1959, only 0.75% of the fish caught were O. mortimeri, while in 1962, as the lake filled, the contribution had increased to 35% of the catch (Kenmuir 1984; Jackson et

al. 1988). For these species, the lake environment mimicked the period of plenty in

a flooded river when food and shelter were plenty resulting in high survival and reproduction, and consequently high catches. The genera that prefer flowing water and were abundant in the river before impoundment, Hydrocynus, Distichodus,

Barbus and Labeo, became dominant only in the more riverine uppermost two lake

basins and in the estuaries of inflowing streams (Jackson 1960; Begg 1974). The rheophilic species, Chiloglanis neumanii, Opsaridium zambezense (and possibly also Leptoglanis rotundiceps) are now confined to the tributaries of the two upper-most basins (Balon 1974). All the former river fishes are restricted to the inshore shallow (< 15m depth) area, except Clarias gariepinus, Mormyrus longirostris and

Syndontis zambezensi that can live in water down to 30 m depth when the water is

well oxygenated (Jackson 1960; Coke 1968; Sanyanga et al. 1995).

The inshore fish species are the basis of a gill-net artisanal fishery that started in 1958 and 1962 in Zambia and Zimbabwe, respectively. Between 1964 and 1972, three species dominated the inshore catches: the predator H. vittatus and two cichlids,

Oreochromis mortimeri and Serranochromis condringtoni (Karenge & Kolding

1995). The Shannon diversity index for the Lakeside fish increased between 1972 and 1990 as a result of natural introduction from the upper Zambezi and tributaries, as well as fish introductions (Karenge & Kolding 1995). To monitor the inshore fish population, the Lake Kariba Fisheries Research Institute (LKFRI) in Zimbabwe established in the 1960s a Lakeside Experimental Sampling Programme where

(18)

gill-528794-L-bw-Chifamba 528794-L-bw-Chifamba 528794-L-bw-Chifamba 528794-L-bw-Chifamba Processed on: 6-2-2019 Processed on: 6-2-2019 Processed on: 6-2-2019

17

1

net nets are set routinely at a site named Lakeside, situated on the shores of Kariba

town. Data from this programme were used in my research.

Introduction of fish species

Although at least five species are known to have been introduced into Lake Kariba, only the clupeid L. miodon and the tilapiine cichlid O. niloticus have been success-fully established and are now widespread throughout the lake. Oreochromis macrochir is found only sporadically; on average about 10 specimens per year were caught out of 5 – 10 000 total number sampled in Zimbabwe at Lakeside between 1974 and 2001 and only three specimens in 1985, 1992 and 1996 in the Zambian Experimental Gillnet Survey (Kolding et al. 2003; Marshall 2006). The introduced Micropterus salmoides was caught only once and has probably not established a viable population. Two of the introduced species, Tilapia rendalli and Serranochromis robustus, may have invaded the lake naturally and the former might have been in the system pre-impoundment (Kolding et al. 2003).

Limnothrissa miodon (sardine), a zooplanktivorous pelagic freshwater clupeid,

was deliberately introduced into Lake Kariba from Lake Tanganyika from 1967 to 1968 in order to fill a vacant pelagic niche and increase fish production (Bell-Cross & Bell-Cross 1971). About 30% of the lake is shallower than 17 m and only the shallow area less than 20 m is used by most of the indigenous Zambezi River fishes, because they are not adapted to a pelagic environment (Begg 1970; Coke 1968). By 1969, there was evidence that the sardine had become established (Kenmuir 1971). The introduction is considered a success because L. miodon has the largest single fish stock in Lake Kariba. Annual commercial catches landed reached a maximum of about 31 000 tonnes in 1990, and a minimum of 15 000 t in 2003 (Kinadjian 2012). From the beginning of the fishery, the Lake Kariba Fisheries Research Institute in Zimbabwe and the Department of Fisheries in Zambia collected data on the catches and fishing effort. Data collected by the two institutions and from other research, show that catches of sardines vary seasonally and annually (Marshall 1988b, 2012a). Each year catches usually reflect two peaks that differ in magnitude. A small peak occurs during April – May and a larger peak during August – September.

The exotic Nile tilapia (Oreochromis niloticus) was first caught in gillnets set routinely at Lakeside in 1993 (Chifamba 1998). Up to August 1994, O. niloticus were only caught close to the fish farms where they may have escaped. Even then, the abundance of O. niloticus was low, constituting a mere 0.4% of the catch by mass and 0.17 % by numbers. Judging from the range of fish size caught (3 – 30 cm and 1 – 1 069 g) and the presence of both sexes in the sample, these fish were by then already established in the lake.

The farming of O. niloticus on the shores of Lake Kariba is responsible for the introduction of this species into the lake. Oreochromis niloticus was selected for farming in Kariba because it is widely used in aquaculture all over the world. This

(19)

18

is due to its superior growth rate and adaptation to aquaculture condition, compared to other cichlids (Philippart & Ruwet 1982; Blow & Leonard 2007). Prior to this fish introduction, there was no assessment of its suitability and of the probable impacts. Impact of fish introductions

Oreochromis niloticus

Since the introduction of O niloticus resulted from escapees from fish farms, it was not carefully monitored (Chifamba 1998). However, with the introduction of O.

niloticus, the once abundant O. mortimeri, an endemic to the Zambezi system,

de-clined and disappeared in many parts of Lake Kariba (Chifamba 2006; Zengeya & Marshall 2008). Several factors may have contributed to this species displacement, and a key aim of this thesis is to evaluate the major potential causes.

Competition is a potential driver of local species displacement (MacArthur & Levins 1967). Because O. mortimeri and O. niloticus show similarities in their diets and reproduction strategies (Chifamba 1998; Mhlanga 2000; Marshall 2011), competition between these species can be expected. Both species feed on algae and organic detritus, plant material, insects and zooplankton, varying with availability (Lowe-McConnell 1958; Moriarty 1973). In both species, the male constructs a large nest in an arena, which it defends (Jubb 1974; Marshall 2011. Therefore, competition for food, nesting and nursery space may have stimulated antagonistic behaviour. Aggression is one mechanism that O. niloticus may have used to displace native O.

mortimeri in Lake Kariba.

Life history trade-offs are reported to be strong determinants of competitive abilities, under both stable and changing ecological conditions (Lancaster et al. 2017). A higher growth rate or an ultimate large size of O. niloticus could be another mechanism to displace O. mortimeri. Many studies have shown that large fish tend to have a larger number of eggs (Schemske 1974, Baglin & Hill 1977, Schenck & Whiteside 1977, Bagenal 1978; Wanink & Witte 2000; Barneche et al. 2018). High growth rates contribute to fitness when large size has benefits such as higher fecundity and reduced mortality. Fast growth would therefore result in a larger number of eggs at an earlier fish age. In addition, fast growth may also reduce the time an animal spends at a vulnerable size because smaller animals tend to be more vulnerable to predation (Sutherland 1996). Hence, a comparison of growth rate and maximum size of the introduced O. niloticus and the native O. mortimeri will give an indication of potential relative competitiveness of these species.

(20)

19

1

Limnothrissa miodon

Limnothrissa miodon was deliberately introduced into Lake Kariba to utilize the

plankton production in the newly formed pelagic area. This may have prevented population expansion of the native zooplanktivore Brycinus lateralis (Woodward 1974). Early catches of L. miodon from the open water contained 20.5% B. lateralis, showing the capability of the latter species to expand from the Upper Zambezi River and fill the vacant pelagic niche in the lake. Therefore, the expansion of B. lateralis into the pelagic area may have been prohibited by competition from L. miodon (Marshall 1991).

The introduction of L. miodon also affected the habits of the native tigerfish (Hydrocynus vittatus). This predator soon added L. miodon to its diet and began to inhabit open water in pursuit of its new prey, where it occurred in the developing sardine fishery (Cochrane 1976; Marshall 1987b, 1991). From April 1969 to March 1970, only 1.5% of the stomach content of H. vittatus consisted of L. miodon, whereas from April 1970 to March 1971 the amount rose to 41.4% and remained high thereafter In this thesis I evaluate the potential for competition between Oreochromis

mortimeri and Oreochromis niloticus, by comparing their reproductive effort,

aggression levels, growth rates and diet. To compare their diet, I analysed the stomach contents of the two species and estimated the Schoener similarity index (Schoener 1970) and the Pianka overlap coefficient (Pianka 1973) to assess the level of potential competition between the two species. Fish of known size and weight were aged by counting annual increments on scales, opercula and otoliths, and the results used to estimate the growth parameters. The mean length of each age group in the fish sample was estimated and, together with the growth para-meters, used to determine which species grows faster and thus can confer size advantage in a contest.

In order to assess the relative aggression of the two species, pairwise contests were setup in an aquarium. The number of aggressive acts such as ‘biting’ and ‘ramming’ were used to score aggression. Aggression indicates which of the two species is likely to be outcompeted in the event of a contest arising during competition for a resource such as food and a breeding site.

Reproductive effort was estimated using the monthly proportion of fish in the samples of both species, that were ready to release gametes (eggs or milt). Having a higher reproductive effort may help a species to outcompete the other by increasing its own population rapidly. The monthly proportions of breeding fish were correlated to rainfall and temperature, both known to trigger repro-duction (Clark et al. 2005; Taranger et al. 2010; Quintana et al. 2004), to find out if the influence of those abiotic on the two species is different.

(21)

20

(Kenmuir 1973; Mhlanga 2003). Hydrocynus vittatus was an important bycatch species from the beginning of the L. miodon fishery. Between 1973 and 1975 the catches of H. vittatus increased with increasing catches of L. miodon (Cochrane 1976). A decline in catch per unit effort (CPUE) of less than one year old H. vittatus between 1974 and 1986 correlated with a decrease in CPUE of L. miodon (Marshall 1987b). Already before its introduction into Lake Kariba it was known that, in Lake Tanganyika, L. miodon is not an obligate pelagic species but that it inhabits the inshore area for a substantial part of its life (Poll 1953; Matthes 1968). In the inshore area L. miodon competes with small cichlids, Brycinus sp. and Barilius sp.

Fishing on Lake Kariba

Although power generation is the most important economic function of Lake Kariba, it is also the largest source of fish in the country. Two fisheries evolved, one based on the introduced L. miodon, operating in deep/pelagic water, and an inshore fishery. During 1994, the major economic activities on and around Lake Kariba, combined for Zambia and Zimbabwe, generated revenue of about 124 million USD, of which 54% was from power generation and 37.9% from the fisheries (Tumbare 2000).

The pelagic fishery is semi industrial and of greater value than the artisanal inshore fishery, contributing 33.9% against 4.0% of the overall revenue and landing 28 423 and 2 473 metric tonnes of fish, respectively. Dried sardines are an important source of protein, particularly in the rural areas, because of the long shelf life of the dried fish. However, while the sardine catches rose with fishing effort at the beginning of the fishery, they have steadily declined since 1990 (Magadza 2011).

In this thesis, I investigate the causes of the decline in the pelagic catches by analysing the relationships between Limnothrissa miodon catches, fishing effort, and several key environmental variables: air temperature and the hydrological factors, rainfall, river flow, and lake level.

In this thesis, the impact of the sardine fishery on the inshore fish species, through the capture of juveniles, will be explored by analysing the bycatch. I also assess the relationship between Limnothrissa miodon and Hydrocynus vittatus, to elucidate the response of H. vittatus to changes in the population of L. miodon.

(22)

21

1

Objectives of the thesis

PART I – BIOLOGY AND IMPACTS OF OREOCHROMIS NILOTICUS

In the first part of the thesis, I evaluate the contribution of the increase in Oreochromis

niloticus to the decline of its native congener Oreochromis mortimeri, a species

endemic to the Middle Zambezi River.

First, to determine the suitability of the new environment for O. niloticus, I estimate its growth parameters and compare them to those of the same species in other water bodies. This is because environmental suitability can be judged by individual growth rate and the maximum size the fish attains in that environment, as compared to other environ-ments. Such information can be used for the estimation of fishable biomass.

Furthermore, I look at possible interactions that could have contributed to the displacement of O. mortimeri, by comparing the growth rates, the degree of diet overlap, and aggression levels in O. niloticus and O. mortimeri. Fast growth may confer competitive advantage to the introduced O. niloticus if this translates to more surviving offspring. A large diet overlap can result in a strong competition for food resources (Hanson & Leggett 1985). Higher aggression levels offer advantage in the form of access to better nesting and brooding sites (Philippart & Ruwet 1982; Seppänen et al. 2009).

PART II – BIOLOGY AND IMPACTS OF LIMNOTHRISSA MIODON

In the second part of the thesis, I evaluate the declining catches of the freshwater sardine Limnothrissa miodon, by investigating growth rate and age at first maturity as potential causes of the decline. Other potential causes are overfishing and environmental changes. These are assessed by correlating the catches with fishing effort, air temperature, rainfall, river flow, and lake level.

Information on growth rate and age at maturity should reveal whether the fishery catches too small, immature fish. Limnothrissa miodon in Lake Kariba is considered stunted, and the bulk of the catches consists of small fish (< 6 cm total length) compared to those in the fisheries on Lakes Tanganyika and Kivu (Marshall 1987a). In an exploited fish population, environmental and fishing effort simultaneously affect the fished population. I use multiple regression analysis to determine the relative contribution of environmental factors and fishing effort to the declining catches. I also estimate the Maximum Sustainable Yield from two fisheries models. That provides a guideline for the level of fishing effort, to be used together with considerations such as environmental variation and the biology of the target species, in order to achieve a sustainable fishery on Limnothrissa miodon in Lake Kariba.

(23)

22

Outline of the chapters

PART I – BIOLOGY AND IMPACTS OF OREOCHROMIS NILOTICUS

 Chapter 2 describes the displacement of the indigenous tilapia Oreochromis

morti-meri by the introduced Oreochromis niloticus, and it explores the possible role of

interspecific differences in reproductive potential. Monthly variation in the gonadal activity of the two species is presented in relation to rainfall and temperature.  In Chapter 3, growth rates of O. niloticus and O. mortimeri are determined and

compared. Ages are estimated from reading scales, opercula and otoliths. Ages from these body parts as well as ages from three time periods are compared.  Chapter 4 shows the results of stomach content analyses performed on O. niloticus

and O. mortimeri. Diet overlap was estimated to inform on the degree of competi-tion between the two species.

 Chapter 5 deals with the comparison of aggression levels and dominance of O.

niloticus and O. mortimeri, observed in paired contests in an aquarium. The working

assumption is that a high aggression level will infer competitive advantage. PART II – BIOLOGY AND IMPACTS OF LIMNOTHRISSA MIODON

 In Chapter 6, catch trends of Limnothrissa miodon are evaluated to determine to what extent they can be explained by fishing effort and temperature and how they relate to the catches of their predator, Hydrocynus vittatus. The occurrence of large sized sardines in the inshore waters is discussed in terms of feeding and breeding. Impact of the sardine fishery on other fish species is explored by analysing incidental catches.

 Growth of L. miodon and spatial differences therein are evaluated in Chapter 7. Fish were aged using daily increments in otoliths. Deposition rate of the in-crements was validated using electron microscopy. Implications of age and size at first maturity for fishery management are discussed.

SYNTHESIS

 Chapter 8 gives an integrated discussion of the results from all the previous chapters. Here the achieved answers to the research questions are evaluated.

Oreochromis niloticus (left; from Boulenger 1907)

(24)

PART

I

(25)

(26)

Chapter

2

Replacement of the indigenous

Oreochromis mortimeri by the

invader Oreochromis niloticus in

the Southern‐African Lake Kariba:

in relation to differences in their

reproductive potential

Portia C. Chifamba

Han Olff

(27)

A female Nile tilapia partly eaten by an African fish eagle. An impression of the reproductive potential can be obtained by relating the size of the tiny ripe eggs (scattered inside and outside the fish) to the size of the buccal cavity.

The Nile tilapia Oreochromis niloticus is one of the worst invaders in African lakes and rivers and can cause a massive decline of abundance and diversity of other cichlid species but is also valued for its economic importance in aqua-culture. However, the invasion process and extend of environmental impacts are not well understood. Similarity of ecological niche of Oreochromis niloticus, introduced into Lake Kariba in the 1990s, placed it in direct completion with an endemic congeneric cichlid, Oreochromis mortimeri. This makes the study of the joint development of their populations (based on catches) both ecological and economically interesting. The resulting patterns in catches found were explained by studying differences in the reproductive potential of the two species.

Oreo-chromis niloticus was found to have a lower length at maturity (17.63 ± 0.70 cm)

than O. mortimeri (19.19 ± 0.99 cm). In addition, the proportion of mature fish with ripe gonads was always higher for O. niloticus compared to O. mortimeri, indicating a higher reproductive potential of the former. The introduction of O.

niloticus in Lake Kariba appears not to have improved the fish catches because

the combined catches of the two species remained constant during the period of analysis. This displacement of an endemic species and driving it into the critically endangered species category, demonstrates how unplanned introductions can pose a threat to biodiversity.

(28)

27

2

Introduction

Nile tilapia (Oreochromis niloticus) is an economically important species for African freshwater aquaculture and fisheries (Brummett & Williams 2000), and for that reason has been introduced (planned or unplanned) in many rivers and lakes throughout the continent outside its native range (Schwanck 1995; Zengeya et al. 2011). However, in some ecosystems, for example in Lake Victoria, this has led to an ecological catastrophe, where the native congenerics were reduced in their abundance (Balirwa 1992; Bbole et al. 2014; Deines et al. 2014). The mechanism of this exclusion is not clear due to the high cichlid diversity of the large East-African lakes, which makes it difficult to study pair-wise species interactions. For this, simpler systems with less species can provide insight.

Lake Kariba is a man-made lake in the Gwembe rift valley, formed by the con-struction of a dam across the Zambezi River at the Kariba Gorge in 1958. The lake is bordered by Zambia on the north and Zimbabwe on the south and at the time of its construction it was the world largest man-made lake (5 820 km2_{) (Coche 1974).} The formation of Lake Kariba converted an existing riverine ecosystem into a lacustrine one and it affected the fish species composition and abundance.

Oreochromis mortimeri (Trewavas 1983), one of the native riverine fish species

in Lake Kariba, is a cichlid endemic to the middle Zambezi river system from the Victoria Falls to Cahora Bassa Gorge (Marshall 2011). Before the Kariba dam was built, catches of O. mortimeri in the river were small, compared to those of other species. Following impoundment, catches of O. mortimeri in the Lake Kariba reservoir in-creased rapidly, making up 38% of the inshore catches in the 1970s (Kenmuir 1984). Although the catches decreased during the 1980s, O. mortimeri was still important to the fishery (Karenge & Kolding 1995).

Oreochromis niloticus (Linnaeus, 1758) invaded Lake Kariba from fish farms

along the shore, first appearing in fisheries catches in 1989, and subsequently being caught throughout the whole lake (Chifamba 1998; Zengeya & Marshall 2008). Its introduction was followed by a reduction in the catches of O. mortimeri (Chifamba 2006). An analysis of the catches taken from the Sanyati basin of Lake Kariba showed a spatial and temporal gradient in the proportion of O. niloticus to O.

mortimeri (Chifamba 2006).

The natural distribution of O. niloticus includes the Nile basin, Rift Valley lakes and some West African rivers (Skelton 2001). In locations where it is introduced,

Oreochromis niloticus tends to displace other fish (Balirwa 1992; Marshall 1999). It

displaced Oreochromis esculentus (Graham) and Oreochromis variabilis (Boulenger) in Lake Victoria and native cyprinids in Lake Luhondo (Balirwa 1992). In Lake Chivero, Zimbabwe, it replaced the previously introduced tilapia Oreochromis

(29)

28

1999; Tiki 2011). Oreochromis niloticus may adversely affect resident fish species through competition for food, hybridisation and competition for space (Zengeya & Marshall 2007; Weyl 2008; Zengeya et al. 2011).

To explain the competitive success of O. niloticus over O. mortemeri, it is important to assess the reproductive potential of different species. Techniques for assessing this, include the determination of the onset of maturity, the duration of the breeding season, the fecundity and the fraction of spawners (Murua et al. 2003). In both O. niloticus and

O. mortimeri, breeding takes place throughout the whole year and their breeding peaks

are associated with the rains (Lowe-McConnell 1982; Donnelly 1969).

In this study, we describe the trends in the population of O. niloticus after it was introduced and the resulting change in the population of the indigenous O. mortimeri, using catch data. We try to explain the differences by comparing the reproductive potential of the two cichlids using the length at maturity and the proportion of mature fish throughout the year. The impact of the introduction on the catches was assessed by analysing the trend in the combined catches of O. niloticus and O. mortimeri and the size of fish in the catches.

Materials and Method

Site description

Lake Kariba is separated into five basins, marked by chains of islands and narrows. The data used in this study was collected in the Sanyati basin, closest to the dam. The Kariba area has one rainy season (November to April) and receives between 400 and 800 mm of rainfall annually (Begg 1970). The Zambezi River floods reach Lake Kariba in March – May, causing the lake level to rise until June or July, followed by a decline until November. Lake Kariba is a warm lake with a surface water temperature of between 28 and 30 0_{C, and a hypolimnion temperature of about 22}0_{C when}

strati-fied. When mixed (turnover between June and July), the value for the whole water column is 22 0_{C (Marshall 2012b). Maximum temperatures are recorded in October}

and November. They decrease gradually during the rains to a minimum in June and July (Balon & Coche 1974).

Data collection

Fish catch data was obtained from a routine gill net sampling programme at Lakeside in the Sanyati basin of the lake, conducted by the Lake Kariba Fisheries Research Institute. Since the 1960s, at least one fleet of gill nets was set overnight, at Lakeside, every fortnight, resulting in a long time series for this site, interrupted only in times of socio-economic crisis. A fleet of gill nets consists of twelve joined 45 m long panels, of stretched mesh sizes 38, 51, 64, 76, 89, 102, 114, 127, 140, 152, 165 and 178 mm.

Each fish caught was identified, its standard length (SL; to the nearest mm), mass (to the nearest gram) measured and the sex and stage of gonad development

(30)

29

2

determined. Both the standard and total length (TL) were measured in 2005, so that

the results obtained in this study could be compared with those from studies that only used total length. Artisanal fishery data was obtained from Nyamhunga, Dandawa, Mudzimu and Nematombo fishing villages from December 2003 to October 2004. Only species name, weight and length of the fish were recorded for the artisanal catch.

Trends in the catches

Trends in the catch per set of O. niloticus and O. mortimeri were compared to assess the impact of the introduction of the former on the latter species. A Loess smoother, set at a sampling proportion of 0.2 and at weights from a Gaussian density function, was used for smoothing the trends plotted using Sigma Plot 12 software.

Sex and length at first maturity

Each specimen was dissected to determine sex and the stage of development of the gonads. Seven stages of gonad development (Ricker 1968; Witte & van Densen 1995; Table 1) were used. Classification of the ovary developmental stages was based on the magnitude of gonad distension, and on the size, yolk content and colour of the ova. For testis, the magnitude of gonad distension and colour were used. Fish too spoilt to be sexed were excluded from the analysis of maturity and gonadal stage.

All fish with gonads at the inactive stage were considered immature and all the other stages beyond this as mature. Error in classification may occur when fish recovering from spawning are indistinguishable from earlier stages (Witte & van Densen 1995). If these fish are included, the 50% maturity level can be overestimated and the percentage of mature fish cannot reach 100%, because some females at resting stage are considered immature. This source of error can result in erroneous classifi-cation and should be considered when estimating and interpreting length at maturity.

Length at maturity is the length at which 50% of the individuals in a length class are mature. It was determined by fitting a sigmoid function to the percentage of male and female mature fish in 1-cm length classes, using a Sigma Plot 12 sigmoid equation with the asymptote set at 100% (mature fish in the population). Percentage sexual maturity was described by a sigmoid function:

PL= 100 / (1+ exp (- (L - L50) / b))

where PL is the percentage of mature fish at length L, L50 = length at sexual maturity, and b determines the steepness of the curve.

Seasonality of breeding

The breeding seasonality was deduced from changes in the proportion of mature fish that were in the ripe, ripe running and spent stages of gonad development (Table 2.1).

(31)

30

Table 2.1 The stages of gonad development used to identify mature and the ripe Oreochromis

niloticus and O. mortimeri.

Gonad stage Code Condition

Immature I Young fish which have not yet reproduced, characterized by very small gonads.

Inactive/active IA Gonads bigger than in inactive fish, about half their ripe size; translucent eggs visible with magnification.

Active A Testis and ova opaque, reddish with blood capillaries; eggs visible to the naked eye as whitish and granular.

Active/ripe AR Gonads 2/3 of their ripe size; ovaries white reddish; individual eggs easy to see without magnification. Ripe R Sexual organ fills the cavity; eggs completely formed; milt is white, but not extruded under light pressure. Ripe/running RR Sexual products are extruded when light pressure is applied. Spent S Sexual products extruded, aperture inflamed; gonads like deflated sacs; residual eggs/sperm may be present.

Fish with gonads in this stage of development were considered breeding. In order to investigate the environmental triggers of breeding, the monthly proportions of breeding fish were correlated with monthly rainfall and air temperature in the Kariba area using Spearman’s rank correlation in SPSS 17.0.

Results

A total of 1 169 O. niloticus and 1 822 O. mortimeri were used in the analysis. The minimum and maximum length of O. niloticus was 7.5 and 62 cm, and that of O.

mortimeri 7.8 and 51.8 cm, respectively. Trends in the catches

Ever since O. niloticus appeared in the Lakeside experimental gill net fishing programme catches in 1993, the catches of this species increased, whilst those of O. mortimeri declined. The latter species disappeared in 2006, except for an occasional catch (Figure 2.1a). About 50% of the variation in the catches of O. mortimeri could be explained by the variation in O. niloticus (Figure 2.1b). The catches of O. niloticus rose to a peak in 2000, remaining stable thereafter with a mean catch of 2.8 kg per set. Combined catches of these two species show the replacement of the O. mortimeri by O. niloticus with the trend in catches strongly fluctuating between years but without a clear trend in the total catch during the period of displacement from 1992 to 2012 (Figure 2.2).

(32)

31

2

Figure 2.1 a) Trends in the

catch per set of nets and (with the trend shown as a Loess smoother) b) regression of

Oreochromis niloticus and O. mortimeri catches from the

Lakeside experimental gill‐ netting site from1968 to 2012.

Figure 2.2 Combined catches

of Oreochromis niloticus and

O. mortimeri per set of gill

nets per night from 1992 to 2012. Year 1995 2000 2005 2010 Tot al c at ch of O . ni lot ic us + O. mo rtime ri (K g/ set ) 0 2 4 6 8 10 O. mortimeri O. niloticus a) Year 1960 1970 1980 1990 2000 2010 2020 C at ch p er s et (K g) 0 10 20 30 40 O. niloticus O. mortimeri O. niloticus O. mortimeri b) O. niloticus (Kg) 0 2 4 6 8 10 O . m or tim er i (K g) 0 2 4 6 catches exponential model

(33)

32

Gillnet size selection

The mesh size of the net that resulted in the largest catch of fish was higher for O.

niloticus than for O. mortimeri (Figure 2.3). Peak catch for O. niloticus over the

whole fishing period was reached with the 152-mm mesh at a total number of 194 fish, with a mass of 181.8 kg (Figure 2.3a). The mean length of O. niloticus caught in the 152-mm mesh was 29.3 cm. In comparison, O. mortimeri catches peaked in the 114-mm mesh nets and the mean length caught in this mesh was 23.4 cm (Figure 2.3b). The net with the maximum catch is above the legal smallest mesh size (102 mm) allowed in Lake Kariba.

Figure 2.3 Total number, mass and mean length of a) Oreochromis niloticus and b) O. mortimeri caught at Lakeside in gill nets of different mesh sizes from 1992 to 2012. Error bars for mean length are standard deviations. a) O. niloticus Tot al n um ber /M as s (k g) 0 100 200 300 400 M ean len gt h (cm ) 0 10 20 30 40

Mesh size (mm) vs Nilo no fish wt Mesh size (mm) vs Nilo wt kg total Mesh size (mm) vs Nilo mean length (cm)

b) O. mortimeri Mesh size (mm) 20 40 60 80 100 120 140 160 180 200 Tot al n um ber /M as s (k g) 0 200 400 600 800 M ean len gt h (c m ) 0 10 20 30 40

Total number of fish Total mass (kg) Mean length (cm)

(34)

33

2

The size distributions of these two species caught in the experimental gill nets are

shown in Figure 2.4 and compared with the length frequency of O. niloticus from the artisanal fishery. For the experimental gill nets, the modal catch of O. niloticus is higher and size distribution wider than that of O. mortimeri. The catches from the fishing villages reflect that their nets had a larger mesh size than those used in the experimental gill-netting programme (Figure 2.4).

To enable size comparison of our fish, measured in standard length, with reported values of total length, the following function that describes the relationship between standard length and total length was derived:

Standard length (SL) = 0.838 total length (TL) – 0.2408 (R2_{= 0.99; n =25)}

Figure 2.4 Length frequency of a) Oreochromis niloticus and b) O. mortimeri caught in the gill nets at Lakeside and c) O. niloticus caught in gill nets at Nyamhunga, Dandawa, Mudzimu and Nema‐ tombo artisanal fishing villages from December 2003 to October 2004. Ages are from scale readings (Chapter 3 – Chifamba & Videler 2014). a) O. niloticus Fr equ en cy 0 50 100 150 200 250 300 b) O. mortimeri Fr equ en cy 0 100 200 300 c) Artisanal fishery Length (cm) 0 10 20 30 40 50 60 Fr eq ue nc y 0 100 200 300 400 500 600 2 yrs 3 yrs 4 yrs 5 yrs

(35)

34

Figure 2.5 Length at first maturity for Oreochromis niloticus (○) and O. mortimeri ( ) calculated

using Lakeside data from 1992 to 2012.

Length at first maturity

The smallest mature O. niloticus female caught was 8.0 cm, versus 13 cm for the smallest mature male. Figure 2.5 shows that the length at 50% maturity was smaller for

O. niloticus than for O. mortimeri. Length at 50% maturity for O. niloticus was 17.63

cm, obtained from the sigmoidal equation below (R2_{= 0.95):}

% mature fish = 100 / (1 + exp (- (fish length - 17.6266) / 3.3132))

There was little temporary change in the length at maturity. During the 1993 – 2002 and 2003 – 2012 time periods, the length at maturity was on average 17.91 cm and 18.28 cm respectively. The model explains 80% (R2_{= 0.80) of the variation in the} data from the first period, compared to 95% for the second period and 95% for the aggregated data from both periods (Table 2.2).

The smallest mature O. mortimeri were of almost the same size in females and males, being 11.5 and 11.0 cm, respectively. The length at first maturity of O.

mortimeri was 19.19 cm for the whole data set. The sigmoidal function below

describes the relationship between the percentages of mature fish in each age group and fish length (R2_{= 0.93):}

% mature fish = 100 (1 + exp (- (fish length - 19.1884) / 4.2728))

The length at first maturity for the period 1993 – 2002 (17.77 cm) was less than that for the 2003 – 2012 period (18.52 cm) though the estimate for the first period was estimated with greater error than that for the second and the whole data set (Table 2.2).

Standard length (cm) 0 20 40 60 % M at ur ity 0 20 40 60 80 100

(36)

35

2

Table 2.2 Results from fitting the sigmoidal function to the proportion (%) of mature fish at a given length to estimate the length at first maturity of Oreochromis niloticus and O. mortimeri using data collected from 1992 to 1993. (CI = 95% confidence interval calculated from the standard error (SE)).

Period Length (cm) SE CI R2 _F _p _n

O. niloticus 1992 ‐ 2002 17.91 0.86 ± 1.69 0.80 152.0 0.0001 41 2003 ‐ 2012 18.28 0.34 ± 0.66 0.95 578.5 0.0001 38 1992 ‐ 2012 17.62 0.36 ± 0.70 0.95 775.8 0.0001 47 O. mortimeri 1992 ‐ 2002 17.77 0.39 ± 0.77 0.91 339.5 0.0001 32 2003 ‐ 2012 18.52 0.32 ± 0.63 0.95 546.4 0.0001 28 1992 ‐ 2012 19.18 0.50 ± 0.99 0.93 399.2 0.0001 34 The combined data show a slightly higher length at first maturity for O. mortimeri compared to O. niloticus. The length at first maturity of O. mortimeri is closer to the size where the highest catches (mean length = 23.4 cm, caught in the 114-mm meshed nets) are made than that of O. niloticus (mean length = 29.3 cm, caught in 152-mm meshed nets) (Figures 2.3 and 2.4).

Seasonal trend in gonad activity

The analysis of the annual trend in gonad activity shows that the proportion for O.

niloticus with ripe and beyond stages of gonad development was always higher than

that for O. mortimeri (Figure 2.6). The trend in females is more distinct compared to that in males. In both species, breeding takes place throughout the year. Breeding of female O. niloticus peaks in December and is lowest in May. In O. mortimeri the peak is in November and the lowest in June.

The annual breeding pattern followed broadly the annual variation in rainfall and temperature that is shown in Figure 2.7. Figure 2.8 shows the relationships between rainfall and minimum temperature, as well as the proportion of breeding for both species. For O. niloticus, the minimum temperature (rs = 0.84; p = 0.001; n = 12) is significantly correlated to the breeding pattern, whilst rainfall (rs = 0.50; p = 0.098;

n = 12) is not (Table 2.3). Both minimum temperature (rs = 0.80; p = 0.002; n = 12) and rainfall (rs = 0.82; p = 0.001; n = 12) are significantly correlated to the proportion of breeding O. mortimeri, with rainfall showing a higher correlation.

(37)

36 Figure 2.6 Annual variation in the proportion of fish classified as ripe (ripe, ripe/running, and spent) stage of gonad development in Oreochromis niloticus and

O. mortimeri. Figure 2.7 Mean month‐ ly rainfall ( ), minimum, mean and maximum air temperature (○) in Kariba measured between 1992 and 2008. (Data from Zambezi River Authority).

Figure 2.8 Relationship between the proportion of ripe gonads of Oreochromis niloticus (○) and

O. mortimeri (●) and a) rainfall and b) minimum temperature.

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

P ro po rti on o f rip e go na ds (% ) 0 10 20 30 40 50 60 70 O. niloticus female O. niloticus male O. mortimeri female O. mortimeri male

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

Tem per at ur e (0 C) 10 15 20 25 30 35 40 Ra in fa ll ( m m ) 0 20 40 60 80