<%BANNER%>

Life History Variation in Tilapia Populations within the Crater Lakes of Western Uganda

Permanent Link: http://ufdc.ufl.edu/UFE0021401/00001

Material Information

Title: Life History Variation in Tilapia Populations within the Crater Lakes of Western Uganda The Role of Size-Selective Predation
Physical Description: 1 online resource (189 p.)
Language: english
Creator: Efitre, Jackson
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: crater, deforestation, fisheries, fishing, history, lakes, life, management, otoliths, pressure
Zoology -- Dissertations, Academic -- UF
Genre: Zoology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Life history theory seeks to describe and explain the evolution of adaptive responses in fitness-related traits such as reproduction, survival, age and size at maturity, and growth as a result of environmental variation. Interpopulational variation in life history characters is well documented in a wide variety of fish taxa, both marine and freshwater. Variability in fish life history characters may be a result of phenotypic plasticity in response to environmental changes and/or genetic changes associated with evolutionary response to size-selective fishing. For effective fisheries management, it is important to consider existing knowledge about fish life history strategies, biotic and abiotic factors simultaneously in predicting population responses to changes in the environment. I examined effects of size-selective fishing and other environmental factors on key life history characters of tilapia populations in the crater lakes of western Uganda with the hope of providing critical information needed for management of the fisheries in these lakes. I used tilapia as a model species because life history variation in tilapiine fishes is well-documented in tropical freshwaters. Tilapia also display high levels of phenotypic plasticity and can tolerate a wide range of environmental conditions due to their ability to vary allocation of resources to reproduction and growth depending on the environmental condition. The occurrence of both unexploited and exploited tilapia populations in a large number of the lakes that also differ in extent of catchment deforestation and fishing pressure further provided suitable systems and the replication needed to explore life history variation across broad environmental gradients. I determined a range of environmental characters in 19 crater lakes and found a wide variation in environmental features among lakes. There was a strong negative relationship between water transparency and Chl-a concentration with deforested lakes having a lower transparency and higher Chl-a concentration compared to forested lakes. I also explored the effects of deforestation and fishing on the condition of two introduced species (Oreochromis leucostictus and Tilapia zillii) in 17 of the crater lakes and found O. leucostictus in severely deforested and heavily fished lakes were in a better condition compared to similar fish in lakes with low productivity and low to medium fishing. Differences in condition of T. zillii were only detectable between lakes with high and low fishing effort. I also developed an appropriate protocol for aging T. zillii in Lake Nkuruba and validated the periodicity and timing of opaque zone formation in otoliths of T. zillii using marginal-increment and edge analyses. Tilapia zillii in Lake Nkuruba deposited two opaque zones on their otoliths per year corresponding to bimodal rainfall that is characteristic of this equatorial region. Tilapia zillii grew rapidly in the first 2 years after which growth slowed down considerably. Gillnetted Tilapia zillii grew faster and attained a larger size-at-age than the trapped fish. Finally, I examined the effects of size-selective predation and environmental factors on life history traits of T. zillii in the 8 crater lakes. I found striking differences in life history traits in T. zillii between and within lakes, suggesting that environmental factors, density-dependence, and size-selective fishing influence variation in key life history characters of T. zillii populations in the crater lakes of western Uganda. These results provide useful baseline data for the management of heavily exploited and unexploited T. zillii populations in the crater lakes of western Uganda.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jackson Efitre.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Chapman, Lauren J.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021401:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021401/00001

Material Information

Title: Life History Variation in Tilapia Populations within the Crater Lakes of Western Uganda The Role of Size-Selective Predation
Physical Description: 1 online resource (189 p.)
Language: english
Creator: Efitre, Jackson
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: crater, deforestation, fisheries, fishing, history, lakes, life, management, otoliths, pressure
Zoology -- Dissertations, Academic -- UF
Genre: Zoology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Life history theory seeks to describe and explain the evolution of adaptive responses in fitness-related traits such as reproduction, survival, age and size at maturity, and growth as a result of environmental variation. Interpopulational variation in life history characters is well documented in a wide variety of fish taxa, both marine and freshwater. Variability in fish life history characters may be a result of phenotypic plasticity in response to environmental changes and/or genetic changes associated with evolutionary response to size-selective fishing. For effective fisheries management, it is important to consider existing knowledge about fish life history strategies, biotic and abiotic factors simultaneously in predicting population responses to changes in the environment. I examined effects of size-selective fishing and other environmental factors on key life history characters of tilapia populations in the crater lakes of western Uganda with the hope of providing critical information needed for management of the fisheries in these lakes. I used tilapia as a model species because life history variation in tilapiine fishes is well-documented in tropical freshwaters. Tilapia also display high levels of phenotypic plasticity and can tolerate a wide range of environmental conditions due to their ability to vary allocation of resources to reproduction and growth depending on the environmental condition. The occurrence of both unexploited and exploited tilapia populations in a large number of the lakes that also differ in extent of catchment deforestation and fishing pressure further provided suitable systems and the replication needed to explore life history variation across broad environmental gradients. I determined a range of environmental characters in 19 crater lakes and found a wide variation in environmental features among lakes. There was a strong negative relationship between water transparency and Chl-a concentration with deforested lakes having a lower transparency and higher Chl-a concentration compared to forested lakes. I also explored the effects of deforestation and fishing on the condition of two introduced species (Oreochromis leucostictus and Tilapia zillii) in 17 of the crater lakes and found O. leucostictus in severely deforested and heavily fished lakes were in a better condition compared to similar fish in lakes with low productivity and low to medium fishing. Differences in condition of T. zillii were only detectable between lakes with high and low fishing effort. I also developed an appropriate protocol for aging T. zillii in Lake Nkuruba and validated the periodicity and timing of opaque zone formation in otoliths of T. zillii using marginal-increment and edge analyses. Tilapia zillii in Lake Nkuruba deposited two opaque zones on their otoliths per year corresponding to bimodal rainfall that is characteristic of this equatorial region. Tilapia zillii grew rapidly in the first 2 years after which growth slowed down considerably. Gillnetted Tilapia zillii grew faster and attained a larger size-at-age than the trapped fish. Finally, I examined the effects of size-selective predation and environmental factors on life history traits of T. zillii in the 8 crater lakes. I found striking differences in life history traits in T. zillii between and within lakes, suggesting that environmental factors, density-dependence, and size-selective fishing influence variation in key life history characters of T. zillii populations in the crater lakes of western Uganda. These results provide useful baseline data for the management of heavily exploited and unexploited T. zillii populations in the crater lakes of western Uganda.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jackson Efitre.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Chapman, Lauren J.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021401:00001


This item has the following downloads:


Full Text

PAGE 1

1 LIFE HISTORY VARIATION IN TILAPIA POPULATIONS WITHIN THE CRATER LAKES OF WESTERN UGANDA: THE ROLE OF SIZE-SELECTIVE PREDATION By JACKSON EFITRE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2007

PAGE 2

2 2007 Jackson Efitre

PAGE 3

3 To my parents for their belief in hard work a nd for encouraging me to pursue my scientific interests

PAGE 4

4 ACKNOWLEDGMENTS I owe thanks to very many people, whose he lp was indispensable in completing this dissertation even though I will not list all of their names here. I start by especially thanking my advisor, Pr of. Lauren J. Chapman, for persevering with me throughout the time it took me to complete the research and writing of the dissertation. Throughout my doctoral work, she guided and encour aged me to develop scientific skills and greatly assisted me with statistical techniques. I appreciate her gene rosity with her time, understanding, and encouragement during the pe aks and troughs of the writing of the dissertation. She was not only read ily available to help me, as sh e generously is for all of her students, but she always read and responded to dr afts of each chapter of my work especially when we had to communicate over emails. I am extremely grateful to all the members of my dissertation committee: Dr. Debra J. Murie, Dr. Mike Allen, Prof. Colin A. Chapman, and Dr. Alan Bolten, for their generous contributions in the form of advice, expert ise, and good-natured support to better my work. I express my deep appreciation to Dr. De bra Murie and Dr. Daryl Parkyn for kindly allowing me to use their fish-agi ng laboratory facilities, and for th eir generous logistical support, much needed advice, and encouragement during th e laboratory phase of my research. I am grateful too to Edward Leonard for introducing me to the techni que of sanding and polishing that greatly enhanced reading of growth rings on tilapia otoliths. My sincere gratitude goes to Prof. Lauren J. Chapman and Prof. Colin A. Chapman for providing the infrastructure and other logistical support throu gh the Kibale Fish and Monkey project in Uganda, which made the research phase of my work possible. My thanks also go to James Kyomuhendo (Abwooki), James Magaro (Apuu li) and John the fisherman, for their invaluable field assistance during data collection. Their enthusiasm and dedication to the work

PAGE 5

5 as well as the humorous stories made the data coll ection such an enjoyable experience. I am also grateful to Judy Dumont, Cassa ndra McEwan, and Ruoqi Wang of McGill University, Canada, for their help with laboratory analysis. Prof Frank Nordlie of the Department of Zoology, University of Florida and Prof. Jaap Kalff of McGill University, Canada offered helpful comments and suggestions on the limnology chapter. I extend my thanks to all friends, colleagues, fe llow graduate students, staff of Kibale Fish and Monkey projects, and Makerere University Biological Field Sta tion for their advice, friendship, hospitality, and support during the course of my work. I thank my parents, brothers, sisters, nieces and nephews for being a constant source of love, support, and encouragement that have kept ringing in my ears over the years. Finally, I must also acknowledge funding sour ces without which my Ph.D. program would not have been successful. My stay in USA wa s supported by the University of Florida through a Teaching Assistantship at the Department of Zoology. Funding for my field research was provided by the Program for Studi es in Tropical Conservation (PSTC)-COMPTON fellowship at the University of Florida, the International Foundation for Science (IFS), Sweden, The Whitley Laing Foundation for Internationa l Nature Conservation/Rufford Small Grant and Niddrie Small Grants through the Centre for African studies, for which I am grateful.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ........10 LIST OF FIGURES................................................................................................................ .......11 ABSTRACT....................................................................................................................... ............14 CHAPTER 1 OVERVIEW OF LIFE HISTORY THEO RY, ENVIRONMENTAL VARIATION, AND SIZE-SELECTIVE FISHING.......................................................................................16 Background..................................................................................................................... ........16 Life History Theory (LHT).....................................................................................................16 Fishing-A Driver of Life History Change..............................................................................18 Fishing and Concomitant Environmental Stressors................................................................19 Tilapiine Fishes............................................................................................................... ........22 The Crater Lakes-Experimental Test Tubes for Life History Studies....................................23 Overview of the Study.......................................................................................................... ..25 2 ANTHROPOGENIC DISTURBANCES AND PREDICTIVE MODELS OF CHLOROPHYLL a IN VOLCANIC CRA TER LAKES OF WESTERN UGANDA..........29 Introduction................................................................................................................... ..........29 Methods and Materials.......................................................................................................... .33 Study Area..................................................................................................................... ..33 Climate, Vegetation, and Soils........................................................................................34 Bathymetry and Morphometry........................................................................................35 Physical and Chemical Characters..................................................................................35 Statistical Analyses..........................................................................................................36 Results........................................................................................................................ .............37 Environmental Data-Lake Morphometry........................................................................37 Temperature Profiles.......................................................................................................38 Dissolved Oxygen Concentration Profiles......................................................................38 pH............................................................................................................................. .......38 Specific Conductance......................................................................................................39 Water Transparency (Secchi Depth)...............................................................................39 Chlorophyll a, Total Phosphorus, and Total Nitrogen....................................................39 Chlorophyll a-Water Transparency and Chlorophyll a-Total Phosphorus Relationships................................................................................................................40 Discriminant Function Analysis (DFA)..........................................................................40 Discussion..................................................................................................................... ..........41 Morphometric Features...................................................................................................41

PAGE 7

7 Limnological Characters.................................................................................................41 Chlorophyll a-Water Transparency and Chlorophyll a-Total Phosphorus Relationships................................................................................................................44 3 PREDICTORS OF FISH CONDITION IN INTRODUCED TILAPIAS OF UGANDAN CRATER LAKES IN RELATI ON TO FISHING PRESSURE AND DEFORESTATION................................................................................................................58 Introduction................................................................................................................... ..........58 Methods and Materials.......................................................................................................... .61 Study Area..................................................................................................................... ..61 Environmental Data.........................................................................................................61 Fish Collections...............................................................................................................63 Calculation of Relative Condition (Kn)..........................................................................63 Statistical Analyses..........................................................................................................64 Results........................................................................................................................ .............65 Length-Weight Relationships and Relative Condition....................................................65 Fishing Pressure, Catchment Defo restation, and Condition Factor................................65 Environmental Characters and Relative Condition Factor..............................................66 Discussion..................................................................................................................... ..........66 4 VALIDATION OF PERIODICITY AND TIMING OF OPAQUE ZONE FORMATION IN Tilapia zillii (Pisces: Cichlidae) OTOLITHS FROM CRATER LAKE NKURUBA, WESTERN UGANDA..........................................................................80 Introduction................................................................................................................... ..........80 Materials and Methods.......................................................................................................... .84 Study Area..................................................................................................................... ..84 Fish Collections and Processing......................................................................................85 Length-Weight Relationships..........................................................................................85 Length Frequency............................................................................................................86 Otolith Processing and Interpretation..............................................................................86 Interpretation of Otolith Secti ons and Precision of Age Estimates.................................87 Marginal Increment Analysis (MIA)...............................................................................89 Growth......................................................................................................................... ....90 Estimates of Total Mortality and Total Apparent Survival Rates...................................90 Instantaneous Natural Mortality Rate (M)......................................................................91 Results........................................................................................................................ .............91 Length-Weight Relationships..........................................................................................91 Length Frequency............................................................................................................92 Tilapia zillii Growth Zone Identification........................................................................92 Validation of Periodicity and Ti ming of Increment Formation.......................................93 Age and Growth..............................................................................................................94 Mortality Estimates.........................................................................................................94 Natural Mortality (M)......................................................................................................95 Discussion..................................................................................................................... ..........95 Length-Weight Relationships..........................................................................................95

PAGE 8

8 Length Frequency............................................................................................................95 Otolith Preparation and Interp retation of Growth Zones................................................96 Growth......................................................................................................................... ....98 Mortality Rates................................................................................................................99 5 LIFE HISTORY VARIATION IN Tilapia zillii (Pisces: Cichlidae) IN THE CRATER LAKES OF WESTERN UGANDA.....................................................................................119 Introduction................................................................................................................... ........119 Variation in Life Hist ory Characters in Fish.................................................................119 Size-Selective Fishing...................................................................................................120 Fishing and Phenotypic Plasticity.................................................................................122 The Environment...........................................................................................................123 The Tilapias-A useful Model for Unde rstanding Life History Variation......................124 The Crater Lakes-Suitable Experimental Test-Tubes for Life History Studies.........124 Materials and Methods.........................................................................................................126 Study Lakes...................................................................................................................126 Environmental Data.......................................................................................................126 Fish Collections.............................................................................................................127 Size Frequency..............................................................................................................128 Batch Fecundity.............................................................................................................128 Size at 50 % Maturity (L50)...........................................................................................128 Age Determination........................................................................................................129 Growth......................................................................................................................... ..129 Estimates of Total Mortality (Z) and Total Apparent Survival Rates (S).....................130 Statistical Analyses........................................................................................................130 Results........................................................................................................................ ...........131 Size Frequency..............................................................................................................131 Sex Ratio...................................................................................................................... .131 Batch Fecundity.............................................................................................................132 Size at Maturity.............................................................................................................132 Age and Growth............................................................................................................132 Instantaneous Total Mort ality (Z) and Total Apparent Survival (S).............................133 Relationships Among Life-History Traits.....................................................................134 Environmental Predictors of Life -History Variation among Lakes..............................134 Discussion..................................................................................................................... ........136 Sex Ratio...................................................................................................................... .136 Batch Fecundity.............................................................................................................137 Size at Maturity.............................................................................................................138 Age and Growth............................................................................................................138 Instantaneous Total Annual Mortality Rates (Z)...........................................................142 Relationships among Life-History Traits......................................................................142 Environmental Predictors of Life-history Variation among Lakes...............................144 Potential Fish Yield Models..........................................................................................146

PAGE 9

9 6 GENERAL DISCUSSION AND CONCLUSION...............................................................163 General Conclusions......................................................................................................163 The Way Forward (Recommendations)........................................................................167 Future Studies................................................................................................................167 LIST OF REFERENCES.............................................................................................................168 BIOGRAPHICAL SKETCH.......................................................................................................188

PAGE 10

10 LIST OF TABLES Table page 2-1 Some morphometric featur es of the Kasenda cluster of Crater Lakes in western Uganda......................................................................................................................... ......47 2-2 Mean euphotic values ( S. E) of physical and chemical parameters analyzed for the crater lakes of western Uganda..........................................................................................48 2-3 Chlorophyll a-Secchi and Chlorophyll a-total phosphorus relationships for 19 western Uganda crater lakes..............................................................................................49 3-1 Lake catchment deforestation level, fi shing pressure, range in total length and degonaded weight, and lengthweight regressions for Oreochromis leucostictus .............73 3-2 Lake catchment deforestation levels, fi shing pressure, range in total length and degonaded weight, and lengthweight regressions for Tilapia zillii ..................................74 4-1 Physical and limnological characters of surface water of Lake Nkuruba, Uganda.........103 4-2 Mean total length at age (mm), standard error (S.E.), and 95% confidence intervals (CI) of Tilapia zillii captured with gillnets and mi nnow traps in Lake Nkuruba............104 4-3 Results of 2-way ANOVA testing for the e ffect of gear type and age on mean total length of Tilapia zillii from Lake Nkuruba, western Uganda..........................................105 5-1 Life history charaters of Tilapia zillii determined in the Crater Lakes of western Uganda from July 2004-June 2005..................................................................................149 5-2 Macroscopical criteria used to stage gonadal development in Tilapia zillii (Modified from Siddique, 1977) from crater lakes of western Uganda............................................150 5-3: Number of females and males and sex ratios of T. zillii caught in 14 crater lakes of western Uganda ( -significant at 5% level, ** -significant at 1% level)..........................151 5-4 Estimated total annual instan taneous mortality rate (Z year-1), regression statistic (r and P-values) and estimated annua l apparent survival (S) for Tilapia zillii. ...................152 5-5 Results of principal component analysis (PCA) describing the major environmental gradients of variation in Tilapia zillii life history charaters among the lakes..................153

PAGE 11

11 LIST OF FIGURES Figure page 2-1 Map of Kasenda cluster of cr ater lakes in western Uganda (0o23-0o33N, 30o10-30o 20E) showing the lakes sampled from July 2004-July 2005.............................................50 2-2 Temperature and dissolved oxygen profile s for crater lakes of western Uganda sampled from July 2004-July 2005....................................................................................51 2-3 pH and conductivity profiles for crater la kes of western Uganda sampled from July 2004-July 2005................................................................................................................. ..53 2-4 Relationship between chlorophyll a and S ecchi depth for crater lakes of western Uganda and subtropical lakes............................................................................................55 2-5 Relationship between chlorophyll a and tota l phosphorus for crater lakes of western Uganda, other African lakes, and subtropical lakes...........................................................56 2-6 Results of discriminant analysis, illustra ting the ability to categorize crater lakes of western Uganda using their physical, ch emical, and morphometric characters................57 3-1 Location of Kasenda cluster of crater lakes in western Uganda (0o23 -0o33 N, 30o10 30o 20 E) where Oreochromis leucostictus and Tilapia zillii were sampled.....................75 3-2 Regional relationships between degonaded weight (g) and total length (mm) for A) Oreochromis leucostictus. B) Tilapia zillii derived by pooling data from 13 lakes.........76 3-3 Mean relative condition of Oreochromis leucostictus and Tilapia zillii in crater lakes of western Uganda exposed to high, medium, and low fishing mortalities.......................77 3-4 Mean relative condition of Oreochromis leucostictus and Tilapia zillii in crater lakes of western Uganda exposed to moderate and severe deforestation levels.........................78 3-5 Comparison of length wei ght-regression curves. A) O. leucostictus from crater lakes and Lake Naivasha, Kenya. B) for T. zillii from crater lakes and other African lakes.....79 4-1 Map of Lake Nkuruba (bottom right ) showing its location in Uganda............................106 4-2 Photograph of a polishe d transverse section of Tilapia zillii otolith from Crater Lake Nkuruba, Western Uganda...............................................................................................107 4-3 Transverse section of a sagi ttal otolith from a 6-year old Tilapia zillii showing the appearance of opaque zones translucent z ones, and core region of the otolith...............108 4-4 Relationship between total weight (g) a nd total length (mm) for male (open circle) and female (closed circles) Tilapia zillii from Lake Nkuruba, western Uganda.............109

PAGE 12

12 4-5 Sagittal otoliths of Tilapia zillii from Crater Lake Nkuruba western Uganda. A is distal view and B is dorsal (medial) view........................................................................110 4-6 Length-frequency distribu tion by 10-mm size class for male, female and combined sexes (male + female) Tilapia zillii collected from Lake Nkuruba in western Uganda,.111 4-7 Length frequencies (% of total catch) of Tilapia zillii collected in minnow traps and gillnets from Lake Nkuruba, western Uganda, from July 2005-June 2006.....................112 4-8 Mean marginal increment of thin-sectioned Tilapia zillii otoliths from Lake Nkuruba, western Uganda over a 12-month period.........................................................................113 4-9 Percentage of otoliths of Tilapia zillii with opaque edges (lin e) from Lake Nkuruba and the mean monthly rainfall (area)...............................................................................114 4-10 Observed length-at-age for Tilapia zillii fitted to the von Bertalanffy growth model forced through zero for fish sampled w ith traps and gillnets from Lake Nkuruba..........115 4-11 Age frequency of Tilapia zillii collected with traps and gillnets expressed as a percentage of the overall catch fr om Lake Nkuruba, western Uganda............................116 4-12 Age frequency and linearized catch curv es for estimating total mortality (Z) for T. zillii caught with traps and gillnet s from Lake Nkuruba, Uganda...................................117 4-13 Size frequency of Tilapia zillii from Lake Nkuruba, west ern Uganda, by month from July 2005 and June 2006..................................................................................................118 5-1 Length-frequency distribu tion by 20-mm size classes for Tilapia zillii collected from 2004 to 2005 from 12 crater la kes in western Uganda....................................................154 5-2 Length frequency of Tilapia zillii sampled with gillnets and minnow traps (as a percent of total catch) from eight crater lakes, western Uganda......................................155 5-3 Relationship between batch fecundity (# no of eggs at stage VI) and A) total length. B) total weight for Tilapia zillii from 10 crater lakes in western Uganda.......................156 5-4 Maturity ogives and size at 50% maturity (L50) of female and male Tilapia zillii from crater lakes of western Uganda........................................................................................157 5-5 Age frequencies of Tilapia zillii from eight crater lakes in western Uganda collected with minnow traps and gillnets between July 2004 and June 2005.................................158 5-6 Mean length-at-age for Tilapia zillii captured with gillnets (open circle) and minnow traps (closed circles) from eight crater lakes in western Ugandan from June 2004 to July 2005. Vertical bars are standa rd errors of the mean (SE).......................................159

PAGE 13

13 5-7 Relationship between mean total length-a t-age 3 (growth) and total annual apparent survival (S) for Tilapia zillii captured with gillnets in eight crater lakes, western Uganda......................................................................................................................... ....160 5-8 Relationship between total annual apparent survival (S) and PC1 for eight crater lakes with varying levels of de forestation and fishing pressure......................................161 5-9 Relationship between mean total length-a t-age 3 (growth) and PC1 for eight crater lakes with varying levels of deforestati on and fishing pressure (low, medium, and high).......................................................................................................................... .......162

PAGE 14

14 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy LIFE HISTORY VARIATION IN TILAPIA POPULATIONS WITHIN THE CRATER LAKES OF WESTERN UGANDA: THE ROLE OF SIZE-SELECTIVE PREDATION By Jackson Efitre August 2007 Chair: Lauren J. Chapman Major: Zoology Life history theory seeks to describe and explain the evolu tion of adaptive responses in fitness-related traits such as reproduction, survival, age and size at maturity, and growth as a result of environmental variati on. Interpopulational variation in life history characters is well documented in a wide variety of fish taxa, both ma rine and freshwater. Va riability in fish life history characters may be a resu lt of phenotypic plasticity in re sponse to environmental changes and/or genetic changes associated with evolutio nary response to size-selective fishing. For effective fisheries management, it is important to consider existing knowledge about fish life history strategies, biotic and abiotic factors simultaneously in predicting population responses to changes in the environment. I examined effects of size-selective fishing and other environmental factors on key life history characters of tilapia popul ations in the crater lakes of western Uganda with the hope of providing critical information needed for management of the fisheries in these lakes. I used tilapia as a model species because li fe history variation in tilapiine fishes is welldocumented in tropical freshwaters. Tilapia also display high levels of phenotypic plasticity and can tolerate a wide range of environmental conditions due to their ability to vary allocation of resources to reproduction and growth depe nding on the environmental condition. The occurrence of both unexploited and exploited tilapia populations in a large number of the lakes

PAGE 15

15 that also differ in extent of catchment deforestat ion and fishing pressure further provided suitable systems and the replication need ed to explore life history vari ation across broad environmental gradients. I determined a range of environmen tal characters in 19 crater lakes and found a wide variation in environmental f eatures among lakes. There wa s a strong negative relationship between water transparency and Chl-a concentr ation with deforested lakes having a lower transparency and higher Chl-a concentration compar ed to forested lakes. I also explored the effects of deforestation and fishing on the condition of two introduced species ( Oreochromis leucostictus and Tilapia zillii ) in 17 of the crater lakes and found O. leucostictus in severely deforested and heavily fished lakes were in a be tter condition compared to similar fish in lakes with low productivity and low to medium fishing. Differences in condition of T. zillii were only detectable between lakes with high and low fish ing effort. I also developed an appropriate protocol for aging T. zillii in Lake Nkuruba and validated th e periodicity and timing of opaque zone formation in otoliths of T. zillii using marginal-increment and edge analyses. Tilapia zillii in Lake Nkuruba deposited two opaque zones on th eir otoliths per year corresponding to bimodal rainfall that is characteristic of this equatorial region. Tilapia zillii grew rapidly in the first 2 years after which growth slowed down considerably. Gillnetted Tilapia zillii grew faster and attained a larger size-a t-age than the trapped fish. Finall y, I examined the effects of sizeselective predation and environmental factors on life history traits of T. zillii in the 8 crater lakes. I found striking differences in life history traits in T. zillii between and within lakes, suggesting that environmental factors, density-dependence, and size-selective fishing influence variation in key life history characters of T. zillii populations in the crater lakes of western Uganda. These results provide useful baseline data for the ma nagement of heavily e xploited and unexploited T. zillii populations in the crater lakes of western Uganda.

PAGE 16

16 CHAPTER 1 OVERVIEW OF LIFE HISTORY THEORY, ENVIRONMENTAL VARIATION, AND SIZESELECTIVE FISHING Background Predicting fish population respons es to environmental change continues to be a major challenge to fisheries scientists. Fish populati ons are increasingly faced with both natural and anthropogenic changes in the environment, such as fluctuations in rainfall and temperature, deforestation of watersheds, and overexploitation. Predicti ng population responses to such changes is critical, as many fishes tend to display life history variation across divergent environments and in response to environmen tal change (Lowe McConnell, 1982; Reznik and Bryga, 1987; Reznik et al., 1990; Winemiller and Rose, 1992). Meeting the challenges of prediction, conservation, restorat ion, and effective management of fish populations requires an integration of existing knowledge about fish life hi story strategies with co mponents of the biotic and abiotic environment. Life History Theory (LHT) The focus of life history theory (LHT) has tradi tionally been on traits related to apparent survival and reproduction and tradeoffs between traits that maximize reproductive rate (Roff, 1992; Stearns, 1992). Life history theory seeks to describe and ex plain the evolution of adaptive responses in fitness-rela ted traits such as reproduction, a pparent survival, age and size at maturity, and growth as a result of environmen tal variation (Roff, 1992; Stearns, 1992). Life history theory is based on the con cept of trade-offs. That is, you ca nnot have it all; a trait that is optimal in one environment may be suboptimal in another. For a given environment, evolution helps to optimize life history stra tegies by balancing tradeoffs of key traits (Stearns, 1976; Roff, 1984; Wootton, 1992). Trade-offs in traits such as growth and reproduction are especially critical in animals with indeterm inate growth such as fishes wher e batch fecundity of female fish

PAGE 17

17 is strongly correlated with size, and allocati on of resources to repr oduction may lead to a decrease in growth rate (Stear ns, 1992). Variation in life hist ory characters induced by such tradeoffs can have extreme consequences for fish population dynamics. Interpopulational variation in life history characters is co mmon in many fish species and includes, as examples, variation in batch fecundi ty (Peters, 1963; De Silva, 1986; Duarte and Alcaraz, 1989; Legendre and Ec otin, 1989, 1996; Dupochelle et al ., 2000), egg size and energy (Blaxter, 1986; Chambers and Leggett, 1996; Cham bers, 1997), size and age at maturity (Reznik et al., 1990; Dupochelle and Panfili, 1998), and growth (Leggett and Power, 1969; Pauly, 1980; Roff, 1984; Hutchings, 1993). Variation in many of these characters is interrelated. fish populations in stable environments with high in traspecific competition tend to have lower batch fecundity, larger eggs, and highe r size at maturity (Lowe McConne ll, 1982; Reznik and Briga, 1987; Reznik et al., 1990), and higher juvenile mortality (Reznik, 1982; Reznik and Briga, 1987; Reznik et al., 1990) than populat ions in variable environments although there are exceptions. Differences in growth rates can al so influence size and age at ma turity, and variation in size and age at maturity of individuals affects growth of fish. Empirical population dynamics studies have mainly been based on conventional stock assessment models, which do not account for plasticity of life history traits exhibited by fish species such as the tilapias. An understanding of evolution of fish life history is critical to fish eries management as reduction in fish growth and maximum size may lead to stunted populations with low economic or recreational value. Stunted growth or the production of a large number of individuals having a low maximum size, is a very widespread phenomenon in freshw ater fish populations and has been observed across a broad range of phylogenetically distin ct lineages including as examples: salmonids (Leggett and Power, 1969) coregonids (Ridgewa y and Chapleau, 1994), cyprinids (Burrough and

PAGE 18

18 Kennedy, 1979), Arctic charr (Parker and Johnson 1991), and perch (Rask, 1983; Ridway and Chapleau, 1994). Clearly, stunting in fish popula tions has very important economic implications because the commercial value of stunted fish populat ions is greatly diminished and there is great interest in understanding and mitigating stunting in harvested populations. Stunting in fishes has been attributed to several f actors including: intraspecific competition due to overcrowding (Sandheinrich and Hubert, 1984; Roff, 1992), re duced availability of food (Rask, 1983; Roff, 1992; Post and McQueen, 1994), and increased juvenile and adult fish survival (Roff, 1992) as a result of absence of predators and low fishing pressure, respectively. Many of these factors reflect resource limitati on, but direct links between ec ological factors and fish life histories are not well understood. Fishing-A Driver of Life History Change Fishing is typically size-selec tive and non-random in that fishi ng gear is designed to target larger, presumably faster grow ing and older individuals in a population (Law, 2000). And, it is often assumed that harvest-induced reductions in population density of the target stock will lead to increased yield because of reduction in intr aspecific competition that releases populations from density-dependence resulting in faster growth (Jennings and Kaiser, 1998; Hall, 1999; Rochet et al., 2000). High populat ion density (low fishing effort ) should increase competition for resources (e.g., food) leading to density-dependent growth. Under reduced and unpredictable food availability, one would pr edict increased allocation of resources towards reproduction, resulting in early maturation at a smaller size. A decrease in population density through predation or fishing should reduce competition for resources and lead to faster growth. When resources are adequate, one would predict increa sed allocation to growt h, resulting in delayed maturity at a larger size. Implicit in these pred ictions is the assumption that changes in fitness are selectively neutral i.e. ther e is no evolution in response to size-selective ha rvest (Law, 2000;

PAGE 19

19 Walsh et al., 2006). If a signifi cant component of variation in lif e history characters reflects environmentally-induced plasticity, then altering predator pressure (through selective harvest) may be an effective way of changing life history characters to increase yield. Interestingly, many fisheries show changes in lif e history traits that are not consistent with releasing the target stock from intraspecific competition (Law 2000), and suggest that genetic change in response to selectiv e effects of fishing may be de trimental to recovery of the population when fishing pressure is reduced. Strong directional sel ection by size-selective fishing can result in a shift in the spawni ng stock towards smaller and slower-growing individuals. If there is a genetic basis for the phenotypi c variation (e.g., size) between individuals, then size-selectiv e fishing may lead to evoluti on towards smaller size-at-age (Law, 2000; Walsh, 2006). Walsh et al. (2006) subject ed the Atlantic silver side to experimental size-selective predation over five generations and found that populations subjected to harvest of large fish exhibited declines in batch fecund ity, egg size, larval si ze at hatching, larval survivorship and growth rate, a nd declined in food consumption and conversion efficiency. And, in a long-term study of grayling populations (Hau gen and Vllestad, 2001) also found that sizeselective gillnet fishing resulted in a reduction in age and length at maturity in populations with recent common ancestors. Fishing and Concomitant Environmental Stressors The effects of fishing on the target stocks ma y also be impacted by concurrent changes in the biotic and abiotic environment. Lakes that are heavily fished ar e often lakes that are accessible (roads, trails, development) and subj ect to other anthropogenic pressures such as deforestation and eutrophication. A critical step in beginnin g to understand variation in life history traits is to explore th e interaction between si ze-selective predation and other biotic and abiotic factors that may affect resource availabi lity or otherwise alter the selective regime.

PAGE 20

20 Human activities such as land use modificati on (e.g., animal grazing, deforestation, and other reductions in vegetation cover) greatly accelerate changes in aquatic ecosystems; and the recent expansion and accelerating rate of deforestation has caused widespread co ncern, particularly in the tropics where the impacts of de forestation and forest degradat ion on aquatic systems remain largely unknown (Chapman and Chapman, 2003). In Africa, it is estimated that rain forests once covered 3,620,000 km2 before anthropogenically-induced habitat alterations (Martin, 1991); now the continent ranks second only to South America in net rate of forest loss (FAO, 2005). According to the Food and Agricultural Organizations Global Forest Resources Assessment (FAO, 2005), the global estimate of net change in forest area betw een 2000-2005 was about 7.3 million ha per year ( 200 km2 of forest per day). Between 1990 and 2005, forests cover in Africa decreased from 655.6 million ha to 635.4 million ha, a consequence of deforestation, selective logging, as well as other human interventions (FAO, 2005). Before thes e anthropogenic disturba nces, rain forests of Africa covered an estimated 74% of Central Africa, 19% of West Africa, and 7% of East Africa (Martin, 1991). By 1985, remaining forest had b een reduced to 55% of the original area in Central Africa, and 28% of the original area in both West and East Africa (FAO, 1985). In Uganda, the focus of this study, it is estimated th at 86% of the tropical moist forest has been cleared (WRI, 1994). Within these deforested areas lie several many heavily exploited fisheries, emphasizing the link between intense fish harvest and other anthropogenic stressors. Within the Lake Victoria watershed, the rapid rise in hum an population has put sign ificant pressure on the lake environment. It is estimated that human population growth rate ar ound Lake Victoria basin is about 3-4 % per annum (Bugenyi and Magumba 1996; Ogutu-Ohwayo et al., 1997). Forests in the watershed are being rapidly cleared for agriculture, firewood, charcoal and human

PAGE 21

21 settlements, leading to increased deforestation and associated erosion, siltation, sedimentation, and nutrient loading into the lake creating a noxic conditions during st ratification (Hecky, 1993; Ogutu-Ohwayo et al., 1997); yet, the lake represents Africas most important inland fishery. Although fish stocks have been heavily explo ited in the lake since the early 1900s (OgutuOhwayo, 1990a; Ogutu-Ohwayo et al., 1997), there is increasing evidence that water quality is also impacting the fish fauna (Hecky, 1993; Mugi dde, 1993). Water quality in Lake Victoria has declined greatly in the past few decades, owi ng chiefly to eutrophication arising from increased inflow of nutrients into the la ke. In addition, changes in sp ecies composition and abundance of benthic invertebrates, zoopla nkton, and phytoplankton have also been reported in the lake (Hecky, 1993; Ogutu-Ohwayo et al., 1997). More attention needs to be directed towards investigating the effects of defo restation on this important ecosys tem and other smaller tropical lakes, as degradation of watersheds may result in deterioration of fish habitats and loss of biodiversity. Although impacts of deforestation on aquati c systems are better unde rstood for temperate regions than tropical regions, recent studies in East Africa emphasize the dramatic impacts of deforestation (Cohen et al., 1993; Al in et al., 1999). An effect of deforestation readily apparent is increased siltation and sedimentation. Re moval of the forest decreases the rate of evapotranspiration and the inte rception of rain and increase s runoff and sediment yield. Increased sedimentation and higher turbidity can lead to the decline of plankton through a reduction in light penetration, a nd the disappearance of many benthi c rheophilic animals that are sensitive to mud on their integument and gills or lose thei r interstitial habitats to clogging by silt (Chutter, 1968; Burns, 1972; Ma rlier, 1973; Welcomme, 1983). C ohen et al., (1993) contrasted undisturbed, moderately disturbed, and highly disturbed areas to de tect effects of sedimentation

PAGE 22

22 on the fauna of Lake Tanganyika. They found that the species richness of ostracods and fishes appeared to be heavily affected by sediment, while diatoms showed little change (Cohen et al., 1993). In a later study of the same region, Alin et al., (1999) showed that the species richness and density of fishes, mollusks, and ostracods were negatively correlated with sedimentation. Deforestation impacts may partic ularly be acute in areas wi th high topographic relief that accelerates runoff and sedimentation in response to forest loss. High-altitude forests that once suffered reduced human pressures because of th e challenges of land conversion on steep slopes are now experiencing accelerating forest clearance. This has been partic ularly severe in the high-altitude forests of East Africa where human migration and population growth contribute to accelerating human impacts. Concomitant with increasing deforestation is also increasing exploitation of high-altit ude lake resources dominated by tilapiine fishes. Tilapiine Fishes Tilapiine fishes are members of the Family Ci chlidae (Order Percifor mes)-one of the three largest vertebrate families with estimated spec ies richness ranging from 1300 to 1900 (Froese and Pauly, 2006). Tilapiines are the most importa nt cichlids in terms of their exploitation as food fishes. Their rapid growth, tolerance to st ocking density, and high adaptability has led to widespread introduction and extens ive farming in different part s of the world. Production of farmed tilapiines is estimated at about 1.5 millio n tonnes annually with an estimated value of US$1.8 billion (De Silva et al., 2004) making them the ideal "aquatic chic kens" of the trade (Barlow, 2000). Their ecological and economic success has been attr ibuted to their ability to tolerate a wide range of environmental conditions (Wootton, 1984) and their ability to vary allocation of resources to reproduction and grow th depending on environmental conditions (Fry er and Iles, 1972; Lowe-McConnell, 1982). Several studies have described life hi story variation in tilapiines

PAGE 23

23 (Lowe McConnell, 1958, 1982; Welcomme, 1970; Fryer and Iles, 1961, 1972; Gwahaba, 1973; Noakes and Balon, 1982; Stewart, 1988; Ribbi nk, 1990; Kolding, 1993; Dupochelle and Panfili, 1998; Dupochelle et al., 1999, 2000), which has been attr ibuted, at least in pa rt, to their evolution in variable environments (Low e-McConnell, 1982; Noakes and Ba lon, 1982). This variation in life history characters is ofte n reflected in stunted tilapia populations characterized by low maximum size as a result of early maturation (Iles, 1973; Ridgeway and Chapleau, 1994). Understanding the links between eco logical factors and tilapia life history variation is important for fishery management. Apart from the work on Oreochromis niloticus (Lowe-McConnell, 1958), Oreochromis mossambicus (De Silva, 1986; James and Bruton, 1992), Sarotherodon melanotheron and Tilapia guinensis (Legendre and Ecotin, 19 89, 1996), data on interdemic variation in important life hist ory characters (growth, size and ag e at maturity, batch fecundity, and condition) are lacking. The occurrence of both unexploited (no fishing) and exploited (high fishing) tilapia populations in a large number of diverse crater lakes in Uganda provides an excellent model system to explore life history va riation across broad environmental gradients. The Crater Lakes-Experimental Test Tubes for Life History Studies While the African Great Lakes (e.g., lakes Vi ctoria, Tanganyika, and Malawi) have been the focus of numerous inves tigations (Beadle 1981, Lowe-M cConnell et al., 1992), much less attention has been paid to the minor lakes. Up to 18% of Ugandas to tal land area is covered by lakes, rivers and swamps. These lakes and ri vers are sources of fish and they provide employment, income, foreign exchange earning, and account for over 50% of animal protein supply in Uganda (Ogutu-Ohwayo, 2000). There ar e approximately 160 small lakes scattered in different parts of Uganda, mainly in centra l, eastern, and western Uganda. These lakes collectively contribute about 1.9% of the national fish production (FRD, 1999). Besides providing fish to the local communities, some of these lakes also harbor endemic flocks of

PAGE 24

24 haplochromine cichlids. Effective management and conservation of fish species in these lakes is important given their economic, ecologi cal, and scientific significance. In western Uganda, there are approximatel y 80 small volcanic crater lakes along the foothills of the Rwenzori Mountains. The first person to investigate these lakes limnologically was Beadle (1963, 1981). He described their meromi ctic stratification patt erns and also noted zooplankton living in virtuall y oxygen-free hypolimnic water laye rs. Later on, Melack (1978) surveyed 16 of the lakes in th e region morphologically, hydrol ogically, and chemically. He emphasized the striking limnological diversity of the lakes. Contrary to early speculation (Beadle, 1981) that small sheltered tropical lakes are unproductive, recent research (Kizito et al., 1993; Chapman et al., 1998) suggests that some of the crater lakes are very productive (e.g., net productivity of Crater Lake Nkuruba is 1.3 g Carbon/m2/day, Chapman et al., 1998). Tilapia species ( Oreochromis niloticus, Oreochromis leucostictus, and Tilapia zillii) were introduced into several of these crater lakes in the 1940s to increase available protein to the local communities. Most of the lakes were fish-free prior to the introductions, but some harbored a variety of native species. Currently, many of these lakes are produc ing stunted tilapia populations, the cause of which remain unknown but may reflect resource limitations associated with the Crater Lake environment and/or low mo rtality leading to high levels of intraspecific competition. Observations collected from over 10 years of work in the crater lakes region of western Uganda suggest that the gr eatest threat to the cr ater lakes ecosystem is related to clearing of the forested crater rims associated with the need for fuelwood, building materials, and conversion to agricultural land (Chapman et al., 2003 ) Deforestation causes soil erosion that leads to increased turbidity, si ltation, and eutrophication (C risman et al., 2001); and these

PAGE 25

25 anthropogenic perturbations may al so contribute to variation in water quality and fish production in the lakes. The variation among the lakes in fishing effort and othe r environmental characters provides an excellent opportunity to identify predictors of lif e history variation. My Ph.D. research contributes directly to our understanding of the interplay between environmental factors and life history characters of tilapia species in the crater lakes, and more generally, the interplay between life history traits and size-selective predation. Overview of the Study The overall goal of my study was to examine effects of size-selective fishing and other environmental factors on life histor y traits of populations of tilapia in crat er lakes of western Uganda. As there are both unfis hed and heavily fished populations in these lakes, fishing may be an important factor in the evoluti on of life history traits in the lake s. As in any natural system, it is difficult to control other factor s while examining the effect of a target variable; the effect of fishing on tilapia populations may be influenced or swamped by variation in environmental conditions. If environmental factors such as water temper ature, dissolved oxygen or food availability were to change so that fish gr ew more slowly, these factors would confound the effect of fishing that tends to select against larger and fast er growing indivi duals. It was important to consider fishing and a suite of other environmental factors simultaneously in assessing life history variation in tilapia populations in the cr ater lakes. I estimated and compared key life history traits of Tilapia zillii (growth, age and size at maturity, condition, and batch fecundity) among 8 crater lakes that differ in fishing pressure (high vs. low effort), deforestation (categorical), and vary (continuously) in other key environmental factors that may affect life history characters (e.g., di ssolved oxygen, chlorophyll a, etc.)

PAGE 26

26 Aquatic ecosystems are under increasing threat from deforestation, a nd volcanic crater lakes may be particularly sensitive to such pert urbation because of their small size, steep-sided slopes, and limited flushing. I described the physi cal and chemical characteristics of 19 volcanic crater lakes in western Uganda, tested for diffe rences in physico-chemi cal conditions of lakes with varying extent of defore station, and developed predictive models of primary productivity (chlorophyll a) for the crater lakes ecosystem. The limnological and morphometric characters measured were: dissolved oxygen concentration, wa ter temperature, pH, electrical conductivity, water transparency (Secchi), ch lorophyll a, total n itrogen, total phosphorus, lake area, maximum depth, and mean depth. Since water quality may have a direct effect on the biology of tilapia populations in these lakes, I explored the effects of anthropogenic factors (i.e., defo restation and fishing) on the condition of two introduced species ( Oreochromis leucostictus and Tilapia zillii ) in 17 of the crater lakes that differed in extent of land c onversion and fishing pressure. The two tilapia species were selected because of their divergen t trophic niches as adul ts (planktivore versus herbivore). The specific questi ons I addressed were: 1) Is the condition of the two tilapia species significantly different among lakes with low, medium, and high fish ing pressure? 2) Are there any predictive relationships between condi tion of the two tilapia species and biotic and abiotic factors in these lakes?, and 3) if there are relationships do they reflect differences in levels of deforestation among the lakes for both species? The result s were discussed with respect to management of tilapia populat ions in the lakes where they sustain artisanal fisheries. As this was a pioneering study of tilapia life hi story traits in the crat er lakes, there was no background information to assist in developing gr owth models for these systems. Although there is a growing literature on age and growth studies in a wide range of tropical environments and

PAGE 27

27 species, otoliths and other calcified structures of tropical fish species ha ve historically been assumed not to form increments that relate to variation in growth resulting from seasonal fluctuations in the environment. Prior to using ca lcified structures such as otoliths for aging fish, the timing and periodicity of growth zone formati on must be validated to ascertain the accuracy and feasibility of the aging criteria. It was im portant to develop an a ppropriate aging criterion for tilapia populations in the crater lakes before obtaining age estimates used for deriving growth and other life history data. I developed an appropriate crite rion for interpreting growth at the edge of thin-sectioned Tilapia zillii otoliths in one Crater Lake (Nkuruba). I used the criteria developed to validate the periodicity and timing of opaque zone formation w ithin otolith sections using the technique of marginal increment analysis (MIA). Finally, I used the validated age estimates to determine growth rate, age at first matu rity, and mortality rates of T. zillii populations within Lake Nkuruba. Since all the Kasenda crater lakes lie in close proximity to Lake Nkuruba, I used the same ageing criteria developed for Lake Nkuruba to estimate a range of age-based population parameters such as growth, size at maturity, and mortality in the rest of the lakes based on the assumption that the crater lakes are in the same c limatic zone and that the pattern of growth ring formation on otoliths of T. zillii populations in these lakes is driven by the same climatic factor(s). Size-selective fishing is incr easingly recognized as a major selective force driving the evolution of life history traits. The effects of fi shing on the target stocks may also be masked by concurrent changes in the biotic and abiotic environment. It is important to consider the two factors simultaneously in life hist ory studies. I quantified life hi story characters (growth, age and size at maturity, mortality, GSI, and batch fecu ndity) and simultaneously examining effects of

PAGE 28

28 size-selective predation and enviro nmental factors on life history char acters of tilapia species in the crater lakes. Given the varia tion that exists in fishing pressu re and environmental features of these lakes, and the strong evidence for developmental plasticity in life history traits of tilapias, I predicted that: 1) The un-fished lakes with hi gh population density (low fishing effort) would have increased competition for resources (e.g., food) leading to density-dependent growth; 2) Under reduced and unpredictable food availabi lity that results from intraspecific competition, fish should allocate more resources towards re production, resulting in a smaller size and age at maturity. 3) A decrease in population density through fishing should reduce competition for resources and lead to increased growth. 4) When resources are adequate, as in heavily fished populations, there should be incr eased allocation to gr owth, resulting in delayed maturity at larger size. Depending on the strength of the si ze-selective fishing, we may also see the opposite (i.e., selection agains t larger and faster growing fish), dispr oving the 'bigger is better', 'fatter is better' and 'faster is better' hypotheses of life-history theory. I synthesized the result s of the above four chapters and discussed their implications for development of long-term manage ment initiatives aimed at mitigating potential impacts of human activities on the crater lakes, protecting the crater lakes ecosystems from further degradation, and sustainability of the artisanal fisheries in the crater lakes. I also discussed the implications of my study to our understanding of mechanisms underlying stunting in tilapias more generally.

PAGE 29

29 CHAPTER 2 ANTHROPOGENIC DISTURBANCES AND PRED ICTIVE MODELS OF CHLOROPHYLL A IN VOLCANIC CRATER LAKES OF WESTERN UGANDA Introduction Globally, aquatic ecosystems are under increas ing threat from rapidly expanding human populations. Human activities such as land use modification (e.g., animal grazing, deforestation, and other reductions in vegetati on cover) greatly accelerate change s in aquatic ecosystems; and the recent expansion and accelera ting rate of deforestation ha s caused widespread concern, particularly in the tropics where the impacts of deforestation and forest degradation on aquatic systems remain largely unknown (Chapman and Chapman, 2003). In Africa, it is estimated that rain forests once covered 3,620,000 km2 before anthropogenically-induced habitat alterations (Martin, 1991); now the continent ranks second only to South America in net rate of forest loss (FAO, 2005). According to the Food and Agricultural Organizations Global Forest Resources Assessment (FAO, 2005), the global estimate of net change in forest area betw een 2000-2005 was about 7.3 million ha per year ( 200 km2 of forest per day). Between 1990 and 2005, forest cover in Africa decreased from 655.6 million ha to 635.4 million ha, a consequence of deforestation, selective logging, as well as other human interventions (FAO, 2005). Before thes e anthropogenic disturba nces, rain forests of Africa covered an estimated 74% of Central Africa, 19% of West Africa, and 7% of East Africa (Martin, 1991). By 1985, remaining forest had b een reduced to 55% of the original area in Central Africa, and 28% of the original area in both West and East Africa (FAO, 1985). In Uganda, the focus of this study, it is estimated th at 86% of the tropical moist forest has been cleared (WRI, 1994). Given the small extent of ra inforests in East Africa and the high intensity of conversion, studies of defore station effects on aquatic system s are increasingly important. Research on the effects of deforestation on aquati c systems has historically been biased towards

PAGE 30

30 temperate regions. The few studies of effects of deforestation in tropical Africa include a general survey of forested catchments in Kenya (Dunne 1979), the effect of deforestation on species richness of fish, ostracods, and diatoms in Lake Tanganyika (Cohen et al., 1993), and degradation of Ethiopian Ri ft-valley lakes (Elizabeth et al., 1992, 1994; Brook, 1994; Zinabu, 1994). Deforestation impacts may partic ularly be acute in areas wi th high topographic relief that accelerates runoff and sedimentation in response to forest loss. High-altitude forests that once suffered reduced human pressures because of th e challenges of land conversion on steep slopes are now experiencing accelerating clearance. Volcan ic crater lakes may be particularly sensitive to land conversion because of their small size, st eep-sided slopes, and limited flushing. Africa is home to numerous volcanic crater lakes concen trated in the western (Cameroon) and eastern (Ethiopia, Kenya, Uganda) regions of the continent. Early limnological studi es in tropical Africa focused primarily on large lakes such as lakes Victoria, Tanganyika, and Malawi (e.g., Graham, 1929; Beadle, 1932), with far less attention direct ed towards small lakes and wetlands (Crisman and Streever, 1996). Despite the f act that the smaller water bodies are the systems most affected by the ever increasing human populations (Crism an and Streever, 1996), it was not until the 1960s that more detailed studies on small lakes such as the African crater lakes were conducted. These included studies of seasonal dynamics of physico-chemical paramete rs of the Bishoftu crater lakes and zooplankton associ ations in Ethiopian crater lake s (Baxter et al., 1965; Prosser et al., 1968; Wood et al., 1976, 1984; Green, 1986; Green and Mengestou, 1991), degradation of Ethiopian Rift-valley lakes (Elizabet h et al., 1992, 1994; Brook, 1994; Zinabu, 1994), investigation of crat er lakes in the Mount Cameroon ar ea (Green, 1972; Green et al., 1973, 1974), the tremendous explosion of carbon diox ide from lakes Nyos and Monoun, western

PAGE 31

31 Cameroon, that asphyxiated more than 1,700 people and livestock up to 25 km away (Kling, 1987, 1988), and the absence of fish in the crater lakes on the island of Bioko, neighboring the Cameroon mainland (Castelo, 1994). Because these lakes often occur in clusters, and are small and manageable in size, they offer natural test tubes for comparative study, a powerful tool for predicting and understanding anthropogenic impacts. In western Uganda alone, there are about 89 small volcanic crater lakes along the foothills of the Rwenzori Mountains that are divided into four clusters (Fort Portal area, Kasenda area, Katwe-Kikorongo area, and Bunyar uguru area). Beadle (1932, 1963, 1981) was the first person to study the limnology of the crater lakes in west ern Uganda. His work focused on a description of meromictic stratification pa tterns as well as documenting the existence of zooplankton in virtually anoxic (oxygen deficient) hypolimnetic waters of one dilu te lake and three concentrated saline lakes. Later, Visser ( 1965) carried out a limnological de scription of one dilute lake followed by Melack and Kilham (1972) who inves tigated the limnology of a peculiar sulfatochloride, saline lake. Other studies have incl uded that of Melack (1978), who investigated the morphological, hydrological, and chem ical characteristics of 16 cr ater lakes in the region and found striking limnological diversity among the lakes. Recently, a limnologi cal survey of some of the crater lakes and detailed studies on succes sions in the plankton communities in crater lakes Nkuruba and Nyahirya were conducted (Kizito et al., 1993; Kizito a nd Nauwerck, 1995, 1996). Contrary to early speculation (B eadle, 1981) that small sheltered tropical lakes that are deep relative to their surface area ar e unproductive, recent research (K izito et al., 1993; Chapman et al., 1998) suggests that some of the crater lakes are very productive. Observations collected from over 10 years of work in the crater lakes region of western Uganda suggest that the greatest threat to the cr ater lakes ecosystem is clearing of the forested

PAGE 32

32 crater rims associated with the need for fu el wood, building materials, and conversion to agricultural land (Chapman and Chapman, 2003). In general, 99.2% of energy used for cooking in Kabarole district of Uganda that hosts many of the crater lakes is derived from wood fuel (NEMA, 1997). Such high levels of forest clear ing and associated soil erosion may lead to increased turbidity, siltation, nutrient and orga nic matter load, and cultural eutrophication (Crisman et al., 2001). Currently many of the lake s are not protected from degradation, and there is little awareness of potential im pacts of human activity on these lake s. There is need to put in place measures that will enable monitoring of anthropogenic changes and prevention of future negative consequences of human activities on the cr ater lakes. We need to know what physical and chemical characteristics are altered by deforestation, and how such changes may affect lake productivity. One mechanism to ach ieve the latter is th e development of empirical models that can predict changes in lake productivity from physico-chemical characteristics. Empirical models for predicting lake produc tivity from total phosphorus (TP) and chlorophyll a (Chl-a) (a surrogate for phytoplankton abundance) have been developed mainly for temperate lakes (e.g., Dillion and Rigler, 1974 ; Canfield and Bachmann, 1981; Smith and Shapiro, 1981; Smith, 1982; Peters, 1986; Mazumd er, 1994; Kalff, 2002). Evaluation of nutrient limitation in tropical lakes has been hampered by the paucity of data on their trophic states (Kilham and Kilham, 1990; Fisher et al., 1995; Lewi s, 1996), as relatively few studies have been carried out to examine Chl-a-nu trient relationships in tropical lakes (but see Salas and Martino, 1991; Sarnelle et al., 1998; Jone s et al., 2000). Chlorophyll a-nut rient relationships developed from temperate data sets may not apply to tropical systems because of the inherent differences in their system functions. First, it has been argued that the grow ing season tends to be longer, thermal stratification more persistent, and nutrien t recycling more efficient in tropical lakes

PAGE 33

33 compared to temperate ones (Lewis, 1996). Second, Kilham and Kilh am (1990) and Lewis (1996) suggested that tropical lake s tend to be at a higher trophi c state compared to temperate lakes for the same nutrient supply. Third, phytop lankton biomass limitation by nitrogen is more commonly observed in tropical lakes compared to temperate lakes (Kilham and Kilham, 1990; Fisher et al., 1995; Lewis, 1996, 2002; Talling an d Lemoalle, 1998). Finally, seasonal variation in phytoplankton biomass is less pronounced in tropical lakes compared to their temperate counterparts (Melack 1978; Lewis, 1996). The development of predictive models of lake productivity (e.g., Chl-a) in tropical waters is es sential to validating the generality of temperate models, and perhaps more importantly, to providing a mechanism to predict and mitigate effects of human activity on tropical waters. The objectives of this study we re to: 1) describe the physical and chemical characteristics of 19 of the volcanic crater lake s in western Uganda; 2) detect differences in physico-chemical conditions of lakes with different histories of land conversion; 3) develop predic tive models of primary productivity (chlorophyll a) for the cr ater lakes ecosystem, and 4) compare the predictive models developed for the crater la kes to those developed for other African and subtropical lakes. Methods and Materials Study Area This study was conducted in the Kasenda cluste r of crater lakes of western Uganda (0o23-0o33N, 30o10-30o 20E), at an altitude of 925 to 1520 m (F igure 2-1). Most of the area is comprised of characteristic rolling hills that fo rm part of the extensive Precambrian basement complex (Government of Uganda, 1967). Most of the crater lakes are small, with surface area ranging from 2 ha (Lake Nyanswiga) to 50 ha (Lake Ntanda). Maximum depth ( zm) ranges from 5 m (Lake Kifuruka) to 259 m (Lake Ntanda) with the majority of the lakes being greater than 30

PAGE 34

34 m deep. Mean depth ( z ) of the lakes varies from 2.9 m (Lake Kifuruka) to 59.7 m (Lake Rukwanzi). Owing to their volcanic origin, most of the lakes have steep sides, with little littoral vegetation except some shallow lakes or shallower pa rts of the deeper lakes that have a diversity of aquatic macrophytes such as Nymphaea spp., Ceratophyllum spp. and Potamogeton spp. (Kizito et al., 1993). Baseline environmental data existed for a large number of the crater lakes (Kizito et al., 1993; Chapman and Chapman unpublished data). The lakes were selected to cover a large range of areas with varying anthropoge nic disturbances, and to repres ent as broad an environmental gradient as possible. The physical and chemical ch aracteristics determined in each lake included lake morphometry, vertical profiles of di ssolved oxygen (DO) concentration and water temperature, pH, conductivity, water transparen cy, total phosphorus (T P), and total nitrogen (TN). Climate, Vegetation, and Soils Mean annual rainfall in the region averaged 1.7 m between the period 1990 to 1999 or 1.5 m from 1903-2000 (Chapman and Chapman, unpublishe d data). Rainfall is bimodal, with two wet seasons from March-May and Se ptember-November. Between 1990 and 1999, the mean daily minimum temperature was 15.5oC, and the mean daily maximum temperature was 23.7oC (Chapman and Chapman, unpublished data). So ils in the catchments of the crater lakes have volcanic ash and lavas that have weathere d to produce very fertile brown soils (Uganda Department of Lands and Surveys, 1965). The na tural vegetation in this area is low-altitude mountain forest (Langdale-Brown et al., 1964; Taylor et al., 1 999). The area was originally forested, but the rapidly expa nding human population has led to land use and cover changes characterized by subsistence farming. There are s till remnants of forest fragments, mostly on

PAGE 35

35 crater rims where agricultural activ ities have been limited by the st eep crater slopes. Outside of the National park, widespread land clearance has re placed natural forest and shrub, and the area is dominated by small-scale plots of bananas, maize, cassava, and millet as well as settlements, with some large scale plantations of pine, cedar, Eucalyptus, and tea. Bathymetry and Morphometry To determine the percentage of the lake that occurs at each depth, 6-15 transects were established in each crater lake depending on la ke size. Transects were spaced approximately evenly over the lake, their orientation determin ed using land, and their point of origin and destination mapped with a Garmin GPS 12 un it. Depth profiles were obtained along each transect using a depth-sounder (EAG LE) to determine maximum depth ( zm) and mean depth ( z) of each lake. Lake area (LA) and catchment ar ea (CA) were also mapped using Garmin GPS 12 unit. Lake area, the greatest length on the lake surface ( l ), and the greatest width ( b ) of the lake perpendicular to the greatest length of the la ke were calculated using ArchGIS (version 9.1). Physical and Chemical Characters Each lake was sampled once between Ju ly 2004 and July 2005. One station was established near the lake center of the lake a nd mid-day DO concentration and water temperature measurements were taken in situ through the water column at 1-m intervals up to 30 m using a portable YSI oxygen/temperature meter (Model 57). Water samp les for conductivity and pH measurements were collected from 0 m to 30 m at 2-m intervals using a 3-liter Van Dorn water sampler. The samples were stored in Nalgene bottles covered with duct ta pe. The bottles were first washed with distilled wate r and then rinsed with lake wa ter prior to sample collection. Conductivity and pH were determined in the labo ratory on the same day of sampling using a YSI (model 30) conductivity meter (corrected to 25oC) and an OAKTON pH Testr 1, respectively. Water transparency was estimated with a 20-cm Secchi disk and readings averaged over two

PAGE 36

36 measurements. Chlorophyll a was determined by filtering 200 ml of lake water through Whatman GF/C filters (particle retention size 47 mm). Chlor ophyll a samples were kept in a freezer at -20oC until analyzed at the Zoology Departme nt, Makerere University. In the laboratory, chlorophyll a was extracted with 90% methanol solution during a 24-h period, and the concentration was determined using HAC H 2010 DR Spectrophotometer. For total phosphorus and total nitrogen analyses, water sa mples were collected from 0 m-30 m at 6-m intervals at the same station and placed in 0.3 lite r plastic bottles previous ly washed with dilute hydrochloric acid and then rinsed with lake water. In the field all samples were stored in a cooler containing ice and later re frigerated until the time of laborat ory analysis at the Fisheries Resources Research Institut e (FIRRI), Jinja, Uganda. Statistical Analyses Data analyses were done with SPSS for Windows (version 12.0). Limnological and morphometric data were log10 transformed to normalize the data and stabilize variance. To detect changes that might have occurred in the water quality of the crat er lakes in the region, a paired t-test was used to determine if differe nces in conductivity, Secchi depth, Chl-a, and TP values in nine of the lakes sampled in the pres ent study were significantly different from results of a previous study by Kizito et al., (1993). The total light extincti on coefficient was calculated following the widely used equation kd = 1.7/zSD (Wetzel, 1983), where ; kd is the extinction coefficient; zSD is the depth at which Secchi disk disa ppears from view; and 1.7 is a conversion factor for the fraction of surf ace irradiance remaining at the de pth of disappearance of Secchi disc and commonly assumed to be 10% (Wetzel, 1983) Linear regression analysis was used to develop predictive models of the concentrati on of Chl-a (phytoplankton abundance) in the lakes from water transparency (Secchi depth) and to tal phosphorus. The Chl-a-TP predictive model

PAGE 37

37 for the crater lakes was compared to those deve loped for other African la kes (Huszar et al., 2006) and subtropical lakes in Florida, USA (Mazumd er and Havens, 1998). Discriminant function analysis (DFA) was used to determine if th e lakes could be classified on the basis of deforestation category (moderate, severe, and complete). Discri minant function analyses (DFA) are usually conducted for classification purposes and to determine whic h continuous variables discriminate between two or more naturally o ccurring groups. We grouped the lakes using the deforestation categories devel oped by Pomeroy and Seavy (2003) where minimal deforestation has 50-100% of the area still forested, modera te deforestation has 25-49% forested, severe deforestation has only small forest patches or sca ttered trees left, and comp lete deforestation has no forest trees. Because of small sample size, the minimal and moderate deforestation categories were combined for the DFA analysis. Results Environmental Data-Lake Morphometry The crater lakes exhibited great variation in their morphometr ic and limnological features. Lake Ntanda was the largest, followed by Mwamba, and Nyabikere (Table 2-1). Surface area ranged from 2 ha (Lake Nyanswiga) to 50 ha (Lake Nta nda), while maximum depth ranged from 5 m (Lake Kifuruka) to 259 m (Lake Ntanda). Mean lake depth was similarly variable, ranging from 2.9 m (Lake Kifuruka) to 59.7 m (Lake Rukw anzi). The ratio of maximum length to maximum width for most crater lakes was great er than 1.0 except Kanyamukale, Kasenda, and Lyantonde. The ratio of maximum length: maximu m width for the latter lakes was close to 1.0 (Table 2-1). Volume development (Dv) was calculated from the ratio of mean depth: maximum depth ( z / zm) following Hutchinson (1957). The z/ zm values for crater lakes Kanyamukale, Kanyango, Kasenda, Kerere, Kifuruka, Lugembe, Wa nkenzi, and Wandakara were greater than 0.5. For the rest of the crater lakes, z/ zm values were less than 0.5 (Table 2-1). Most lakes had

PAGE 38

38 experienced moderate or severe catchment deforestation with the exception of lakes Nkuruba, Ntanda, and Kerere (Table 2-1). Temperature Profiles With the exception of the shallow Lake Ki furuka where no thermal stratification was detectable, the temperature profiles of all deep lakes showed a str ong vertical stratification of the water column (Figure 2-2) despite a small cha nge in water temperature. Thermocline depth varied among lakes, ranging between 8 and 20 m depth (Figure 2-2). Mid-day surface water temperatures ranged from 24oC in Lake Lyantonde (Figure 2-2g) to 27.1oC in Lake Kanyamukale (Figure 2-2a). The temperature difference between the surface waters and near bottom (shallow lakes) or 30 m depth (deep lakes) ranged from 10C in Lake Kifuruka (Figure 2-2e) to 4.20C in Lake Kanyango (Figure 2-2b). Dissolved Oxygen Concentration Profiles The euphotic zones of all lakes were highly oxygenated. Dissolved oxygen concentrations diminished with depth, with anoxic or near-anox ic conditions evident in the hypolimnion of most crater lakes (Figure 2-2). Surface oxygen ranged from 9.5 mgL-1 in Lake Kanyango (Figure 22b) to 4.7 mgL-1 in Lake Nyaherya (Figure 2-2m). The depth of the oxycline was highly variable among lakes, ranging from 4 m in Lake Mwamba to 26 m in Lake Nyinabulitwa (Figures 2-2i and 2-2o). Below the oxycline, a strong odor of H2S was detected while obtaining water samples in several of the lakes. pH The water of the epilimnion was alkaline in all lakes. Surface pH ra nged from 7.5 in Lake Nyanswiga (Figure 3n) to 9.3 in La ke Kerere (Figure 2-3d). Ther e was a gradual decline in pH with depth in all lakes with hypolimnetic laye rs being slightly acidic in some lakes (Figure 2-3a-r).

PAGE 39

39 Specific Conductance Specific conductance (conduc tivity, corrected to 25 oC) exhibited a gradient, gradually increasing towards the deeper part of the wate r column (Figure 2-3a-r). Conductivity varied widely among lakes with the highest m ean euphotic conductivity value of 1201.2 1.7 Scm-1 recorded in Lake Wandakara and the lowest mean euphotic conductiv ity value of 211.3 0.2 Scm-1 in Lake Kerere (Table 2-2). Water Transparency (Secchi Depth) Secchi-disk depth varied widely among lakes (Table 2-2). Lake Wandakara had the least transparent water, and Lake Kerere was the most transparent (Table 2-2). The light extinction coefficient was calculated for each crater lake, and the values ranged from 0.35 m-1 (Lake Kerere, June 2005) to 3.47 m-1 (Lake Wandakara, June 2005). Chlorophyll a, Total Phosphorus, and Total Nitrogen Mean values of Chl-a, total phosphorus, and tota l nitrogen integrated over the euphotic zone are presented in Table 2-2. Chlorophyll a (produc tivity) varied greatly among lakes but was generally higher in severely and completely deforested lakes compared to moderately deforested lakes (Table 2-2). Chlorophyll a ranged from 3.5.70 gL-1 in Lake Rukwanzi to 200.7.7 gL-1 in Lake Wakenzi (Table 2-2). Mean TP and TN were also highly variable among lakes. TP ranged from 30.3.0 gL-1 in Lake Nyinabulitwa to 580.9.3 gL-1 in Lake Kerere (Table 2-2). Total nitrogen values were very high in the crater lakes with mean euphotic values ranging from 1251.9.8 gL-1 in Lake Lyantonde to 96742.0.1 gL-1 in Lake Wandakara (Figure 2-2). Paired t-tests were used to de tect differences in the mean values of Chl-a, TP, and TN between the current study and results of a study done in the early 1990s (K izito et al., 1993). Data for these time periods were only availabl e for nine lakes. Average conductivity across

PAGE 40

40 lakes was significantly higher in the present study compared to the average value recorded in 1993 (t = 2.637, p = 0.034). Chlorophyll a concentr ation (phytoplankton abundance) was also significantly higher in th is study (t = 3.217, p = 0.012). TP and water transparency did not differ significantly between the two study periods. Chlorophyll a-Water Transparency and Chlorophyll a-Total Phosphorus Relationships Water transparency was highly negatively correl ated with Chl-a concentration (Table 2-3, Figure 2-4). A comparison of th e crater lakes curve with the Chl-a-Secchi relationship for subtropical lakes in Florida reve aled a generally higher Chl-a per unit of water transparency in the crater lakes than for subtr opical lakes, and a steeper slope in the crater lakes relationship (Figure 2-4). There was a weak positive relations hip between Chl-a concentration and TP in the crater lakes (Table 2-3, Figure 2-5). Total phosphorus explaine d only 12% of the variation in Chl-a among the lakes. For a given unit of TP, Ch l-a was generally lower in the crater lakes than in the other African lakes and subt ropical lakes; an effect that wa s greater with higher levels of TP (Figure 2-5). Discriminant Function Analysis (DFA) Discriminant function analys es indicated signifi cant differences among deforestation categories. Our DFA predicted deforestation cat egory for each lake based on physico-chemical characteristics, with the null hypot hesis being one-third co rrectly classified. This DFA classified 66.7% of the lakes into the correct category (Wilks = 0.29, p = 0.036). There was some overlap between severely deforested lakes and completely deforested lakes with one of the former category of lakes classifi ed as the latter (Figure 2-6). Based on the structure matrix, the discriminant functions most important for classifying the lakes were log10 transparency and log10 conductivity.

PAGE 41

41 Discussion Morphometric Features The Kasenda cluster of crater lakes exhibi ted wide variation in their morphometric characteristics, but fell into two major categories: 1) deep lakes with high mean depth: maximum depth ( z/ zm), parabolic shape, and very steep slop es, and 2) shallow lakes with lowz / zm ratio, having more or less conical basin shape, less st eep rims, and some aquatic vegetation along some areas of the shoreline. The results of the pres ent study are in agreement with that of Melack (1976) who foundz / zm values close to 0.6 for the deep lakes, whereas shallow lakes (e.g., Lake Saka) had lower z/ zm values, a result he attributed to gent le sloping peripheral regions. In addition, the elliptical parabola sh ape of basins of the deep lake s is in agreement with those reported in studies of crater la kes in other regions (e.g., Vazquez et al., 2004). Kasenda crater lakes also varied in shape with the la kes having maximum length: maximum width ( l : b ) ratio of 1.0 characterized by approximately circular shape and the lakes w ith l:b ratios greater than 1.0 having irregular shapes. Limnological Characters The deep crater lakes of the Ka senda cluster exhibited clear th ermal stratification. This is in agreement with studies of other crater lakes from Mexico (Vzquez et al., 2004), Costa Rica (Uma a and Jimnez, 1995), Cameroon (Kling, 1988) and Uganda (Melack, 1978). Depth of thermocline varied among lakes. The deep ther moclines (greater than 20 m) in lakes Ntanda, Nyinabulitwa, Kerere, and Rukwanzi may reflect their high transparency of water that permits penetration of solar radiation to a greater depth compared to similar lakes with less transparent water. The steep crater slope s and vegetation (forest) around th e crater rims possibly dampen wind action that affects water circul ation patterns. This is common in highly stratified lakes that

PAGE 42

42 are protected from wind acti on by the high rims and vegetati on (Wetzel, 2001). Kling (1988) reported a thermocline depth of 10-14 m for Mbal ang Crater Lake in Cameroon and Vzquez et al., (2004) found a similar thermocline depth for Lake Majahual in Mexico. The lack of stable thermal stratification in Lake Ki furuka can be explained by its small maximum depth (5 m) and daily mixing of the water column by wind action. The thermal behavior of Lake Kifuruka is similar to that reported for other shallow crater lakes such as Lake Nyamisingire in Uganda (Melack, 1978), Kotto and Bambili in Cameroon (K ling, 1988), as well as Los Tuxtlas crater lakes Mogo and Verde in Mexico (Vzquez et al., 2004). The saturation of the epiliminetic water laye rs of the Kasenda crater lakes may indicate high photosynthetic rates by phytoplankton. The well-defined metalimnetic oxyclines in the deep lakes are possibly a result of the decom position of allochtonous and autochtonous organic material that enters the lakes es pecially during the rainy season, le ading to depletion of available DO, hence the anoxic hypolimnetic conditions. O xycline depths varied among lakes and may be attributed, at least to some degr ee, to the influence of lake mor phometry. In crater lakes Ntanda, Nyinabulitwa, Kerere, and Rukwanzi, all deep, sheltered lakes, water became anoxic between depths 18-26 m, whereas oxycline depth was ex tremely shallow in Kanyamukale, Lugembe, Mwamba, Nyabikere, Nyanswiga, and Wandakara. The variation in depth of oxycline and the apparent lack of vertical mixing is in agreem ent with Melacks study. Melack (1978) concluded that morphometric features, particularly basin di ameter and minimum height of the crater rim above the lake are important determinants of water mixing. He proposed the ratio DH-1; as an index for determining wind caused mixing (where D is the maximum diameter of the lake and H is the minimum height of the crater rim above th e lake). He then described the relationship

PAGE 43

43 between this ratio and oxycline depth for 17 tropi cal and temperate crater lakes by the regression equation lnAX = 0.99 + 0.35 ln (DH-1) where; AX is depth of oxycline. The Kasenda crater lake surface waters were al kaline, which is in agreement with earlier studies on crater lakes in the area (e.g., Melack, 1976; Kizito et al ., 1993). The alkaline water of the epilimnion is possibly a result of carbondioxide fixation by photosynthetic phytoplankton, whereas respiration and decomposition proce sses may have accounted for the pronounced decrease in pH of the hypolimnion. Differences in conductivity among lakes may reflect, at least to some degree, differences in extent of defore station along the crater rims. The results of the discriminant analysis further strengthen this suggestion, as conductivity and water transparency were the two most important factors determining th e classification of the la kes into deforestation categories. The lakes with severely or comple tely deforested slopes we re characterized by low Secchi depth and high conductivity values. Lake Wandakara, which is completely deforested, had the highest conductivity and very low tran sparency (Secchi dept h) compared to wellsheltered Lake Kerere that had the lowest c onductivity and highest transparency. In addition, differences in sampling season may have contri buted to inter-lake vari ation in conductivity levels, as some of the lakes were sampled in dry season whereas others were sampled during the rainy season characterized by highe r levels of run-off from surr ounding lands. With exception of Lake Wandakara that had high cond uctivity values, the rest of the Kasenda cluster of crater lakes can be categorized as dilute lakes with surface water conductivity ranging from 219-929 S cm-1 at 25oC. Melack (1976) reported that the electrical conductivity of the Kasenda cluster of the crater lakes ranged from 288 to 533 Scm-1 at 20oC. Kizito et al., ( 1993) reported a range of 320-511 S cm-1 for the same group of lakes. Thes e earlier studies did not cover lakes Wandakara and Kanyamukale.

PAGE 44

44 Chlorophyll a-Water Transparency and Chlorophyll a-Total Phosphorus Relationships Water transparency varied widely among the crat er lakes and appears to be driven in part by dense populations of phytoplankton (blue green alg ae, and green algae) in some of the lakes. Introduction of particulate material from erosion of deforested steep crater rims especially during the wet season, further contri butes to decreased water transparency in some of the lakes. This is evidenced by the high transparency values of lakes Ntanda and Ke rere that are protected from erosion by the forests around the crater rims. In contrast, lakes Wandakara and Mwamba that have experienced intense deforestation had very low water transparency. In addition, some of the lakes (e.g., Lake Mwamba and Lake Rwanke nzi) have streams flow ing into them. The streams pass through agricultural and deforested areas and may introduce suspended particulate matter and other nutrie nts into the lakes. Regression analysis indicated that water transp arency relates better to Chl-a concentration than TP. The strong association between wa ter transparency (Secchi depth) and Chl-a concentration (a surrogate of primary productivity) suggests that variation in deforestation levels is a contributor to variation in primary producti vity among the crater lake s of western Uganda. Completely or severely defore sted lakes were characterized by lower transparency and high chlorophyll a concentration whereas the lakes ex periencing moderate deforestation had higher transparency and lower chlorophyll a concentration. The increased er osion from deforested crater rims and associated nutrient loading seems to be contributing to eutrophic ation of some of the lakes. Water transparency is known to be a ffected by both particulate matter and dissolved organic carbon (Wetzel, 2001). In addition, a major fr action of the incident light is also scattered by algae, depending on their size distri bution (Kirk, 1986; Mazumder et al., 1990b). Water transparency, in turn, influences several physi cal, chemical, and biologi cal properties of lakes such as dissolved oxygen concentration of th e hypolimnion (Cornett and Rigler, 1980), fish

PAGE 45

45 predation and associated graze r communities (Pace, 1984; Capent er et al., 1985; Hansson, 1992; Sarnelle, 1992), thermal structure and mixing (Mazumder et al., 1990b; Mazumder and Taylor, 1994), lake morphometry (Straska ba, 1980), as well as grazer-nut rient interactions (Mazumder and Lean, 1994). In comparison with results of studies of subtropi cal lakes in Florida (Mazumder and Havens, 1998), the Chl-a-Secchi rela tionship in our crater lakes study suggests a higher water transparency per unit of Chl-a th an does the predictive model developed for subtropical lakes. The slope of the Chl-a-Secchi regression was also a bit steeper for the crater lakes compared to the subtropical lakes of Fl orida. Mazumder and Havens (1998) found higher Secchi depth in temperate lakes compared with s ubtropical lakes, a resu lt they attributed to temperate systems being dominated by larger zoop lankton compared to subtropical systems. In addition, regulation of nutrientphytoplankton biomass (Chl-a) relationships through grazing by large zooplankton has been documented in temperate lakes (Muzumder, 1994). Small zooplankton such as cladocer ans and copepods form the domi nant group in subtropical and tropical lakes (Lewis, 1996; Bran co, et al., 2002; Jeppensen, et al ., 2005), and may not have the same grazing effect on phytoplankton as in temper ate lakes. The differen ce between the tropical and subtropical predictive models may be a result of factors other than zooplankton grazing. The positive relationship between chlorophyll a and total phosphorus that we observed in the crater lakes is in agreem ent with other studies where phosphorus has been found to be a major determinant of algal biomass. Despite a near ly 20-fold range in TP across the crater lakes, the relationship was weaker than has been ge nerally observed for temperate systems (Smith, 1982; Mazumder, 1994; Nrnberg, 1 996). This trend supports the findings of Huszar et al., (2006) in their eval uation of 192 lakes from tropical and s ubtropical regions. They also found a weak Chl-a-TP relationship (r2=0.42). They argued that N limitation is not a likely explanation

PAGE 46

46 for the weak relationship in tropical-subtropical systems, because they observed high average TN:TP ratios. In addition, review s of nutrient addition experiment s in tropical systems cited by Huszar et al., (2006) suggeste d that N and P limitation does no t only vary between systems but also varies seasonally within the same system. Similarly, in our study of the Ugandan crater lakes, we found very high average TN values an d high TN:TP values. The very high TN values we observed are generally in the range presente d in Huszar et al., ( 2006), suggesting that high TN may be characteristic of these systems. Our evaluation of long-term changes in selected chemical features a nd Chl-a concentration in the crater lakes over the last 15 years indicated a significant increase in conductivity and Chl-a concentration in the lakes. The increased conductivity and Chl-a concentration may reflect increased deforestation over the years that ha s accelerated soil erosi on leading to increased organic matter load, and cultural eutrophication. It should be noted though that this point-in-time sampling done in both 1993 and 2006 limits the stre ngth of the comparison because of potential effects of season on the selected characters.

PAGE 47

47Table 2-1. Some morphome tric features of the Kasenda cluster of Crater Lakes in western Uganda. Crater Lake Deforestation Altitude (m) Area (ha) Max. depth zm (m) Mean depth ( z) (m) Ratio ( z/ zm ) Max. length l (m) Max. width b (m) Ratio l:b(Dv) Kanyamukale Complete 1161 2 12 6.8 0.6 169.8 166.4 1.0 Wandakara Complete 1166 3 12 6.0 0.5 266.5 132.1 2.0 Murigarime Complete 1196 22 58 25.3 0.4 708.7 588.5 1.2 Nyanswiga Moderate 1474 2 66 12.5 0.2 226.6 126.0 1.8 Nkuruba Moderate 1499 3 37 16.2 0.4 270.0 160.0 1.7 Kasenda Moderate 1240 6 14 8.2 0.6 283.4 288.5 1.0 Rukwanzi Moderate 1348 8 172 59.7 0.4 388.6 303.8 1.3 Kerere Moderate 1187 28 77 36.6 0.5 734.5 534.0 1.4 Nyinabulitwa Moderate 1426 41 182 37.9 0.2 969.6 624.5 1.6 Ntanda Moderate 1343 50 259 80.2 0.3 917.4 698.4 1.3 Nyahirya Severe 1434 3 98 30.4 0.3 220.4 192.0 1.2 Lugembe Severe 1290 8 18 10.5 0.6 421.8 260.5 1.6 Lyantonde Severe 1396 14 187 42.6 0.2 457.0 440.3 1.0 Kanyango Severe 1274 15 68 37.1 0.5 491.5 374.4 1.3 Kifuruka Severe 1404 15 5 2.9 0.6 519.6 388.8 1.3 Wankenzi Severe 1160 18 62 30.1 0.5 674.2 402.5 1.7 Nyabikere severe 1393 44 57 19.8 0.4 951.5 602.4 1.6 Mwamba Severe 1308 45 203 18.4 0.1 1059.2 603.4 1.8

PAGE 48

48Table 2-2. Mean euphotic values ( S.E) of physical and chemical parameters analyzed for the crater lakes of western Uganda. Crater lake Temp. (oC) pH DO (mgL-1) CND at 25oC (Scm-1) Secchi (m) Chl. a (gL-1) Ext. Coeff. TP (gL-1) TN (gL-1) Wandakara 25.2.5 8.6.1 7.4.1 1201.2.7 0.5.0 171.7.1 3.47 107.6.8 96742.0.1 Kanyamukale 26.6.3 8.6.1 6.9.1 926.8.2 1.7.0 28.5.0 1.02 566.0.7 6991.6.6 Murigarime 25.6.2 9.1.0 6.7.1 441.4.2 1.9.0 20.9.8 0.89 71.8.1 19073.2.1 Nyanswiga 24.1.3 7.8.2 6.2.1 346.7.3 1.8.1 22.9.7 0.96 332.6.7 10373.1.6 Kasenda 26.2.1 8.6.0 8.1.1 307.4.4 2.5.0 16.7.2 0.68 81.6.4 19491.9.2 Nkuruba 23.9.2 8.1.1 4.9.1 379.1.7 3.2.0 12.5.6 0.53 426.4.1 14886.2.7 Rukwanzi 25.3.2 8.8.2 5.3.7 400.1.7 4.0.1 3.5.70 0.43 41.6.5 49010.2.1 Nyinabulitwa 24.2.2 8.1.2 6.8.1 262.4.2 4.3.1 8.6.93 0.39 30.3.0 10997.5.5 Ntanda 24.1.2 8.7.1 6.6.0 442.7.6 4.5.3 3.9.6 0.38 42.6.2 5140.1.3 Kerere 26.3.1 9.3.0 5.7.0 211.3.2 4.8.0 9.2.9 0.35 580.9.3 14886.2.0 Wakenzi 24.1.0 8.6.3 8.9.3 368.5.3 0.6.1 200.7.7 2.79 173.6.6 16770.3.1 Nyabikere 23.7.6 7.8.0 5.3.1 279.4.5 0.8.0 45.2.9 2.14 91.0.7 2300.4.8 Lugembe 24.5.2 8.6.1 7.3.0 406.9.8 1.0.1 69.2.4 1.69 206.6.2 25877.1.5 Mwamba 25.3.4 8.7.0 7.7.3 447.2.0 1.1.0 23.3.0 1.55 37.3.5 1390.8.9 Kifuruka 24.0.1 8.0.2 5.5.0 495.7.6 1.1.0 25.7.2 1.55 438.0.9 4908.6.9 Kanyango 26.1.4 8.8.1 9.7.2 450.4.9 1.1.1 38.6.1 1.60 76.7.9 19387.2.6 Nyahirya 24.3.4 8.0.1 4.5.1 481.9.2 1.7.1 18.4.9 1.02 33.8.5 2131.4.3 Lyantonde 23.9.0 8.6.0 6.9.0 508.7.3 3.2.3 10.2.8 0.54 44.3.5 1251.9.8

PAGE 49

49Table 2-3. Chlorophyll a-Secchi and Chlorophyll a-total phosphorus relationships for 19 western Uganda crater lakes compared t o nutrient-chlorophyll a relationships developed fo r other African lakes a nd subtropical lakes. Lake type Regression model n r2 P Source Secchi Vs Chl.a Crater lakes Log secchi = 1.07 + 0.60 l og chl-a 19 0.83 p < 0.001 Present study Subtropical lakes Log secchi = 0.59 0.43 log Chl-a 59 0.76 p < 0.001 Mazumder and Havens, 1998 TP vs Ch-a Crater lakes Log chl-a = 0.57+ 0.37 log TP 19 0.12 p = 0.05 Present study Subtropical lakes Log chl-a = -0.36 + 1.06 l og TP 59 0.42 p < 0.001 Mazumder and Havens, 1998 African lakes Log chl-a = 1.19 + 0.49 log TP 14 0.79 p < 0.001 Regression data from Huszar et al.2006

PAGE 50

50 Figure 2-1. Map of Kasenda cluster of crater lakes in western Uganda (0o23-0o33N, 30o10-30o 20E) showing the lakes sampled from July 2004-July 2005.

PAGE 51

51 Figure 2-2. Temperature and dissolved oxygen pr ofiles for crater lakes of western Uganda sampled from July 2004-July 2005.

PAGE 52

52 Figure 2-2. Continued

PAGE 53

53 Figure 2-3. pH and conductivity profiles for crater lakes of western Uganda sampled from July 2004-July 2005.

PAGE 54

54 Figure 2-3. Continued

PAGE 55

55 -0.50.00.51.01.52.02.53.0 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 Secchi depth (log10 m)Chlorophyll a (log 10 gL-1)C r at e r l a kes of U ga nda Subt ropi c al l a ke s Figure 2-4. Relationship between chlorophyll a a nd Secchi depth for crater lakes of western Uganda and subtropical lakes. Data for subtropical lakes were adapted from Mazumder and Havens (1998).

PAGE 56

56 1.01.52.02.53.0 0 1 2 3 Chlorophyll a (log10 g/L)Total phosphorus (log10 g/L) S u b t r o p i c a l l a k e s O t h e r A f r i c a n l a k e s c r a t e r l a k e s o f Ug a n d a Figure 2-5. Relationship between chlorophyll a and total phosphorus for crater lakes of western Uganda, other African lakes, and subtropical lakes. Data for subtropical lakes were adapted from Mazumder and Havens (1998) and for other African lakes from Huszar and colleagues (2006).

PAGE 57

57 Figure 2-6. Results of discriminant analysis, illustrating the ability to categorize crater lakes of western Uganda using their physical, chem ical, and morphometric characters. Inputs to the discriminant analysis were log tr ansformed values of (Secchi depth, chlorophyll a, dissolved oxygen, conductivity, total phosphorus, total nitrogen, lake area, maximum depth, and mean depth). -4-2 024Function1 -2 0 2 4 Function 2 Deforestation ModerateSevereCompleteCentroidFunction 2

PAGE 58

58 CHAPTER 3 PREDICTORS OF FISH CONDITION IN INTRODUCED TILAPIAS OF UGANDAN CRATER LAKES IN RELATION TO FISH ING PRESSURE AND DEFORESTATION Introduction Fish condition, defined as the robustness or well-being of an individual fish (Le Cren, 1951; Bulow et al., 1981; Blackwell et al., 2000), is an essential component of fishery biology used to assess the general health of populations (Gulland, 1983; Sparre et al., 1989). Condition can vary dramatically both within and among populations depending on s easonal changes and/or other environmental factors. It is critical to iden tify environmental predictors of this variation to optimize fishery production. Two types of cond ition indices are generally used by fisheries scientists as surrogates of fi sh health and growth (Wootton, 1990; Pauly, 1993; Petrakis and Stergiou, 1995; Binohlan and Pauly, 1998). The first are somatic indices such as calorific and proximate (e.g., lipid content) indices (Brown and Murphy, 1991), and liver-, fat-, and gonadosomatic indices (Adams et al., 1982 ). The second group is length -weight based indices such as Fultons condition factor (Le Cren, 1951; Rick er, 1975), relative condi tion factor (Le Cren, 1951), and relative weight (Wege and Anderson, 1978). Length-weight based condition indices are the most frequently applied because the data are relatively easy, efficient, and cost-effective to collect. Length-weight relationships have thus been used extensively in fisheries biology to convert growth-in-length to growth-in-weight equa tions for use in stock assessments (Oscoz et al., 2005) and to provide an index of condi tion (Le Cren, 1951; Bolger and Connolly, 1989). Relationships between fish condition and populat ion structure, batch f ecundity, life history adaptations, environmental conditions, and/or ma nagement actions have been studied for a variety of fish species in temperate regi ons (e.g., Brown and Murphy, 1991; Gabelhouse, 1991, Blackwell et al., 2000). Studies in tropical fisheries are far less prev alent, but there is a growing body of literature on the condition of tilapiine fishes in the larger tropical freshwater lakes and

PAGE 59

59 reservoirs (Lowe McConnell, 1958, 1982; Frye r and Iles, 1961, 1972; Welcomme, 1970; Gwahaba, 1973; Siddique, 1977; Kolding, 1993). Thes e empirical studies suggest great variation in tilapia condition within and among systems. Given the ecological and commercial importance of tilapiine fishes in tropical fresh waters a nd in aquaculture systems, an understanding of environmental predictors of tilapia condition is becoming increasingly important. In particular, we need to more fully explore the impact of si ze-selective harvest on tilapia condition. Several studies have linked size-selective fishing to change s in life history traits such as growth rate, sizeand age-at maturity, a nd batch fecundity (e.g., Gadgil and Bossert, 1970; Gwahaba, 1973; Law, 1979; Ricker, 1981; Anon, 1988). Effects of size-selective harvest on fish condition are less well known, although studies suggest that fishing pressure can improve fish condition by minimizing density-dependent growth (Law, 2000; Sass et al., 2004). Comparative study of tilapia condition across gradients of anthropogeni c disturbance (e.g., fishi ng pressure, catchment deforestation) and other envir onmental factors offers the opport unity to identif y significant predictors of fish condition. In this study, we explore po ssible effects of anthropogenic disturbances on the condition factor of introduced tilapia species in small volcanic crater lakes of Uganda and identify significant environmental pr edictors (biotic and abiotic) of fish condition. In western Uganda, three tilapia species: Oreochromis niloticus (Linnaeus, 1758) Oreochromis leucostictus (Trewavas, 1933) and Tilapia zillii (Gervais, 1874) were introduced into a large number (approx. 89) of volcanic crater lakes in the 1940s, and subsequent years, to increase available protein to the local communities. Currently, many of these lakes seem to be producing stunted tilapia populations, causes of which remain unknown but may reflect resource limitations associated with the cater lake environment a nd/or low mortality lead ing to high levels of

PAGE 60

60 intraspecific competition. There is great variation in fishing pressure among the lakes from almost no fishing, to individual based hook-and-lin e fishing, to small-scale commercial gillnet fisheries run by communities around th ese lakes. In addition, many of the crater lakes lie outside protected areas and are increasingl y threatened with deforestation of the crater rims due to the rapidly expanding human populations. Information on the effects of anth ropogenic disturbances and fishing pressure on the biology of tilapiine speci es in these lakes is non-existent, challenging the development of effective management goals for fish in these very abundant lakes. The aim of the present study was to explor e the relationships between anthropogenic factors (e.g., deforestation and fishing) and the condition of two introduced species ( Oreochromis leucostictus and Tilapia zillii) in 17 crater lakes in western Uganda with varying extent of catchment deforestation and fishing pressure. We select ed two tilapia species that are widely introduced in these lake s and have divergent trophic nich es as adults (planktivore and herbivore, respectively). In particular, I asked whether fish condition differs among lakes characterized by differences in fishing pressure developed predictive models of fish condition for the two tilapia species base d on a suite of biotic and abio tic factors, and related these relationships to differences in extent of catchment defore station among the lakes and the management of the tilapia populations in the la kes where they sustain artisanal fisheries.

PAGE 61

61 Methods and Materials Study Area The study was conducted in the Kasenda cluste r of crater lakes of western Uganda (0o23-0o33N, 30o10-30o 20E) at an altitude of 925 to 1520 m (Melack, 1976, Figure 3-1). Most of the Kasenda cluster crater la kes are small, and the 17 lakes us ed in this study had surface areas ranging from 2 ha (Lake Nyanswiga) to 50 ha (Lake Ntanda), maximum depth ( zm) ranging from 5 m (Lake Kifuruka) to 259 m (L ake Ntanda), and mean depth ( z ) ranging from 2.9 m (Lake Kifuruka) to 59.7 m (Lake Rukwanzi). In addition, the lakes also differ in their level of fishing pressure and the extent of deforestation of their catchments. Owing to their volcanic origin, most of the lakes have steep sides, with little l ittoral vegetation except a few shallow lakes or shallower parts of the deep la kes that have a diversity of aquatic macrophytes such as Nymphaea spp., Ceratophyllum spp. and Potamogeton spp. (Kizito et al., 1993). Rainfall is bimodal, with two wet seasons from March-Ma y and September-November, with 1.7 m of total rainfall received annually (1990-2005). Between 1990 a nd 2004, the mean daily minimum temperature was 14.9oC, and the mean daily maximum temperature was 20.2oC. Environmental Data Baseline environmental data existed for a large number of the crater lakes (Kizito et al., 1993; Chapman and Chapman unpublished data). The lakes were selected to cover a large range of areas with varying levels and types of anthro pogenic disturbance, and to represent as broad an environmental gradient as possible. The physic al and chemical characteristics determined in each lake included: lake area, maximum dept h, mean depth, vertical profiles of dissolved oxygen (DO) concentration and water temper ature, pH, conductivity (corrected to 25oC), water transparency, chlorophyll a (C hl-a) concentration, total phosphorus (TP) and total nitrogen (TN). Water temperature and dissolved oxygen concentration were determined in situ with a

PAGE 62

62 YSI oxygen/temperature meter. Conductivity an d pH were measured using a YSI conductivity meter and an OAKTON pH Testr 1, respectively. Water transparency wa s estimated with a 20-cm Secchi disk. Chlorophyll a was determined using a spectrophotometric technique. Total phosphorus and total nitrogen were determined using standard limnological techniques following APHA (1988). The lakes were grouped into th ree categories of fishing pressure (low, medium, and high) based on the average nu mber of fishermen observed during a 6-day sampling period in each lake, and presence or absen ce of gillnets in a lake. The lakes with low fishing effort either had no fish ing pressure or a small number (<20) of hook-and-line fishermen. Lakes with medium fishing had a large num ber (30-45) of hook-and-line fishermen and occasionally gillnets. The lakes with high fish ing pressure had an established gillnet fishery. Large predatory fish species are absent in mo st of the lakes except Lake Kasenda where the catfish ( Clarias gariepinus) was recently introduced. In addition, Pomeroy and Seavy (2003), in a recent survey of the crater lakes, found low populations of waterfowl (e.g., mean number of fish eating per lake was 14). I assumed that fishing is the main source of predation in these lakes and that mortality caused by the fish-e ating birds may be minimal. I also grouped the lakes using a modification of the deforesta tion categories developed by Pomeroy and Seavy (2003), where minimal deforestation has 50-100% of the lake area still forested, moderate deforestation has 25-49% forested, severe deforest ation has only small forest patches or scattered trees left, and complete deforestation has no forest trees along the crater rim. Due to the low number of lakes in some categories, the mini mal and moderate (moderate) and severe and complete (severe) deforestation categories were combined for the statistical analyses. Not surprisingly, lakes that were severely defore sted were often charac terized by high fishing pressure. This precluded quant itatively exploring the interac tion between deforestation and

PAGE 63

63 fishing pressure. As the correlation between th e two factors was not 100% (e.g., some severely deforested lakes were characteri zed by moderate or low fishing pr essure) I looked at each factor independently. Fish Collections A total of 2114 O. leucostictus and 2697 T. zillii were collected between July 2004 and July 2005 from the 17 crater lakes. Each lake was sampled once over a 5-6 day period. Significant effort was made to sample fish in di fferent size classes by using a variety of fishing gears, including: 1) tw o experimental monofilament gillnets each with 4 panels 60-m long and 1.0-m deep with stretched mesh sizes of 25.4 mm, 50.8 mm, 76.2 mm and 101.6 mm; 2) 1 artisanal fishermen nets comprised of mesh sizes 25.4 mm, 50.8 mm 63.5 mm, and 76.2 mm with a length of 1,800 m; and 3) 20 metal minnow traps (450mm long and 7-mm square wire mesh). The nets were set between 1400-1600 hr and pulled the following morning between 0800 hr -1000 hr. To obtain samples representative of fi sh populations in each la ke, nets were set in all major habitats, including open water, at the e dge of shoreline vegetation, and on rocky shores. Minnow traps, baited with bread, were set at the shallow littoral ar eas to obtain additional samples of smaller fish. Individual fish were measured to the nearest 1 mm for both total (TL) and standard (SL) length, and total weight a nd fish weight with the gonads removed were determined to the nearest 0.1 g using an Ohaus hand-held electric scale. Fish >320 g were weighed using a spring balance ( 1 g). Gonads were removed and weighed to the nearest 0.1 g and sex was determined by examination of the gonads. Calculation of Relati ve Condition (Kn) To facilitate comparisons among lakes and for the development of predictive models of fish condition, we calculated the relative condition factor of fish, which is the ratio of observed individual fish weight to expected weight of an individual of a given length, using the formula:

PAGE 64

64 Kn = Wi/aLi b (Le Cren, 1951); where Wi is observed individual fish weight, Li is observed individual fish total length, and a and b are sp ecies-specific constant s. These regression constants were obtained from the regi onal length-weight relationship (W = aLb) derived by pooling data for all lakes for each species. Length and weight data were log transformed and the resulting linear relationships fitted by least s quare regression using we ight as the dependent variable. Since the fishermen do not separate their landings based on sex, I did not determine sex-specific length-weight regressions for each species. To minimize error arising from seasonal fluc tuations in body weight due to reproduction, I used fish weight devoid of gonads for calcula ting relative condition factor. A minimum of 30 individuals per lake was considered acceptable fo r computing the length-weight regressions, but sample size (n) varied from 49 fish (Lak e Marusi) to 341 fish (Lake Mwamba) for O. leucostictus; and from 63 fish (Lake Wakenz i) to 317 (Lake Mwegenywa) for T. zillii. Only fish in similar size ranges were selected for the analyses. For O. leucostictus fish between 80-340 mm TL were included in the analyses, whereas a broader size range (65-300 mm TL) was used in computing length-weight relationships for T. zillii Not all lakes contained both species, and relative condition factor was determined in 13 lakes for O. leucostictus and 13 lakes for T. zillii Statistical Analyses Step-wise multiple regression analyses were used to assess the relationship between mean relative condition and a suite of continuous environmental variable s. These variables included: water transparency (Secchi depth), Chl-a, cond uctivity, pH, temperature, DO, TP, TN, maximum depth, mean depth, and lake area. Mean condition of fish in lakes of varying fishing pressure and deforestation was computed, and the values for both species were approximately 1.0. Oneway ANOVA was used to test for differences in mean relative condition of each fish species in lakes exposed to different fishing pressures (hi gh, medium, and low) and different levels of

PAGE 65

65 crater rim deforestation (moderate versus seve re). The Scheff post-hoc test was used for multiple comparisons. Analyses were performe d with Fishery Analysis and Simulation Tools software (FAST version 2.0) a nd SPSS for widows (version 12.0). Results Length-Weight Relationships and Relative Condition The size of the two tilapia species was hi ghly variable among lakes ranging from 80-340 mm TL and 7-420 g degonaded weight for O. leucostictus (Table 3-1) and 68-287 mm TL and 6-418 g degonaded weight for T. zillii (Table 3-2). The regional relationship between mass and length for each species, derived by pooling data across all lakes, was significant (both p < 0.001, Figure 3-2a, b). The weight-length allometry between the two species was also significantly different (ANCOVA, p < 0.018), with T. zillii being on average heavier at a given length than O. leucostictus Fishing Pressure, Catchment Defo restation, and Condition Factor Mean relative condition vari ed among lakes with differe nt fishing pressure for O. leucostictus (F3,13 = 5.05, p = 0.031). Oreochromis leucostictus in lakes with high fishing pressure had a higher relative condition compar ed to lakes with low and medium fishing pressure, which did not differ significantly from each other (Scheff: high versus low, p = 0.05; high versus medium, p = 0.05; Figure 3-3). Similarly, mean relative condition of T. zillii also varied significantly among lakes w ith different fishing pressure (F2,13 = 6.138, p = 0.018). Mean relative condition of T. zillii was also higher in lakes with high fishing mortality compared to lakes with low fishing mortality, but did not di ffer from lakes with medium fishing mortality (Scheff: high versus low p = 0.02; Figure 3-3). Mean condition of O. leucostictus also differed between moderately and severely deforested lakes (F1,13 = 4.74, p = 0.05), with fish exhibiting a higher condition than average in

PAGE 66

66 severely deforested lakes and conversely a lower condition than average in moderately deforested lakes (Figure 3-4) Mean relative condition of T. zillii did not differ between severely and moderately deforested lakes (F1,13 = 2.45, p = 0.15, Figure 3-4). Environmental Characters and Relative Condition Factor Stepwise multiple regressions identified two environmental factors, water transparency and conductivity, as significant predictors of relative condition of O. leucostictus with increased water transparency and conductivity associated with a lower fish condition. Water transparency (Secchi depth) was the most important determin ant of fish condition in the crater lakes (r2 = 0.48, p = 0.01), but in combination with conductivity, it accounted for 61% of the variation in fish condition (p = 0.02). For T. zillii, maximum lake depth, lake area, water temperature, conductivity, and water transparency were all identified as importa nt environmental correlates of fish condition (r2 = 0.88, p = 0.009). Water transparency, maximum depth and conductivity were negatively correlated with condition of T. zillii, but larger lake area and higher water temperature were associated with better fish condition. Discussion The size of O. leucostictus varied among the 13 crater la kes from which it was collected, ranging from 124 mm TL in Kerere to 161 mm TL in Kasenda. The largest O. leucostictus were a 340 mm TL male in Nyinabulitwa and a 340 mm TL female from Lake Murigarime. This is greater than the maximum size of 300 mm TL for male and 280 mm TL for female O. leucostictus reported for the open waters of Lake Victoria (Welcome 1967). In Lake Naivasha, Kenya, O. leucostictus was reported to reach a maximu m size of 310 mm TL for males and 280 mm TL for females (Siddique, 1977). Thus, in a few crater lakes, O. leucostictus seemed to reach a size comparable to some larger system s, though in many of the lakes the maximum size was much smaller. In our study, the allometric coefficient (b) for the regression between fish

PAGE 67

67 mass and length was 2.81 for O. leucostictus for fish between 80 mm and 340 mm TL, close to the value of 2.90 reported by Siddiqu e (1977) for Lake Naivasha, but lower than the coefficient of 3.06 reported in a later study of the same species in the same lake by Britton and Harper (2006, Figure 3-5a). Thus, for a fish of a gi ven length, mass increased more quickly in the O. leucostictus of Lake Naivasha collected in 2004-2005, than O. leucostictus from the crater lakes region of Uganda, and than O. leucostictus from a much earlier study of Lake Naivasha. The size of T. zillii also differed markedly among lakes ranging from 116 mm TL in Kasenda to 154 mm TL in Nyanswiga. The largest T. zillii individual caught in this study was a 287 mm TL male from Rukwanzi. For T. zillii collected in our study, th e value of the regression coefficient b was 2.99 for fish between 65 mm and 287 mm TL. This was very similar to the length-weight regression coefficients for T. zillii populations in Lake Zwai in Ethiopia (2.98 for fish between 55 320 mm TL) (Negasa and Getahun, 2003), but slightly less than for some West African lakes (3.2 for fish between 70-150 mm TL) (Teugels and Thys van den Audenaerde, 1991) and Lake Naivasha (3.2 for fish between 37 210 mm TL, Figure 5b) (Britton and Harper, 2006). Thus, T. zillii in the crater lakes of western Uganda are reaching relatively large size in some lakes; the condition of the fish seem s lower relative to some other systems. Interestingly, the maximum size of both O. leucostictus and T. zillii did not show a pattern between lakes with high vs. low fishing pressure One might expect to see a smaller maximum size in lakes with high fishing pres sure, unless the gear was not size-s elective. It is possible that fishing mortality is not severe enough in these systems to truncate the ag e or size distribution, although lakes with high fishing pressure did have established gi llnet fisheries. With higher fishing mortality one would initially expect greater growth in the fi sh (and potentially older ages if increased growth positively aff ects apparent survival). With increasing and consistent fishing

PAGE 68

68 pressure, it is possible that reduced intraspeci fic competition would continue to promote fast growth and good condition, but produce few large fis h. I am currently expl oring this idea using growth analyses and size freque ncy distributions across a subset of these lakes that vary in fishing pressure. Mean relative condition of the two tilapia species differed among la kes characterized by different fishing pressures. In addition, the condition of O. leucostictus differed between severely and moderately deforested lakes, while no such a relationship was detected for T. zillii A limitation of this study is the fact that lake s that were heavily deforested were rarely characterized by low fishing pressure. Deforesta tion often coincides with increased accessibility and/or human population density that promotes exploi tation of lakes. It is difficult to tease apart the interaction between deforestation and fishin g pressure. Nonetheless, there was sufficient variation across the lakes to speculate on the effect s of these two anthropoge nic pressures, and to identify significant limnological and morphometric predictors of fish condition for both species. Although studies examining the effects of fish ing on condition are scarce, the effect of fishing on other life history traits is well doc umented, especially for major commercial fish stocks (e.g., Ricker, 1981; Olsen et al., 2004, 2005; Grift et al., 2002). It is often assumed that fishing-induced reductions in popula tion density of target stock le ads to increased yield because of reduction in intraspecific competition that releases populations from density-dependence resulting in faster growth and earlier maturati on (Jennings and Kaiser, 1998; Hall, 1999; Law, 2000; Rochet et al., 2000). A reduction in intraspeci fic competition is also likely to foster better fish condition. In support of this, the growth and condition of largemouth bass ( Micropterus salmoides ) in smaller water bodies was also found to be affected by density-dependence (Wege and Anderson, 1978; Schindler et al., 1997) Diet studies on the two tilapiine species

PAGE 69

69 indicate that the food of O. leucostictus consists predominantly of phytoplankton (Trewavas, 1983), whereas T. zillii is known to feed mostly on macrophytes (Buddington, 1979) and macrobenthos (Greenwood, 1966). In the pres ent study, the differences in condition of O. leucostictus among lakes with high fishing pressure a nd the lakes with low to medium fishing pressure may reflect a greater availability of phytoplankton in re sponse to reduction in population density caused by fishing. The abundan ce of phytoplankton, as indicated by the high chlorophyll a values in some lakes, may have contributed to faster growth and improved condition of O. leucostictus in lakes with high fishing pressure. Similarly, for T. zillii, it is also likely that a decrease in populat ion density associated with he avy fishing may have decreased intraspecific pressure on food resources. Other environmental factors were also si gnificant predictors of condition of both O. leucostictus and T. zillii. For the phytoplanktivorous O. leucostictus water transparency (Secchi depth) was the most significant predictor of fish condition. In these lakes water transparency is highly correlated with Chl a (an index of phytoplankton abundan ce) and also characteristic of severely deforested lakes (Efitr e, 2007). One possible scenario is that in lakes experiencing severe deforestation, increased erosion and nut rient loading leads to accelerated primary productivity that may enhance the food base for the phytoplanktivorous O. leucostictus In addition to water transparency, condition of O. leucostictus was also correlated with water conductivity, with lake s having higher conductivity associated with poorer fish condition. This contrasts with studies of conditi on of other freshwater fish sp ecies, such as spotted bass Micropterus punctulatus (Rafinesque) (DiCenzo et al., 1995) and brown trout Salmo trutta L. (Dennis et al., 1995). These studies report that higher water conductivity (a surrogate of productivity) is associated with better fish condi tion and attribute this to the high metabolic of

PAGE 70

70 ion regulation by fish in low conductivity streams with a steeper gradient against which ions are transported into the body fluid compared to fish in streams with high c onductivity. Studies of Mediterranean Barbus meridionalis Risso (Vila-Gispert a nd Moreno-Amich, 2001) and 14 species of 0+ year fishes from the River Grea t Ouse in the United Kingdom (Copp, 2003) did not find any relationship between conductivity and fish condition. It has been argued that the relationship between water conductivity and fish condition may be a simplistic and coincidental one and probably a reflection of geographical pa tterns in ecosystem function, such as ecological succession state or productivity (DiCenzo et al., 1995; Dennis et al., 1995; Fortin et al., 1990). In addition, the conductivity-f ish condition relationship also ignores biotic and other environmental factors such as dissolved oxygen concentratio n and riparian cover which influence fish condition (Vila-Gispert and Moreno-Amich, 2001). In contrast, primary productivity was not correl ated with condition of T. zillii and condition of this species did not differ between groups of lakes characterized by di fferent levels of deforestation. Instead, the condition of T. zillii was correlated with maximum lake de pth, lake area, water temperature, water transparenc y, and conductivity. Tilapia zillii is characteristically a macrophyte feeder (Buddington, 1979). In the crater lakes, in gene ral, macrophytes are quite rare, but if they do occur, it is often in shallow lakes or large lakes with a better developed littoral zone. One could imagine macrophytes should be more of a limite d resource than phytoplankton in these steepsided craters. It is possible that T. zillii in the crater lakes has br oadened its dietary niche to include food items other than macrophytes. I routinely observe T. zillii grazing on the epiphytic layer that carpets the inshore rocky boulder areas of the submersed crater rim in Lake Nkuruba. Fish condition is known to vary seasonally de pending on changes in gonadal development, food availability, and other environmental factor s (Pope and Willis, 1996). It would be best to

PAGE 71

71 compare condition of fish populations sampled at the same time of the year. In this study, the populations were sampled during different seasons of the year and we did not include gonad weight in our estimates of condition. In addition, both O. leucostictus and T. zillii are known to spawn throughout the year with slight variation in intensity of breeding (Welcomme, 1967; Siddique, 1977). In a study of O. leucostictus in Lake Naivasha, Siddique (1977) found all stages of gonad maturation all year round and di d not note any seasonal fluctuation in relative condition, which he attributed to a constant proportion of fish with gonads in different stages of development. In the present study, I also observe d both male and female fish of both species at various stages of gonad development during the di fferent seasons. Although seasonal changes in condition may certainly occur, th e strength of our predictive re lationships suggest that bias associated with seasonal variation is likely minimal. Measures of fish condition are generally intende d to be indicators of tissue energy reserves and may characterize components of the environmen t in which the fish lives (e.g., habitat, prey availability, and competition) (Vila-Gispert et al., 2000; Vila-Gispert and Moreno-Amich, 2001). In this way, indices of fish condition are of value to fishery managers who must assess population status, the impact of management actions, and anthropogenic influences on the resource they are managing (Brown and Murphy, 1991). In this study, I have explored how two tilapiine species respond to two t ypes of anthropogenic perturbati ons (i.e., deforestation that reflects differences in primary pr oductivity and fishing pressure th at may decrease intraspecific competition). Results of the present study demonstrate that O. leucostictus in severely deforested lakes and heavily fished lakes were in a better condition compared to similar fish in lakes with low productivity and low to medi um fishing. Differences in condition of T. zillii were only detectable between lakes with high and low fishing but no relationship was observed

PAGE 72

72 between condition and extent of deforestation, although I was unable to quantitatively explore the interaction between the two factors. From a conservation biology view point, it is interesting that O. leucostictus showed higher condition values in seve rely deforested lakes compared to moderately deforested ones. This may in part reflect the association between deforestation and fishing pressure. It may also reflect changes in water quality induced by deforestation of the watershed. A strong negative relationship between water transparency a nd Chl-a concentration was reported in the crater lakes with lakes with deforested catchments having a lower transparency and higher Chl-a concentration comp ared to lakes with forested catchments (Efitre, 2007). In addition, a weak positive rela tionship between Chl-a concentration and total phosphorus concentration was also found in these lakes with incr eased input of total phosphorus (TP) leading to increase in Chla concentration (Efitre, 2007). These results should be integrated into development of innovative approaches for management of tilapiine populations in these lakes.

PAGE 73

73Table 3-1. Lake catchment deforestation le vel, fishing pressure, range in total le ngth and degonaded wei ght, and length-weight regressions (W=aLb) for Oreochromis leucostictus from 13 crater lakes in western Uganda, where b=slope of regression and log10a=intercept of regression. Lake Deforestation Fishing pressure Sample size (n) Range in total length (mm) Range in degonaded weight (g) Log10 a b r2 Wankenzi Severe High 181 113-172 27-86 -3.59 2.5 0.90 Mwamba Severe High 316 106-177 22-88 -3.82 2.6 0.88 Kanyango Severe High 277 91-210 13-160 -3.88 2.6 0.85 Lugembe Severe High 284 89-205 12-163 -3.98 2.6 0.87 Murigarime Severe High 112 80-340 8-612 -4.81 3.0 0.98 Lyantonde Severe Medium 156 87-188 13-115 -3.54 2.5 0.90 Marusi Severe Medium 49 80-275 8-270 -4.52 2.8 0.97 Wandakara Severe Medium 114 118-238 28-230 -4.65 2.9 0.90 Nyinabulitwa Moderate Medium 169 93-340 13-420 -4.52 2.9 0.93 Nkuruba Moderate Low 126 102-262 18-281 -3.72 2.5 0.90 Ntanda Moderate Low 98 85-160 23-285 -3.89 2.6 0.92 Kerere Moderate Low 91 80-235 7-207 -4.58 2.9 0.97 Kasenda Moderate Low 130 80-290 7-411 -4.65 3.0 0.97

PAGE 74

74Table 3-2. Lake catchment deforestation le vels, fishing pressure, range in total leng th and degonaded weight, and length-weigh t regressions (W=aLb) for Tilapia zillii populations from 13 crater lakes in western Uganda, wh ere b=slope of regression and log10a=intercept of regression. Lake Deforestation Fishing pressure Sample size (n) Range in total length (mm) Range in degonaded weight (g) Log10 a b r2 Kanyango Severe High 146 78-197 9-134 -4.87 2.8 0.95 Lugembe Severe High 205 77-265 8-300 -4.33 2.8 0.98 Wankenzi Severe High 62 73-253 7-297 -4.45 2.9 0.97 Kifuruka Severe High 262 69-243 5-278 -4.66 3.0 0.97 Lyantonde Severe Medium 232 68-211 5-163 -4.72 3.0 0.98 Wandakara Severe Medium 163 73-234 7-226 -4.80 3.0 0.99 Mwegenywa Severe Medium 332 71-233 6-235 -4.99 3.1 0.97 Rukwanzi Severe Low 136 75-287 7-418 -4.78 3.0 0.99 Nyinabulitwa Moderate Medium 220 70-272 6-353 -4.98 3.1 0.99 Nyanswiga Moderate Low 270 68-270 6-290 -4.33 2.8 0.99 Nkuruba Moderate Low 187 68-176 6-384 -4.71 3.0 0.99 Kasenda Moderate Low 112 75-204 7-151 -4.98 3.0 0.98 Ntanda Moderate Low 271 76-222 8-180 -4.98 3.1 0.98

PAGE 75

75 Figure 3-1. Location of Kase nda cluster of crater lakes in western Uganda (0o23 -0o33 N, 30o10 30o 20 E) where Oreochromis leucostictus and Tilapia zillii were sampled along with environmental variables from July 2004-July 2005.

PAGE 76

76 0 100 200 300 400 500 600 700 050100150200250300350400 0 100 200 300 400 500 600 700 Degonaded weight (g)a)W = 0.000042L2.81(R2 = 0.92, n = 2114) Total length (mm) b) W = 0.00018L2.99(R2 = 0.98, n = 2697) Figure 3-2. Regional relationships between degonaded weight (g) and total length (mm) for A) Oreochromis leucostictus. B) Tilapia zillii derived by pooling data from 13 crater lakes in western Uganda.

PAGE 77

77 0.9 1.0 1.1 Low Medium High b ab a b aRelative condition (Kn)Fishing pressure O. leucostictus T. zillii a Figure 3-3. Mean re lative condition of Oreochromis leucostictus and Tilapia zillii in crater lakes of western Uganda exposed to high, medium and low fishing mort alities. Vertical bars represent 1 standard error. Diffe rent letters above the fishing pressure categories indicate a significant difference at p 0.05.

PAGE 78

78 0.9 1.0 1.1 a a b Moderate SevereRelative condition (Kn)Deforestation O. leucostictus T. zillii a Figure 3-4. Mean re lative condition of Oreochromis leucostictus and Tilapia zillii in crater lakes of western Uganda exposed to moderate a nd severe deforestation levels. Vertical bars represent 1 standard error. Differe nt letters above the deforestation levels indicate a significant difference at p 0.05.

PAGE 79

79 0 100 200 300 400 500 080160240320 0 100 200 300 400 500 600 700 Weight (g) L. Naivasha (Britton & Harper 2006) L. Naivasha (Siddique 1977) Lake Zwai (Negassa & Getahun 2003) Ugandan crater lakes(a) Total length (mm)(b) Figure 3-5. Comparison of lengt h weight-regression curves for O. leucostictus populations from A) Ugandan crater lakes and La ke Naivasha, Kenya. B) for T. zillii populations from Ugandan crater lakes, Lake Naivasha, Ke nya, and Lake Zwai, Ethiopia. Data for Lake Naivasha were adapted from Siddique 1977 and Britton and Harper 2006 and for Lake Zwai from Negassa and Geta hun 2003 with sampling periods representing the years 1974-1975, 2004-2005, and 2001, respectively.

PAGE 80

80 CHAPTER 4 VALIDATION OF PERIODICITY AND TIMI NG OF OPAQUE ZONE FORMATION IN Tilapia zillii (PISCES: CICHLIDAE) OTOLITHS FROM CRATER LAKE NKURUBA, WESTERN UGANDA Introduction Knowledge of the age and growth of fishes is critical for assessing the effects of fishing on stocks, understanding fish life history, determinin g the impact of management practices, and for maximizing yield (Jones, 1992). Information on ag e structure is also vital for understanding how environmental variation affects the growth, a pparent survival, and recruitment success of juvenile fish (Jones, 1992). Ag e-structured stock assessments form the basis for the management of many fish species (Hilbron and Walters, 1992; Campana, 2001); and accurate age estimates are critical as underestimation of age and result ant overestimation of gr owth can lead to the collapse of major commercial fisheries (Campana et al., 1990; Beamish and McFarlane, 1993). Fish age and growth can be estimated thr ough various methods including the examination of calcified structures for growth rings. Fish calcified structures, such as scales, vertebrae, fin rays, spines, opercules, clavicles, and otoliths often show daily, sub-annual, or annual growth zones or rings, and provide an important means of aging fishes. Amongst the structures used for fish aging, the counting of periodic increm ents in otoliths is considered to be the most reliable procedure (Campana, 2001), because they can be used over the broadest range of ages in many species (Secor et al., 1995a). Otoliths or ear bones are small biogenic structures that form part of the hearing and balance system of teleost fishes. In addition to their physiological functions, otoliths also act as nat ural data loggers recording information in their microstructure and chemistry in relation to the growth a nd environment of the fish (Campana, 1999). Since Raibisch first observed annular ring formation in Pleuronectes platessa in 1899 (as reported in Ricker, 1975), the use of otoliths to age temperate fish species has gained wide

PAGE 81

81 acceptance. Aging of tropical fishes using ot oliths only became common in the early 1990s (e.g., Morales-Nin and Ralston, 199 0; Francis et al., 1992; review s by Fowler, 1995). The lack of early otolith studies in tropical fishes refl ected a widespread assumption that growth of tropical fish is rather consistent due to the weak temperature fluc tuations characteristic of many tropical environments. Otoliths and other calcified structures of tropi cal fish species were not expected to form increments that relate to variation in growth resulting from seasona l fluctuations in the environment. It has become evident that variatio ns in climatic conditions in tropical regions may lead to the deposition of one or more growth z ones in fish calcified st ructures (Brothers, 1979; Yosef and Casselman, 1995, Bwanika et al., In pre ss). Although the physiological basis for the formation of growth zones in tropical fish species has not been conclusively determined (Ferreire and Russ, 1994), bot h abiotic and biotic factors have been suggested as possible causes of formation of these growth zones (Beckma n and Wilson, 1995). For example, wet-and-dry seasonal cycles (Bwanika, 2005, Bwanika et al., In press), regular seasonal variation in temperature in some tropical regions (Blake and Blake, 1978; Panella, 1980; Yosef and Casselman, 1995), variation in total dissolved solids (Fagade, 1974), photoperiod, feeding patterns, and spawning period (e.g., Morales-Nin and Ralston, 1990), and spawning activity and associated body condition (Nekrasov, 1980; Booth et al., 1995) have been suggested as possible causes of growth zone formation in tropical fish species. This has led to a growing literature on age and growth studies in a wide range of tr opical environments and species including as examples, studies of tropical co ral reef fishes (Fowler, 1995; Ne wman et al., 1996) and studies of cichlid fishes in Africa (Yosef and Casselm an, 1995; Panfili and Toms, 2001; Egger et al., 2004; Bwanika et al., In press).

PAGE 82

82 Prior to using calcified structures such as ot oliths for aging fish, the timing and periodicity of growth zone formation need to be validated (Beamish and McFarlane, 1983) to ascertain the accuracy and feasibility of aging criteria (Campana, 2001) as well as differentiating indeterminate increments" (increments not related to the annual time scale) that occur in all fish calcified structures (Fowler a nd Doherty, 1992). Although several validation techniques such as mark-recapture of chemically tagged wild fi sh, bomb-radiocarbon, and marginal increment analysis have been used for ag e validation (Campana, 2001), techni ques that validate the age of all age-groups rather than only young fish are ge nerally considered the most successful (Beamish and McFarlane, 1983). Of these techniques, marg inal increment analysis (MIA) is the most commonly used validation method because of its modest sampling requirement and low cost (Campana, 2001). The underlying premise of MIA is that if a growth increment is formed on a yearly cycle, the average state of completion of the outermost in crement should display a yearly sinusoidal cycle when plotted ag ainst months (e.g., Hyndes et al ., 1992; Fowler and Short, 1998). In this Chapter, I use marginal-increment analys is to validate periodic ity and timing of opaque zone formation in otoliths of T. zillii in an equatorial tropical crater lake. Validation of periodicity and timing of opaque zone formation seen in the otoliths of th is species is important to the management of introduced ti lapias that dominate the fish fauna of approximately 89 crater lakes in western Uganda and serve as an important local protein resource. Tilapia zillii commonly referred to as the red-belly t ilapia, is an herbivorous species that naturally occurs in West Africa through the Chad ba sin to the Nile River, Lake Albert, and Lake Turkana into Israel and the Jordan Valley (T rewevas, 1933). The success of this species undoubtedly reflects the high degree of plasticity that it exhibits in its diet, life-history traits, and physiology. And, the ability of T. zillii to tolerate a wide range of temperature and salinity, in

PAGE 83

83 addition to its ability to feed on aquatic vegeta tion, makes it a suitable species for aquaculture (Siddiqui, 1979). The plasticity in life history characters of T. zillii is often associated with production of stunted populations which is a major concern to both aquaculture and capture fisheries. An understanding of variation in im portant life history traits of T. zillii such as growth and size/age at maturity is of great value to ma nagement of introduced and native populations, and may have broader application to other substrate spawning tilapias that have received far less attention than mouth-brooding tila pias. Several studies have documented aspects of the ecology and life history of T .zillii in African lakes (e.g., El Zark a, 1962; Welcomme, 1967; Siddiqui, 1979; Negassa and Getahun, 2003; Britt on and Harper, 2006). There is paucity of information on age and growth of the species in natural popula tions. In particular, va lidated age and growth studies of T. zillii in African natural waters using otolith microincreme nt structure are lacking. The aim of this Chapter was to validate the pe riodicity and timing of opaque zone formation in otoliths of T. zillii in Crater Lake Nkuruba using margin al increment analysis. The specific objectives were: 1) to develop a criterion for interpre ting growth at the edge of the otoliths; 2) to validate the periodicity and timing of opaque zo ne formation within otolith sections using marginal increment analysis (MIA); and 3) to us e the validated age data to estimate growth and mortality rates of T. zillii population within Lake Nkuruba.

PAGE 84

84 Materials and Methods Study Area Crater Lake Nkuruba, a small (surface area 3 ha) mesotrophic freshwater lake, belongs to the Kasenda cluster of volcanic crater lakes in western Uganda, located at 0o 32 N and 30o 19 E (Figure 4-1). Maximum and mean depths of th e lake are 37 m and 16 m, respectively. Surface dissolved oxygen concentration ranges from 4-8 mg L-1, but the lake is anoxic below 8-15 m depending on the season (Kizito et al., 1993; Chapman et al., 1998). Water transparency fluctuates between 1.4-5 m, electrical conductiv ity of surface water ranges from 341438 S cm1, and surface water temperature ranges from 22-24.3 oC (Kizito et al., 1993; Chapman et al., 1998). Total phosphorus concentration of su rface water ranged from 38 to 43 g L-1 during the period 1990-1992 (Kizito et al., 1993) during which time chlorophyll a concentration of surface water was low (range 9-22 g L-1( Table 4-1). Aquatic macrophytes are limited to a very small area on the eastern shore of the la ke; and the crater wall, averag ing about 48 m above the water surface, is characterized by a rich forest of large tropical trees. Lake Nkuruba has no surface inflow or outflow but changes in water levels have been attributed to ground water exchange (Kizito et al., 1993). Two tilapia species, Oreochromis leucostictus (Trewavas, 1933), and Tilapia zillii (Gervais, 1874), as well as the guppy Poecilia reticulata (Peters, 1859) were introduced into the lake to improve protein sources for the local people and to control malaria-causing mos quitoes. Of the two tilapia species, T. zillii is the most abundant in the lake. There is no gillnetting fishery in Lake Nkuruba, although a small number of hook fishermen (less than 30) fish in the lake. Rainfall is bim odal, with two wet seasons from March-May and September-November, with 1.7 m of total rainfall received annually (1990-2005). Between 1990 and 2004, the mean daily minimum temperature was 14.9oC, and the mean daily maximum temperature was 20.2oC (Chapman and Chapman unpublished data).

PAGE 85

85 Surrounding areas are largely ag ricultural land with small patc hes of forest around the steep crater valleys. Fish Collections and Processing I collected a total of 526 Tilapia zillii during a 12-month period between July 2005 and July 2006, with approximately 2 w eeks between sampling periods. Fi sh were captured using two sets of experimental monofilament gillnets and mi nnow traps. Each gillnet consisted of 4 panels 60-m long and 1.0-m deep with stretched me sh sizes of 25.4 mm, 50.8 mm, 76.2 mm, and 101.6 mm. The nets were set between 1400-1600 hr and pulled the following morning between 0800 hr-1000 hr. To obtain samples representative of fish populations in each lake, I set the nets in all major habitats, including open water, at the edge of s horeline vegetation, and on rocky shores. A total of 20 minnow trap s, baited with bread, were also set at the shallo w littoral areas to obtain additional samples of smaller fish. I measured individual fish to the nearest 1 mm for both total (TL) and standard (SL) length. Total weight was measured to the nearest 0.1 g using Ohaus hand-held electric scale. I weighed fish >320 g using a spring balan ce ( 1 g). Finally, I removed gonads and weighed them to the nearest 0.1 g and determined sex of individual fish by examining the gonads. Length-Weight Relationships I used the regression equation: W = aLb, to describe the length-weight relationships for the gillnet and minnow trap samples separately and then for the combined data; where; W = fish weight in grams, L = total length in mm, a = a regression constant, and b = an exponent usually ranging from 2.5-3.5, that describes the curve of the relationship (e.g., a fish maintaining the same shape across length categories will have an exponent (b) of 3). Length and weight data were log transformed, and the resu lting linear relationships were fi tted by least square regression using body mass as the dependent variable.

PAGE 86

86 Length Frequency To detect potential effects of gear on mean total length, I plotted length frequencies for T. zillii caught with traps and nets separately. I also evaluated the effect of gear on gender by plotting length frequencies for female and male fish separately for each gear. A visual assessment of the length frequency plots revealed no differences between female and male size frequencies for each gear. I pooled length-frequency data fo r all fish (males, females, immatures) by gear type. I compared length fr equencies of gillnet and minnow trap samples using Kolmogorov-smirnov test (significance = 0.05). Otolith Processing and Interpretation In order to estimate age and growth of T. zillii sagittal otolith pairs from subsampled fish in different size classes were removed through an incision on the cranium slightly above the eyes. Both otoliths were wiped with small hand towels to remove attached membrane tissue, airdried, and stored in labeled plastic vials until pro cessed for age determination. Later, the otoliths were prepared using thin-sectioning. Various me thods have been used to prepare otoliths for examination of periodic increment structure in fish otoliths, includi ng: 1) whole otolith examinations, where the otolith is immersed in a liquid medium and read under a stereoscope (e.g., Hyndes et al., 1992); 2) br oken and burnt method in which a transverse broken section of the otolith is heated to intensify the incremen t structure (e.g., Francis et al., 1992); and 3) thinsectioning method, where transverse sections are sliced through the premodial region of the otolith (e.g., Beamish, 1979). Of the three met hods, thin-sectioning is considered the best for otolith examination and interpretation, because it improves detection of increment structures especially for older fishes that may be under-aged using whole otoliths (Beamish, 1979; Campana, 1984; Hyndes et al., 1992).

PAGE 87

87 In the laboratory, the left sagittal otolith from each fish was secured to a frosted glass slide using melted Thermoplastic Quartz Cement (the ri ght sagittal otolith was processed if the left otolith was broken or damaged during preparation) The mounted otolith was then thin-sectioned (~0.5 mm thick) through the core region of the ot olith along a transverse plane using a Buehler Isomet 1000 digital sectioning saw (Buehler IL) with a diamond wafering blade (7.6 cm diameter X 0.15 mm blade width) at a speed of 300 rpm. The resu lting thin sections from the otolith were rinsed in ethyl al cohol and dionized wate r and then mounted on glass slides with thermoplastic glue (crystal bond). To improve read ability of the microincrements, sections were ground with Gatorgrit wet-dry sandpaper (grit sizes 400 and 600 micron) down to the core on one side. The samples were then finished by polishing with Buehler micro cloth and 0.3 micron alpha Buehler micropolish alumina polishing powder paste. Interpretation of Otolith Sections and Precision of Age Estimates Otolith sections were examined with a ME IJI EMZ-TR stereomicroscope (10-40 X) under transmitted light to count opaque growth zones. A combination of one narrow opaque zone (dark brownish) with one broad translucent zone (clear light brown) was interpreted as a complete annulus under transmitted light. Annuli were read and counted along the sulcus that was chosen as the primary reading axis because the annuli were most visible along this axis (Figure 4-2). The narrow opaque zones were counted to determine the number of annuli and fish aged following Egger et al., (2004). The c hoice of opaque zones for aging was based on previous studies that provide evidence in suppor t of seasonal deposition of one or more growth zones in tropical fish calcified structures (e.g., Brothers, 1979; Yosef and Casselman, 1995). In a study of Oreochromis niloticus in Ethiopian lakes, Yosef (1990) and Yosef and Casselman, (1995) found that the opaque zone formation was a ssociation with periods of faster growth fast

PAGE 88

88 growth, whereas the corresponding translucent zo ne were formed during periods of slower growth. Aging criteria were then esta blished taking into a ccount the type of growth zone at the edge and the reading quality of the otolith. Readab ility codes were assigned as follows: i) otolith section was very difficult to read (= 3); ii) good readability (= 2); and iii ) excellent readability (= 1). For edge type, the following codes were used: i) the opaque zone was at the age (edge = 0); ii) narrow translucent zone at the e dge (width less than about 30% of the previous increment = 1); iii) medium translucent zone at the edge (width about 30-60% of the previous increment = 2), and iii) wide translucent zone at the edge (width more than 60% of the previous increment = 3). Following the establ ished aging criteria, the primary reader (JE) read all otoliths twice, without knowledge of collection date, fi sh length, and other sampling information. The second reading was done after approximately a mont h had passed. Number of annuli, edge type, and readability quality were recorded for each fish. To determine between reader-agreement, a random subsample of 100 otoliths covering the complete age range of were read by a secondary reader (DJM) experienced in examination of transverse sagittal otolith sections. Otoliths were read with no reference to fish length, date of collection, or resu lts of previous age estimates by the primary reader. Counts of annuli from the pr imary and secondary reader were compared, and otolith sections that yielded di fferent counts were re-examined by both readers, and a consensus reached after a discussion. Edge type and coll ection date were then used to assign ages to individual fish through advancing or not advancing counts of annuli. For example, fish with 10 opaque zones and having a wide translucent zone (edge = 3) would be advanced and assigned the same age as another fish with 11 opaque zones but with ultimate opaque zone just forming at the edge (i.e., edge = 0).

PAGE 89

89 Marginal Increment Analysis (MIA) Edge interpretation and marginal-increment anal ysis were used to validate the periodicity of opaque zone formation by analysis of temporal pattern of marginal increment. For edge analysis, the type of zone (opaque or translu cent) at the growing edge of the otolith was identified in each month. The percentage of otol iths with opaque zones at the growing edge was then calculated and plotted against month ove r a 12-month cycle. Marginal-increment was determined quantitatively by measuring the marg inal increment (MI), and previous increment (PI) of the translucent zones at th e otolith edge on digitized images along the same axis used for counts of opaque increments (i.e., from the proxi mal growing edge, ventral to the sulcus, Figure 4-3). Measurements were made using Motic Image System (Version 2.0, Motic, Inc., Richmond B.C.) under transmitted light at magnif ication ranging from X20-X40 To minimize any potential errors in increment measurement, otol iths were measured randomly without any prior knowledge of fish age, size, and sampling date The periodicity of marginal-increment deposition was then determined by calculating the index of completion (C) for transverse thinsectioned otoliths using the follo wing formula (Tanaka et al., 1981): C = Wn/Wn-1 Where; Wn is the width of the marginal increment (distance from the otolith growing edge to the center of the outermost complete opaque zone) Wn-1 is the width of the previous complete increment (distance from the center of the outermost opaque zone to the center of the second from outermost opaque zone). Marginal increment ratio (MIR) was used to expr ess the MI of each otolith as a percentage of the PI (previous increment) by calculating the marginal increment ratio using: MIR = MI/PI*100. The MIR is a measure of the percentage of otolith growth since the last increment was formed relative to the previous year of otolith growth. An assumption of the method is that

PAGE 90

90 the widths of the outermost two increments are re latively similar, such that when the forming increment is complete, the MIR will be ~ 100%. Growth Because of lack of smaller fish in the ag ed sample, it was not possible to determine to using the von Bertalanffy growth equation (Ricker 1975). To better estimate growth, I forced the Von Bertalanffy growth model through zero (Fisher et al., 2004) using non-linear regression analysis (SAS Institute Inc., 1987): Lt = L (1-e -k [t to] ), Where: Lt = predicted total length (mm) at time t (age, in yrs); L = estimate of average maximum length (asymptotic length) (mm) ; K = Brodys growth coefficient; and to = theoretical age (yrs) when fish length would be 0 (assu med to be zero in this analyses). I fitted models independently for gillnet and trap samples to explore potential gear biases. Estimates of K obtained from the models forced through 0 seemed unrealistic. I also estimated growth by calculating mean length of each age class for each gear type and plotted mean length at age against age to determine average growth rates. I used a two-way ANOVA to test for differences in mean length at age between gears, where age and gear (gillnet and minnow traps) were main effect and length at age the res ponse variable. Age related growth differences between gears were only determined for ages 3 to 7. Ages less than 3 and greater 7 were excluded from the analysis due to low sample size. Analyses were performed using SPSS (version 12.0) and at significance level = 0.05. Estimates of Total Mortality and Total Apparent Survival Rates I estimated instantaneous total annual mortality (Z) and total annual apparent survival (S) rates for T. zillii for the trap and gillnet samples separate ly using standard cat ch-curve analysis (Ricker, 1975) based on the assumption that the samples from each gear represented the actual age distribution from the populat ion across the sizes selected by the gears. Mortality and

PAGE 91

91 apparent survival estimates were computed usin g the regression method of plotting ln(N) versus age; where; ln is the natural logarithm of the num ber of fish at each age. Apparent survival was then estimated from the formula; S = e-Z; where Z is the instante nous total annual mortality. Instantaneous Natural Mortality Rate (M) I determined instantaneous rate of natural mortality using tw o different methods. The first natural mortality estimate was computed us ing FISAT software (version 1.0.0) following the empirical method developed by Rikhter and Efa nov (1976) with the equation described by: M = 1.52/ tmass0.72 ; Where; tmass is the age at massive maturation [I used the Fishbase upper age at maturity value = 3 (range of 2 3 yrs) reported for T. zillii ]. The second approximation of M was based on Hoenigs (1983) longevity-mortality relationship where the mortality rate is computed from the oldest individual encountered in the data set following the equation: ln(M) = 1.46 1.01 ln(tmax); Where; ln is natural logarithm, and tmax is the maximum age for the species (observed maximum age for lake Nkuruba T. zillii population was 8 yrs) Results Length-Weight Relationships The length-weight relationships for the mi nnow trap sample (size range 51-138 mm TL and 4.1-48.8 g total weight) and gillnet sample (size range 73-284 mm TL and 6.0-425 g total weight) from Lake Nkuruba were both curvilinear and highly significant (Trap: W = 0.00002TL2.98, R2 = 0.96, n = 362, p < 0.0001; gillnet: W = 0.000016 TL3.01 R2 = 0.99, n = 78, p < 0.001, Figure 4-4). A gene ral ANCOVA model indicated that slopes (p = 0.07) and intercepts (p = 0.09) of the regre ssion lines for the trap and gillnet fish were not different. All data were pooled and a single regression e quation obtained for both gears. The common equation for the combined data was: W = 0.00021TL2.97, R2 = 0.99, n = 440.

PAGE 92

92 Length Frequency Length frequencies for both gears showed distinct separation in sizes of fish caught by the two gears. For each gear, the lengt h frequency distributions of fema le and male fish were similar (Figure 4-6) and were combined to de tect gear effects. A total of 450 T. zillii were collected during the course of this study, including 370 (8 2.2%) from minnow trap s and 80 (17.8%) from gillnets (Figure 4-7). Length frequencies of T. zillii collected in gillnet and minnow trap samples were significantly different (Kolmogorov-smirnov test; Z = 3.94; p = 0.0001). Gillnets consistently collected larger fish than did the use of minnow traps with traps catching mostly fish less than 130 mm TL and nets ca tching bigger fish, although there wa s some degree of overlap in sizes of fish caught by the two g ears (Figure 4-7). A plot of T. zillii size frequency by month over a 1-year period also showed a bimodal pattern with gillnetted fish being bigger compared to trapped fish in most lakes (Fi gure 4-13), supporting the trend obser ved in the combined plots of all months. Tilapia zillii Growth Zone Identification For Tilapia zillii opaque and translucen t zones were not readily identifiable in whole otoliths (Figure 4-5). Transver se sagittal otolith thin secti ons showed faint and indistinct alternating translucent and opaque zones when viewed under transmitted light. Sanding of transverse otolith sections with wet-dry sand paper and subsequent polishing with 0.3 m alumina polishing powder clearly improved the readab ility of the sections and identification of the zones (Figure 4-2). The core of thin-sections of T. zillii otoliths was easily recognizable as a fairly dense large opaque struct ure (Figure 4-2), separated by a large translucent zone that corresponds to the first complete annulus. Thin sections of T. zillii otoliths displayed considerable variation in size a nd structure of the opaque zone; from narrow and distinct opaque zones extending throughout the otolith sections (Figure 4-3) to broad and split-opaque zones

PAGE 93

93 with cross-checking or false a nnuli along the dorso-ventral axis of the sulcus (Figure 4-2). These split annuli not only made reading of annuli difficult, but were also a potential source of aging error as they increased the number of annuli. As a result, annuli were followed along the core-ventral axis to the sulcus region, which was chosen as the standard reading and counting axis because the annuli were the cl earest in this region (Figure 4-2) If false annuli and other accessory growth structures joined with the true annuli at the sulcus region, then it was counted as one annulus. Annuli were disregarded when rings were faint and incomplete. For older fishes, the annuli became compact and less di stinct at the edges making aging difficult. Validation of Periodicity and Ti ming of Increment Formation Results of marginal-increment and edge analyses indicated that T. zillii in the crater lakes of western Uganda deposit two annuli (two opaque zones and two translucent zones) in their otoliths annually. Mean indice s of completion (width of translucent zones at the otolith edge) were highest during the dry seasons between Janu ary-February (peaking in January) and between June-August (peaking in August), corresponding with translucent zone form ation (Figure 4-8). In contrast indices of completion were lowest between September and November and between March and May, indicating deposition of two opaque zones per year (Figure 4-8). Edge analysis results further confirmed the bimodal pattern, w ith the highest percentage of opaque zones at otolith edges occurring between September and November and between March and May (Figure 4-9). The highest values of mean monthly index of comp letion in August and January corresponded with peaks of the dry season, and the lowest values of mean index of completion in November and May were detected during the peak of the first wet season and about one month after the peak of the second wet season, respec tively (Figure 4-9). Th ere was a significant positive correlation between mean monthly rainfall and percentage of otolith sections with

PAGE 94

94 opaque edges (r = 0.70, p = 0.005, n = 12). There wa s no significant relationship between mean monthly rainfall and the mean index of completion (r = 0.12, p = 0.35, n = 12). Age and Growth The age of T zillii collected from Lake Nkuruba ranged from 2 to 8 years (Figure 4-10). Growth increment in T. zillii peaked between age 2 and age 3 and decreased at ages greater that 3 (Figure 4-10). The maximum observed age for T. zillii in Lake Nkuruba wa s 8 years (Figure 410). There was a significant difference in mean to tal length-at-age between gears, with gillnets consistently collecting larger fish at age co mpared to those collected with minnow traps (ANOVA F = 40.63; df = 8; p = < 0. 0001, Table 4-3), indicating a fast er growth rate in the gill net samples compared to the minnow trap fish. On average, age 3T. zillii were 137.8 mm TL in the gillnet samples and 87.1 mm TL in the minnow trap samples (Table 4-2). There was no significant difference in age distributions betwee n gillnet and minnow trap samples (F = 1.04; p = 0.4) nor was there an interact ion between age and gear type (F = 0.7; p = 0.65; Table 4-3). Both gears exhibited unimodal distributions in age frequency with the minnow trap sample dominated by fish at age 4 or 16.7% of total catch and gillnet m odal age being 5 or 4.1% of the total catch (Figure 4-11). Mortality Estimates Full recruitment of T. zillii to the fishery started at age 3 for both the trap and gillnet samples. Peak recruitment occurred at age 4 for the trap sample and at age 5 for the net sample (Figure 4-12). Instantenous tota l annual mortality (Z) was estimated as 0.74 (age range: 4-8 yrs) for the gillnet samples and 0.71 for the trap sa mples (Figure 4-12) corresponding to apparent survival rates of 0.36 and 0.39, respectively

PAGE 95

95 Natural Mortality (M) Estimates of natural mortality (M) for T. zillii in Lake Nkuruba varied depending on the method used. Rikhter and Efanovs (1976) equa tion yielded the lowest natural mortality estimate of 0.52 based on age of massive maturation of 3 years for T. zillii Hoenigs (1983) longevity-mortality relationship prod uced a higher estimate of natural mortality rate 0f 0.54. The values considered more representative were those estimated from Hoenigs, since they were estimated from the actual sample. Discussion Length-Weight Relationships The value of the regression coefficient (b) for the combined sexes was 2.97 for fish between 51 and 284 mm TL, and 4.1 and 425 g total weight. This is less than the assumed theoretical value of the coefficient (i.e., b = 3, Le Cren 1951); a nd, indeed, overall the Tilapia zillii in Lake Nkuruba are in a slightly poorer co ndition compared to populations in other African lakes. For example, (Negassa and Getahun, 2003) reported a regression coefficient value of 2.98 for T. zillii between 55-320 mm TL in Lake Zwai, Ethiopi a. In West African lakes, a value of the coefficient b of T. zillii recorded was 3.2 for fish betw een 70-150 mm TL (Teugels and Thys van den Audenaerde, 1991). The maximum size of T. zillii caught in Lake Nkuruba was 284 mm (TL), a male weighing 425 g, suggesting a smalle r maximum size in Nkuruba compared to other African lakes such as Lake Zwai in Ethiopia wh ere the species was reported to attain a maximum length of 320 mm TL (Negassa and Getahun, 2003). Length Frequency The use of minnow traps consistently resulted in collection of smaller fish compared to gillnets. This can be attributed to gear selectiv ity, habitat differences, and/or fish behavior. The narrow openings of minnow traps selects for onl y smaller fish, while the gillnets (mesh size

PAGE 96

96 range 25.4 mm-101.6 mm) collected larg er size classes. Smaller T. zillii mostly inhabit the shallower littoral zones within Lake Nkuruba; and this is where the traps were primarily set. Larger fish exhibit greater movement offshor e making them more vulnerable to capture by gillnets. Otolith Preparation and Interpretation of Growth Zones Despite the labor-intensive and time-consuming process of otolith preparation and analysis, sanding and polishing of tr ansverse thin-sectioned sagittal otoliths seems to be the best way to age both juvenile and adult T. zillii in the crater lakes. Op aque and translucent zones were not readily identifiable in w hole otoliths, and transverse thin-s ections of sagittal otoliths of T. zillii showed weak and not easily visible opaque a nd translucent zones. Sanding and polishing of the transverse sections clearl y improved visibility and readability of otolith sections. Results of marginal-increment and edge analyses i ndicated that depositi on of opaque zones in T. zillii otoliths followed a bi-annual pattern with one opaque zone deposited between March-May and the second between September-November. Based on these results, I consid ered the annuli seen in the otoliths of T. zillii validated as biannual deposits with formation of translucent zones coinciding with the peaks of two dry seasons (A ugust and January). This is consistent with studies of other tilapia species in the tropical regions that have reported opaque zone formation associated with wet-and-dry seasonal cycles (Warburton, 1978; Bwanika et al., In press), and more generally, studies that support seasonal depos ition of one or more growth zones in tropical fish calcified structures (Bro thers, 1979; Yosef and Casselma n, 1995). Opaque zone deposition is thought to occur during peri ods of increased growth, wher eas the correspondi ng translucent zone is formed during periods of low metabo lic activity (Beckman and Wilson, 1995). Studies of other tilapia species (e.g., Oreochromis niloticus ) in African lakes also associate opaque zone formation with faster growth (Yosef 1990; Yose f and Casselman, 1995). In reviews of otolith

PAGE 97

97 studies in tropical latitudes, Beckman and Wilson (1995) and Fowler (1995) found that, for the majority of tropical species, the period of opa que zone formation coincided with spring and summer months. Of the 29 tropical species re viewed by Fowler (1995), only four showed opaque zone formation in the winter period. There are several factors suggest ed to influence growth zone formation in tropical fishes including seasonal variation in temperature, rainfall, photoperiod, f ood quality, reproductive activity, and changes in other environmental pa rameters (Blake and Blake, 1978; Warburton, 1978; Nekrasov, 1980; Morales-Ni n and Ralston, 1990; Booth, Merron and Buxton, 1995; Yosef and Casselman, 1995; Bwanika et al., In press). In Lake Nkuruba, the mean annual surface water temperature over the study period was 23.9oC and the range was relatively small (21.7oC-25.1oC). Day length in the crater lakes region is also relativ ely constant, since the lakes lie astride the equator; and phot operiod change is unlikely to trigger growth zone formation. Despite the fact that Tilapia zillii allocate high energy to re production, with both parents participating in nest building, ventilation, and defense of the y oung (El-Zarka, 1956), the effect of spawning activity on the forma tion of growth zones was not ev ident in this study. I observed annulus in otoliths of immature as well as mature T. zillii In addition, Tilapia zillii are also known to breed all year round in the equatorial regions with slightly higher breeding intensity during wet months March-May and between July-August (Welcomme, 1967; Siddique, 1977). In the present study, I also observed both male and female T. zillii at various stages of gonad development during the different se asons in the 1-year cycle. It is unlikely that opaque zone formation is a consequence of temperature vari ation, spawning activity, or photoperiod. It is possible that opaque zone formation in otoliths of T. zillii from Lake Nkuruba is driven by rainfall and associated availability of food. If opaque zone formation is associated with

PAGE 98

98 increased food availability, then the bimodal rain fall experienced in the crater lakes region of Uganda combined with nutrient influx, may contribute to seasonal fluctuation in lake productivity and associated feeding conditions. Several marginal-increment analysis studies ha ve suggested that error in measurement of otolith growth zones may contribut e to apparent variation in the periodicity and timing of opaque zone formation in otoliths (e.g., Smith and Deguara 2003). In addition, potential errors in the observation of completed opaque increments may al so occur due to optical artifacts of sectioned otoliths (Francis et al., 1992). I believe the m easures I took to validate the timing and periodicity of opaque zone formation, the careful attention given to interpretation of edge type, and the rigorous protocol I used in this study minimized possible marginal-i ncrement analysis errors. If such errors did occur, then th ey were consistent across size-a nd age-classes; and these results should provide robust estimates of T. zillii age in Lake Nkuruba. Growth Tilapia zillii collected with the gillnets were larger at age and had a faster growth rate compared to those collected with minnow traps; smaller individuals composed the major (82.2%) portion of the catch. These differences in grow th rates between gears may be influenced by shifts in diet and dietary differe nces between the gears and/or dens ity-dependent factors. Growth of tilapia species is known to be affected by the quality and quantity of food (Lowe-McConnell, 1982). A qualitative assessment of the diet of T. zillii in Lake Nkuruba (N = 25 fish) showed that detritus (Macrophyte sp. ) constituted the greatest percentage (> 50%) of the food items in the stomachs of fish of all sizes caught in both gill nets and minnow traps. Other food items included phytoplankton ( Microcystis sp., Navicula sp., Scenedesmus sp., Oedogonium, Oscillatoria sp. and Flagillaria sp .), fish eggs, fish s cales, and gastropods ( Sphaerium sp .). Zoobenthos ( Chironomidae larvae ) and newly hatched fish larvae were only found in the stomachs of fish

PAGE 99

99 collected with gillnets. This difference in diet may contribute to a better growth in gillnet-fish because of their more varied diet composed of food of both plant and animal origin. Density-dependence due to intraspecific comp etition for food resources may also have contributed to the observed diffe rence in growth rate between gears especially in the minnow trap samples. Shallow littoral areas containing macrophytes that form the major diet of the minnow trap fish are limited in Lake Nkuruba; and this may c ontribute to density-dependent growth in this region of the lake. The role of density-dependent factors in growth of fishes is well documented (e.g., Lowe-McConnell, 1982), and pr olonged restriction in food availability has been reported to lead to T. zillii sacrificing somatic growth to maintain reproductive investments (Coward and Bromage, 1999). In this study, the greatest differ ences in growth rates were observed between 2year and 3-year old T. zillii It is possible that la rger fish are released from density-dependent competition for food re sources when they move more offshore. The oldest T. zillii recorded in this study was 8 yrs. This is within the range of maximum ages reported for the species in previous studies. For example, maximum life span for T. zillii has been reported to be between 8 years (De Si lva, 1991) and 11 years old (Fryer and Iles, 1972; James 1989). Lvque (1997) also repor ted a maximum age of 7 years for T. zillii in Lake Kinnert in Israel. Longevity of tilapia has been reported to be i nversely related to environmental quality (James 1989). Hodgkiss and Man (1977) reported that most til apia species in a thermally harsh reservoir were less than 5 years old. Similarly, Hecht and Zway (1984) reported stunted tilapia mostly 5 years old or less in a hot spring. The low level of explo itation in Lake Nkuruba may contribute to their intermediate longevity compared to those reported for other lakes. Mortality Rates Estimates of mortality rates for T. zillii in natural systems are to my knowledge not available. The estimated total instantaneous mortality (Z) and natural mortality (M) rates for

PAGE 100

100 Lake Nkuruba are at the low end of the range of values reported for other tilapia species. For example, Getabu (1992) estimat ed total mortality rate (Z ) of 0.818 yr-1 for Oreochromis niloticus in the Kenyan part of Lake Victor ia using length-based methods. In a study of the same species in Lake Victoria, Okemwa et al., (1994) repor ted total and natural mo rtality rates of 1.71 yr-1 and 0.72 yr-1, respectively. Other estimates of to tal and natural mortality rates for O. niloticus in Lake Victoria were at 1.67 yr-1 and 0.63 yr-1 by LVFRD (UNECIA, 2002); and 1.84 yr-1 and 0.72 yr-1, respectively (Muhoozi, 2002). In Kandulla Reservoir, Amarasinghe and de Silva (1992) estimated Z = 1.399 yr-1 and 1.707 yr-1 for O. mossambicus and O. niloticus respectively, using the non-seasonalized catch curve. R ecently, King and Etim (2004) computed total mortality and natural mo rtality rates of 1.75 yr 1 and 0.99 yr 1, respectively, for Tilapia mariae in a Nigerian wetland stream using the seas onalized length-converted catch curve method. Despite biases associated with the different me thods used for estimation of mortality rates, Z from seasonalized length-converted catch curve is often comparable with Z from age-structured catch curve (Pauly 1990). Our estimates of Z obtained for Nkuruba population using the linearized catch curve are extremely low due to th e low fishing mortality (f ishing is currently not a significant mortality factor for the T. zillii population in Lake Nkuruba). Natural mortality accounts for most of the reduction in population dens ity in Lake Nkuruba. A number of factors may have contributed to the natu ral mortality in Lake Nkuruba. Fluctuation in environmental factors, in particular, low levels of surface dissolved oxygen concentration caused by periodic overturns, has caused fish kills in Lake Nkuruba in recent years. During these overturns, the surface water becomes anoxic and schools of T. zillii are commonly observed gulping for oxygen at the lake surface. Intestinal infestation by nematode pa rasites represents another potential source of natural mortality in T. zillii (pers. observ.). Predation by fish-eating waterfowl may

PAGE 101

101 also contribute to natural mortality in lake Nkur uba, but this is likely not significant, since their number is low in the lake (pers. observ.). Estimates of instantaneous natural mortality rate (M), apart from showing the proportion of death caused by all factor s except fishing, is a required input in the computation of many models in fish population dynamics, e.g. the relative yiel d per recruit and relative biomass per recruit (Newman, 2002). Overestimates of M can lead to high estimates of the potential yield of fish stocks, and this may lead to overexploitation a nd ultimately recruitment overfishing (Newman et al., 2000a). Based on my estimate of Z, I feel th at the Hoenig (1983) method (M = 0.54) yielded the most realistic estimate of M for T. zillii in lake Nkuruba because it was closer to (Z = 0.74 for the gillnet; Z = 0.71 for the minnow trap samples) which should be the case in a lightly exploited population. This is the first otolith-based study of T. zillii in this region. The present study adds to the growing list of tropical species fo r which age estimates have been successifully validated. Using marginal-increment and edge analyses, I validate d that opaque zone formation in otoliths of T. zillii from Lake Nkuruba is biannu al and associated with the wet-dry seasonal cycles. These results also represent the first estim ates of growth and mortality for T. zillii in the crater lakes and provide an essential first step in appropriate management of th e small scale fishery for local communities within the vicinity of these lakes. Fish populati on dynamics models require the best possible data for accurate prediction of key characters such as growth and mortality. If a significant size bias exists due to gear selectivity, then the predicted results may be incorrect. In this study, minnow traps appeared to be biased toward smaller, slower growing fish inhabiting inshore habitats while gillnets collected larger, faster growing fish in near shore and offshore areas. For example, if estimates of populati on characters were ba sed only on minnow trap

PAGE 102

102 samples, then it would lead to underestimati on of growth and apparent survival of the T. zillii population in Lake Nkuruba. These differences in gear selectivity dictate that estimates of T. zillii key population characters be gear specific, as combining data from both gears may result in incorrect estimates of growth, mortality, and apparent survival. Gillnets may be part icularly effective for describing larger and faster growing fish ; while data from minnow traps may be suitable for modeling growth and mortality in smaller, slower growing fi sh. I suggest that the use of multiple gears be applied and periodically evalua ted in future studies of a T. zillii populations in the crater lakes to ensure the data obtained accurately reflect T. zillii population parameters.

PAGE 103

103 Table 4-1. Physical and limnological character s of surface water of Lake Nkuruba, Uganda. Parameter Kizito et al., (1993) Ch apman et al., (1998) Present study Surface area (ha) 3.0 3.0 Maximum depth (m) 37.0 38.0 37.0 Mean depth (m) 16.0 16.0 Chlorophyll a (g L-1) 8.8 4-12 22.2 Dissolved oxygen (mg L-1) 7.4 4-8 5.1 Water transparency (m) 1.4 1.5-5 3.2 pH 8.2 8.2 8.6 Conductivity (S cm-1) 341 325.0 438.0 Temperature (oC) 22.9 22-24 24.3 Total phosphorus(g L-1) 38.4

PAGE 104

104 Table 4-2. Mean total length at age (mm), standard error (S.E.) and 95% confidence intervals (CI) of Tilapia zillii captured with gillnets and minnow traps in Lake Nkuruba, western Uganda from July 2005-June 2006. S.E. 95% CI Age Gear Mean TL (mm) Lower Upper 3.00 Gillnet 137.8 6.9 124.2 151.3 Trap 87.1 2.4 82.4 91.8 3.50 Gillnet 132.9 4.9 123.3 142.4 Trap 90.9 1.9 87.1 94.7 4.00 Gillnet 142.4 4.9 132.8 151.9 Trap 92.7 1.7 89.3 96.0 4.50 Gillnet 146.3 4.9 136.7 155.8 Trap 90.4 1.9 86.6 94.2 5.00 Gillnet 145.3 3.7 138.1 152.5 Trap 91.7 2.4 87.1 96.3 5.50 Gillnet 145.4 6.1 133.3 157.5 Trap 91.8 2.7 86.5 97.1 6.00 Gillnet 142.7 5.6 131.6 153.7 Trap 93.8 3.3 87.3 100.4 6.50 Gillnet 149.3 5.6 138.3 160.4 Trap 91.4 3.3 84.9 98.0 7.00 Gillnet 141.5 9.7 122.4 160.6 Trap 83.2 6.1 71.1 95.3

PAGE 105

105 Table 4-3. Results of 2-way ANOVA testing for th e effect of gear type and age on mean total length of Tilapia zillii from Lake Nkuruba western Uganda. Source SS df MS F p Corrected model 139212.2 17 8188.9 43.4 < 0.0001 Intercept 1997740.2 1 1997740.2 10578.6 < 0.0001 Age 1567.3 8 195.9 1.0 0.407 Gear 100608.7 1 100608.7 532.8 < 0.0001 Age*gear 1121.7 8 140.2 0.7 0.654 Error 64774.4 343 188.8 Total 3798603.0 361 Corrected total 203986.6 360

PAGE 106

106 Figure 4-1. Map of Lake Nkuruba (botto m right) showing its location in Uganda.

PAGE 107

107 Figure 4-2. Photograph of a po lished transverse section of Tilapia zillii otolith from Crater Lake Nkuruba, Western Uganda. Opaque zones (OZ) are narrow dark-brown areas and were counted for age determination. Photo taken with Motic Image System Version 1.3, Mag. X20, under transmitted light.

PAGE 108

108 Figure 4-3. Transverse section of a sagittal otolith from a 6-year old Tilapia zillii viewed under transmitted light showing the appearance of opaque zones (OZ), translucent zones (TZ), core region of the ot olith, the reading (measuring) axis, marginal increment (MI), and previous increment (PI) used for marginal increment analysis (MIA). The number of opaque zones is indicated by the numerals.

PAGE 109

109 4080120160200240280320 0 100 200 300 400 500 Net TrapsTotal weight (g)Total length (mm) Figure 4-4. Relationship between total weight (g) and total lengt h (mm) for male (open circle) and female (closed circles) Tilapia zillii from Lake Nkuruba, western Uganda.

PAGE 110

110 Figure 4-5. Sagittal otoliths of Tilapia zillii from Crater Lake Nkuruba western Uganda. A is distal view and B is dorsal (medial) view (Photo: Motic Image System Version 1.3, mag.X20). A B

PAGE 111

111 0 10 20 30 40 50 NET TRAPFEMALES 050100150200250300 0 10 20 30 40 Total length (mm) ALL FISH 0 10 20 30 40 Percent frequencyMALES Figure 4-6. Length-frequency di stribution by 10-mm size class fo r male, female and combined sexes (male + female) Tilapia zillii collected from Lake Nkuruba in western Uganda, from July 2005-June 2006.

PAGE 112

112 050100150200250300 0 5 10 15 20 25 Percent frequencyTotal len g th ( mm ) Gillnet Trap(N = 80) (N = 370) Figure 4-7. Length frequencie s (% of total catch) of Tilapia zillii collected in minnow traps and gillnets from Lake Nkuruba, western Uganda, from July 2005-June 2006.

PAGE 113

113 J u l y A u g S e p t O c t N o v D e c J a n F e b M a r A p r M a y J u n e0 20 40 60 Index of completion (%)Months 2005 200647 3068383212 47 26 37 32 22 11 Figure 4-8. Mean marginal increment of thin-sectioned Tilapia zillii otoliths from Lake Nkuruba, western Uganda over a 12-month pe riod. Vertical bars denote standard errors and the numbers above the bars indicate sample size (n).

PAGE 114

114 0 20 40 60 80 100 JulyAug.Sept.Oct.Nov.Dec.Jan.FebMar.Apr.MayJune 0 20 40 60 80 100 2005Opaque zone (%)Months 2006 Mean monthly rainfall (mm) Figure 4-9. Percentage of otoliths of Tilapia zillii with opaque edges (lin e) from Lake Nkuruba and the mean monthly rainfall (area) meas ured at Makerere University Biological Field Station, Kibale National Park during the study period.

PAGE 115

115 0123456789 0 50 100 150 200 Observed age (yrs) Gillnet TrapObserved total length (mm) Figure 4-10. Observed length-at-age for Tilapia zillii fitted to the von Bertalanffy growth model forced through zero for fish sampled with traps and gillnets from Lake Nkuruba, western Uganda, from July 2005-June 2006.

PAGE 116

116 012345678 0 5 10 15 20 Percent frequencyObserved a g e (y rs ) Net Trap Figure 4-11. Age frequency of Tilapia zillii collected with traps and gillnets expressed as a percentage of the overall catch fr om Lake Nkuruba, western Uganda.

PAGE 117

117 0 1 2 3 0123456789 0 1 2 3 4 5 ln (Number of fish)GILLNET (Z = 0.74; r2 = 0.72) 0 3 6 9 12 15 Number of fishGILLNET 0123456789 0 10 20 30 40 50 60 70 TRAP Observed age (yrs) TRAP (Z = 0.71; r2 = 0.91) Figure 4-12. Age frequency and linearized catch curves for estimating total mortality (Z) for T. zillii caught with traps and gillnets from La ke Nkuruba, Uganda, between July 2005 and June 2006. Age frequency was derived from the standard von Bertalanffy growth analysis. Only the black solid points we re included in the regression analyses.

PAGE 118

118 0 20 40 60 80 Net Trap August '05 September '05 October '05 0 20 40 60 80 Percent frequencyNovember '05 December '05 January '06 0 20 40 60 80 February '06 March '06 050100150200250300 0 20 40 60 80 Total length (mm) May '06 050100150200250300 June '06 050100150200250300 Total length (mm) April '06 Figure 4-13. Size frequency of Tilapia zillii from Lake Nkuruba, western Uganda by month from July 2005 and June 2006.

PAGE 119

119 CHAPTER 5 LIFE HISTORY VARIATION IN Tilapia zillii (PISCES: CICHLIDAE) IN THE CRATER LAKES OF WESTERN UGANDA. Introduction Variation in Life History Characters in Fish Life history theory seeks to describe and explain the evolu tion of adaptive responses in fitness-related traits such as re production, apparent survival, age a nd size at maturity, and growth as a result of environmental va riation (Roff, 1992; Stearns, 1992). For a given environment, evolution helps to optimize life hi story strategies by ba lancing tradeoffs of key traits (Stearns, 1976; Roff, 1984; Wootton, 1992). Tradeoffs in tr aits such as growth and reproduction are especially critical in an imals with indeterminate, growth such as fishes, because batch fecundity of female fish is strongly correlated with size and allocation of resour ces to reproduction may lead to a decrease in growth rate (Stearns, 1992). Interdemic varia tion in life history characters induced by such tradeoffs can have extrem e consequences for fish population dynamics. Variation in fish life history traits is well documented in the field in both temperate and tropical species and includes, as examples, variat ion in batch fecundity (P eters, 1963; De Silva, 1986; Duarte and Alcaraz, 1989; Legendre and Ec otin, 1989; 1996; Dupochelle et al., 2000), egg size and energy (Blaxter, 1986; Chambers and Leggett, 1996; Chambers, 1997), size and age at maturity (Reznik et al., 1990; Dupochelle and Panfili, 1998), and somatic growth (Leggett and Power, 1969; Pauly, 1980; Roff, 1984; Hutchings, 1993). Variation in many of these traits is interrelated. For example, fish populations in stable environm ents with high intraspecific competition tend to have lower batch fecundity, la rger eggs, and higher size at maturity (Lowe McConnell, 1982; Reznik and Briga, 1987; Reznik et al., 1990) compared to populations in variable environments, although th ere are exceptions. Similarly, di fferences in growth rates can influence size and age at matur ity, and variation in size and age at maturity of individuals may

PAGE 120

120 also affect growth. An understanding of evolutio n of fish life history is critical to fisheries management, as reduction in fish growth and maximum size may lead to stunted populations with low economic or recreational value (Miller and Kapuscinski, 1994). Stunted growth or production of a large number of individuals having a low maximum size, is a very widespread phenomenon in freshw ater fish populations and has been observed across a broad range of phylogenetically distin ct lineages including as examples: salmonids (Leggett and Power, 1969), coregonids (Ridge way and Chapleau, 1994), cyprinids (Burrough and Kennedy, 1979), Arctic charr (P arker and Johnson, 1991), and perch Perca fluviatilis (Rask, 1983; Ridway and Chapleau, 1994). Clear ly, stunting in fish populations has very important economic implications because the co mmercial value of stunted fish populations is greatly diminished, and there is great interest in understanding and mitigating stunting in fish populations as harvesting may infl uence stunting. Stunting in fi shes has been attributed to several factors including: intraspecific spatia l competition due to overcrowding (Sandheinrich and Hubert, 1984; Roff, 1992), reduced availabili ty of food (Rask, 1983; Roff, 1992; Post and McQueen, 1994), and increased juven ile and adult fish apparent su rvival (Roff, 1992) as a result of absence of predators and low fishing pressu re, respectively. Alt hough many of these factors reflect resource limitation, direct links between ecologica l factors and fish lif e histories are not well understood. An evaluation of consequences of ecological factors and size-selective fishing on evolution of life history of fishes is critical. Size-Selective Fishing Fishing is typically size-selec tive and non-random in that fishi ng gear is designed to target larger, presumably faster growi ng and older individuals in a popul ation (Law, 2000). It is often assumed that fishing-induced reductions in populati on density of a target stock leads to increased yield because of reduction in intraspecific comp etition that releases populations from density-

PAGE 121

121 dependence resulting in faster growth (Jennings and Kaiser, 1998; Hall, 1999; Rochet et al., 2000). As a result, fishing is increasingly rec ognized as a major sele ctive force driving the evolution of life history traits (Stokes et al., 1993; Ashley et al., 2003). Descriptive empirical studies have shown that strong si ze-selective fishing mortality lead s to a downward shift in size at age and age at maturity in wild commercial species such as the Northern cod, Gadus morhua (Olsen et al., 2004, 2005) and North Sea Plaice, Pleuronectes platessa (Grift et al., 2002). In addition, experimental laboratory studies have also demonstrated that selective fishing of typically adult size-cl asses results in incr eased reproductive investment in guppies, Poecilia reticulata (Reznick, 1990), and evolution towards slower somatic growth in Atlantic Silversides, Menidia menidia (Conover and Munch, 2002). Furthermore, theoretical models (e.g., Ernande et al., 2004) show that selective fish ing of mature fish induces late maturation at older ages and larger sizes whereas fishing immature fish lead s to early maturity at younger ages and smaller sizes. The weakness with such theoretical models is that they examine onl y one evolving trait, usually maturation schedules. Fishing may induce evolution in multiple traits that are evolving simultaneously. As life history e volution in fishes is known to occur rapidly in natural fish populations (Reznick et al., 1990; Rijnsdorp, 1993; Haugen and Vollestad, 2001; Reznick and Ghalambor, 2005), studies examining the effects of size-selective fishing and environmental factors are important as fishi ng may not only influence demogra phics of a fish population, but also lead to genetic changes in their life history (St okes et al., 1993; Ashley et al., 2003). Yet, little attention has been paid to the effect of fishing and other environmental factors on patterns of life history evolution in fishes.

PAGE 122

122 Fishing and Phenotypic Plasticity Fishing-induced changes in life history traits may not only repr esent a genetic variation but also phenotypic plasticity, or an interaction between the two (Trippel, 1995; Rochet, 1998; Law, 2000). For example, low fishing levels may result in high population density that increases competition for resources (e.g., food) leading to density-dependent growth. Under reduced and unpredictable food availability, one would predict increased a llocation of resources towards reproduction, resulting in early maturation at a smaller size. A decrease in population density through fishing may reduce competition for resources and lead to faster growth. When resources are adequate, one would predict increased allocati on to growth, resulting in delayed maturity at larger size. Implicit in these predictions is the assumption that changes in fitness are selectively neutral i.e. that there is no e volution in response to size-selec tive fishing (Law, 2000; Walsh et al., 2006). Even if resources are adequate, fishing alone may also cause evolution in life history traits if there is by selecting ag ainst faster growing fish. If a significant component of variation in life history characters refl ects environmentally-induced plas ticity, then altering predator pressure (through selective fishing) may be an effective way of changing life history characters to increase yield on annual or short-time scale since phenotypically plasti c responses can usually be mitigated within one single gene ration as demonstrated in guppies Poecilia reticulata (Reznik and Yang, 1993). Interestingly, many fisheries show changes in lif e history traits that are not consistent with releasing the target stock from intraspecific competition (Law 2000), and suggest that genetic change in response to selectiv e effects of fishing may be de trimental to recovery of the population when fishing pressure is reduced. By targeting the largest, oldest, and fastestgrowing individuals, fisheries crea te a strong directional selection favoring the apparent survival of smaller, younger individuals (Walsh et al., 200 6). Consequently, size and age show truncated

PAGE 123

123 distributions skewed towards smaller and younger individuals (Conover and Munch, 2002). If there is a genetic basis for the phenotypic vari ation (e.g., size) between individuals, then sizeselective fishing may lead to evolution towa rds smaller size-at-age (Law, 2000; Walsh, 2006). For example, Walsh et al., 2006 subj ected the Atlantic silverside to experimental size-selective fishing over five generations and found that populati ons subjected to fishi ng of large individuals exhibited declines in batch fecundity, egg size, larval size at hatching, la rval survivorship and growth rate, and a decline in food consumption and conversion efficiency. In a long-term study of graylings, Thymallus thymallus populations, Haugen and Vllestad (2001) also found that size-selective gillnet fishing resulted in a reductio n in age and length at maturity in populations with recent common ancestors. The two studies do not elucidate if ea rly maturity in the respective populations is a result of growth potential or conditi on of individual fish within the populations. The Environment The effects of fishing on the target stocks ma y also be impacted by concurrent changes in the biotic and abiotic environment. Lakes that are heavily fished are of ten accessible and subject to other anthropogenic pressures such as deforestation and eutr ophication. A critical step in beginning to understand variation in life history traits is to explore the inte raction between sizeselective fishing and other biotic and abiotic f actors that may affect resource availability or otherwise alter the sel ective regime. The goal of this Ch apter was to quantify interdemic variation in life history charac ters (growth, age and size at matu rity, and batch fecundity) and to examine the degree to which this variation can be explained by effects of size-selective fishing and other environmental factors such as deforest ation. To do this, I use the introduced tilapia, Tilapia zillii in a suite of crater lakes in western Uganda that provide a range of fishing pressures, environmental degradation, and limnological features.

PAGE 124

124 The Tilapias-A useful Model for U nderstanding Life History Variation Tilapiine fishes (Cichlidae, commonly referred to as tilapias) provide a suitable model group of species for testing the effects of size-se lective fishing and envi ronmental factors on lifehistory traits of freshwater fishes because they have the ability to vary the allocation of resources to reproduction and growth depending on environm ental conditions (Fryer and Iles, 1972; LoweMcConnell, 1982). In addition, tilapias display high levels of phenotypic plasticity and can tolerate a wide range of envi ronmental conditions (Lowe-McConne ll, 1982). Furthermore, life history variation in tilapiine fi shes is well-documented in tropi cal freshwaters (Lowe McConnell, 1958, 1982; Fryer and Iles, 1961, 1972; Welcomme, 1970; Gwahaba, 1973; Noakes and Balon, 1982; Stewart, 1988; Ribbink, 1990; Koldi ng, 1993; Dupochelle and Panfili, 1998; Dupochelle et al., 1999, 2000; Bwanik a et al., In press). Fina lly, in western Uganda, both unexploited (no fishing) and exploi ted (high fishing) tilapia popul ations occur in a large number of small crater lakes, providing suitable systems and the replication needed to explore life history variation across broad environmental gradients. The Crater Lakes-Suitable Experimental Test-Tubes for Life History Studies There are about 89 small volcanic crater la kes along the foothills of the Rwenzori Mountains in western Uganda. Three tilapia species: Oreochromis niloticus (Linnaeus 1758), Oreochromis leucostictus (Trewavas, 1933), and Tilapia zillii (Gervais 1874) were introduced into a large number of these lakes to increase available protein to the local communities. Currently, many of these lakes seem to be pr oducing stunted tilapia populations, causes of which remain unknown but may reflect resource lim itations associated with the Crater Lake environment and/or low mortality leading to high levels of intr aspecific competition. There is great variation in fishing pressure among the lake s from almost no fishing, to individual based hook-and-line fishing, to small-scale commercial gillnet fisheries operated by communities

PAGE 125

125 around these lakes. The intensity of fishing in these lakes also va ries from low where there is either no fishing or a small number (less than 20) of hook-and-line fishermen to medium where there is a large number of hook-and-line fishermen (bet ween 30 and 45) and occasionally gillnets, to high where there are established gilln et fisheries. Large predatory fish species are largely absent in most of the lakes except Lake Kasenda where the catfish ( Clarias gariepinus ) was recently introduced. In addition, many of the cr ater lakes lie outside protected areas and are increasingly threatened with defo restation of the crater rims du e to the rapidly expanding human populations. The extent of defore station of crater rims ranges from minimal (50-100% of the crater rim still forested) to moderate (25-49% of the crater rim forested) to severe (only small forest patches or scattered trees left), a nd complete (no forest tr ees along the crater rim) (Pomeroy and Seavy, 2003). The variation am ong lakes in fishing effort and other environmental characters provides an excellent oppor tunity to identify predictors of life history variation. A second important feature of thes e systems is their amenability to large-scale experimental manipulations in the future that can shed light on th e source of life history variation in the tilapias. To explore the effect of sizeselective fishing and deforest ation on the life history of Tilapia zillii, I: (1) estimated key life hi story traits (growth, age and size at maturity, and batch fecundity) of Tilapia zillii in a series of crater lakes that di ffer in fishing pressure (high vs. low fishing effort) and deforestation (severe vs. m oderate) and vary (conti nuously) in other key environmental factors; (2) compared the estima ted life history characte rs among lakes; and (3) explored predictive relationships between life history characters and environmental factors (e.g., dissolved oxygen, chlorophyll a, water transparency, etc.).

PAGE 126

126 Materials and Methods Study Lakes The study was conducted in the Kasenda cluste r of crater lakes of western Uganda (0o23-0o33N, 30o10-30o 20E) at an altitude of 925 to 1520 m (Melack, 1976; see figure 2-1, chapter 2). With the exception of a few lakes, th e Kasenda crater lakes are typically small, with surface area ranging from 2 ha (Nyanswiga) to 50 ha (Ntanda), maximum depth (zm) ranging from 5 m (Kifuruka) to 259 m (L ake Ntanda), and mean depth ( z) ranging from 2.9 m (Kifuruka) to 59.7 m (Rukwanzi). Owing to their vo lcanic origin, most of the lakes have steep sides, with little littora l vegetation except a few shallow lakes or shallower parts of the deep lakes that have a diversity of aquatic macrophytes such as Nymphaea spp., Ceratophyllum spp ., and Potamogeton spp. (Kizito et al., 1993). Rainfall is bimodal, with two wet seasons from March-May and September-November, with an aver age of 1.7 m of total rainfall received annually (1990-2005). Between 1990 and 2004, th e mean daily minimum temperature was 14.9oC, and the mean daily maximum temperature was 20.2oC. Environmental Data To predict the effect of environmental factor s on life history charac ters such as growth, size at maturity (male and females), apparent survival, and batch fecundity (females), I determined a suite of physical and chemical char acteristics in each lake including: lake area, maximum depth, mean depth, vertical profiles of dissolved oxygen (DO) concentration and water temperature, pH, conductivity (corrected to 25oC), water transparenc y, chlorophyll a (Chl-a) concentration, total phosphorus (TP) and total nitrogen (TN). Wa ter temperature and dissolved oxygen concentration were determined in situ with a YSI oxygen/temperature meter. Conductivity and pH were measured using a YS I conductivity meter and an OAKTON pH Testr 1, respectively. Water transparency was estimated with a 20-cm Secchi disk. Chlorophyll a was

PAGE 127

127 determined using a spectrophotometric technique. Total phosphorus and total nitrogen were determined using standard limnological techni ques following APHA (1988). All the above variables were determined once in each lake a nd represent point in time estimates. Details of limnological procedures are in Chapter 2. Fish Collections I collected a total of 2,697 T. zillii between July 2004 and June 2005 from 14 crater lakes in western Uganda (Figure 2-1, Chapter 2). I sampled each lake once over a 5-6 day period making a significant effort to sample fish in di fferent size classes by using a variety of fishing gears, including: 1) tw o experimental monofilament gillnets each with 4 panels 60-m long and 1.0-m deep with stretched mesh sizes of 25.4 mm, 50.8 mm, 76.2 mm, and 101.6 mm; 2) artisanal fishermen nets comprised of mesh sizes 25.4 mm, 50.8 mm 63.5 mm, and 76.2 mm with a length of 1,800 m; and 3) metal minnow traps (450-mm long and 7-mm square wire mesh, with opening size 25.4-38.1 mm). The nets we re set between 1400-1600 hr and pulled the following morning between 0800 hr -1000 hr. To obtain samples representative of fish populations in each lake, I set the nets in all major habitats, incl uding open water, at the edge of shoreline vegetation, and on rocky shores. A tota l of 20 minnow traps, baited with bread, were also set at the shallow littoral areas to obtain additional samples of smaller fish. I measured individual fish to the nearest 1 mm for both tota l (TL) and standard (SL) length, total weight, and fish weight with the gonads removed to the neares t 0.1 g using an Ohaus handheld electric scale. I weighed fish >320 g using a spri ng balance ( 1 g). Mature fi sh were then dissected, sexed, and their gonads removed to the nearest 0.1 g, weighed, and examined macroscopically and allocated a stage of gonadal development according to a macroscopic staging key modified from Siddiqui (1979) for T. zillii Finally, sagittal otoliths were removed from 150-200 fish in each lake, air-dried, and stored in labeled plastic vials until processed for age determination.

PAGE 128

128 Size Frequency To quantify variation in size distribution of T. zillii among lakes, I first plotted size frequency of T. zillii (for males and females separately) in 12 of the lakes where the species existed (Figure 5-1). To detect potential effects of gear on mean to tal length, I then plotted size frequencies for T. zillii caught with traps and gilln ets separately for the 8 lakes used in the aging analysis (Figure 5-2). I also evaluated the effect of gear on gender by plotting length frequencies for female and male fish separately for each gear. Batch Fecundity Estimates of batch fecundity (defined as the num ber of oocytes to be released in the next spawning event) were determined in 10 lakes (Table 5-1) using oocytes in ripe and running stage (stages VI) only. Ripe female fish were care fully dissected and the ovaries taken out and preserved in 40% formaldehyde. The ovary sample s were later transferred to 75% ethanol for storage. In the laboratory, sections of ovaries of known mass were teased apart and weighed. The number of eggs in each section were then counted and multiplied by the total weight of the ovaries to calculate absolute batch fecundity. The relationship between ba tch fecundity and fish length and between batch fecundity and fish mass was determined for T. zillii using batch fecundity data for all lakes combined. Size at 50 % Maturity (L50) I classified staged gonads (Table 5-2) as eith er immature (Stages I, II, and III) or mature (stages IV, V, and VI). I then estimated size at sexual maturity by fitting a binary logistic regression equation to the proportion of mature i ndividuals in each size category (mm, bins) with non-linear regression (SPSS software), using the formula: L50 = {[e(-a+bTL)]/[1+e(-a + bTL)]};

PAGE 129

129 Where L50 is the proportion of mature individuals, TL is the total length, a and b are constants of the logistic regression model. Length at maturity for T. zillii was determined in 12 crater lakes for which there wa s sufficient data (Table 5-1) Age Determination In Chapter 4, I used marginal-increment and edge analyses to validate opaque zone formation in otoliths of T. zillii from Lake Nkuruba. Tilapia zillii were observed to deposit two growth rings per year, between September-Nove mber and between March-May, periods that correspond with the bi-modal rainfa ll patterns that are ch aracteristic of this equatorial region (Chapter 4). These results provi ded a validated aging method that was then used to determine age, growth, and mortality in the crater lakes desc ribed in this chapter, since they lie in close proximity of Lake Nkuruba and are in the same climatic zone. In this Chapter, age was determined for T. zillii populations in eight lakes (Nkuruba Nyanswiga, Mwegenywa, Kifuruka, Ntanda, Kanyango, Lugembe, and Wandakara). The eight lakes were chosen to maximize the range of variation in deforestati on and fishing pressure. Details of otolith preparation procedures and aging criteria are presented in Chapter 4. Growth Growth data were collected for 8 of the crater lakes. Due to low number of smaller fish in the aged sample, it was not po ssible to model growth of T. zillii in the 8 (Table 5-1) lakes using the von Bertalanffy growth equation (see Chapter 4). I estimated growth by calculating mean length-at-ages 3, 4, and 5 as index of growth becau se these were the ages comprising fish fully recruited to the fishery for both the gillnet and trap, and these age groups had sufficient samples of gillnetted fish in all 8 lakes. Lastly, there was minimal change in mean length after age 4 in some lakes and age 5 in other lakes.

PAGE 130

130 Estimates of Total Mortality (Z) and Total Apparent Survival Rates (S) I estimated instantaneous total annual mortality (Z) and total annual apparent survival (S) rates for T. zillii for the trap and gillnet samples separate ly using standard cat ch-curve analysis following Ricker, 1975 under the assumptions ra ndom sampling, constant recruitment, and constant mortality. Only age groups that are fully recruited to the gear type were considered for the analysis, and only age groups with a minimum nu mber of 2 individuals were included in the analysis. Apparent survival was then estimated from the formula; S = e-Z; where Z is the instantaneous total annual mortality. Mortality an d survival were estimated in 8 of the lakes (Figure 5-1). Statistical Analyses I compared the size at maturity between fish captured in gill nets and those captured in minnow traps for 4 lakes where there was suffi cient data from both gears (Nkuruba, Ntanda, Kanyango, and Wandakara). In two of the lake s (Nkuruba and Kanyango), size at maturity was smaller in the trapped fish compared to the gillnetted fish. For the other two lakes (Ntanda and Wandakara), it was not possible to fit maturity og ives to the trap data due to low number of mature fish. I used only the gill net data (that are available for all lakes) to compare size at maturity among lakes. To determine if differences in fish condition (C hapter 3) were correl ated with the observed differences in growth between gillnetted and trapped fish, I calculated condition for four of the lakes with sufficient data for both gears. To take a more integrative approach, princi pal components analysis (PCA) was used to describe the major environmental gradients of variation among the lakes. Only principal components (PCs) with Eigenvalues greater than 1 were retained for interpretation. Loadings were qualitatively designated as high for absolute values greater than 0.60. The factor scores in

PAGE 131

131 PCA were saved and used as independent variable s in a linear regression analysis for each life history trait. I used two appro aches to explore the relationship be tween key life-history traits of T. zillii and both fishing pressure an d deforestation level of th e lakes. One-way ANOVA was used to detect differences in trait means. I also used the PCA factor scores to plot the relationship between growth and apparent surviv al and PCA according to deforestation category and fishing pressure. Results Size Frequency A visual assessment of the length frequency pl ots revealed no differe nces between female and male size frequencies for each gear. Lengthfrequencies for female and male fish were pooled by gear type for each of the 8 lakes. Overall, the length frequency distribution of T. zillii was highly variable among lakes (Figure 5-1). The modal size class for males ranged from 80 mm in Ntanda to 160 mm in Nyanswiga; while for females modal size ranged from 100 mm in Lyantonde to 180 mm in Kanyango. Maximum size wa s also variable rang ing from 240 mm in Wandakara to 280 mm in Ntanda (Figure 5-1). Size frequencies in the 8 lakes used for aging analysis showed distinct separation in sizes of fish caught by gillnet s and minnow traps although there was some overlap. Lakes Nkuruba, Ntanda, and Wandakara show this trend more than the other lakes (Figure 5-2). These lakes also have very low number of fish in the 120-140 mm size range (Figure 5-2). In general, gillnets cons istently collected larger fish, and minnow traps captured mostly fish less than 130 mm TL (Figure 5-2). Sex Ratio Of the 2753 T zillii individuals sexed, 1583 (57.5%) were male and 1,170 (42.5%) were female (Table 5-3). I used the chi-square goodness of fit test to detect differences in sex ratios from the expected female: male sex ratio of 1:1 in each lake and for the overall sex ratio across

PAGE 132

132 lakes (Table 5-3). Across lakes, the sex rati o was 1:1.4 and was significan tly different from the expected ratio of 1:1 ( 2 = 167.9, p < 0.0001, d.f. 13). Only seven of the 14 lakes had more males than females (Table 5-3); and only two lakes had a sex ratio skewed towards females (Table 5-3). Batch Fecundity Batch fecundity was estimated for 84 T. zillii females pooled from 10 lakes (size range 85 to 210 mm TL and 9.8 to 157.4 g total weight). Batch fecundity of T. zillii ranged from 298 to 6,451 eggs with a mean batch fecundity of 1,509 eggs; and there was a strong relationship between batch fecundity and body size of the fe males represented by the equation Log F = -0.64 ( 0.71) + 1.75 ( 0.33) log TL and Log F = 2.17 ( 0.18) + 0.55 ( 0.11) log WT, respectively (Figure 5-3a, b). Batch fecundity of T. zillii also showed considerab le variation within size classes (Figure 5-3a, b), that could reflect vari ation among lakes. Analysis of covariance was used to calculate adjusted mean batch fecundity for T. zillii populations in 7 lakes, where the sample size was greater than 3 females. Fo r a fish of a given body mass, batch fecundity averaged 1,455 eggs and ranged from 1,023 eggs in Lake Nkuruba to 2,692 eggs in Lake Nyinabulitwa. Size at Maturity Female T. zillii in 8 of the 12 lakes generally exhibited a lower L50 compared to males (Figure 5-4). In Lake Nyinabulitwa, females matu red later than males, but both sexes matured at approximately the same length (149 mm TL) in Kanyango. Length at maturity ranged from 128.0 to 161.7 mm TL for females and from 134. 7 to 170.7 mm TL for males (Figure 5-4). Age and Growth Tilapia zillii collected from the 8 crater lakes ranged in age from 1.5 to 7.5 years (Figure 5-5). Maximum age ranged from 5 yrs in Kifuruka (high fishing) to 7.5 years in lightly

PAGE 133

133 fished Lake Nyanswiga (Figure 5-5). Age struct ure in all lakes exhibi ted unimodal distribution for gillnets with full recruitment to the fishery o ccurring at age 3 in some lake and at age 4 in other lakes (Figure 5-5). For the four lakes for which I was able to compare trap-captured fish to gillnet-captured fish, there are tw o notable trends. First, fish captured in the minnow traps generally did not reach the age of those in the gill nets but they did span a large age range for their size (up to 6 years), and th ere was no abrupt truncat ion of ages. Second, in three of the four lakes, the length-at-age for fishes captured in the minnow traps wa s smaller than fish captured in the gillnets, indicating a slower growth rate in these more in shore fish (Figure 5-6). The exception was Lake Kanyango (a heavily exploited lake) where the trapped and gillnetted fish showed very comparable growth rates (Figur e 5-6). Growth was variable among lakes (Figure 5-6). Tilapia zillii grew fastest between ages 2 a nd age 3 after which growth rate decreased (asymptoted) at ages greater than 3 except in the heavily exploited Lake Kanyango (Figure 5-6). Mean length-at-ages 3 (used as an index of growth rate) varied among lakes. For gillnet data, length-at-age 3 ranged from 129.3 mm TL in moderately exploited Wandakara to 146.8 mm TL in unexploited Ntanda (Figure 5-7). Instantaneous Total Mortality (Z ) and Total Apparent Survival (S) Mortality and apparent survival rates were estimated only from gillnet samples because sample size was sufficient in all lakes for fish captured in the nets. Instantaneous total annual mortality (Z) was low in unexploited and moderate ly exploited lakes except Lake Wandakara. Instantaneous total mortality ranged from 31% in Mwegenywa to 70% in Wandakara. Estimates of annual apparent survival also varied among lakes with the lowest annual apparent survival rate (30%) recorded in Lake Wandakara and the highes t apparent survival rate (79%) registered in Lake Mwegenywa for the gillnet-fish. (Table 5-4).

PAGE 134

134 Relationships Among Life-History Traits I used a Pearson correlation matrix to detect relationships among the following lifehistory traits: size at maturity (L50) (for males and females separately), adjusted mean batch fecundity (log10), length at age 3 (as an index of growth) and apparent survival I did not apply a Bonferonni correction, as I was interested in dete rmining general patterns. Mean length-at-age 3 (index of growth) and apparent survival were positively correlated among lakes (r=0.751, P=0.032, n=8, Figure 5-7) indicating th at fish that grew faster al so exhibited a higher apparent survival rate. There was a marginal positive relationship between the size at maturity (L50) of females and adjusted mean batch fecundity (r=0.725, P=0.065), suggesting that females that mature at a larger size produce a greater batch fecundity. Environmental Predictors of Life -History Variation among Lakes Stepwise multiple regressions were used to detect significant environmental predictors of key life-history traits. The environmen tal variables included: water tr ansparency, Secchi depth, Chl-a, electrical conductivit y, pH, temperature, DO, TP, TN, maximum depth, and lake area. The life history traits used as dependent variable s were: apparent survival, grow th, size at maturity, and batch fecundity (as surrogate of reproductive investment). I used observed length-at-age 3 as an index of average growth rate. Growth and apparent survival rates of gillnet-fish were used in this analysis because sample size was sufficient in all lakes. Conductivity was the only signifi cant predictor for mean lengthat-age 3 (index of growth, r = -0.822, P = 0.023); lakes with lower levels of c onductivity were characteri zed by faster growth. There were several significant pr edictors of apparent survival: total nitrogen (partial r = -0.868, P=0.002), conductivity (partial r = -0.153, P=0.041), ma ximum depth (partial r = 0.249, P=0.010), and lake area (partial r = -0.154, P=0.039) In crater lakes characterized by lower total nitrogen, lower conductivity, deeper depth, and smaller area (when c ontrolling for the linear effects of the other

PAGE 135

135 independent variables), Tilapia zillii exhibited a higher apparent surviv al rate. Total phosphorus was the only significant predictor of adjusted mean batch fecundity (r=0.841, P=0.036); lakes with higher TP tended to harbor females with higher mean batc h fecundity. No significant predictors emerged for the size at maturity (L50) of female and male T. zillii. PCA yielded three principle components with Eigenvalues greater than 1 that explained 87.5% of the variance in the data. Variable loadings of the firs t three principle components are given in Table 5-5, and variables that had loadings of absolute values greater than 0.6 were designated as high (labeled* in Table 5-5). The first three PCs accounted for 47.1%, 26.6%, and 10.9% of the variance, respectivel y. Factors that loaded heavily on PC1 included Chl-a, water transparency, conductivity, lake area, and maximu m depth. Variables that loaded heavily on PC2 included dissolved oxygen concentration, pH Total nitrogen, and total phosphorus (Table 5-3). Linear regression indicated a si gnificant relationshi p between apparent survival and PC1 (r = -0.937, P = 0.001, n = 8, Figure 5-8). Length-at-age 3 was also strongly related to PC1 (r = -0.754, P = 0.031, n = 8, Figure 5-9). This suggest s that water transpar ency, chlorophyll a, and conductivity as well as total nitrogen, la ke area and maximum depth of the lake are important correlates of growth and apparent survival of Tilapia zillii across the crater lakes examined. There was no significant predictor of size at maturity (L50) of female or male T. zillii To examine the validity of sel ecting length-at-age 3 as my i ndex of growth, I redid the PC1length regressions for length-at-a ge 4 and length-at-age 5. In both cases, there was a significant negative relationship between th e index of growth and PC1 (age 4: r = -0.797, P = 0.018; age 5: r = -0.755, P = 0.05), supporting the trend found for mean length-at-age 3.

PAGE 136

136 There was no significant difference in size at maturity(L50), growth, apparent survival, and adjusted mean batch fecundity detected between la kes with different levels of fishing pressure for gillnetted fish (P > 0.29). Heavily deforested lakes showed marginally lower values for T. zillii growth (F = 4.09, P = 0.090) and apparent su rvival (F = 5.04, P = 0.066). When growth and apparent survival were regressed against PC1 with deforestation and fishing categories indicated on the plots, there is a pattern that emerges supporti ng lower apparent survival and lower growth in lakes charac terized by severe deforestati on and high fishing pressure (Figures 5-8, and 5-9). Discussion Interdemic variation in li fe history characters of Tilapia zillii from the crater lakes region of western Uganda was quantified to examine th e degree to which the observed variation could be predicted by environmental variation among lake s. Life history trai ts including size at maturity, length-at-age 3 (index of growth), survivorship, and batch fecundity were variable among populations, but notably, size also showed a strong bimodality within some populations associated with gear bias that may reflect di vergence across habitats. In this discussion I summarize major trends in th e life history traits relati ve to other populations of T. zillii or more generally, tilapiines; and I offer my interpretations of the variation in life history traits observed in light of the degree to which these lakes di ffer in both deforestati on and fishing pressure. Sex Ratio Sex ratios for T. zillii differed among lakes, but in gene ral showed a significant departure from unity was skewed towards males in most of the lakes. Predominance of males over females has also been reported for T. zillii in Lake Quaraun (El-Zarka, 1962) Several factors have been suggested as possible causes of unbalanced sex ratio s in other fish species including: differences in mortality and longevity between sexes, differe ntial catchability arisi ng from size dimorphism

PAGE 137

137 (King and Etim, 2004), differential activity patterns, spawning seas on, type of fishing gear, and sampling site (Demeke Admassu, 1994). It is po ssible that males were more wide-spread over the feeding and spawning areas, while females ma y have been more restricted to submerged vegetation and rocky areas to avoid predators (i ncluding fishermen) esp ecially during peaks of the spawning season. We know that in mouth brooding cichlids such as Pseudocrenilabrus multicolor brooding females tend to be more secret ive and hide amongst dense structures (Reardon and Chapman, 2007). Whether this applies to the substrate-spawni ng tilapias such as T. zillii is not clear, but would be an interesting area for future study. Batch Fecundity The high batch fecundity in T. zillii may be a result of abund ance of food supply due to lack of interspecific competi tion in the crater lakes, as the second tilapia species ( O. leucostictus) found in these lakes are predominantly phytopl anktivorous whereas detritus (macrophytes) constitute the greatest percentage (> 50%) of T. zillii food in the crater lakes. Significant relationships between batch fecundity and female length and weight have been reported for pond-raised T. zillii (Dadzie and Wangila, 1980; Coward a nd Bromage, 1999a); it is notable in the crater lakes that batch fecund ity varied considerably among indi vidual females of similar size both within and among lakes. In a study of reproductive biology of T. zillii in Lake Naivasha, Kenya, Siddique (1979) attributed variation in batch fecundity among similarly sized females to lower egg counts in fish that have already spawned at least once. Other studies of T. zillii biology have associated variati on in batch fecundity with di fferential abundance of food (Dadzie and Wangila, 1980; Coward and Brom age, 1999a), differential feeding success (Fagade et al., 1984), and genetic or othe r environmental influences (Wootton, 1979).

PAGE 138

138 Size at Maturity The observed lengths at first maturity (L50) of 120.8 mm and 134.7 mm TL for female and male T. zillii, respectively in the crater lakes is close to thos e reported for the species in previous studies. For example, Siddique (1979) reported length at maturity of 110 mm TL for female and 120 mm TL for male T. zillii in Lake Naivasha. In Lake Vi ctoria, Welcomme (1976) found that the standard length of the smallest mature female T. zillii was 130 m, and that of the smallest mature male was 150 mm. In lake Quarun, Egypt, th e smallest ripe females of the same species were reported to be 50 mm TL and the smallest males 40 mm TL (El Zarka, 1962). Smaller size at maturity is well-documented in tilapia populat ions under stressed condi tions, especially in populations from highly seasonal ha bitats (Fryer and Iles, 1972; Lowe McConnell, 1987). Sizeselective fishing is also known to correlate with ch anges in length at maturity in other tilapiines. For example, Gwahaba (1973) attributed th e decrease in length at maturity in O. niloticus populations in Lake George to overfishing in the lake over a 15year period. In Lake Magadi, increased desiccation and high alkalinity were cited as possible causes of smaller length at maturity observed in Sarotherodon alcalicus (Stewart, 1988). Fryer and Iles (1972) also reported dwarfing and smaller length at maturity fo r tilapia species in Lake Rukwa in response to desiccation and reduced annual rainfall. Age and Growth Longevity of T. zillii captured with gillnets was variable in the crater lakes. The oldest T. zillii recorded in this study was 7. 5 years in Lake Nyanswiga, but maximum age varied from 5 years in Kifuruka to 7.5 years in Nyanswiga. Th ese differences in longevity may be related to environmental factors or the effect of fishing on the apparent survival of T. zillii in lakes with different levels of exploitation. Longevity of tila pia is known to be inversely correlated with the extremity of environmental factors (James 1989). For example, Hodgkiss and Man (1977)

PAGE 139

139 reported maximum age of 5 years for tilapia specie s in a thermally harsh reservoir. Similarly, Hecht and Zway (1984) reported stun ted tilapia mostly 5 years old or less in a hot spring. The maximum age for T. zillii in the present study is within the range of maximum life span of 8 years (De Silva, 1991) and 11 years old (Fry er and Iles, 1972; James 1989), and 7 years (Lvque, 1997) reported for T. zillii in various lakes. A comparison of trapped and gillnetted fish in four of the eight crater lakes with sufficient data from both gears showed a smaller mean length-at-age for fish captured in the traps compared to gillnetted fish in three of the four lakes, indicating a slower average growth rate in the trapped fish that may reside in more inshor e areas. Average growth rates were comparable for both gears in the fourth lake (Kanyango). Th e striking differences in growth rates between the two gears in three of the four lakes may be influenced primarily by th ree factors; 1) dietary differences between the gears; 2) density-dependent fact ors such as intraspecific competition for food resources; and 3) gear selectivity. A qualitative assessment of the diet of T. zillii in one crater lake (Nkuruba) through stomach content analysis revealed that detritus ( Macrophyte sp. ) constituted the highest percentage (> 50%) of the f ood items in the stomachs of T. zillii of all sizes caught in both gillnets and traps. Other food items included phytoplankton ( Microcystis sp., Navicula sp., Scenedesmus sp., Oedogonium, Oscill atoria sp. and Flagillaria sp .), fish eggs, fish scales, and gastropods ( Sphaerium sp .). Zoobenthos ( Chironomidae larvae ) were only found in the stomachs of fish collected with gillnets. This di fference in diet may contri bute to a faster growth in gillnetted fish because of their more varied diet composed of food of both plant and animal origin. Growth of tilapia species is known to be affected by the qual ity and quantity of food (Lowe-McConnell, 1982). In a recent study, Bw anika and colleagues (2006) found that an

PAGE 140

140 omnivorous diet was character istic of Nile tilapia ( Oreochromis niloticus) in lakes Nabugabo and Victoria where they persist with introduced Nile perch, while an herbivorous diet was characteristics of four other N ile tilapia populations in lakes w ithout Nile perch. Nile tilapia from Lake Nabugabo showed a higher growth rate than a nearby herbivorous population in Lake Wamala. Bwanika et al., (2006) proposed that these different patterns of growth between the two lakes may reflect a more energy-ri ch omnivorous diet in Lake Nabugabo. Density-dependent growth where growth is in hibited by intraspecific competition for food at large stock sizes has been re ported for many marine species (e.g., Helser and Almeida, 1997). In a study of variation in populati on mean length-at-age of the Atla ntic cod, Sinclair et al. (2002) found that size-selective fishing ha d the strongest effect on length -at-age, followed by a negative effect of density-dependence and a weak but positive effect of wate r temperature. The effect of intraspecific competition for food resources on ke y life history characters such as individual growth is also well documented in tilapias (Lowe McConnell, 1958, 1982; Fryer and Iles, 1961, 1972). Prolonged restriction in food availabi lity has been reported to lead to T. zillii sacrificing somatic growth to maintain reproductive investments (Coward and Bromage, 1999). In the present study, spatial sepa ration of the fish populations (i .e., trap fish inshore, gill net fish offshore) and associated density-depe ndence due to intraspeci fic competition for food resources in the inshore habitats may have contri buted to the observed difference in growth rate between gillnetted fish and trapped fish in the three lakes with available data. Shallow littoral habitats containing macrophytes that form the majo r diet of the trapped fish are limited in many of these crater lakes. The high population density in the inshore habitats and the low levels of exploitation in most of these lakes may lead to increased competition fo r food resources and the slower growth rate in trapped fish. The trap ped fish may also be allocating their available

PAGE 141

141 resources towards reproduction as opposed to gr owth, resulting in a slower growth rate. I investigated this by comparing size at maturity between gillnet-captured fish and trap-captured fish in 4 lakes and found a smaller size at maturity trap-fish fish compared to the gill-net fish in two of the lakes, providing some support for th e above prediction. One exception to the above scenario was the heavily exploited Lake Kanyango where growth rates for the gillnet-fish and trap-fish were similar. It is possible that the high exploitation in Kanyango may lead to decrease in overall population density that releases the tr ap-fish from density-dep endence, hence reduced competition for resources and increa sed growth rate. It is well documented that harvest-induced reductions in population density reduces intraspeci fic competition that releases populations from density-dependence resulting in faster growth (Jennings and Ka iser, 1998; Hall, 1999; Rochet et al., 2000). Gear selectivity may also have contributed to differences in growth between the trapped fish and gillnetted fish as fish catches in each ge ar could have resulted from the selectivity of the gears without any spatial separati on of the slow versus fast grow ing sub groups. It is possible that small, slow growing fish were also presen t offshore but not vulnerable to the gillnets, and vice versa for the gillnets (i.e., large fish could ha ve been present in the inshore habitats but not captured by the traps). As predation risk does not seem to be a major player for tilapia in these lakes, it is possible that the trends in fish catches and growth of tilapia in these lakes were due, at least in part, to gear selectivity rather than spatial separation. Within lakes, there appears to be divergence in growth with the gillnetted fish growing faster than the trapped fish. Density-dependen ce seems to be a major factor influencing the growth of trapped fish. Across lakes, T. zillii populations in the unfishe d lakes grew faster than those in lakes with high fishi ng, although all lakes appear to have stunted populations. The

PAGE 142

142 pattern of growth across lakes ma y reflect resource availability (food). On the other hand, it is possible that stunting may be selected for over time in response to heavy fishing on older, larger and faster growing individuals. The existing popu lations in the lakes wi th high fishing may be comprised of predominantly slower growing, smaller individuals. Instantaneous Total Annual Mortality Rates (Z) The variation in estimated instantaneous total mortality rate (Z) for T. Zillii populations in the crater lakes may reflect differences in fish ing pressure among lakes. Although estimates of mortality rates for T. zillii in natural systems are, to my knowledge, unavailable, a comparison of mortality estimates from the present study with those of other tilapia species showed that values for the crater lakes were lower than for other tilapia species. For example, Getabu (1992) estimated total mortality rate ( Z) of 0.818 yr-1 for Oreochromis niloticus in the Kenyan part of Lake Victoria using length-based methods. In a study of the same species in Lake Victoria, Okemwa et al., (1994) reported total mortality rate of 1.71 yr-1. Other estimates of total rate for O. niloticus in Lake Victor ia were at 1.67 yr-1 by LVFRD (UNECIA, 2002); 1.84 yr-1 (Muhoozi, 2002), respectively. In Kandulla Rese rvoir, Amarasinghe and de Silva (1992) estimated Z at 1.39 yr-1 and 1.71 yr-1 for O. mossambicus and O. niloticus respectively, using the non-seasonalized catch curve. Recently, King and E tim (2004) computed total mortality rates of 1.75 yr-1 for Tilapia mariae in a Nigerian wetland stream using the seasonalized lengthconverted catch curve method. The highest mort ality rate recorded in this study was 0.71 yr-1 in moderately exploited Lake Wandakara. Relationships among Life-History Traits Apparent survival was higher in faster-growing T. zillii captured with gillnets. This is in agreement with a current life histor y model that predicts that apparent survival is directly related to growth (Hutchings 1993) especially in un exploited populations, and perhaps suggesting that

PAGE 143

143 fishing pressure in these lakes is not sufficiently intense to re lease fish from intraspecific competition at a level that dramatically enhances growth. There are no natural predators in these lakes. Fishermen are the major source of preda tion in the fished lakes. In heavily exploited populations, apparent survival is often inversely related to growth as a result of harvesting of larger individuals and release of individuals from intraspecific predation pressure (Jennings and Kaiser, 1998; Hall, 1999). Our data do not suppor t this life history pred iction as both apparent survival and growth tended to be low even in the heavily fished lakes. This may arise if selective removal of larger and faster grow ing individuals in heavily fished crater lakes selects for slowergrowing individuals. Comparisons of growth across lakes were limited to gillnetted fish-the component of the population most h eavily targeted by fishers. If there is a genetic basis for growth in these lakes, then size-selective fishi ng may lead to evolution towards smaller size-atage. It has been shown that e xperimental size-selective predation in Atlantic silverside over five generations leads to declines in survivorship and growth rate, among other life history characters (Walsh, 2006). The higher apparent survival and faster growth in some lakes with low fishing (e.g., Lake Nkuruba) may also relate to the existe nce of fastand slow-growing fish within these lakes (e.g., Lake Nkuruba). As me ntioned above, we used the gill net data for inter-lake growth comparisons, data that reflect the faster-growing mode in lakes with bimodality in growth rates. The low levels of fishing in relatively unexpl oited lakes may enhance apparent survival by minimizing mortality rates from fishing, and fast er growth may be advantageous in order to reach a size that enables individuals to escape the intense intraspecific competition in inshore habitats. Size at maturity (L50) of female T. zillii was positively (albeit weakly) related to batch fecundity (an index of reproductive output). Popu lations with females matu ring at a larger size

PAGE 144

144 tended to have a higher batch fecundity or repro ductive output compared to those with females maturing at a smaller size. Life history theory pr edicts that low adult apparent survival and high juvenile growth should be associated with high reproductive effort (Roff, 1984; Stearns and Koella, 1986; Hutchings, 1993). This prediction was not supported by data from my study, as I did not detect any correla tion between size at maturity and grow th or apparent survival. Life history theory also predicts th at low adult apparent survival should favor early maturity (Roff 1984; Stearns and Koella 1986; Hutchings 1993). In my st udy, the size at maturity of females was neither related to the intensity of fishing nor apparent survival. The positive association between size at matu rity and batch fecundity is more difficult to interpret. There could be a trade-off whereby females may mature at a small size with lower batch fecundity or a larger size with higher batch fecundity, but few reproductive events (assuming similar life span). Certainly, there are exceptions to the predictions of life history models. For example, early maturity can be associated with slow growth and delayed maturity associated with faster growth in some fish populations (Hutchings, 1993; Fox, 1994). Environmental Predictors of Life-history Variation among Lakes Stepwise regression indicated significant predicto rs for growth. Lakes with lower levels of conductivity were characterized by fast er growth until age 3. There were several significant predictors of apparent survival. In crater lakes characterized by higher total nitrogen, lower c onductivity, large depth, and smaller area, Tilapia zillii exhibited a higher apparent surviv al rate. Total phosphorus was the only significant predictor of ad justed mean batch fecundity; lakes with higher TP tended to harbor females with higher mean batch fecundity. In general the results of the PCA supported so me of the results of the stepwise regression analyses. Both apparent survival and length-at-a ge 3 were related to PC1, suggesting that water transparency, chlorophyll a, conductivity, lake area, maximum depth of the lake and total nitrogen are

PAGE 145

145 important correlates of growth and apparent survival of Tilapia zillii across the crater lakes examined. In Chapter 2, I explored varia tion in physico-chemical characters among 19 crater lakes, including those studied in this chapter. I found a strong negative relationshi p between water transparency and chlorophyll a (Chl-a) concentration with deforested lake s having a lower transparency, higher conductivity, and higher Chl-a concentra tion compared to forested lakes. The variation in growth and apparent survival across crater la kes may indirectly reflect effect s of deforestation on water quality conditions. Severely deforested lakes showed marginally lower values for T. zillii growth and apparent survival than moderately deforested la kes. There was no significant difference in size at maturity, growth, apparent survival, and adjusted mean batch fecundity dete cted between lakes with different levels of fishing pressu re. When growth and apparent su rvival were regressed against PC1 with deforestation and fishing le vels indicated, a pattern emerges that indicates lower apparent survival and lower growth in lakes characterized by heavy deforestation and heavy fishing pressure. A deeper euphotic zone and greater visibi lity may enhance feeding conditions for T. zillii, a fish that feeds predominantly on macrophytes in these lakes. This general trend of faster growth in m oderately deforested lakes with low fishing compared to heavily fished lakes with severe deforestation is inconsistent with the results reported in Chapter 3 that compared condition in both T. zillii and O. leucostictus I found a higher mean relative condition factor for T. zillii in heavily fished populations compared to populations with low fishing pressu re (Chapter 3). To determine whether differences in growth between fish captured with the two gears is a resu lt of variation in conditi on, I calculated relative condition separately for trap fish and gill net fi sh in 4 lakes. I found no significant difference (p= 0.41) between relative condition of fish from traps and gillnets. It is possible that variation in shape across size range within and among lake s may contribute to the pattern of growth

PAGE 146

146 observed across lakes. For, example, the condition of O. leucostictus is lower than T. zillii (Chapter 3). In part, this reflects the shape differences between the two species, with T. zillii having a deeper body for its length than O. leucostictus. If growth rate of netted fish tends to be lower in deforested, heavily exploited lakes, then the stunted body form may contribute to a better condition of individual fish. This issue could be addresse d by measuring shape variation across sizes within (trap vs. g illnet) and among lakes with different rates of growth using geometric morphometric tools that are powerful at detection of small shape differences among groups. There were some discrepancies in the relatio nship between apparent survival and fishing pressure. Survival estimates for T. zillii in the eight crater lakes usi ng catch-curve an alysis did not completely match my apriori catego rization of lakes based on fishing effort (high, medium, low). For example, Lake Wandakara had the lowest survival estimate and yet it was categorized in the medium fishing group. It is possible that the fishing effort (number of fishermen) in Lake Wandakara was underestimated as the results indicate that the lake is heavily fished. Potential Fish Yield Models The extensive dataset generated in this study will form the basis in efforts to develop predictive fish yield models for the crater lakes. Easy to measure parameters such as lake area, mean depth, conductivity alkalinity, chlorophy ll a concentration, phytoplankton photosynthesis, fishing effort (number of fishers or boats), fish catch have been used to predict fish yield in a variety of tropical lakes. For example, Ha nson and Leggett (1982) found total phosphorus (TP) and total dissolve solids (TDS) to be the best predictors of fish yield and developed the following equation: (Y = 0.066 TP + 0.141z + 0.013 TDS 1.513; n = 21, r = 0.98). In a study of 46 lakes and 25 reservoirs in Africa, Crul (1993) found that lake area was th e best predictor of fish yield (bigger lakes yield more fish!) base d on the relationship: Catch (t yr-1) = 8.32*Area0.92 r2 = 0.93.

PAGE 147

147 Moreau and De Silva (1991) used multiple regres sion models to develop the following fish yield models for lakes and reservoirs in Sri La nka, Thailand, and Philippines, respectively: Y = -63.7 + 0.111*Area + 5.898*Effort n = 20, r2 = 0.857 Y = 112.3 + 0.010*Area + 0.309*Effort n = 19, r2 = 0.786 Y = 436.2 + 0.336*Area + 0.745*Effort n = 17, r2 = 0.929 Where Y = total yield per water body (t yr-1), area (ha), effort (no. of fishermen) Moreau and De Silva (1991) also found mor phometric variables, water transparency, and total alkalinity to be impo rtant predictors of fish yield in the same lakes. In a study of seven African lakes, Melack (1976) found a significant relationship be tween gross photosynthesis (PG) and fish yield described by the equation logY = 0.95 + 0.00034 PG; r = 0.82. Machena and Fair (1983) found Melacks yield models to be reason able for predicting yield in Lake Kariba but resulted in a six-fold yield underestimate for Lake Tanganyika. In temperate lakes, McConnell et al. (1977) found fish producti on in ponds was related to gro ss photosynthesis. Jones and Hoyer (1982) also found the yield of sport fish in US reservoirs to correl ate with chlorophyll a. A general global predictive model of fish pr oduction was developed based on assumption that conversion of phytoplankton into fish production is dependent on trophic status of lakes (Downing et al. 1990). The equation was: log10Pf = 0.600 + 0.575 log10 Pf; n = 19, r = 0.89; where; Pf is fish production and Pf is primary production. The authors suggest that the conversion of primary production may be as high as 100 times in more in hyper-eutrophic lakes. Given the variability in of environmental characters in the crater lakes, a hyper-eutrophic lake such Wakenzi (Chl.a = 200 gL-1) would yield approximately 2,500 Kg/ha of fish and an oligotrophic lake such as Ntanda (Chl.a = 3.9 gL-1) would yield approximately 280 kg/ha of fish.

PAGE 148

148 In this chapter, I have demonstrated differen ces in important life history charaters within T. zillii of the same lake sampled with differe nt gear types and among populations of T. zillii from different lakes. I have also documented relationships between life history characters such as growth and apparent survival as well as relationships between lif e history charaters and environmental variables. Whether the di fference in life hist ory characters of T. zillii observed in the crater lakes are under genetic control and/or related to phenotypic plasticity of individuals as a result of environmental variation can not be determined from our data. Nevertheless, environmental factors, density-dependence, and size-selective fishing provide possible explanations for variation in key life history characters of T. zillii populations in the crater lakes, and significant predictors were detected for most lif e history traits. It is not possible to separate the effects of deforestation and fi shing pressure on the life history traits considered, as lakes with severe levels of deforestation were often characterized by higher fishing pressure. These results provide useful baseline data for the manage ment of heavily exploited and unexploited T. zillii populations in the crater lakes. First, emphasis could be placed on collecting appropriate environmental data for forecasting future yields from the heavily exploited lakes. Second, density-dependent growth in unexploited lakes could be mitigated through mass removal of T. zillii in unexploited lakes (at different densities) and determining their growth rates after 1-2 years. In addition, models used to determ ine biological reference points for management (e.g., yield per recruit and spawning stock biomass pe r recruit) need to take into account densitydependent growth in T. zillii populations within these lakes fo r accurate predictions. Third, to determine the effect of size-selec tive fishing, it will be important to investigate the possibility of selection against faster growing fi sh given the findings of this study and the implications of longterm microevolutionary change for the future yiel d and sustainability of the artisanal fisheries.

PAGE 149

149 Table 5-1. Life hi story charaters of Tilapia zillii determined in the Crater Lakes of western Uganda from July 2004-June 2005. Lake Fishing pressure Deforestation Sex ratioSize frequency Size at maturity Batch fecundity Growth Mortality Survival Kanyango High Severe + + + + + + Kifuruka High Severe + + + + + + + Lugembe High Severe + + + + + + + Wakenzi High Severe + + + Kasenda Low Moderate + + + + Nyanswiga Low Moderate + + + + + + Nkuruba Low Moderate + + + + + + + Ntanda Low Moderate + + + + + + + Rukwanzi Low Severe + + Mwegenywa Medium Severe + + + + + + + Wandakara Medium Severe + + + + + + + Lyantonde Medium Severe + + + + Nyinabulitwa Medium Moderate + + + +

PAGE 150

150 Table 5-2. Macroscopical criteria used to stage gonadal development in Tilapia zillii (Modified from Siddique, 1977) from crater lakes of western Uganda. Gonad stage Gonad state Gonad description I Immature (Inactive) Ovaries buff-colored, translucent, elongated oocytes invisible. Testis dull white and thread-like. II Maturing Ovaries buff-colored, translucent, elongated oocytes invisible. Testis dull white and thread-like III Ripening Ovaries yellowish-green, oocytes visible to the naked eye. Testis creamy-white, enlarged and thickened IV Ripe (Active ripe) Ovar ies olive-green, large and occupy almost the whole of ventral visceral cavity. Testes creamy white, enlarged and more thickened. V Ripe Ovaries olive-green and eggs can be extruded by slightly pressing the belly. Testes creamy white, enlarged and more thickened, heavy with milt. VI Ripe-running Ovaries olive-green. Slight pressure on the belly ruptures the ovary and involuntarily releases eggs into the visceral cavity. Testes creamy white, enlarged and more thickened with milt freely oozing from them. VII Spent condition Ovaries are greenish-yellow, reduced in size due to extrusion of eggs Testes are creamy white, shrunk, and flaccid.

PAGE 151

151 Table 5-3: Number of female s and males and sex ratios of T. zillii caught in 14 crater lakes of western Uganda ( -significant at 5% level, ** -significant at 1% level). Lake Females Males Sex ratio Females: males ( 2) Kanyango 40 97 1:2.4 17.79** Kifuruka 110 134 1:1.2 1.77 Lugembe 118 100 0.8:1 1.11 Wakenzi 142 76 0.5:1 14.99** Kasenda 46 68 1:1.5 3.18 Marusi 41 70 1:1.7 5.68* Nyanswiga 55 195 1:3.5 58.80** Nkuruba 112 75 0.7:1 5.49* Ntanda 144 195 1:1.4 5.75* Rukwanzi 60 76 1:1.3 1.41 Mwegenywa 79 181 1:2.3 30.01** Wandakara 85 76 0.9:1 0.37 Lyantonde 82 118 1:1.4 4.86* Nyinabulitwa 56 122 1:2.2 18.35** TOTAL 1170 1583 1:1.4 167.90**

PAGE 152

152 Table 5-4. Estimated total annual in stantaneous mortality rate (Z year-1), regression statistic (r and P-values) and estimated annua l apparent survival (S) for Tilapia zillii from eight Crater Lakes in western Uganda. Lake Z r p S Kanyango 0.540 0.88 0.064 0.56 Kifuruka 0.347 0.71 0.076 0.65 Lugembe 0.565 0.96 0.001 0.51 Nyanswiga 0.343 0.85 0.027 0.66 Nkuruba 0.421 0.87 0.002 0.61 Ntanda 0.297 0.93 0.002 0.71 Mwegenywa 0.314 0.99 0.001 0.79 Wandakara 0.634 0.89 0.001 0.31

PAGE 153

153 Table 5-5. Results of principal component anal ysis (PCA) describing the major environmental gradients of variation in Tilapia zillii life history charaters am ong the lakes. Highest loadings for each environmental va riable are highlighted in bold. Component Variable (Log 10) 1 2 3 Chlorophyll a 0.974* 0.056 -0.111 Water transparency (Secchi depth) -0.925* -0.120 0.251 Dissolved oxygen 0.197 0.865* -0.241 pH 0.138 0.835* 0.195 Temperature 0.469 -0.033 0.836* Conductivity 0.765* 0.371 -0.100 Total phosphorus 0.458 -0.838* -0.026 Total nitrogen 0.828* 0.343 0.236 Lake area -0.747* 0.461 -0.178 Maximum depth -0.769* 0.358 0.353 Total variance explained 47.1 27.6 10.9

PAGE 154

154 0 20 40 60 80 All fish (mature + immature Males Females) Kanyango Kifuruka Lugembe 4080120160200240280 Lyantonde 4080120160200240280 0 20 40 60 80 Mwegenywa Nkuruba Nyanswiga 4080120160200240280 Nyinabulitwa Wakenzi 4080120160200240280 Total length (mm) Wandakara 0 20 40 60 80 Percent frequencyKasenda Ntanda Figure 5-1. Length-freq uency distribution by 20-mm size classes for Tilapia zillii collected from 2004 to 2005 from 12 crater lakes in western Uganda. The lakes are ordered based on fi shing pressure (High, low, medium, respectively).

PAGE 155

155 04080120160200240280 Nyinabulitwa 04080120160200240280 Lyantonde 0 10 20 30 40 50 Percent frequencyKasenda Wakenzi 04080120160200240280 Total length (mm) Wandakara Nyanswiga Ntanda Nkuruba 04080120160200240280 0 10 20 30 40 50 Mwegenywa Lugembe Kifuruka 0 10 20 30 40 50 Net Trap Kanyango Figure 5-2. Length frequency of Tilapia zillii sampled with gillnets and minnow traps (as a percent of total catch) from eight crater lakes, western Uganda between June 2004-July 2005. The lakes are ordered based on fishing pr essure (High, low, medium, respectively).

PAGE 156

15604080120160200240 0 2000 4000 6000 8000 04080120160200 0 2000 4000 6000 8000 Total length (mm) (a) Batch fecundity (# eggs at stage VI)Total weight (g) (b) Figure 5-3. Relationship between ba tch fecundity (# no of eggs at stage VI) and A) total length. B) total weight for Tilapia zillii from 10 crater lakes in west ern Uganda (pooled data).

PAGE 157

157 04080120160200240280 0.0 0.5 1.0 04080120160200240280 0.0 0.5 1.0 04080120160200240280 0.0 0.5 1.0 04080120160200240280 Male Female Kifuruka Lugembe Mwegenywa Nyanswiga Total length (mm) Wandakara Ntanda Nkuruba Kanyango Lyantonde Proportion matureKasenda Rukwanzi (LOW) Nyinabulitwa Figure 5-4. Maturity ogives and size at 50% maturity (L50) of female and male Tilapia zillii from crater lakes of western Uganda. The lakes are ordered based on fishing pressu re (High, low, me dium, respectively).

PAGE 158

158 0 10 20 30 40 50 Net TrapKanyango Kifuruka Lugembe 0123456789 0 10 20 30 40 50 Observed age (yrs) Mwegenywa Nkuruba 0123456789 Wandakara 0 10 20 30 40 50 Percent frequencyNyanswiga 0123456789 Ntanda Figure 5-5. Age frequencies of Tilapia zillii from eight crater lakes in western Uganda collected with minnow traps and gillnets between July 2004 and June 2005. The lakes are ordered base d on fishing pressure (High, low, medium, respectively).

PAGE 159

159 0 50 100 150 200 0 50 100 150 200 012345678 012345678 0 50 100 150 200 012345678 Kifuruka Mean observed TL (mm)Nyanswiga Net TrapKanyango Ntanda Observed age (yrs) Mwegenywa Wandakara Nkuruba Lugembe Figure 5-6. Mean length-at-age for Tilapia zillii captured with gillnets (open circle) and minnow traps (closed circles) from eight crater lakes in western Ugandan from June 2004 to July 2005. Vertical bars are standard e rrors of the mean (SE). The lakes are ordered based on fishing pressu re (High, low, me dium, respectively).

PAGE 160

160 0.00.20.40.60.8 125 130 135 140 145 150 Mean total length at age 3 (mm)Survival Figure 5-7. Relationship between mean total leng th-at-age 3 (growth) and total annual apparent survival (S) for Tilapia zillii captured with gillnets in eight crater lakes, western Uganda.

PAGE 161

161 Figure 5-8. Relationship between total annua l apparent survival (S) and PC1 (water transparency, chlorophyll a, and conductivity as well as the area and maximum depth) for eight crater lakes with va rying levels of deforestation (low and heavy) and fishing pressure (low, medium, and high).

PAGE 162

162 Figure 5-9. Relationship between mean total length-at-age 3 (growt h) and PC1 (water transparency, chlorophyll a, and conductivity as well as the area and maximum depth) for eight crater lakes with va rying levels of deforestati on (low and high) and fishing pressure (low, medium, and high).

PAGE 163

163 CHAPTER 6 GENERAL DISCUSSION AND CONCLUSION General Conclusions The overall goal of this dissert ation was to examine effects of size-selective fishing and a suite of environmental factors on life history variation in tilapi a populations within the crater lakes of western Uganda, with the hope of providing critical information needed for the monitoring of long-term changes in water quality and the artisanal fishery in these lakes. The existence of both unexploited and heavily exploited tilapia populations in th ese lakes, in addition to the wide variation in environmental conditions permitted me to investigat e the effect of sizeselective fishing and environmental factors on li fe history variation in the tilapia species simultaneously. In this chapter, I highlight the major findings of the study and discuss their implications for the development of long-term management goals aimed at mitigating potential effects of anthropogenic activitie s on the Crater Lake ecosystem and sustainability of the artisanal fisheries in the crater lakes. I al so summarize major conclusions and recommendations arising from the study. Over the last two decades, there has been an ac celeration of clearing of forested crater rims driven by need for more agri cultural land, fuel wood, and build ing materials for the rapidly expanding human population in the region. In 1997, National Environment Management Authority estimated that wood fu el provided 99.2% of energy used for cooking in Kabarole that hosts many of the crater lakes. The results of loss in forest c over and associated soil erosion have been increased turbidity, si ltation, nutrient, organic matter lo ad, and cultural eutrophication. Unfortunately, limnological data for a large num ber of the lakes with varying level of anthropogenic disturbance were lacking, hindering th e development of predictive models of lake productivity that would permit monitoring of en vironmental changes within the lakes. To

PAGE 164

164 address this need, I determined a range of enviro nmental characters in a large number (19) of the lakes and used the data to developed predictiv e models of primary productivity (chlorophyll a) for the crater lakes (Chapter 2). The crater lakes exhibited wide variation in their environmental characters. I found a strong negative relations hip between water tran sparency and Chl-a concentration with deforested lakes having a lo wer transparency and higher Chl-a concentration compared to forested lakes. I also found a pos itive relationship between Chl-a concentration and total phosphorus concentration in th e crater lakes with increased i nput of TP leading to increase in Chl-a concentration in the lakes. A comparison of the above regression relationships with those developed for sub-tropical lakes indica ted that Chl.-a-trans parency and Chl.-a-TP regressions lines for sub-tropical lakes largely lie outside the c onfidence intervals for this study, suggesting that the predictive models of primar y productivity developed for other regions may not be suitable for use in the equatorial tropics or perhaps more specifically, for crater lakes in tropical regions. I hope that the predictive model of chlo rophyll a (as a surrogate of phytoplankton abundance) proposed in this study will pr ovide a useful tool in efforts to monitor, control, and potentially mitigate the effects of de forestation of the crater rims and associated eutrophication of the lakes. However, there is ne ed to consider broader studies covering a larger number of lakes in the tropics in order to evalua te the suitability of the present model for use in other tropical lakes. As human induced changes in environmental feat ures of the crater lakes may have a direct effect on the biology of tilapia sp ecies in these lakes, it was important to identify environmental predictors of key life history charaters (e.g., condition). Thus I explored the effects of anthropogenic factors (i.e., deforestation and fish ing) on the condition of two introduced species ( Oreochromis leucostictus and Tilapia zillii ) in 17 of the crater lakes that differed in extent of

PAGE 165

165 deforestation and fishing pressure (Cha pter 3). Results demonstrated that O. leucostictus in severely deforested lakes and heavily fished lake s were in a better condition compared to similar fish in lakes with low productiv ity and low to medium fishing. In contrast, differences in condition of T. zillii were only detectable between lake s with high and low fishing but no relationship was observed between condition and extent of de forestation, although we were unable to quantitatively explore the interaction between the two factors. From a conservation biology viewpoint, it is interesting that O. leucostictus showed higher condition values in severely deforested lakes compared to moderately deforested ones. This may in part reflect the association between deforestation and fishing pressure. However, it may also reflect changes in water quality induced by deforestation of the watershed. These results should therefore be integrated into development of innovative approaches for manage ment of tilapia populations in these lakes. This was the pioneering study of life history among fiel d populations of Tilapia zillii the dominant fish species in the cr ater lakes. There was therefore no prior information on important life history charaters, particularl y, age, growth, and mortality data that are essential in developing models used for fisheries management. To de rive age information for the dominant tilapia species ( T. zillii ) in the crater lakes, I used marginal-inc rement and edge analyses to validate the periodicity and timing of opaque zone formation in otoliths of T. zillii in one Crater Lake Nkuruba (Chapter 4). I found that deposition of opaque zones in T. zillii populations in lake Nkuruba follows a bimodal pattern corresponding to the two seasonal peaks of precipitation that are characteristic of this equa torial region. I then used the criteria developed above to age individual T. zillii in Lake Nkuruba and to estimate growth, mortality and apparent survival rates. Results indicated that differences in growth be tween gillnet-captured and minnow trap-captured

PAGE 166

166 fish with the gillnet-fish growing faster and a ttaining larger size-at-age than the minnow trapfish. Estimated total mortality (Z) and natural mortality (M) values were also lower for Lake Nkuruba compared to values reported for other til apia species in larger lakes. These results represent the first estimates of growth and mortality for T. zillii in the crater lakes and therefore provide an essential first step in appropriate management of th e small scale fishery for local communities within the vicinity of these lakes. Differences in gear selectivity observed in this study dictate that estimates of T. zillii key population characters must be gear specific, as combining data from both gears may result in incorrect estimates of growth, mortality, and apparent survival that could lead to developm ent of inaccurate population models. Gillnets may be particularly effective for describing larger and faster growing fish; while data from minnow traps may be suitable for modeling growth and mortality in smaller, slower growing fish. Finally, simultaneous examination of effects of size-selective predat ion and environmental factors on life history of T. zillii in the crater lakes (growth, size at maturity, mortality, and batch fecundity) indicated differences in im portant life history charaters in T. zillii between gillnets and traps and among populations of T. zillii from different lakes. I al so documented relationships between life history characters such as growth and apparent survival as well as relationships between life history charaters a nd environmental variables. Whet her the difference in life history characters of T. zillii observed in the crater lakes are unde r genetic control and/or related to phenotypic plasticity of individual s as a result of environmental va riation can not be determined from our data. Nevertheless, environmental f actors, density-dependence, and size-selective fishing provide possible explanations for va riation in key life history characters of T. zillii populations in the crater lakes, and significant predictors were detected for most life history traits. It is not possible to separate the effects of deforestation and fishing pressure on the life

PAGE 167

167 history traits considered, as lake s with severe levels of deforest ation were often characterized by higher fishing pressure. However, these results provide useful baseline data for the management of heavily exploited and unexploited T. zillii populations in th e crater lakes. The Way Forward (Recommendations) It is important to monitor the deforestation levels in the crater lakes and encourage the local communities to undertake reforestation projects to reduce the pressure on the remaining forest patches around the crater rims. Emphasis should also be placed on collecting appropriate environm ental data for forecasting changes in water quality and future yields from the heavily exploited lakes. In light of the profound growth differences be tween trap-fish and gill net-fish, the use of multiple gears for sampling populations of tila pia species in these crater lakes is recommended, and estimation of key life history characters should be gear-specific. There is also need to periodically evaluate the effec tiveness of these gears in future studies of a T. zillii populations in the crater lakes to ensu re the data obtained accurately reflect T. zillii population parameters. Density-dependence as a result of intraspe cific competition is an important factor influencing key life history char acters in the crater lakes. Models used to determine biological reference points for management (e.g., yield per recruit and spawning stock biomass per recruit) should therefore take into account densitydependent growth in T. zillii populations within these lakes for accurate predictions. Future Studies Comparison of key life history ch aracters among lakes for trap-captured fish was limited to 4 lakes in this study due to low number of ag ed trap-captured fish. Given the growth differences that exist between the two gears used in this study, a comprehensive age and growth study should be conducte d covering both inshore and offshore habitats in a wide range of the lakes. It is possible that part of th e variation in fish condition ma y reflect differences in shape between tilapia species and potentially among po pulations. Studies should be conducted to measure shape variation across sizes within (trap vs. gillnet) and among lakes with different rates of growth using ge ometric morphometric techniques. Based on results of the present study, it was not possible to determine if life history variation in T. zillii within the crater lakes is a ph enotypic response and or genetically controlled. Density studies should be designe d where mass removal (at different densities) of T. zillii in unexploited lakes is carried out and growth rates estimated in the populations after 1-2 years. Genetic studi es investigating the possibility of selection against faster growing fish should also be conducted to assess implications of long-term microevolutionary change for the future yield an d sustainability of the artisanal fisheries.

PAGE 168

168 LIST OF REFERENCES American Public Health Association (1988) Stan dard methods for the examination of water and wastewater, 17th edition-Washington DC: American Public H ealth Association. Adams S. M., McLean R. B. and Parrotta J. A. (1982) Energy partitioning in largemouth bass under conditions of seasonally fluc tuating prey availability. Transactions of the American Fisheries Society 111 549-558. Alin S. R., Cohen A. S., Bills R., Gashagaza M. M., Michel E., Tiercelin J. J., Martens K., Coveliers P., Mboko S. K., West K., Soreghan M., Kimbadi S., and Ntakimazi G. (1999). Effects of landscape disturbance on animal co mmunities in Lake Tanganyika, East Africa: Conservation Biology 13 1017 Amarasinghe U. S. and De Silva S. S. (1992) Population dynamics of Oreochromis mossambicus and O.niloticus (Cichlidae) in two rese rvoirs of Sri Lanka. Asian Fisheries Science 5 37-61. Amon K. and Pauly D. (editors.) The Third Inte rnational Symposium on Tilapia in Aquaculture, ICLARM Conf. Proc. 41, Manila. Anderson R. O. and Neumann R. M. (1996) Length, weight, and associated structural indices. In: Fisheries techniques, B. R., Murphy, and Willis, D. M (Eds.). American Fisheries Society Maryland, USA, pp.447-481. Anon. (1988) Fish evolving in res ponse to commercial fishing? BioScience 38 319-20. Ashley M. V., Willson M. F., Pergams O. R. W., O'Dowd D. J., Gende S. M. and Brown J. S. (2003) Evolutionarily enlightened management. Biological Conservation 111 ,115-123. Barlow G. W. (2000) Cichlid Fishes: Nature's Gr and Experiment in Evolution Perseus Books, Cambridge. Baxter R. M., Prosser M. V., Talling J. F. a nd Wood R. B. (1965) Stra tification in tropical African lakes at moderate altitudes (1500-2000 m). Limnology and Oceanography 10 510-529. Beadle L. C. 1932. Scientific results of the Camb ridge expedition to the East African lakes in relation to their flora and fauna. Zoological Journal of the Linnean Society 38 157-211. Beadle L. C. (1963) Anaerobic lif e in a tropical Crater Lake. Nature 200 1223-1224. Beadle L.C. (1981) The Inland Waters of Tropical Africa. Longman, London. Beamish R. J. (1979) New information on the longevity of Pacific ocean perch ( Sebastes alutus) Journal of the Fisheries Resources Board of Canada 36 1395-1400.

PAGE 169

169 Beamish R. J. and McFarlane G. A. (1983) Th e forgotten requirement for age validation in fisheries biology. Transactions of the Amer ican Fisheries Society 112 735-743. Beckman D. W. and Wilson C. A. (1995) Season al timing of opaque zone formation in fish otoliths. Pages 27-44 In : Secor, D. H., J. M. Dean, and S. E. Campana (editors). Recent Developments in Fish Otolith Research. University of South Carolina Press, Columbia. Beverton R. J. H. and Holt S. J. (1957) On the dynamics of exploited fish populations. Fisheries Investigation Series II Vol. 19, Ministry of Agricu lture, Fisheries and Food, London, 533 pp. Binohlan C. and Pauly D. (1998) The length-weight table, In: Fr oese R. and Pauly D. (eds.) Fishbase 1998: concepts, design and data sources. ICLARM, Manila, pp. 121-123. Blackwell B. G., Brown M. L. and Willis D. W. (2000) Relative weight (Wr) status and current use in fisheries assessment and management. Reviews in Fisheries Science 8 1-44. Blake C. and Blake B. F. (1978) The use of operc ular bones in the study of age and growth in Labeo senegalensis from Lake Kainji, Nigeria. Journal of Fish Biology 13 287-295. Blaxter J. H. S. (1986) Development of sense orga ns and behavior of Teleost fishes with special reference to feeding and predator avoidance. Transactions of the American Fisheries Society 115 89-114. Bolger T. and Connolly P. L. (1989) The selec tion of suitable indices for the measurement of fish condition. Journal of Fish Biology 34 171. Booth A. J., Merron G. S. and Buxton C. D. (1995) The growth of Oreochromis andersonii (Pisces: Cichlidae) from the Okavango Delta, Bo tswana, and a comparison of the scale and otolith method of ageing. Environmental Biology of Fishes 43 171-178. Branco C. W. C., Rocha M. I., Pinto G. F. S ., Gomara G. and Filipo R. D. (2002) Limnological features of Funil reservoir (RJ Brazil) and i ndicator properties of rotifers and cladocerans of the zooplankton community. Lake Reservoirs and Resources Management 7 87-92. Britton J. R. and Harper D. M. (2006) Length-weight relationshi ps of fish species in the freshwater rift valley lakes of Kenya. Journal of Applied Ichthyology 22 334-336. Brook L. (1994) Changes in the limnological behavi or of a tropical Afri can explosion crater lake: L. Hora-Kilole, Ethiopia. Limnologica 24 57-70. Brothers E. B. (1979) Age and growth studies on tropical fishes. Pp. 119-136. In: S. B. Saila and P. M. Roedel (editors). Stock assessment for tropical small scale fisheries. Prceedings of an international workshop, September, 1979, International Center for Marine Resources Development, University of Rhode Island, Rhode Island, USA.

PAGE 170

170 Brown M. L. and Murphy B. R. (1991) Relationship of relative weight (Wr) to proximate composition of juvenile stripe d bass and hybrid striped bass. Transactions of the American Fisheries Society 120 509-518. Buddington R. K. (1979) Digesti on of an aquatic macrophyte by Tilapia zillii Journal of Fish Biology 15 449-456. Bugenyi F.W.B. and Magumba K. M. (1996) Th e present physico-chem ical ecology of Lake Victoria, Uganda, pp.141-145. In: Johnson T. C. and Odada E. (Eds.). The limnology, chemical, and paleolimnology of East Afri can Lakes. Gordon and Bready, Amsterdam. Bulow F. L., Zeman M. E., Winningham J. R. a nd Hudson W. F. (1981) Seasonal variation in RNA-DNA ratios and in dicators of feeding, reproduction, energy storage and condition in a population of bluegill, Lepomis macrochirus Rafinesque. Journal of Fish Biology 18 237-244. Burns, J. W. (1972) Some effects of logging and associated road construction on northern California streams Transactions of the American Fisheries Society 101 1-17. Burrough R. J. and Kennedy C. R. (1979) The occurre nce and natural alleviat ion of stunting in a population of roach. Rutilus rutilus (L). Journal of Fish Biology 15 93-109. Bwanika N. G. (2005) Feeding Ecology a nd Bioenergetics of Nile Tilapia ( Oreochromis niloticus ) in the Inshore Habitats of Lake Nabugabo: Predicting the future of lake Victoria Basin Nile Tilapia. Ph.D. Dissert ation, Makerere University, Uganda. Bwanika G. N., Murie D. J. and Chapman L. J. (2007) Comparative age and growth of Oreochromis niloticus LINNE (Nile tilapia) in lakes Nabugabo and Wamala, Uganda. Hydrobiologia ( In Press) Campana S. E. and Neilson J. D. (1985) Microstructure of fish otoliths. Canadian Journal of Fisheries and Aquatic Science 42 1014-1032. Campana S. E. (1984) Interactive effects of age and environmental modifiers on the production of daily growth increments in ot oliths of plainfin midshipman, Porichthys notatus Fishery Bulletin 82, 165-177. Campana S. E. (1999) Chemistry and composition of fish otoliths: pathways, mechanisms and applications. Marine Ecology Progress Series 188 263-297. Campana S. E. (2001) Accuracy, precision and quality control in age determination, including a review of the use and abuse of age validation methods. Journal of Fish Biology 59, 197-242. Campana S. E., Zwanenburg K.C. and Smith J.N. (1990) 210Pb/226Ra determination of longevity in redfish. Canadian Journal of Fisheries and Aquatic Sciences 47 163-165.

PAGE 171

171 Campana, S. E. (1984) Interactive effects of age and environmental modifiers on the production of daily growth increments in ot oliths of plainfin midshipman, Porichthys notatus Fishery Bulletin 82, 165-177. Canfield D. E. and Bachmann R.W. (1981) Prediction of total phosphorus concentration, chlorophyll a, and Secchi depths in natural and artificial lakes. Canadian Journal of Fisheries and Aquatic Sciences 38 414-423. Carlstrom D. (1963) A crystallographi c study of vertebrate otoliths. Biological Bulletin (Woods Hole) 125 441-463. Carpenter S. R., Kitchell J. F., and Hodgson J.R. (1985) Cascading trophic interactions and lake productivity. Bioscience 35 634-639. Castelo R. (1994) Biogeographical consid eration of fish diversity in Bioko. Biodiversity and Conservation 3 808-827. Chambers R. C. and Leggett W. C. (1996) Mate rnal influence on variation in egg sizes in temperate marine fishes. American Zoologist 36 180-196. Chambers R. C. (1997) Environm ental influence on egg and propagul e sizes in marine fishes. Pp. 63-102. In: Chambers R. C. and Trippel E. A. (editors). Early Life History and Recruitment in Fish Populations. Chapman and Hall, London. Chambers R. C. and Leggett W. C. (1996) Mate rnal influence on variation in egg sizes in temperate marine fishes. American Zoologist. 36 180-196. Chapman C. A. and Chapman L. J. (2003) Defore station in tropical Africa: Impacts on aquatic ecosystems. Pp. 229-246. T. L. Crisman, L. J. Chapman, C. A. Chapman, and L. S. Kaufman (Eds.), Conservation, Ecology, and Ma nagement of African Freshwaters. University Press of Florida, Gainesville, Florida: 229-246. Chapman L. J., Chapman C. A., Crisman T. L. and Nordlie F. G. (1998) Dissolved oxygen and thermal regimes of a Ugandan crater lake. Hydrobiologia 385 210-221. Chilton D. E., and Beamish R. J. (1982) Age de termination methods for fishes studied by the ground-fish program at the p acific Biological Station. Canadian Special Publication of Fisheries and Aquatic Sciences 60 102 pp. Chutter F. M. (1968) On the ecology of fauna of stones-in-the-current supporting a very large Simulium (Diptera) population. Journal of Applied Ecology 5 531-561. Cohen A., Bills R., Conquyt C. Z., and Caljon A. G. (1993) The impact of sediment pollution on biodiversity in Lake Tanganyika. Conservation Biology 7 667-677. Conover D. O. and Munch S. B. (2002) Sustaining fi sheries yields over evolutionary time scales. Science 297 94-96.

PAGE 172

172 Copp G. H. (2003) Is fish conditi on correlated with water conductivity? Journal of Fish Biology 63 263-266. Cornett R. G. and Riggler F. H. (1980) The ar eal hypolimnetic oxygen deficit. An empirical test of the model. Limnology and Oceanography 25 672-679. Coward K. and Bromage N. R. (2000) Repr oductive physiology and endocrinology of female tilapia broodstock. Reviews in Fish Biology and Fisheries 10 1-25. Crisman T. L., Chapman L.J., Chapman C.A., and Printer J. (2001) Cultur al eutrophication of a Ugandan highland crater lake: A twenty-five year comparison of limnological parameters. Verhandlungen der Internationale Vere inigung fr theoretische und angewandte Limnologie 17 272-282. Crisman T. L. and Streever W. J. (1996) Th e legacy and future of tropical limnology. In: Schemer, F. and Boland, K. T. (Eds.), Persp ectives in Tropical Limnology. SPB Academic Publishing, Amsterdam: 27-42. Dadzie S. and Wangila B. C. C. (1980) Repr oductive biology, length-we ight relationship and relative condition of pond raised Tilapia zillii (Gervais). Journal of Fish Biology 17 ,243-253. De la Cruz-Aguero G. (1993) A preliminar mode l of Maninga lagoon, Veracruz, Mexico. In: V. Christensen and D. Pauly, Trophi c modles of Aquatic Ecosys tems. ICLARM Conf. Proc. 26 193-196. Demeke Admassau (1994) Maturity, batch fecu ndity, brood size and sex ratio of tilapia ( Oreochromis niloticus L .) in Lake Awassa. SINET: Ethiopian J ournal of Science 17 53-69. Dennis T. E., MavAvoy S. E, Steg M. B. and Bo lger A. J. (1995) The association of water chemistry variables and fish condition in streams of Shenandoah National Park (USA). Water, Air and Soil Pollution 85 365-370. De Silva S. S (1985) Body condition and nutritional ecology of Oreochromis mossambicus (Pisces: Cichlidae) populations of man-made lakes in Srilanka. Journal of Fish Biology 27 621-33. De Silva S. S. (1986) Reproductive Biology of Oreochromis mossambicus populations of the man-made lake in Sri Lanka: a comparative study. Aquaculture and Fisheries Management 17 31-47. De Silva S. S. (1991) Grow th rate and the role of Oreochromis mossambicus in the fishery of a tropical, upland, deep reservoir in Sri Lanka. Fisheries Research 12:125-138. De Silva, S. S., Subasinghe R., Bartley D. and Lo wther A. (2004) Tilapias as exotics in the Asia-Pacific: A review. FAO Fi sheries Technical Paper, 453 pp.

PAGE 173

173 DiCenzo V. J, Maceina M. J. and Reeves W. C. (1995) Factors related to growth and condition of the Alabama subspecies of spotted bass in reservoirs. North American Journal of Fisheries Management 15 794-798. Dillon P. J. and Rigler F. H. (1974) The phosphorus-chlorophyll relationship in Lakes. Limnology and Oceanography 19, 767-773. Downing J. A., Plante C. and Lalonde S. ( 1990) Fish production corr elated with primary productivity, not the morphoedaphic index. Canadian Journal of Fisheries and Aquatic Sciences 47 1929-1936. Duarte C. M. and Alcaraz M. (1989) To produce many small or few large eggs: a size independent tactic of fish. Oecologia 80 401-404. Dunne T. (1979) Sediment yield a nd land use in tropi cal catchments. Journal of Hydrology 42 281-300. Dupochelle F. and Panfili J. (1998) Va riation in age and size at matu rity of female Nile tilapia, Oreochromis niloticus populations from man-made la kes of Cote d Ivoire. Environmental Biology of Fishes 52 453-465. Dupochelle F., Cecchi P., Corbin C., Nunez J. and Legendre M. (1999) Spawning season variations of female Nile tilapia, Oreochromis niloticus from man-made lakes of Cote D Ivoire. Environmental Biology of Fishes 56 375-387. Dupochelle F., Cecchi P., Corbin C., Nunez J ., and Legendre M. (2000) Variation in batch fecundity and egg size in female Nile Tilapia, Oreochromis niloticus from man-made lakes of Cote D Ivoire. Environmental Biology of Fishes 57 155-170. Edley M. T. and Law R. (1988) Evolution of life hi stories and yields in ex perimental populations of Daphnia magna Biological Journal of the Linnean Society 34 309-26. Efitre J (2007) Life history vari ation in tilapia populat ions within the crater lakes of western Uganda: the role of size-selective predation. Ph.D. dissertation, University of Florida, Gainesville, USA. Egger B., Meekan M., Salzburger W., Mwape L., Shapola R. and Sturmbauer C. (2004) Validation of the periodicity of increment formation in the otoliths of a cichlid fish from Lake Tanganyika, East Africa. Journal of Fish Biology 64 1272-1284. El Bolock A. and Koura R. (1960) Age, growth and breeding season of Tilapia zillii (Gerv.) in Egyptian experimental fish ponds. Notes and Memoirs of Alexandria Institute of Hydrobiology 49 1-36. El Zarka S. (1962) Tilapia Fisher ies investigation in Egyptian lake s. III. Maturity, spawning and sex ratio of Tilapia zillii (Gervais) in Lake Quaran. Notes and memoirs Alexandria Institute of Hydrobiology 67, 1-24.

PAGE 174

174 Elizabeth K., Getachew T., Taylor W. D. and G. M. Zinabu G. M. (1992) Eutrophication of Lake Hyaq in the Ethopian highlands. Journal of Plankton Research 14 1473-482. Elizabeth K., Zinabu G. M. and Ahlgren I. (1994). The Ethiopian Rift Valley lakes: chemical characteristics of a salin ity-alkalinity series. Hydrobiologia 288 1-12. Ernande B., Dieckmann U. and Heino M. (2004) Adaptive changes in harvested populations: plasticity and evolution of ag e and size at maturation. Proceedi ngs of the Royal Society of London Series B-Biological Sciences 271 415-423. Fagade S. O. (1974) Age determination in Tilapia melanotheron (Rppel) in the Lagos Lagoon, Lagos, Nigeria, with a discussion of the envi ronmental and physiologi cal basis of growth markings in the tropics. Pp. 71-77 In: T. B. Bagenal (editor) The Ageing of Fish. Unwin Brothers Ltd, Old Woking, Surrey, England. Fagade S. O., Adebisi A. A. and Ata nda A. N. (1984). The breeding cycle of Sarotherodon galilaeus in the IITA lake, Ibadan Nigeria. Archiv fr Hydrobiologie 100 493-500. FAO (1985) Report of the sixth session of the FAO panel of experts on forest gene resources, including forest genetic resource s priorities. FAO, Rome, Italy. FAO (2005) Global Forest Resources Assessmen t 2005. Progress towards sustainable forest management. FAO Forestry Paper 147, Rome, Italy, 320p. Ferreira B. P. and Russ G. R. (1994) Age validati on and estimation of grow th rate of the coral trout, Plectropomus leopardus (Lacepede 1802) from Lizard Island, Northern Great Barrier Reef. Fishery Bulletin 92 46-57. Fisher T. R., Melack J.M., Grobbelaar J.U. and Howarth R.W. (1995) Nutrient limitation of phytoplankton and eutrophication of inland, estuar ine, and marine waters. In: Tiessen H. (ed), Phosphorus in the Gl obal Environment. John Wiley and Sons Ltd., New York, NY, pp. 301 322. Fortin R., Dumont P. and Guentte S. (1990) De terminants of growth and body condition of lake sturgeon ( Acipenser fulvenscens ). Canadian Journal of Fisher ies and Aquatic Sciences 53 1150-1156. Fowler A. J. and Doherty P. J. (1992) Validation of annual growth increments in the otoliths of two species of damselfish from the southern Great Barrier Reef. Australian Journal of Marine and Freshwater Resources 43 1057-1068. Fowler A. J. and Short D. A. (1998) Validation of age determination from otoliths of the King George whiting Sillaginodes punctata (Perciformes). Marine Biology 130 577-587. Fowler A. J. (1995) Annulus formation in otoliths of coral reef fish-a review. Pp. 46-63.In: D. H. Secor, J. M. Dean, and S. E. Campana (edito rs). Recent Developments in Fish Otolith Research, vol. 19. The Bella W Baru ch Library in Marine Sciences.

PAGE 175

175 Francis M. P., Williams M. W., Pryce A. C., Po llard S., and Scott S. G. (1992) Uncoupling of otolith and somatic growth in Pargus auratus (Sparidae). Fishery Bulletin 91 159-164. Froese R. and Pauly D. (2006) Ed. Tilapia. Fishbase Fryer G. and Iles T.D. (1961) Observa tions on the biology of the cichlid fish O. variabilis Boulenger, in the north waters of Lake Victoria (East Africa). Revue de Zoologie et de Botanique Africaines 64 1-33. Fryer G. and Iles T. D. (1969) Alternative routes to evolutionary succe ss as exhibited by African cichlid fishes of the genus Tilapia and the species flocks of the Great lakes, Evolution 23 359-369. Fryer G. and Iles T. D. (1972) The Cichlid Fish es of the Great Lakes of Africa. Oliver and Boyd, London 641 pp. Gadgil M. and Bossert W. H. (1970) Life hist orical consequences of natural selection. American Naturalist 104 1-24. Gauldie R. W. (1988) Function, form and timekeeping properties of fish otoliths. Comparative Biochemistry and Physiology 91 395-402. Graham M. (1929) The Victoria Nyanza and its fisheries. Crown Agents for the Colonies, London. Getabu A. (1992) Growth parame ters and total mortality in Oreochromis niloticus (Linnaeus) from Nyanza Gulf, Lake Victoria. Hydrobiologia 232 91-97. Green J. (1972) Ecological studies on crater lakes of West Cameroon. Zooplankton of Barombi Mbo, Mboandong, Lake Kotto and Lake Soden. Journal of Zoology London 166 283-301. Green J. (1986) Zooplankton associations in some Ethiopian crater lakes. Freshwater Biology 16, 495-499 Green J., Corbet S.A., and Betney E. (1973) Ecological studies on cr ater lakes in West Cameroon. The blood of endemic cichlids in Ba rombi Mbo in relation to stratification and their feeding habits. Journal of Zoology London 170 299-308. Green J., Corbet S. A., and Betney E. (1974) Ecological studies on cr ater lakes in West Cameroon. Debundasha Lake. Journal of Zoology London 173, 199-223. Green J. and Mengestou S. (1991) Specific diversity and community structure of Rotifera in a salinity series of Ethiopian inland waters. Hydrobiologia 209 95-106. Greenwood P. H. (1966) The Fishes of Ug anda. The Uganda Society, Kampala.

PAGE 176

176 Gulland J. A. (1983) Fish stock assessment: a ma nual of basic methods. FAO/Wiley Series, Food and Agriculture Organization, Rome. Gulland J. A. (Editor) 1988. Fish Population Dynamics. Second Edition. John Wiley and Sons, London. 422 pp. Gwahaba J. J. (1973) Effects of fishing on the Tilapia nilotica population in L. George, Uganda, over the past twenty years. East African Wildlife Journal 11 317-328. Hall S. J. (1999) The Effects of Fishing on Ma rine Ecosystems and Communities. Blackwell Science, Oxford, UK. Hanson J.M. and Leggett W. C. (1983) Empiri cal prediction of fish biomass and field. Canadian Journal of Fisheries and Aquatic Sciences 39 257 263. Hansson L. H. (1992) The role of food chain co mposition and nutrient ava ilability in shaping algal biomass development. Ecology 73 241-247. Haugen T. O. and Vollestad L. A. (2001) A cen tury of life-history evolution in grayling. Genetica 112 475-491. Hecky R. E. (1993) The eutroph ication of Lake Victoria. Verhandlungender Internationale Vereinigung fr Theoretische und Angewandte Limnologie 25 39-48. Hecht T. and Zway P. (1984) On the stunted mozambique tilapia Sarotherodon mossambicus of the Matiovila Hot Spring, Kr uger National Park. Koedoe 27 25-38. Helster T. E. and Almeida F. P (1997) Density-dep endent growth and sexual maturity of silver hake in the northwest Atlantic. Journal of Fish Biology 51 607-623. Hilborn, R. and Walters C. J. (1992) Quantitative fisheries stock assessment, Choice, Dynamics and Uncertainty. Chapman and Hall, New-York, London. 570 pp. Hodgkiss J. and Man H. (1977) Age composition, growth, and body condition of the introduced Sarotherodon mossambicus (Cichlidae) in Plover C ove reservoir, Hong Kong. Environmental Biology of Fishes 2 35-44. Huszar V. L. M, Caraco N. A., Roland F. a nd Cole J. (2006) Nutrient -chlorophyll relationships in tropical-subtropical lakes: do temperate models fit? Biogeochemistry 79 239-250. Hutchings J. A. (1993) Adaptive life histories affected by age specific apparent survival and growth rate. Ecology 74 673-684. Hutchinson G. E. (1957) A Treatise in Lim nology. John Wiley and Sons, Inc., New York. Hyndes G. A., Loneragan N. R. and Potter I. C. (1992) Influence of sectioning otoliths on marginal increment trends and age and growth estimates for the flathead Platycephalus speculator Fisheries Bulletin 90 276-284.

PAGE 177

177 Jalabert, B. and Zohar Y. (1982) Reproduction physiology in cich lid fishes, with particular reference to Tilapia and Sarotherodon. P p. 129-140. In: R.S.V. Pullin and R.H. LoweMcConnell (eds.) The biology and culture of tilapias. ICLARM Conf. Proc. 7. 432p. James N. P. E. (1989) A life-hist ory approach to the biology of Oreochromis mossambicus in the Eastern Cape, South Africa. Ma ster of Science Thesis, Rhodes University, Grahamstown, 206 p. James N. P. E and Bruton M. N. (1992) Altern ative life-history trai ts associated with reproduction in Oreochromis mossambicus (Pisces: Cichlidae) in small water bodies of the Eastern Cape, South Africa. Environmental Biology of. Fishes 34 379-392. Jennings S. and Kaiser M. J. (1998) The e ffects of fishing on marine ecosystems. Advances in Marine Biology 34, 201-352. Jeppesen E., Sndergaard M, Mazzeo N., Meerof M ., Branco C. W. C, Huszar V. L. M. and Scasso F. (2005) Lake restora tion and biomanipulation in te mperate lakes: relevance for subtropical and tropical la kes. In Reddy V. (ed), Tropi cal Eutrophic Lakes: Their Restoration and Management. Oxford and I.B.H Publ. New Hampshire: 331-259. Jones J. R. and Hoyer M. V. (1982) Sportf ish harvest predicted by summer chlorophyll a concentration in midwestern lakes and reservoirs. Transactions of the American Fisheries Society 111 176-7. Jones C. M. (1992) Development and applica tion of the otolith increment technique. In : D.K. Stevenson and S.E. Campana (editors) Otolith Microstructure Examination and Analysis. Canadian Special Publication of Fisheries and Aquatic Sciences 117 :1-11. Jones J. R. B., Perkins D., Witt A., Kaiser M. S., Thamasara S., Siriworakul M., and Banyasut P., (2000). Limnological characteristics of some reservoirs in Thailand. Verhandlungen der Internationale Vereinigung fr theoretische und angewandte Limnologie 27 2158-2166. Kalff J. (2002) Limnology. Pren tice Hall, Upper Saddle, NJ. Keast A. and Eadie J. M. (1985) Growth compensa tion in year-0 large mouth bass: The influence of diet. Transactions of the American Fisheries Society 114 204-213. Kilham P. and Kilham S.S (1990) Endless summer: internal loading proc esses dominate nutrient cycling in tropical lakes. Freshwater Biology 23 379-389. King R. P. and Etim L. (2004) Repr oduction, growth, mortality and yield of Tilapia mariae Bouleger 1899 (Cichlidae) in a Nige rian rainforest wetland stream. Journal of Applied Ichthyology 20 502-510. Kirk J. T. (1986) Optical limnology a manife sto. In Limnology Australia, (Eds. P. De Deckker and W.D. Wlilliams) 33-36 (CSIRO: Melbourne. Dr. Junk Publishers: Dordrecht).

PAGE 178

178 Kizito Y. S., Nauwerck A., Chapman L. J., and Koste W. (1993) A limnological survey of some western Uganda crater lakes. Limnologica 23 335-347. Kizito Y. S. and Nauwerck A. (1995) Temporal a nd vertical distribution of planktonic rotifers in a meromictic crater lake, Lake Nyaherya (Western Uganda). Hydrobiologia :, 313/314 303-312. KizitoY. S. and A. Nauwerck A. (1996). Temporal and vertical distributi on of planktonic rotifers in Lake Nkuruba, Western Uganda. Limnologica 26, 263-273. Kling G. W. (1987) Seasonal mixing and the cata strophic degassing in trop ical lakes, Cameroon, West Africa. Science 237 1022-1024. Kling G. W. (1988) Comparative transparency, dept h of mixing, and stability of stratification in lakes of Cameroon, West Africa. Limnology and Oceanography 33 27-40. Kolding J. (1993) Population dynamics and life-history styles of Nile tilapia, Oreochromis niloticus in Fergusons Gulf, Lake Turkana, Kenya. Environmental Biology of Fishes 37, 25-46. Kolher N. Casy J. and Turner P. (1995) Length-we ight relationships of 13 species of sharks from the western North Atlantic. Fisheries Bulletin 93 412-418. Langdale-Brown, I., H. A. Osmaston and Wilson J.G. (1964) The Vegetation of Uganda and its Bearing on Land-use. Uganda G overnment Printer, Entebbe. Law R. (1979) Optimal life histor ies under age-specific predation. American Naturalist 114 399-417. Law R. (2000) Fishing, selec tion, and phenotypic evolution. ICES Journal of Marine Science 57 659-669. Le Cren E. D. (1951) The lengt h-weight relationship and seas onal cycle in gonad weight and condition in Perch ( Perca fluviatilis ). Journal of Animal Ecology 20 201-219. Legendre M. and Ecountin J. M. (1989) Suitabi lity of brackish water tilapia species from the Ivory Coast for lagoon aqu aculture. 1-Reproduction. Aquatic Living Resources 2 71-79. Legendre, M. and Ecountin J.M. (1996) Aspects of the reproductive strategy of Sarotherodon Melanotheron : a comparison between a natural populat ion (Ebrie Lagoon, Cote d Ivoire) and different cultured populations. Pp. 320 325. In R. S. V Pullin and J. Lazard, M. Legendre, J. B. Amon Kothias and D. Pauly (e ds.) The Third Inte rnational Symposium on Tilapia in Aquaculture, ICLA RM Conf. Proc. 41. Manila. Leggett W.C and Power G. (1969) Differences be tween two populations of landlocked Atlantic salmon (Salmon salar) in Newfoundland. Journal of the Fisheries Resources Board of Canada 26 1585-1596.

PAGE 179

179 Lvque C. (1977) Biodiversity dynamics and cons ervation: The freshwater fishes of tropical Africa.Cambridge, UK, Cambridge University Press. Lewis W. M. Jr. (1996) Comparis ons of phytoplankton biomass in te mperate and tropical lakes. Limnology and Oceanography 35 1838-1845. Lewis W. M. Jr. (2002) Causes for the high freq uency of nitrogen limitation in tropical lakes. Verhandlungen der Internationale Vere inigung fr theoretische und angewandte Limnologie 28, 210-213. Longhurst A. R. and Pauly D. (1987) Ecology of Tropical Oceans. Academic Press, San Diego, California. 407 pp. Lowe-McConnell R. H. (1958) Observations on the biology of Tilapia niloticus Linne, in East African waters. Revue de Zoologie et de Botanique Africaines 57 129-170. Lowe-McConnell R. H. (1982) Tilapias in fish co mmunities. Pp.83. In R. S. V. Pullin and R. H. Lowe-McConnell (editors). The Bi ology and Culture of Tilapias, ICLARM Conference Proceedings 7, Manila, Philippines. Lowe-McConnell R. H. (1987) Ecological Studies in Tropical Fish Communities. Cambridge University Press, Cambridge. Lowe-McConnell, R. H., Crul R. C. M and Roest F.C. (editors) (1992) Proceedings of the International Symposium on Resource Use a nd Conservation of Afri can Great Lakes, 29 November 2 December 1989, Bujumbura, SIL Mittelungen. No. 23:128 pp. Machena C. and Fair P. (1986) Co mparison of fish yields from pr ediction models between Lakes Tanganyika and Kariba. Hydrobiologia 137 29-32. MacIntyre S. and Melack J. M. (1982) Meromi xis in an equatorial African soda lake. Limnology and Oceanography 27 595-609. Marlier G. (1973) Limnology of the Congo and Amazon rivers. In : B.J. Meggers, S.A. Ayensu and W.D. Duckworth (Eds.) Tropical forest ecosystems in Africa and South America: A comparative review, Washington, D.C ., Smithsonian Institution Press, pp. 223-38 Martin C. (1991) The Rainforests of Western Africa: Ecology, Threat s, and Conservation. Basel: Borkhauser Verlag. Mazumder A., Taylor W. D., McQueen D. J., and Lean D. R. S. (1990) Effects of fish and plankton on lake temperat ure and mixing depth. Science 247 312-315. Mazumder A. (1994) Phosphorus-chlorophyll rela tionships under contrasting herbivory and thermal stratification: pr edictions and patterns. Canadian Journal of Fisheries and Aquatic Sciences 51 390-400.

PAGE 180

180 Mazumder A. and Lean D. R. S. (1994) Consum er-dependent responses of lake ecosystems to nutrient loading. Journal of Plankton Research 16, 1567-1580. Mazumder A. and Havens K. E. (1998) Nu trient-chlorophyll-Secchi relationships under contrasting grazer communities of temperate versus subtropical lakes. Canadian Journal of Fisheries and Aquatic Sciences 55 1652-1662. McConnell W. J., Lewis S., and J. E. Olsen. (197 7) Gross photosynthesis as an estimator of potential fish production. Tran sactions of the American. Fisheries Society 106 417-423. Melack J. M. and Kilham P. (1972) Lake Mahe ga: A mesothermic, sulphato-chloride lake in Western Uganda. African Journal of Tropica l Hydrobiology and Fisheries 2 141-150. Melack J. M. (1978) Limnology and dynamics of phytoplankton in Equatorial African lakes. Ph.D. Dissertation, Duke Univer sity, Durham, North Carolina. Melack J. M. (1978) Morphometric, physical and chemical features of the volcanic crater lakes of Western Uganda. Archiv f r Hydrobiologie 84 430-453. Miller L. M. and Kapuscinski A. R. (1994) Estima tion of selection differen tials from fish scales: a step towards evaluating genetic alterati on of fish size in exploited populations. Canadian Journal of Fisheries and Aquatic Sciences 51 774-783. Morales-Nin B. and Ralston S. (1990) Age and growth of Lutjanus kasmira (Forsskal) in Hawaiian waters. Journal of Fish Biology 36 191-203. Moreau J. and De Silva S. S. (1991) Predictive fi sh yield models for lakes and reservoirs of the Philippines, Sri Lanka and Thailand. FAO Fisheries Technical Paper No. 319. FAO, Rome. 42p. Moutopoulos D. K. and Stergiou K. I. (2002) Length-weight rela tionship of fishes from the Aegean Sea (Greece). Journal of Applied Ichthyology 18 200-203. Mugidde R. (1993) The increase in phytoplankt on primary productivity and biomass in Lake Victoria (Uganda). Verhandlungen der Internationale Ve reinigung fr Theoretische und Angewandte Limnologie 25 846-849. Munro J. L. (1983) Caribbean coral reef fisheries. International Center for Living Aquatic Resource Management Study Reviews 7 1-276. Muhoozi L. I. (2002) Exploitation and management of artisanal fisheries of the Ugandan waters of Lake Victoria, Ph.D. dissertation, University of Hull, UK. Negassa A. and Getahun A. (2003) Breeding s eason, length-weight relationship and condition factor of introduced fish, Tilapia zillii Gervais 1948 (Pisces: Cichlidae) in Lake Zwai, Ethiopia. SINET: Ethiopian J ournal of Science 26 115-122. Nekrasov V.V. (1980) The causes of annul us formation in tropical fishes. Hydrobiological Journal 2 35.

PAGE 181

181 NEMA (1997) Kabarole distri ct environment profile, National Environment Management Authority Report. Newman S. J., Williams D. M. and Russ G. R. (1996) Age validation, growth and mortality estimates of the tropical snappers Lutjanus adetii and L. quinquelineatus from the Central Great Barrier Reef. Australia Journal of Marine and Freshwater Resources 47 575-584. Newman S. J., Cappo M., Williams D. M. B. (2000) Age, growth, mortality rates and corresponding yield estimates using otol iths of the tropi cal red snappers, Lutjanus erythropterus L. malabaricus and L. sebae from the central Great Barrier Reef. Fisheries Research 48, 1-14. Nikolsky G. V. (1963) The Ecology of Fishes. Academic Press, London,UK. Noakes D. I. G. and Balon E. K. (1982) Life histor ies of tilapias: an evol utionary perspective. pp. 66-82. In : Pullin R.S.V. and Lowe-McConnell R. H. (editors). The Biology and Culture of Tilapias, ICLARM Conference Proceedings 7, Manila. Nrnberg G. K. (1996) Trophic states of clear an d colored, softand hardwater lakes with special consideration of nutrients, a noxia, phytoplankton and fish. Lakes and Reservoirs Management 12 432-447. Ogutu Ohwayo R. (1990).The decline of the nati vefishes of LakesVictoria and Kyoga (East Africa) and the impact of introduced species, especially the Nileperch, Lates niloticus, and Niletilapia, Oreochromis niloticus. Environmental Biology of Fishes 27 81-96. Ogutu-Ohwayo R., Hecky R. E, Cohen A. S., and Kaufman L. (1997) Human impacts on the African Great Lakes. Envir onmental Biology of Fishes 50 117-131. Ogutu-Ohwayo R. (2000) The status of Aquatic Biodiversity in Uganda: Its conservation and sustainable use. NARO, Uganda. Oscoz J., Campos F. and Escala M. C. (2005) Wei ght-length relationships of some fish species of the Iberian Peninsula. Journal of Applied Ichthyology 21 73-74. Osmaston H. A.(1967) Plant fossils in volcanic tuffs near Ruwenzori. Palaecology of Africa 2 25-6. Pace M. L. (1984) Zooplankton community structur e but not biomass influences the phosphorus chlorophyll a relationship. Canadian Journal of Fisheries and Aquatic Sciences 41 1089-1096. Panella G. (1980) Growth patterns in fish sagittae.Pp. 519-560. In: D.C. Rhoads and R.A. Lutz (editors). Skeletal growth of aquatic organisms. Plenam Press, New York, NY. Panfili J. and Toms J. (2001) Validation of age estimation and back-calculation of fish length based on otolith microstructures in tilapias (Pisces, Cichlidae). Fisheries Bulletin 99 139-150.

PAGE 182

182 Parker H. H and Johnson L. (1991) Population st ructure, ecological segr egation and reproduction in non-anadromous Arctic Charr, Salvenilus alpinus (L), in four unexploited lakes in the Canadian high Arctic. Journal of Fish Biology 38 123-147. Pauly D. (1980) Length-converted catch curves and the seasonal growth of fishes. Fishbyte 8 :33-38. Pauly D. (1990) Length-converted catch curves and the seasonal growth of fishes. ICLARM Fishbyte 8 33-38. Pauly D. (1993) Fishbyte section. Editorial. Naga ICLARM Quarterly 16 26. Peters H. M. (1963) Batch fecundity, egg weight and oocyte development in tilapias (Cichlidae, Teleosti). Translated and edited by Pa uly D., 1983. ICLARM, Translation 2, 28 pp. Peters R. H. (1986) The role of prediction in limnology. Limnology and Oceanography 31 1143-1159. Petrakis G. and Stergiou K. I. (1995) Weight-l ength relationships for 33 fish species in Greek waters. Fisheries Research 21 465-469. Pomeroy D. E. and Seavy N. E. (2003) Survey of Great Crested Grebes Podicepus cristatus and other water birds on the Kasenda cluster of crater lakes in western Uganda. Journal of East African Natural History 92, 49-62. Pope K.L. and Willis D. W. (1996) Seasonal infl uences on freshwater fisheries sampling data. Reviews in Fisheries Science 4 57-73. Post, J. R. and McQueen D. J. (1994) Variabil ity in first-year growth of yellow perch ( Perca fluviatilis ): Predictions from a simple mode l, observations and an experiment. Canadian Journal of Fisheries and Aquatic Sciences 51 2501-2512. Prosser M.V., Wood R. B., and Baxter R. M. (1968) The Bishoftu crater lakes: a bathymetric and chemical study. Archiv f r Hydrobiologie 65 309-324. Rask M. (1983) Differences in growth of perch ( Perca fluviatilis L) in two small forest lakes. Hydrobiologia 101 139-144. Reardon E. E and Chapman L. J. (2007) Reproduc tive seasonality in a swamp-locked African cichlid. Ecology of Freshwater Fish (In Press) Reznik D. A. and Bryga H. (1987) Life histor y evolution in guppies: phenotypic and genotypic changes in an introduction experiment. Evolution 41 1370-1385. Reznik D. A., Bryga H. and Endler J. A. (1990) Experimentally induced life-history evolution in a natural population. Nature 346, 357-359.

PAGE 183

183 Reznik D. A. and Yang A. P. (1993) The influe nce of fluctuating resources on life history: Patterns of allocation and plasticity in female guppies. Ecology 74 2011 2019. Ribbink A. J. (1990) Alternative life history styles of some African cichlid fishes. Environmental Biology of Fishes 28 87-100. Ricker, W. E. (editor) (1971) Method for Assessment of Fish Pr oduction in Freshwaters. IBP Handbook No. 3. Blackwell Scientific Publications, Oxford, UK. Ricker W. E. (1975) Computation a nd interpretation of biological sta tistics of fish parameters in the von Bertalanffy equation. Canadian Journal of Fish eries and Aquatic Sciences 47 1416-1426. Ricker W. E. (1975) Computation and Interpretation of th e Biological Statistics of Fish Populations Bulletin of the Fisheries Research Board of Canada Number 191. Ottawa: Department of the Environment, Fisheries and Marine Services. Ricker W. E. (1981) Changes in average size and average age of Pacific Salmon. Canadian Journal of Fisheries and Aquatic Sciences 38 1636-1656. Rikhter V. A. and Efanov V. N. (1976). On one of the approach es to estimation of natural mortality of fish populations. ICNAF Research Document, 79/VI/8, 12p. Ridgeway L. L and Chapleau F. (1994) Study of a stunted populati on of yellow perch ( Perca fluviatilis ) in a monospecific lake in Gatineau Park, Quebec. Canadian Journal of Zoology 72 1576-1582. Rochet M. J., Cornillon P. A., Sabatier R. a nd Pontier D. (2000) Comparative analysis of phylogenetic and fishing effects in life history patterns of teleost fishes. Oikos 91 255-270. Roff D. A. (1984) An allocation model of growth and reproduction in fish. Canadian Journal of Fisheries and Aquatic Sciences 40 1395-1404. Roff, D. A. (1992) The Evolution of Life Hi stories. New York, L ondon: Chapman and Hall. Salas H. J. and Martino P. (1991) A simplifie d phosphorus trophic state model for warm-water tropical lakes. Water Research 25 341-350. Sandheinrich M. B. and Hubert W. A. (1984) In traspecific resource partitioning by yellow perch, Perca flavescens in a stratified lake Canandian Journal of Fisheries and Aquatic Sciences 41, 1745-1752. Sarnelle O. (1992) Nutrient enrichment and grazer effects on phytopl ankton in lakes. Ecology 73 551-560.

PAGE 184

184 Sarnelle O., Cooper S. D., Wiseman S. and Ma vuti K. M. (1998) The relationship between nutrients and trophic-level bioma ss in turbid tropical ponds. Freshwater Biology 40 65-75. Sass G. G., Hewett S. W., Beard Jr. T. D., Fayram A. H., and Kitchell J. F., (2004) The role of density dependence in growth patterns of ceded te rritory walleye populations on northern Wisconsin: Effects of changing management regimes. North American Journal of Fisheries Management 24 1262-1278. Schindler D. E., Hoggson J. R. and Kitchell J. F. (1997) Density-dependent changes in foraging specialization of large-mouth bass. Oecologia 110 592-600. Secor D. H., Dean J. M. and Campana S. E. (1995) Recent Developments in Fish Otolith Research. Proceedings of the First International Symposium. 762 Pp. Belle W. Baruch Library in Marine Science, University of South Carolina Press, Columbia, SC. USA. Siddiqui A. G. (1977) Lake Naivasha (East Afri ca, Kenya) fishery and it s management together with a note on the food habits of fishes. Biological Conservation 12 217-228. Sinclair A. F., Swain D. P. and Hanson J. M. (2 002) Disentangling the e ffects of size-selective mortality, density, and temper ature on length-at-age. Canadian Journal of Fisheries and Aquatic Sciencces 59 372-382. Smith K. A. and Deguara K. (2003) Formation a nd annual periodicity of opaque zones in sagittal otoliths of Mugil cephalus (Pisces: Mugilidae). Marine and Freshwater Research 54 57-67. Smith V. H. and Shapiro J. (1981) Chlorophyllp hosphorus relations in individual lakes. Their importance to lake restoration strategies. Environmental Science and Technology 15 444-451. Smith V. H. (1982) The nitrogen and phosphorus dependence of algal biomass in lakes: an empirical and theoretical analysis. Limnology and Oceanography 27 1101-1112. Smith V. H., 1990. Phytoplankton responses to eutr ophication in inland waters. In Akatsuka, I. (ed), Pages 231-249 in I. Akatsuka, editor. Introduction to Applied Phycology. SPB Academic Publishing, The Ha gue, The Netherlands: 231-249. Sparre P. and Venema S. C. (1998) Introduction to tropical fish stock assessment. Part 1. Manual. FAO Fisheries Technical Paper. No. 306/1, Rev.2. Rome, FAO 407pp. Steams S. C. (1976) Life-history t actics: a review of the ideas. Quarternly Review of Biology 51, 347. Stearns S. C. (1992) The evolution of life hist ories. Oxford University Press, Oxford, UK. Stewart K. M. (1988) Change in condition and maturity of the Oreochromis niloticus population of Fergusons Gulf, Lake Turkana, Kenya. Journal of fish Biology 33 181-188.

PAGE 185

185 Straskraba M. (1980) The effects of physical variables on freshwat er production: Analyses based on models. In The Functioning of Freshwater Ecos ystems. E. D. and R. H. McConnell Eds., 13-31. Cambridge University Press. Cambridge, England. Talling J. F. and Lemoalle J. (1998) Ecological Dynamics of Tropical Inland Waters. Cambridge University Press, Cambridge, UK. Taylor D. Merchant R. A. and Robertshaw P. (1999) A sediment-based history of medium altitude forest in central Africa: A reco rd from Kabata swamp, Ndale volcanic field, Uganda. Journal of Ecology 87 303-315. Teugels G. G. and Thys van den Audenaerde D. F. E. (1991) Tilapia. In: J. Daget, J. P. Gosse, G. G. Teugels and D. F. E. Thys van den A udenaerde (eds.) Check-list of the freshwater fishes of Africa (CLOFFA). ISNB, Brussels ; MRAC, Tervuren; and ORSTOM, Paris. Trewavas E. (1983) Tilapiin e fishes of the genera Sarotherodon Oreochromis and Danakilia British Museum of Natura l History, London, UK. 583 p. Trippel E. A. (1995) Age at maturity as a stress indicator in fisheries. BioScience 45 759-771. Uganda Department of Lands and Surveys ( 1965) Fort Portal-soils sheet NA-36-13, 1:250,000. Department of Lands and Surveys Uganda, Entebbe, Uganda. Umaa V. G. and Jimnez G. (1995) The basic lim nology of a low altitude tropical crater lake: Cerro Chato, Costa Rica. Revista de Biologia Tropical 43 131-138. UNECIA (2002) Lake Victoria Fi sheries Research Project Phas e II, 1997-2001. Final Report, UNECIA Ltd. 84 pp. Vzquez G., Favila M., Madrigal R., Carlos D ., Baltans A. and Miguel A. B. (2004) Limnology of Crater Lakes in Los Tuxtlas, Mexico. Hydrobiologia 523 59-70. Vila-Gispert A., Zamora L. andMoreno-Amich R. (2000) Use of the condition of Mediterranean barbel ( Barbus meridionalis ) to assess habitat quality in stream ecosystems. Archiv fr Hydrobiologie 148 135-145. Vila-Gispert A. and Moreno-Amich R. (2001) Ma ss-length relationship of Mediterranean barbel as an indicator of environmen tal status in south-west European stream ecosystems. Journal of Fish Biology 59 824-832. Visser S. A. (1965) Conditions affecting lif e in a stratified tropical crater lake. West African Journal of Biology and Applied Chemistry 8 48-52. Warburton K. (1978) Age and growth determinati on in marine cat-fish using an otolith check technique. Journal of Fish Biology 13 429-434. Walsh M. R., Munch S. B., Chiba S. and Conover D. O. (2006). Maladaptive changes in multiple traits caused by fishing: imped iments to population recovery. Ecology Letters 9 142-148.

PAGE 186

186 Wege G. J. and Anderson R. O. (1978) Relative weight (Wr): a new index of condition for largemouth bass. In: G. D. Novinger and J. G. Dillard (Eds) New Approaches to the Management of Small Impoundments Bethesda, USA: American Fisheries Society Special Publication Number 5 79-91. Welcomme R. L. (1970) Studies on the effects of abnormally high water levels on the ecology of fish in certain shallow regions of L. Victoria. Journal of Zoology 160 405-436. Welcomme R. L. (1967) Observations on the biol ogy of the introduced speci es of Tilapia in the Lake Victoria. Revue de Zoologie et de Botanique Africaines 76 249-279. Wetzel R. G. (1983) Limnology, 2nd ed. Saunders College Pu blishers, Philadelphia. Wetzel R. G. (2001) Limnology: Lake and Ri ver Ecosystems. Academic Press, San Diego. Winemiller K. O. and Rose K. A. (1992). Patter ns of life-history diversification in North American fishes: implications for population regulation. Canadian Journal of Fisheries and Aquatic Sciences 49, 2196-2218. Winemiller K. O. (1992) Life-history strategies and the effectiveness of sexual selection. Oikos 63 318-327. Wood R. B., Prosser M.V. and Baxter R. M. (1976) The seasonal pattern of thermal characteristics of four of the Bi shoftu crater lakes, Ethiopia. Freshwater Biology 6 519-530. Wood R. B., Baxter R. M. and Prosser M. V. (1984) Seasonal and co mparative aspects of chemical stratification in some tr opical crater lakes, Ethiopia. Freshwater Biology 14 551-573. Wootton R. J. (1984) Introduction: strategies and tactics in fish repr oduction. Pp.1-12. In: G.W Potts and R.J. Wooton (editors). Fish Reproduction: Strategies and Tactics. Academic Press, London. Wooton R. J. (1990) Ecology of Teleost Fishes Chapman and Hall, New York. Wootton R. J. (1992) Fish ecology. Chapman and Hall, New York. Wootton R. J., Elvira B. and Baker J.A. (2000) Life history evoluti on, biology and conservation of stream fish. Ecology of Freshwater Fishes 9 90-91. World Resources Institute (1994) World Re sources 1994-1995: A Guide to the Global Environment, World Resources Institute, Washington. Yosef T. G. and Casselman J. M. (1995) A proc edure for increasing the precision of otolith age determination of tropical fish by differenti ating biannual recruitment. Pp. 247-269. In Recent Developments in Fish Otolith Research(Secor, D. H., Dean, J. M. and Campana, S. E., editors), University of South Carolina Press.

PAGE 187

187 Zinabu G. M. (1994) Long-term changes in indices of chemical and prod uctive status of a group of tropical Ethiopian lakes with differing exposure to human influence. Archiv fr Hydrobiologie 132 115-125.

PAGE 188

188 BIOGRAPHICAL SKETCH Jackson Efitre was born in Arua District, Ug anda. He went to St. Josephs College Ombaci for his ordinary and advanced level s econdary education where he obtained principal passes in his A level (Physics, Chemistry, Bi ology, and Subsidiary Maths) in 1991. He then graduated with a Bachelor of Science (BSc. Hon.) in Zoology in 1994 from Makerere University, Kampala, where he specialized in wildlife ma nagement and hydrobiology. Upon graduating, he was employed as a graduate research assistant to the Rwenzori Fisheries Biodiversity Project based at then Uganda Institute of Ecology, Queen Elizabeth National Park, Uganda. While there, he became actively involved in field data colle ction and upkeep of project information. With funding from the Government of Uganda a nd the Norwegian Universities Funding Union (NUFU), he graduated with a Mast er of Science (M.Sc.) in Fish eries and Aquatic Sciences, at Makerere University, Kampala, Uganda, in 1999. His masters thesis examined aspects of the ecology of benthic macroinvertebrates in Lake Nabugabo, Uganda. The major contribution of this work was the improvement of knowledge of the limnology and food base of fish species in Lake Nabugabo. After earning his M.Sc. degree, he was hired as a water quality consultant, for the Royal Netherlands funded Ecological Monitori ng Program (EMP) of the Institute of Tropical Forest Conservation (ITFC) based at Bwindi Impenetrable National Park-Uganda, where he established a protocol and baseli ne data for long-term ecological monitoring of aquatic habitats in the park. He also trained one ITFC sta ff who has gone onto pursue a Ph.D. project focusing on effects of anthropogenic disturbance on fish assemblages of Bwindi Impenetrable National Park streams. Between December 1999 and July 2002, Jackson served as Project Manager for the then University of Florida Research Program based at Makerere University Biological Field Station (MUBFS), Kibale National Park, Uganda, where he directed a team of field staff in conducting research on both terrestrial and aqua tic ecosystem in and around Kibale National

PAGE 189

189 Park. In addition, he also worked closely with the Uganda Wildlife Authority (UWA) and Makerere University Biological Field Stati on (MUBFS) staff. In August 2002, Jackson was accepted into a doctoral degree program at the De partment of Zoology, University of Florida, and Gainesville, USA, where he received a Teach ing Assistantship for five years. He also received funding from Program for Studies in Tropical Conservation (PSTC)-COMPTON fellowship at the University of Florida, the In ternational Foundation for Science (IFS), Sweden, The Whitley Laing Foundation for International Na ture Conservation/Rufford Small Grant, and Niddrie Small Grants to work on his dissertation research that fo cused on life history variation in tilapia populations within the crater lakes of western Uganda. In addition to his academic qualifications and professional expe rience, he has participated a nd taught in international field courses. These courses have st rengthened his interest in field research as a crucial tool for effective conservation and manageme nt of natural resources. In particular, the experience of working as part of an intern ational research team has give n him insight into conservation problems and their solutions in different countries.