phyilogenetic relationships etween morphs logically

PHYILOGENETIC RELATIONSHIPS ETWEEN

MORPHS LOGICALLY SIMILAR ,tARBUS SPECIES,

WITH REFERENCE TO TF MR TAXONOMY,

DISTRII1 UTI N AND CONSERVATION.

by

JOHANNES SCHALK ENGEL RIECHT

Thesis presented

In fulfilment of the requirement

For the degree

PHILOSOPHY D '0) CTOR

in

ZOOLOGY

in the

FACULTY OF NATURAL SCIENCES

at the

RAND AFRIKAANS UNIVERSITY

PROMOTOR: DR. F. H. VAN DER BANK

CO-PROMOTOR: PROF. J. T. FERREIRA

AUGUST 1996

CONTENTS:

Table of contents Abstract Samevatting Acknowledgement Foreword

Chapter 1: Introduction - a rationale for the selection of goldie and

chubbyhead barb populations for phylogenetic studies. 1.1 - 1.8

Chapter 2: Morphological characteristics of the goldie and chubbyhead

barb populations studied. 2.1 - 2.10

Chapter 3: Isozyme and allozyme differences in four shortfin barb (Barbus

brevipinnis Jubb, 1966) populations with reference to an

undescribed Barbus species from the Transvaal, South

Africa. 3.1 - 3.16

Chapter 4: Allozyme differences between populations of chubbyhead barb

(Barbus anoplus Weber, 1897) and Marico barb (B. motebensis

Steindachner, 1894). 4.1 - 4.17

Chapter 5: Genetic relationships between Marico barb (Barbus

motebensis, Steindacher 1894), Redtail barb (B. gurneyi

GOnther, 1868) and Amatola barb (B. amatolicus Skelton,

1990). 5.1 - 5.19

Chapter 6: Genetic relationships between seven species within the

chubbyhead and goldie barb groups of minnows (Cyprinidae). 6.1 - 6.23

Chapter 7: A key based on biochemical genetic data, morphology and

distribution to identify species-groups within the chubbyhead

and goldie barb groups of minnows. 7.1 - 7.10

Chapter 8: Summary. 8.1 - 8.14

Chapter 9: References. 9.1 - 9.11

Appendix 10.1 - 10.4

i

ABST CT

A genetic study of seven fish species within the chubby and goldie barb groups

of minnows was done by means of starch-gel electrophoresis. These two groups

of minnows are widely distributed within the temperate rivers of southern Africa.

Because of little morphological differentiation between the species within these

two groups of barbs, their taxonomy has not yet been settled. Subsequently,

species within these two groups of minnows could not always be identified with

certainty and caused some problems for conservation bodies concerning the

distribution and status of the different species. To make a contribution towards

the taxonomy and conservation of these fish species, 330 specimens of goldie

and chubbyhead barbs were analysed to determine the genetic variation within..

and between 17 populations comprising seven species.

The heterozygosity values obtained in the present study within populations

is similar to the values obtained for fish in other studies (Nevo et al., 1984; Alves

and Coelho, 1994). The highest heterozygosity values were observed in fish

populations collected from the eastern Cape, which could possibly be an

indication of a relationship between this specific habitat and genetic diversity.

In contrast, relatively low heterozygosity values were observed in some

populations, for example Barbus brevipinnis (Marite River) and B. gurney! (Mgeni

River), which could possibly be attributed to a degraded habitat. This assumption

is supported by the fact that these populations were infested with parasites,

which is often an indication of stress.

ii

The present study made a contribution towards understanding the

phylogenetic relationships and the taxonomy of the chubby and goldie barb

groups of minnows. Although the phylogenetic relationship observed in the

present study largely agree with the present classification of the chubbyhead

and goldie barbs (Skelton, 1993), the following differences were obtained:

according to the present study, the species within the chubbyhead group

of barbs which was formerly thought to be closely related or synonymous

(Groenewald, 1958; Gaigher, 1969, 1973, 1976), consist of two

genetically different groups, namely a chubbyhead (B. anoplus) and a

tubercle barb group (B. motebensis, B. gurney! and B. amatolicus). The

most obvious morphological characteristic that differentiates these two

groups is the presence of tubercles in the latter group.

the genetic distance between three species of tuberculed barbs (B.

motebensis, B. gurney! and B. amatolicus) and a population collected

from the Ohrigstad River (?B. motebensis) suggest that the latter may be

a fourth species.

Barbus anoplus comprise at least two genetic distinctive groups, namely

a Vaal River/eastern Cape group and a north eastern escarpment group

(Crocodile, Pongola and Tugela Rivers). These two groups can be divided

morphologically based on an incomplete lateral line in the former group

in contrast to a complete lateral line in the latter group.

The goldie barb populations analysed in this study can be subdivided into

three phylogenetic distinctive species-groups, namely a B. brevipinnis

(Sabie and Pongola Rivers) a ?B. brevipinnis (Mogol and Mogalakwena

uua

Rivers) and a B. pallidusineefi group (Crocodile, Olifants and Tugela

Rivers). The phylogenetic relationships obtained between the species-

groups in the present study differ mainly from the present taxonomic

classification as follows:

little genetic differentiation was observed between the B. pallidus and

B. neefi populations analysed and

B. brevipinnis (Sable and Pongola Rivers) are phylogenetically

different from the Waterberg ?B. brevipinnis (Mogol and Mogalakwena

Rivers).

Besides the fact that this study provided some clarity on certain aspects of

the phylogenetic relationships within the chubby and goldie barb groups of

minnows, fixed allele mobility differences were observed between populations

of the same species or specie-groups. These differences as well as the

relatively large genetic distances between most of the populations studied may

suggest that these populations have developed separately and that genetic

divergence occurred between these populations (inter- and intraspecific). It is

possible that this genetic divergence can in some cases justify species-specific

status. Considering that populations of chubbyhead and goldie barbs are

typically restricted to upper catchments of rivers and are often isolated in

headstreams of rivers, genetically unique populations may occur frequently and

widespread within southern Africa. The phylogenetic relationships between these

populations in combination with their distribution patterns may provide the

opportunity to trace the evolution of these fish species and probable migration

iv

routes. This aspect is briefly discussed in the thesis.

Conservation priorities are often driven by the presence of rare and

endangered species and often ignore genetically-unique populations of more

common species, which could be rare or even endangered. Species orientated

conservation can not succeed to conserve genetic variation within these and

other fish species. Therefore, conservation priorities should not be bound by

species and species diversity, but should also strive to conserve genetic diversity

within common and often ecologically important populations of more common

species such as the chubbyhead, tubercule and goldie barbs. Thus, phylogenetic

studies may also play an important role in the future to identify genetically unique

populations and "hotspots" for conservation purposes.

SA EVATTNG

'n Genetiese studie van sewe visspesies binne die dikkop- en goud-

ghieliemientjie groepe is gedoen deur middel van proteTen jel-elektroforese.

Hierdie twee groepe visse kom wydverspreid voor in die koel en hoogliggende

gedeeltes van riviere in suidelike Afrika. As gevoig van min morfologiese

differensiasie tussen die spesies binne hierdie twee groepe is die taksonomie

van die spesies nog nie behoorlik afgehandel nie. Dit het veroorsaak dat spesies

binne hierdie twee groepe nie altyd met sekerheid bepaal kon word nie en het

dit probleme vir bewaringsorganisasies geskep in terme van die verspreiding en

status van die verskillende visspesies. Om 'n bydrae te maak tot die taksonomie

en bewaring van hierdie vissoorte, is 330 goud- en dikkop-ghieliemientjie

eksemplare ontleed om die genetiese variasie te bepaal tussen en binne 17

bevolkings, bestaande uit sewe spesies.

Die heterosigositeits-waardes verkry vir bevolkings in die huidige studie is

soortgelyk aan die waardes wat deur ander studies vir visspesies aangedui word

(Nevo et al., 1984; Alves en Coelho, 1994). Die hoogste heterosigositeits-

waardes is waargeneem in visbevolkings wat in die Oos Kaap versamel is wat

moontlik 'n aanduiding kan wees van 'n verwantskap tussen hierdie habitat en

genetiese diversiteit. In teenstelling hiermee kan die Iae heterosigositeits-

waardes wat waargeneem is by sommige bevolkings, soos byvoorbeeld Barbus

brevipinnis (Mariterivier) en B. gurneyi (Mgenirivier), moontlik toegeskryf word

aan verswakte habitat-toestande. Hierdie aanname word ondersteun deur die

felt dat die betrokke bevolkings besmet was met parasiete wat dikwels 'n

aanduiding is van omgewingsdruk.

vi

Hierdie studie het 'n bydrae gelewer om die filogenetiese verwantskappe

asook die taksonomie van die goud- en dikkop-ghieliemientjie groepe to

verstaan. Alhoewel die filogenetiese verwantskappe wat verkry is in hierdie

studie grootliks ooreenstem met die huidige kiassifikasie van die goud- en

dikkop-ghieliemientjies (Skelton, 1993), is daar noemenswaardige verskille.

Hierdie verskille is kortliks die volgende:

die spesies binne die dikkop-ghieliemientjie groep, wat voorheen as

naasverwant of selfs as sinoniem beskou is (Groenewald, 1958; Gaigher,

1969, 1973, 1976), bestaan volgens die huidige studie uit twee geneties

uiteenlopende groepe, naamlik 'n dikkop- (B. anoplus) en 'n tuberkel-

ghieliemientjie groep (B. motebensis, B. gumeyi en B. amatolicus). Die

mees opvallende morfologiese kenmerk wat hierdie twee groepe van

mekaar onderskei is die teenwoordigheid van tuberkels in Iaasgenoemde

groep.

die genetiese afstand tussen drie tuberkel-ghieliemientjies spesies (B.

motebensis, B. gumeyi en B. amatolicus) en 'n bevolking wat in die

Ohrigstadrivier versamel is (?B. motebensis), dui daarop dat

Iaasgenoemde moontlik 'n vierde spesie verteenwoordig.

Barbus anoplus bestaan uit ten minste twee geneties onderskeibare

groepe, naamlik 'n Vaalrivier/Oos-Kaap groep en 'n noordoostelike

eskarpement groep (Krokodil-, Pongola- en Tugelarivier). Hierdie twee

groepe kan morfologies van mekaar onderskei word op grond van 'n

onvolledige laterale sylyn in eersgenoemde groep in teenstelling met 'n

volledige laterale sylyn in Iaasgenoemde groep.

die goud-ghieliemientjie bevolkings wat in hierdie studie ontleed is, kan

verdeel word in drie filogeneties onderskeibare groepe, naamlik 'n B.

brevipinnis (Sable- en Pongolarivier), 'n ?B. brevipinnis (Mogol- en

Mogalakwenarivier) en 'n B. pallidus/neefi groep (Krokodil-, Olifants- en

Tugelarivier). Hierdie filogentiese verdeling verskil hoofsaaklik van die

bestaande kiassifikasie as volg :

min genetiese differensiasie is waargeneem tussen die B. pallidus en

B. neefi bevolkings wat bestudeer is en

B. brevipinnis (Sabie- en Pongolarivier) is filogeneties onderskeibaar

van die Waterberg ?B. brevipinnis (Mogol- en Mogalakwenarivier).

Bo en behaiwe dat hierdie studie daarin geslaag het om meer duidelikheid

te kry oor sekere aspekte van die filogenetiese verwantskappe binne die goud-

en dikkop-ghieliemientjie groepe, is daar ook mobiliteit- verskille van vaste alleel

produkte tussen bevolkings van dieselfde spesie of spesie-groepe gevind.

Hierdie verskille en die relatief groot genetiese afstande tussen die meeste van

die bevolkings in hierdie studie kan daarop dui dat hierdie bevolkings onafhanklik

van mekaar ontwikkel het en dat noemenswaardige genetiese divergensie

tussen hierdie bevolkings plaasgevind het (binne en buite spesieverband). Dit

is moontlik dat hierdie genetiese divergensie in sommige gevalle selfs spesie-

spesifieke status kan regverdig. Met inagneming daarvan dat dikkop- en goud-

gieliemientjies kenmerkend beperk is tot hoogliggende opvanggebiede en dat

hulle dikwels geIsoleer is in die bolope van riviere, kan geneties unieke

bevolkings waarskynlik algemeen en wydverspreid voorkom. Die filogenetiese

verwantskappe tussen hierdie bevolkings in kombinasie met hul

verspreidingspatrone bled 'n geleentheid om die evolusie en moontlike

migrasieroetes van die betrokke visspesies na te spoor. Hierdie aspekte is

kortliks bespreek in die proefskrif. '

Bewaringsaksies word dikwels gerig op skaars en bedreigde spesies, terwyl

groot aantal geneties unieke bevolkings van meer algemene soorte, wat

skaars of selfs bedreig kan wees, dikwels oor die hoof gesien word. Spesie-

gerigte bewaringsaksies kan nie daarin slaag om die genetiese variasie binne

hierdie en ander vissoorte te bewaar nie. Daarom moet bewaringaksies nie net

gebind wees aan spesies en spesie-verskeidenheid nie, maar moet dit ook

strewe om die genetiese-diversiteit te bewaar van meer algemene en dikwels

ekologies belangrike soorte, soos die dikkop-, tuberkel- en goud-ghieliemientjies.

Hierdie tipe studies kan dus 'n belangrike rol in die toekoms speel om geneties

unieke bevolkings vir bewarings-doeleindes te identifiseer.

ix

ACKNOWLEDGE E TS

I would like to express my sincere gratitude to my promotor, Dr Herman van der Bank, for his guidance and support during this study.

This study involved assistance from many quarters and I would also like to express my gratitude to the following people in particular:

Dr Jannie Ferreira for his interest and for initiating this study.

Dr Neels Kleynhans, as a friend and a colleague, for his comments and guidance.

Dr Paul Skelton (J. L. B. Smith Institute for Ichthyology) for sharing my interest and for his valuable contributions towards this study.

Mr Mike Coke (Natal Parks Board) for collecting fish samples.

Mr Andre Hoffman, Mr Tjokkie Pieterse and Mr Francois Roux for technical assistance.

2i4cot important,

praise the ctord /or the privile ye to duly the comp lexity ol

creation.

would like to thanh eriha, Cleome and jean-Pierre„ m y wile and

children, /or their love and su pp ort.

_9 alio would like to thank m y mother Rag, who alwa y3 lead 4 examp le, even durin g times 0/ A.arcLhi p, my brother and niter, ariw

and Jena ai well as m y mother-in-law -icletena l /or their su pp ort and

willingness to help where

FOREWORD

This thesis is presented as a collection of five research papers summarising the

results after completion at each stadium of the study. This thesis is preceded by

an introduction (Chapter 1), recounting the primary motivation and objectives for

this study which broadly discusses the results obtained at different stages of the

study and its influence on the direction of the study. Chapter 2 deals with the

most commonly used morphometric and meristic characteristics of the species

studied and its taxonomic limitations.

Chapters 3 and 4 deal with the genetic differentiation within B. brevipinnis

(shortfin barb) and two morphological similar barb species from the former

Transvaal (B. anoplus and B. motebensis) respectively, which was the primary

objective of this study. The chapter dealing with the B. brevipinnis (shortfin barb)

has been published in Biochemical Systematics and Ecology (Chapter 3) and the

section dealing with B. anoplus and B. motebensis (Chapter 4) has been

submitted to Water SA for publication.

A subsequent paper dealing with the genetic differentiation between three

species of tubercle barbs (B. motebensis, B. gurneyi and B. amatolicus) in

southern Africa has been accepted for publication in Animal Genetics (Chapter

5). Chapter 6 deals with the genetic differentiation between all 17 populations

analysed for this study and was submitted for publication to African Journal of

Zoology. A short communication relating biochemical data, morphological

characteristics and distribution patterns to the observed phylogenetic groupings

xi

were submitted for publication to African Journal of Zoology (Chapter 7). Finally

in Chapter 8, the results of the present study are summarized and its

implications on conservation and management of fish species are broadly

discussed, with reference to probable migration routes. This chapter also

evaluates the degree to which the original goals were met (see Chapter 1).

The format of these papers has been standardised for this thesis, but the

contents of Chapters 3 to 7 are as published or submitted. Differences in

statistical approaches in papers reflect specific requirements of the journals and

recommendations by various referees. Referees comments included the view

that manuscripts were clear and well written, showing the importance of

biochemical genetics to identify cryptic species of fish, and that the appropriate

statistical tests were used to demonstrate population subdivision and potential

species distinction. Standard methods were used for analysis throughout the

study. Because these methods have been used commonly and extensively

since the 1960's, no elaborate descriptions are given. For further reading refer

to Hillis and Moritz (1990) and Carvello and Pitcher (1995).

The findings of this study have been presented at several national and local

symposiums (South African Association for Aquatic Scientists (1992), Transvaal

Nature and Environmental Conservation annual meetings (1992 and 1993) and

RAU colloquy (1991)). The results of this study were discussed with various

experts and were published or submitted for publication in international and

national journals.

xii

introductio 'Ho ale f r the Sei cti n of

Go di and Chit, yhea P puiations if 10

hyli g® etic Studies.

CONTENTS

INTRODUCTION 1.1

CHUBBYHEAD BARBS 1.1

GOLDIE BARBS 1.2

THE OBJECTIVES OF THIS STUDY 1.4

Figure 1: Map of South Africa depicting sampling sites of goldie

and chubbyhead barbs studied in relation to the major

rivers. 1.8

Rationale for the Selection of Goldie and Chubbyhead Barb Populations for Phylogenetic Studies

DRITRODUCTDOINI

To conserve biotic diversity, it is essential to describe the existing diversity of

organisms in terms of their spatial and temporal distribution and abundance. The

most important tool to describe the diversity of living beings is systematics, which

includes classification and phylogenetics, the latter being the genealogy of biota

in the course of their evolutionary history. For many years, researchers have

been mainly involved in studies surrounding the classification and distribution of

freshwater fish species in southern Africa based on their morphological

characteristics (Gilchrist and Thompson, 1913; Barnard 1938, 1943; Groenewald,

1958; Crass, 1960, 1964; Jubb, 1967; Gaigher, 1969, 1973, 1976; Kleynhans,

1983; Skelton, 1993). A history of uncertainty regarding the different species of

goldie and chubbyhead barbs in most of these studies highlighted the need for

phylogenetic studies of these freshwater fish species in southern Africa. As an

aquatic scientist, attached for more than 13 years to the former Transvaal

conservation body, similar difficulties were often experienced to classify the

different chubbyhead and goldie barb species (Engelbrecht, pers. obs.). This

caused uncertainties regarding the distribution and conservation status of

different species. The purpose of this chapter is to discuss these uncertainties

briefly and to formulate the objectives of the study. The selection of specific

goldie and chubbyhead barb populations for this study was determined by these

objectives.

CHU J

rit, ThE A ID A = S

Chubbyhead barb is a collective name for a group of species endemic to South

Africa and is widely distributed in the headwaters of many rivers south of the

1.1


Limpopo River. This group of minnows are of ecological importance because they

are frequently the only fish species in many of these rivers and are often found

in isolated small stretches of river above waterfalls. According to Skelton (1993),

this group comprise the chubbyhead barb (B. anoplus Weber, 1897), the redtail

barb (B. gumeyi Gunther, 1868), the Marico barb (B. motebensis Steindachner,

1894) and the Amatola barb (B. amatolicus Skelton, 1990). However,

researchers such as Groenewald (1958) and Gaigher (1969, 1976) experienced

difficulty in separating B. anoplus from B. motebensis and therefore suggested

that these species might be synonymous. Based on the above-mentioned

observations as well as personal experience the following question is posed:

Q1. Are B. anoplus and B. motebensis synonymous and is the presence of

tubercules in B. motebensis of taxonomic importance?

ILODE

Goldie barb is a collective name for a group of small minnows, referring to the

colour of the male's breeding livery (Skelton, 1993). The goldie barbs are a

distinct southern African group of minnows, limited to the headwaters of rivers,

where they occasionally occur with chubbyhead barbs. Their distribution patterns

consist mainly of three subdivided areas namely, the Upper Zambezi River and

rivers in southern Zaire (northern area of distribution); rivers in Mpumalanga and

Kwazulu-Natal (central area); and rivers in the Eastern Cape (southern area).

However, they also occur in the rivers of the Waterberg (northwesterly areas of

high topographical elevation in the catchment of the Limpopo River). According

to present taxonomy (Skelton, 1993) this group comprise the goldie barb (B.

pallidus A. Smith, 1841), the shortfin barb (B. brevipinnis Jubb, 1966) and the

1I.2


sidespot barb (B. neefi Greenwood, 1962). Barbus pallidus was originally

described from the eastern Cape but was subsequently listed by Boulenger

(1909-1916) and Gilchrist and Thompson (1913) as B. hemipleurogramma

Boulenger, 1911. Barnard (1938) suggested that the two above-mentioned

species are synonymous and that the latter name should be suppressed, but

Groenewald (1958) preferred to use the name B. hemipleurogramma for a goldie

barb species collected from the Mooi River (Orange River System).

Many taxonomists were puzzled by the goldie barbs, for instance, Crass

(1960) confused B. pallidus with B. viviparus, B. unitaeniatus and B.

lineomaculatus, but in a later publication (Crass, 1964) separates these species.

Jubb (1966) also used B. viviparus to compare the newly described B.

brevipinnis, though he refers to the fact that the closely related B. neefi was

collected from the Ohrigstad River (Limpopo River System). Gaigher (1969)

concluded that B. pallidus, B. neefi and B. brevipinnis represented a single

polytypic species and therefore lumped B. pallidus and B. neefi under B. neefi.

A goldie barb collected from the headwaters of the Mogol and Lephalala Rivers

(Waterberg) by Kleynhans (1983) was classified as B. brevipinnis. Skelton (pers.

comm.*) suggested that there could be four "species" of goldie barbs: B.

brevipinnis with a dashed lateral line, B. pallidus with large irregular spots and the

third species (possibly new: ?B. brevipinnis) with an unbroken lateral line from the

Waterberg. According to Skelton, (pers. comm.), the fourth species, Barbus neefi,

which was collected from the Ohrigstad River (former Transvaal), have a few

characteristics that set it apart from the other three "species" (i.e. shorter head

* Dr P. H. Skelton, J. L. B. Smith Institute of Ichthyology, Grahamstown, South Africa.

1.3


with neurocranium more curved than the others, narrower ethmoid and

infraorbital bones). Externally only pigmentation serves to distinguish these

"species" from each other. A ?B. brevipinnis population isolated in the upper

reaches of the Mogol River showed a pigmentation difference to other

populations from the Waterberg (Kleynhans, pers. comm.*), which suggested that

this population could have evolved independently. Based on these observations,

the following questions arose:

Are B. pallidus, B. neefi and B. brevipinnis a single polytypic species?

Is B. brevipinnis from the headwaters of the Sabie River (Incomati River

System) different from ?B. brevipinnis from the Waterberg rivers?

Is the ?B. brevipinnis population isolated in the upper reaches of the

Mogol River different from other populations from the Waterberg?

THE Z JECTIVES F THIS STUDY

Evident from the preceding discussion, the classification of the different species

of goldie and chubbyhead barbs, based on their external morphological

characteristics only, caused some uncertainties and were inconclusive. To

contribute towards the knowledge-base which could help to answer the four

questions raised above, the following objectives were set for this study.

Objective 1 - determine the phylogenetic relationship within and between

populations of . brevipinnis (S bie over) and ? . brevipinnis (Fivers

of the Waterberg).

To achieve this objective (refer to Q3 and 4), B. brevipinnis specimens (G1-2

in Fig. 1) were collected from the Marite (type locality) and Sand Rivers (Sabie

* Dr C. J. Kleynhans, Institute for Water Quality Studies, Department of Water Affairs, Pretoria, South Africa.

1.4


River, Incomati River System), and ?B. brevipinnis (G3-4 in Fig. 1) from the

Mogalakwena and Mogol Rivers (Limpopo River System).

Objective 2 - determine the phyiogenetic relationships between B.

brevipinnis and the morphometricaiiy and meristicaliy similar . neefi

and f pallidus.

To achieve the second objective (refer to Q2), B. neefi and B. pallidus

specimens (G5-6 in Fig. 1) were collected from the Ohrigstad River (Limpopo

River System) and Crocodile River (Incomati River System) respectively.

Objective 3 - determine the phylogenetic relationships between

nopbus and f. motebensis.

To achieve this objective (refer to Q1), B. motebensis (TU1 in Fig. 1) was

collected from the Ohrigstad River and B. anoplus (CH1 in Fig. 1) from the

Crocodile River (the same localities as for the above-mentioned B. neefi and

B. pallidus samples respectively).

The electrophoretic results based on these eight populations were sufficient

to conclude that B. brevipinnis from the eastern escarpment is phylogenetically

different to the ?B. brevipinnis from the Waterberg region and that they may

represent different species (Chapter 3). Unexpectedly large genetic differences

were observed between B. anoplus collected from the Crocodile River and B.

motebensis collected from the Ohrigstad River, which suggested a large degree

of genetic divergence within the chubbyhead barbs. In complete contrast to these

results, two species (B. neefi and B. pallidus) collected from the same localities

mentioned above, were genetically almost identical. Based on this inconsistency

and discussions with other scientists (i.e. Kleynhans, pers. comm.; Skelton, pers.

11.5


comm.; Coke, pers. comm.*), the following questions were raised:

Why are two species (B. neefi and B. pallidus), collected from two

different river systems (Limpopo and Incomati River Systems)

genetically almost identical?

Does B. neefi and B. pallidus within the former Transvaal and Kwazulu-

Natal represent one or more species?

Is B. motebensis from the Ohrigstad River the same species as the

geographically subdivided B. motebensis from the Marico River (type

locality) and what is its phylogenetic relationship with the other two

"tubercle" barb species (B. gumeyi and B. amatolicus)?

What is the phylogenetic relationship between populations of B. anoplus

within its distribution range?

Why is chubbyhead barbs widely distributed within rivers in South Africa

but absent from the rest of the continent and why are they often the only

species in the headwaters of rivers?

To answer these questions the objectives of the present study were revised

to include the following:

bjec‘ five 4 - determine the phylogenetic relationsHp between

populations off [' neefi and paili us within MpannaDanga and

Kwazulu-fatal.

To achieve this objective (refer to Q5 and 6), B. pallidus (G7-8 in Fig. 1)

specimens were collected for analysis from the Manzaan (Pongola River

System) and Buffels Rivers (Tugela River System).

* Mr M. M. Coke, Natal Parks Board, Pietermaritzburg, South Africa.

1.6


Objective 5 - determine the phylogenetic relationships between

populations of B. anoplus within its distribution range.

To achieve this objective (refer to Q8 and 9), B. anoplus specimens

(CH2-5 in Fig. 1) were collected for analysis from the Vaal (Orange

River System), Buffels (Tugela River System), Xuka (Mbashe River

System) and Cowie Rivers (Cowie River System).

Objective 6 - determine the phylogenetic relationship between

motebensis populations from the Ohrigstad and Mari= Rivers as

well as other tubercule b rb species.

To achieve the last objective (refer to Q7 and 9), B. motebensis

specimens were collected for analysis from the Marico River (TU2 in

Fig. 1) and specimens of B. gurneyi and B. amatolicus (TU3-4 in Fig.

1) were collected from the Mgeni River and Mzimvubu River Systems

respectively.

For practical and logistical reasons it is not possible to sample all the

variations within the goldie and chubbyhead barbs, but the distribution and

selection of samples for the present study were considered to give a fair

indication of the genetic variations within and between the goldie and

chubbyhead barb groups of minnows.

1.7

5 Co E a)

O

C O

a) C

a)

O

U) rn 2 .o -0

a)

_o .o

C.)

C

a)

O

`45

a) U)

C 0_ E U)

C

7)

a)

O

5 6

cp 2

,a2

LL


cJ O

= 1.. ca

-0 et ,c2

7,-, •M ..M

..; L.)

,,..;

E24

-- CV en .cr trl •- ■ rq rn er

= CSC = D D a U U U C.) C..) H E- E- E-

Go

Idle

ba

r bs

0 0 0 0 0 0 , ;„... .. ...,

4-; °) c4

--. csi En et 00 t- .c) kr1 00000000

Sp

ecie

s

.,•, ..) r

.7 .7 .7 .7 Z.; ,Z3 • .. ......

-Z -Z V 0

C Z CZI t:1 CS :-.. :-. 00 C CiC4CiwCIdCCiC

: 2 Er.?

Sp

ecie

s

,., .t.t . ,t ...,

.r...- .::. .Z. .fi .t ,t L, - - .:•_-. .... ..----- .1:3 -C .1:3

cl.) cv to -.- ---

L.. - k. C C C3 CI) -FS .rz, -o -cz z., L.,... :-. C1CiCiC1CiCiC cci

1.8

M • rphoha gic Char ter tic of ths G©i us and

Ch v b yh b P • pulati ns tu as

CONTENTS

INTRODUCTION 2.1

MATERIALS AND METHODS 2.1

RESULTS AND DISCUSSION 2.2

Figure 1: A graphic representation of points obtained by using SO CA

depicting the relationships between the populations

studied and the meristic characteristics measured. 2.6

Table 1: Sample size, species and localities where specimens of

chubbyhead and goldie barbs were collected for

morphometric and meristic measurements. Species listed

according to existing taxonomy. 2.7

Table 2: Morphological characteristics of goldie barb populations

studied. See Table 1 for locality details. 2.8

Table 3: Morphological characteristics of chubbyhead barb

populations studied. See Table 1 for locality details. 2.9

Table 4: Similarity (Jaccard, 1928) between the populations studied

based on meristic characteristics. See Table 1 for locality

details. 2.10

Morphological Characteristics of the Goldie and Chubbyhead Barb Populations Studied

ODUCTDON

The taxonomy of freshwater fish in southern Africa is largely based on meristic

parameters (Gilchrist and Thompson, 1913; Barnard 1938, 1943; Groenewald,

1958; Crass, 1960, 1964; Jubb, 1966, 1967; Skelton, 1993). The parameters most

often used by these authors included counts of scales and fin rays. Morphometric

measurements have also been used to describe species of goldie barbs, e.g.

Jubb (1966) used proportional body measurements to compare B. brevipinnis with

B. viviparus. However, morphometric characteristics of populations may be

influenced by local environmental factors [i.e. the short fins of B. brevipinnis

(shortfin barb) are characteristic only of the Marite River population (Skelton,

1993)]. The purpose of this chapter is to evaluate if genetic studies would be

needed to supplement morphological characteristics to describe variations within

the goldie and chubbyhead barb groups of minnows.

MATE GALS D ETHODS

Morphological measurements and counts used in the present study included those

most often used in literature to describe the relevant species. For morphometric

comparisons within and between the goldie and chubbyhead barb populations and

species studied, the following measurements were taken from specimens: total

and standard length; head length and width; dorsal and anal fin length; orbital

diameter and gut length. Meristic measurements taken from fish specimens

included counts of lateral line and caudal peduncle scales, dorsal and anal fin rays

and the number of barbels.

2.1


Measurements (1/10mm) were taken with a vernier caliper (accuracy 0.02mm)

from 311 specimens representing 14 populations (Table 1). Because three of the

total of 17 populations included in the present study were frozen, no

measurements were taken from these to avoid unnecessary thawing of samples

before electrophoresis. Using existing research as a guideline (Jubb, 1966;

Skelton, 1988), body measurements were expressed as a percentage of standard

length for comparative purposes. SO MC i (Greenacre, 1990) was used to analyse

a contingency table of meristic count frequencies. SO CA projects two sets of

points (populations and meristic characteristics) into a multidimensional space,

after which it determines the position of each point in relation to a lower number

of dimensions. For further details refer to Greenacre (1987). Meristic counts were

transformed into a presence or absence matrix before it was statistically analysed,

using OS (Jaccard's (1928) similarity index).

RESULTS AND DISCUSS1 N

The results of the morphometric and meristic measurements for the goldie and

chubbyhead barbs studied are presented in Tables 2 and 3. These tables suggest

that the morphometric characteristics of the populations studied varied more within

than between populations and that the different species could not be differentiated

based on these values. Even in B. brevipinnis (shortfin barb) where short fins are

characteristic for the Marite River population, no clear separation between this

population and any of the other goldie or chubbyhead barb populations studied

was observed. During sampling, an effort was made to collect the largest

2.2


individuals available in the population. Based on these measurements, a size

difference exists between the goldie and chubbyhead barbs.

Several meristic differences were observed between the populations studied,

which could be of value to differentiate between the goldie and chubbyhead barbs

(Tables 2 and 3). In Fig. 1, SIIVICA differentiates between the goldie and

chubbyhead barbs on the X-axis (positive and negative respectively) based on the

following meristic characteristics:

The number of scales on the lateral line (11-12 and 13) corresponds with the

goldie barbs (G1-8) and (14 and 5) with the chubbyhead barbs (CH1-2 and

5; TU1-3).

The dorsal fin rays can be used to differentiate between the goldie barbs

(d2) and the chubbyhead barbs (dl).

The number of scales on the caudal peduncle is useful to differentiate

between the goldie (q1) and chubbyhead barbs (q2 and 3).

SIB; C differentiates between B. anoplus and the other chubbyhead barb

species on the Y-axis (negative and positive respectively) based on the following

meristic characteristics (Fig. 1):

Caudal peduncle scale counts (q3) correspond with B. anoplus (CH1-2

and 5) and (q2) with the other chubbyhead barb species (TU1-3).

The presence of two pairs of barbels (b2) in B. motebensis and B. gurneyi

(TU1-2 and 3) separates these populations from B. anoplus (CH1-2 and 5)

2.3


which has only one pair of barbels (b1).

Some characteristics did not corrrespond to any of the species studied (i.e. an

incomplete lateral line (Ii) was encountered only in B. anoplus from the Vaal River

(CH5) and B. motebensis from the Marico River (TU2)). The close grouping

between the populations within the goldie barb group (G1-8) in contrast to the

chubbyhead group (CH1-2 and 5; TU1-3) suggest that the species and

populations within the former group are meristically very similar. The similarity

index (Table 4) suggests two groups of goldie barbs consisting populations G1-4

and G8 (B. brevipinnis) and G5-7 (B. pallidus/neefi) respectively. These two

groups were 100% and 88% similar within and between the two groups

respectively. The latter suggests that G8 may be incorrectly classified as B.

pallidus. The chubbyhead group of barbs studied (CH1-2 and 5 ; TU1-3) were not

closely grouped in Fig. 1, which suggested meristic differentiation within and

between populations and species within this group. The similarity index shows OS

values ranging from 39% to 88% between the different populations and species

within the chubbyhead group of minnows and OS values from 7% to 33% between

the goldie and chubbyhead barb groups of minnows. This suggests that the

chubbyhead barbs show meristic differentiation, which cannot be explained in

terms of their present taxonomic status. It also shows that chubbyhead barbs can

easily be separated from the goldie barbs based on these meristic parameters.

Morphometric and meristic characteristics of populations can be powerful tools

2.4


to describe species and specie-strains, as suggested by Teugels (in press).

However, difficulties were experienced to differentiate between the species

because of the meristic differentiation within species of chubbyhead barbs and the

lack thereof between species of goldie barbs. Based on the characteristics

described above, it is evident that additional data is needed to clarify the

phylogenetic relationships of these fishes.

25

CH2 0

CH5 ❑

q2

b2 0

TU3 0

TU2 TU1 ❑

-al -di- 0 0

q1

d 11 2 Ic

0

11:2 3

14 b1

CH1 0

q3 0

li 15 9

Morpholoeical Characteristics of the Goldie and Chubbvhead Barb Populations Studied

LEGEND

11 = 26-27 lateral line scales al = AIII 5 anal fin rays 12 = 28-29 lateral line scales a2 = A1116 anal fin rays 13 = 30-31 lateral line scales li = incomplete lateral line 14 = 32-33 lateral line scales lc = complete lateral line 15 = 34-35 lateral line scales bl = 1 pair of barbels ql = 12 caudal peduncle scales b2 = 2 pairs of barbels q2 = 14 caudal peduncle scales G1-8 = goldie barbs q3 = 16 caudal peduncle scales CH1, 2 & 5 = B. anoplus dl = DIII 7 dorsal fin rays TU1-3 = other chubbyhead barbs d2 = DIII 8 dorsal fin rays

Figure 1 : A graphic representation of points obtained by using SINICA depicting

the relationships between the populations studied and the meristic

characteristics measured.

2.6


Table 1: Sample siz species nd Dc. ceDotB& s wf re speci ens of ch '-byh- -d and gs Idle barbs were c rDDE cted f r morsh.metrlc and meRistic measurrements. Species listed accordi g tw ex sting t xol a t.)my.

N Species Locality Latitude Longitude bbreviation

30 10 20

30 30 10

30 30 21 20

10 10 30 30

Chubbyhead barbs

B. anoplus Buffelskloofspruit A 24°47'S 30°30'E CH2 B. anoplus Ngagagane E 27°59'S 29°52'E CH1 B. anoplus Blesbokspruit C 26°11'S 28°23'E CH5

B. motebensis Ohrigstad River ° 24°53'S 30°36'E TU1 B. motebensis Kaaloog se Loop E 25°47'S 26°24'E TU2 B. gumeyi Msinduzi F 29°38'S 30°25'E TU3

Goldie barbs

B. brevipinnis Marite River' 24°47'S 31 °05'E G1 B. brevipinnis Sand River" 24°09'S 31 °02'E G2 B. brevipinnis Sterk River' 24°32'S 28°31'E G4 B. brevipinnis Grootspruie 24°29'S 27°51'E G3

B. pallidus Manzaan River K 27°38'S 30°53'E G8 B. pallidus Ngagagane E 27°59'S 29°52'E G7 B. pallidus Buffelskloofspruit A 24°47'S 30°30'E G6 B. neefi Ohrigstad River D 24°53'S 30°36'E G5

A

B

C G

E

F G

" I J

K

Tributary of the Crocodile River (Incomati River System). Tributary of the Buffels River (Tugela River System). Tributary of the Vaal River (Orange River System) Tributary of the Olifants River (Limpopo River System). Tributary of the Marico River (Limpopo River System). Tributary of the Mgeni River (Mgeni River System). Type Locality (Jubb, 1966), Tributary of Sabie River (Incomati River System). Tributary of Sabie River (Incomati River System). Tributary of Mogalakwena River (Limpopo River System). Tributary of Mogol River (Limpopo River System). Tributary of the Pongola River (Pongola River System).

2.7


T bk 2: M , rphoDogucall char- cteristics *if g•1cilie barb popullations stuceo d. See T, We 1 for El caHty detaRie.

REFERENCE G1 B.

brevipinnis

G2 B.

brevipinnis

G3 a.

&evil:duals

G4 B.

brevlpinnis

G5 B.

naafi

G6 B.

pallidus

G7 B.

pallidus

G8 a.

pallidus

River Sabie Sand Mogalakwena Mogol Ohrigstad Crocodile Buffets Manzaan

N 30 30 21 20 30 30 10 10

Standard length (1no.)

329 - 453 302 - 403 264 - 334 360 - 427 277 - 480 308 - 424 350 - 374 371 - 408

Lateral line complete complete complete complete complete complete complete complete

Barbels 1 pair 1 pair 1 pair 1 pair 1 pair 1 pair 1 pair 1 pair

Elliorphometrics (% of standard ilength) mean/range

Head length 25 24 25 24 25 24 24 23 21 -29 22 - 30 19 - 29 19 - 27 21 -32 22 - 26 24 - 25 22 - 25

Head depth 18 18 20 18 19 18 18 17 16 - 25 15 - 22 15 - 21 14 - 21 17 - 22 17 - 22 17 - 19 14 - 18

Eye 8 8 8 7 8 8 8 8 diameter 6-10 7 - 9 6 - 9 6 - 8 7-12 6 - 9 7 - 9 7 - 9

Dorsal fin 20 21 22 23 20 22 23 22 16 - 25 17 - 25 16 - 25 19 - 25 16- 18 21 -25 21 -24 20 - 22

Anal fin 15 16 18 18 16 18 17 16 12 - 18 14 - 20 13 - 20 14 - 19 13 - 18 14 - 19 16 - 18 15 - 17

Gut length 106 103 102 115 120 115 108 128 74 - 133 83 - 141 86 - 126 85 - 137 103 - 166 85 - 137 97 - 116 108 - 140

Meristic measurements Count (number of individuals)

Lateral line 29 (11) 29 (10) 29 (4) 29 (5) 31 (2) 31 (3) 31 (1) 28 (9)

scales 28 (12) 28 (14) 28 (10) 28 (9) 30 (5) 30 (5) 30 (1) 27 (1) 27 (6) 27 (6) 27 (5) 27 (6) 29 (5) 29 (8) 29 (3) 26 (1) 28 (12) 28 (14) 28 (4)

27 (6) 27 (1)

Caudal peduncle scales

12 (30) 12 (30) 12 (21) 12 (20) 12 (30) 12 (30) 12 (10) 12 (10)

Dorsal fin DIII 7 (2) DIII 7 (4) DIII 8 (21) DIII 8 (20) DIII 7 (1) DIII 7 (4) DIII 7 (2) DIII 8 (10)

rays DIII 8 (28) DIII 8 (24) DIII 8 (28) DIII 8 (25) DIII 8 (8) DIII 9 (2) DIII 9 (1) DIII 9 (1)

Anal fin AIII 5 (1) AIII 6 (30) AIII 6 (21) AIII 6 (20) AIII 6 (29) AIII 6 (30) AIII 6 (10) AIII 6 (9)

rays Alll 6 (27) Alll 7 (1) AIII 7 (1) AIII 7 (2)

2.8


TabEle 3: ri FphollogicaEl char cat risUcs o chubbyhead barb p(.puDaiioorts sh.ii lied. S T sWBe I fwr Doc Bo dt t Hs.

REFERENCE CH1 B.

anoplus

CH2 B.

anoplus

CH5 B.

anoplus

TUi B.

motebensis

TU2 B.

motebensis

TU3 B.

gurneyi

River Buffels Crocodile Vaal Ohrigstad Marico Msinduzi

N 10 30 20 30 30 10

Standard length (mom)

485 - 584 408 - 571 353 - 523 253 - 710 326 - 527 509 - 763

Lateral line complete complete incomplete complete incomplete complete

Barbels 1 pair 1 pair 1 pair 2 pairs 2 pairs 2 pairs

Mor • hornetrics (% of standard Dength) mean/range

Head length 25 24 - 26

27 22 - 30

26 24 - 27

28 22 - 31

28 26 - 31

26 24 - 27

Head depth 17 17 - 18

19 18 - 21

18 17 - 19

20 17 - 22

18 17 - 20

18 17 - 19

Eye diameter

7 6 - 9

9 6 - 10

7 6 - 8

9 7 - 11

8 7 - 9

7 6 - 8

Dorsal fin length

22 20 - 25

20 17 - 22

21 19 - 23

19 15 - 23

22 20 - 26

22 20 - 23

Anal fin length

17 15 - 19

18 16 - 23

16 15 - 18

16 13 - 26

17 15 - 19

17 15 - 19

Gut length 92 66 -106

113 91 - 136

104 89 - 119

108 83 - 125

114 89 -131

92 78 - 102

Meristics Count (number of individuals)

Lateral line scales

33 (4) 32 (6)

35 (6) 34 (15) 33 (6) 32 (3)

35 (5) 34 (13) 33 (2)

35 (10) 34 (14) 33 (6)

35 (6) 34 (19) 33 (5)

34 (6) 33 (4)

Caudal peduncle scales

16 (10) 16 (30) 16 (20) 12 (5) 14 (25)

14 (30) 14 (10)

Dorsal fin rays

DIII 7 (10) DIII 7 (29) DIII 8 (1)

DIII 7 (20) DII 7 (29) DIII 8 (1)

DIII 7 (30) DIII 7 (10)

Anal fin rays

AIll 5 (10) Alll 5 (30) Alll 5 (20) AIII 5 (30) AIII 5 (30) AIII 5 (10)

2.9


T ?Me 4: Similarry Waccard, 1921 b i - , n pop laticns studied based o meristic characteristics. Sec Tabi 1 for iocaiity det ils.

GI *****

G2 100

G3 100 100

G4 100 100 100 *****

G5 88 88 88 88

G6 88 88 88 88 100 *****

G7 88 88 88 88 100 100

G8 100 100 100 100 88 88 88 ****"

CHI 33 33 33 33 28 28 28 33

CH2 33 33 33 33 28 28 28 33 78 *****

CH5 19 19 19 19 14 14 14 19 68 68

TU1 23 23 23 23 19 19 19 23 45 60 39 *****

TU2 10 10 10 10 7 7 7 10 39 39 60 68

TU3 19 19 19 19 14 14 14 19 52 52 45 88 78

G1 G2 G3 G4 G5 G6 G7 I G8 CHI I CH2 I CH5 TU1 TU2 TU3

2.1

zym z Weres cais Ratliff'

Shortfin Barb (Barbus brevipinnis J b, 1966)

P puiati s with

r nc t descrP d

Barbs Sp ci s from the Transv S • (Loth Afric

CONTENTS

ABSTRACT 3.1

INTRODUCTION 3.1


RESULTS 3.4

DISCUSSION 3.5

Figure 1: Map of the Transvaal, depicting the distribution of B.

brevipinnis in relation to the major rivers, as well as the

positions of the sampling points. 3.11

Figure 2: Phylogenetic trees obtained by using: (a) D1SW G and (b)

FREQPARS, showing the relationships between the

populations within the eastern escarpment group (B.

brevipinnis) and the Waterberg group (?B. brevipinnis).

3.12

Table 1: Localities and coordinates where B. brevipinnis (a, b) and

B. ?brevipinnis (C, d) populations were collected. 3.13

Table 2: Enzyme commission numbers (E.G. no) of the proteins

separated by electrophoretic analysis, abbreviations used for

loci resolved and buffers giving best results. 3.14

Table 3: Relative mobilities, allele frequencies and G-test values of the

polymorphic loci and loci where differences were detected

and H values with standard errors for four B.

brevipinnis populations. 3.15

Table 4: Genetic distances between populations calculated with the

distance coefficients: (D = Nei, 1972, D' = Nei, 1978 and C =

Cavelli-Sforza & Edwards, 1967 Chord Distance) for B.

brevipinnis and ?B. brevipinnis. 3.16

Isozyme and Allozyme Differences in Four Shortfin Barb

A ri STRACT

Starch gel-electrophoresis was used to survey genetic variation among samples

from four different populations of Barbus brevipinnis. Thirteen of 30 loci were

found to be polymorphic. It is suggested, based on the extent of the genetic

differences between populations, that the populations collected from the

Waterberg should be regarded as an undescribed species of Barbus.

DATRODUCTBON

Barbus brevipinnis from the Marite River, a tributary of the Sabie River (eastern

Transvaal escarpment), was described by Jubb, (1966). The name refers to the

very short fins found in specimens of this endemic species. Specimens of this

species normally exhibit a dark broken line above the lateral line, consisting of

three to six dashes. A small barb species collected by Gaigher (1969; 1973)

from the upper reaches of the Mogol, Mogalakwena and Lephalala Rivers

(northwestern Transvaal), was initially referred to as B. neefi. However,

subsequent samples collected by Kleynhans (1983) from the same localities

were referred to as B. brevipinnis. Closer inspection of the barbs collected from

the Waterberg revealed that they differed from B. brevipinnis from the Sabie

River in that they displayed a thick solid line above the lateral line, sometimes

consisting of numerous smaller dashes and spots. Examination by Skelton

(pers. comm.) suggested the possibility that the fishes from the Mogol,

Mogalakwena and Lephalala Rivers may belong to a different species.

3.1


In terms of the hydrobiological regions as defined by Harrison (1959), B.

brevipinnis from the Sabie River is restricted to the eastern escarpment region

whereas populations from the Mogol, Mogalakwena and Lephalala Rivers occur

in the Transvaal mountain region (Waterberg). The headwaters of these two

regions are completely isolated from one another by a distance of approximately

300 kilometres and are also separated by the Olifants River (Fig. 1).

Although linear measurements were taken from several B. brevipinnis

populations which suggested a larger body size for individuals from the Mogol

River as well as shorter fins in some specimens of B. brevipinnis from the

eastern escarpment region, the determination of the taxonomic status of these

fish populations by means of morphometric and meristic measurements was

found to be inconclusive (Chapter 2).

On the other hand, protein gel-electrophoresis has been instrumental in the

discovery of morphologically cryptic species (Grant et al., 1984). In the present

study the genetic variation within and between B. brevipinnis populations were

investigated to determine whether populations from the eastern escarpment

and the Transvaal mountain regions represent one or more species and to what

extent these populations differ from one another within these two regions.

3.2


ATERDALS A ID METH IDS

During September 1990, 60 B. brevipinnis specimens were collected from the

eastern escarpment (Sable and Sand Rivers), and 41 ?B. brevipinnis specimens

from the Mogalakwena and Mogol Rivers (Table 1). The specimens from these

four localities represented the morphological variations observed as discussed

in the introduction. Selected linear and meristic measurements were taken from

all these specimens for comparison, before liver and muscle tissues were

sampled. The muscle tissue of each specimen was individually homogenised in

distilled water. The livers of most specimens were small and groups of five livers

were lumped before analysis. Tissue samples were analysed by starch

gel-electrophoresis according to the methods described by Van der Bank et al.

(1992) and the histochemical methods of Harris and Hopkins (1976) were used

in staining for enzyme activity. The structure of the banding patterns was

deduced from the criteria described by Harris and Hopkinson (1976) and the

locus nomenclature described by Schaklee et al. (1990) was used.

Average heterozygosity (H) was calculated as outlined by Nei (1975). A

G-test for goodness of it was used for non-duplicated loci to determine possible

deviations of allele frequencies from expected Hardy-Weinberg proportions

(Sokal and Rohlf, 1969). The gene diversity analysis (Chakraborty et al., 1982)

was used to test the equality of allelic frequencies among samples within and

between populations. Genetic distances (D=Nei, 1972, D°=Nei, 1978) and C

Cavalli-Sforza and Edwards (1967) chord distance) between selected

3.3


populations were calculated using BOOSYS (Swofford and Selander, 1981).

Standard errors of D and D' were calculated according to the methods of Nei

and Roychoudhury (1974). A modification of the Wagner procedure

(FREQPARS), described by Swofford & Berlocher (1987), was used to construct

a phylogenetic tree by applying the principle of parsimony to allele frequency

data, and DISWAG (Swofford and Selander, 1981) was used to construct a

cladogram from Cavalli-Sforza and Edwards (1967) chord distance values.

ES U LTS

Thirty protein coding loci provided interpretable results, of which 23,3%

displayed polymorphism. The enzyme commission numbers, names of the

proteins examined, locus abbreviations and buffers giving the best results are

presented in Table 2. Relative mobilities, allelic frequencies and G-test values

of the polymorphic loci and loci where isozyme and allozyme differences

between populations occurred, as well as H values and standard errors thereof

are presented in Table 3. In summary, all the loci studied were expressed in

muscle tissue although the PGDH and SOD loci were more conspicuous in liver

tissue samples. The AAT, CK, GPI-2, EDIT and DH-2 protein coding loci were

not observed in liver samples. No mobility differences were found in 18 of the

30 loci (AK, ADH, AAT-1 and -2, CK, EST-2 and -3, PROT-1, -2 and -3, GPI-2,

GAPDH, G3PDH, IDDH, LDH-2, DH-1, 7.1 PO, and PEPS) examined and the

alleles at the EST-4 and PROT-4 loci were not detected in the Mogol River

population. Polymorphism was observed only at the EST-1, GPI-1, ODHP,

3.4


LDH-1, IMiDH-2, ithIE and PGIVi protein coding loci.

Average heterozygosity values for the Barbus populations studied, based on

thirty loci, ranged from 0.015 to 0.064 (Table 3). The average between

population gene diversity, excluding monomorphic loci, accounted for 77% (±

0.09) of the total diversity leaving only 23% (± 0.09) as the within population

fraction of the total diversity.

Genetic distances (Nei, 1972) between B. brevipinnis and ?B. brevipinnis

populations were greater than 0.240. Smaller genetic distances (0.028 to 0.133)

were found between the populations within these two regions respectively (Table

4). Similar trends were observed for Nei's (1978) and Cavalli-Sforza & Edwards'

(1967) cord distance values between these populations.

The phylogenetic tree (Fig. 2a) obtained by using DISWAG separate the B.

brevipinnis and ?B. brevipinnis populations. However, these groupings were not

obtained by using the FREQPARS procedure (Fig. 2b) because the latter

method produces a completely bifurcating tree, which does not allow two equal

groupings.

DISCUSSION

Differences were observed among populations in both the presence as well as

the frequencies of alternative alleles (Table 3). The two populations from the

3.5


Waterberg can be separated from the two escarpment populations based on the

relative mobilities of alleles at the PGDH, SOD and PEPA loci. The EST-4 and

PROT-4 protein coding loci were not detected in the Mogol River population and

the Sand River population could be differentiated from the other populations by

the presence of the PG *110, PG 7,1 *90 and GPI-1*90 alleles. Statistically

significant (P>0.05) deviations of alleles from expected Hardy-Weinberg

proportions occurred at the GPI-1 and a1DH loci for Sand River population

(Table 3). These deviations may have been the result of sampling error since

only 30 individuals could be sampled (Table 1). However, the selection against

a hetro- or homozygote is a common phenomenon within fish populations

(Kirpichnikov, 1981).

Barbus brevipinnis populations are normally isolated in smaller upper

catchment streams as a result of habitat selection which restricts the effective

size of breeding populations. According to Grant and Stahl (1988), species with

small isolated breeding populations will tend to lose alleles and a low H value

is therefore expected in such populations. The lowest values for H would

therefore be expected for the Mogol River population, which is isolated in a part

of a small stream above a waterfall and represents the most restricted of the

populations studied. The Sabie River population, however, reflected the lowest

H value (0.015). The lower H value calculated for this population (Table 3) may

be the result of selection due to siltation and habitat degradation, caused by

forestry and agricultural development in the catchment area of the river.

3.6


Mulder (1989) found that the H values of the nine large Barbus species from

southern Africa ranged from 0.052 for B. andrewi to 0.216 for B. mattozi. The

relatively low H value for B. andrewi was attributed to isolation and this value is

comparable with those found in the present study for the Sand and

Mogalakwena River populations. Our estimate is also in agreement with the H

values calculated from the allelic frequencies obtained by Berrebi et al. (1990),

for Saudi Arabian as well as small Barbus species from West Africa (range=

0.028 to 0.278).

The range of D values (0.240 to 0.280) found in the present study between

B. brevipinnis and ?B. brevipinnis populations (Table 4) compare well with D

values reported by Mulder (1989) for nine large Barbus species from South

Africa. Similar D values were also indicated by Agnese et al. (1990) for two small

barbs (0.128) and seven large barbs (0.086 to 0.274). Berrebi et al. (1990)

reported [D values of 0.112 to 0.565 for five small barbs and 0.460 to 1.837 for

three large barb species. Agnese et al. (1990) found D values between two

populations of large Barbus species to be in the region of 0.01. For fish,

Schaklee et al. (1982) found that D values between pairs of conspecific

populations ranged from 0.002 to 0.07 (average 0.05) and for congeneric

species it ranged from 0.03 to 0.61 (average 0.3). Therefore the values of D

(Table 4) between B. brevipinnis and ?B. brevipinnis suggest that these two

groups are congeneric species. This observation is supported by the large (77%)

between-population relative gene diversity compared to the within-population

3.7


relative gene diversity fraction (23%) of the total diversity. These differences also

indicate congeneric species rather than conspecific populations. This

assumption is substantiated by pigmentation differences between the

populations as mentioned in the introduction. On the other hand, the values of

D between the B. brevipinnis (0.028) and the ?B. brevipinnis populations (0.133)

are considerably lower, as can be expected for conspecific populations.

The phylogenetic relationship of the barbs in Fig. 2b do not conform to

geological events but the cladogram depicted in Fig. 2a clearly shows the

presence of two groups of barbs, namely a B. brevipinnis and a ?B. brevipinnis

group. The divergent branch lengths (Fig. 2a), in conjunction with the present

distribution pattern of the populations, seems to indicate that ?B. brevipinnis

populated the rivers of the Waterberg region only after the divergence of B.

brevipinnis and ?B. brevipinnis occurred. ?Barbus brevipinnis probably

populated the rivers of the Waterberg region via a linkage between the Limpopo

and the Okavango/Upper-Zambesi Rivers, a migration route suggested by

Gaigher and Poll (1973). According to these authors, many species reached the

Cape via the Nossob and Orange Rivers, including the morphologically similar

B. pallidus and B. neefi which populated the rivers south of the Limpopo River.

The above-mentioned route or stream capturing events on the north eastern

escarpment are the most likely routes whereby the ancestors of B. brevipinnis

could have reached the Sabie River. According to Partridge and Maud (1987),

the capture of the upper Orange River by the lower Orange River dated to the

3.8


late Pliocene, which represents a major change in the drainage systems and

could support such a migration route. Based on the assumption that these fish

migrated via the Nossob and Orange Rivers, and that this migration coincided

with the capture of the upper Orange River by the lower Orange River, as

suggested above, the divergence of the Waterberg and eastern escarpment

populations could have happened nearly 3 million years ago (Pliocene).

It was shown that less than 15% of all living organisms are known (according

to some estimates) and that biological systematic research is therefore badly

needed (Raven and Wilson, 1992). Although Bruton (1989) points out that the

adoption of alternative phenotypic states in nature is probably widespread and

that many populations which are currently recognized as species may be no

more than ecophenotypes of one or another homeorhetic state, Scholl (1973)

stated that morphological similarity does not necessarily imply little genetic

variation. The present study has shown that both allozyme and isozyme

differences occur between ?B. brevipinnis and B. brevipinnis (Table 3). They can

thus be viewed as different species, even though little morphological

differentiation has occurred. It is also evident that several genetically unique

populations occur within the two regions which deserve specific conservation

attention. Barbus neefi and B. pallidus, which are morphometrically and

meristically similar to B. brevipinnis (Skelton, pers. comm.), should however be

included in successive studies to determine the phylogenetic relations among

these species. The phylogenetic relationships of populations, representing

3.9


species from the major catchment areas, will not only increase our

understanding of the biology and evolution of such species, but may also give

an indication of possible migration routes, as has been demonstrated in this

study.

3.10

Isozvme and Allozyme Differences in Four Shortfin Barb

07 tv

U) to

a)

Co E a)

O

Co

O

O

U)

• cs. . _ (1.)

cci

'Es u)

o) = Q_ ".t..) a)ca3_ =c1

E -- fB co u) co > cn _c C O

I— 2u) CD 0

in 0 0 Q. RS a)

a)

IL

2 -.5

3.11

0 0.4 0.08 0.12 0.16 0.20 0.24 0.28 0.32

SAME (B. brevipinnis)

SAND (B. brevipinnis)

MOGOL (?B. brevipinnis)

MOGALAKWENA (?B. brevipinnis)

Distance from root

DISWAG

(a)

0 1.0

Character state changes FREQ'' ARS

(b)

MOGOL (?B. brevipinnis)

MOGALAKWENA (?B. brevipinnis)

SAND (B. brevipinnis)

SABLE (B. brevipinnis)


Figure 2: Phylogenetic trees obtained by using: (a) DISINAG and (b) FREQPARS, showing the relationships between the populations within the eastern escarpment group (B. brevipinnis) and the Waterberg group (?B. brevipinnis).

3.12


Table 1: Localities and coordinates where B. brevipinnis (a, b) and ?B. brevipinnis (c, d) populations were collected.

N Locality Latitude Longitude Reference in text

30 Marite Rivera 24°47'S 31 °05'E Sabie River

30 Sand River b 24°09'S 31°02'E Sand River

21 Sterk Rivier c 24°32'S 28°31'E Mogalakwena River

20 Grootspruit d 24°29'S 27°51'E Mogol River

a Type locality (Jubb, 1966), tributary of the Sabie River.

b Major tributary of the Sabie River (Incomati River System).

C Small tributary of the Mogalakwena River (Limpopo River System).

d Isolated population in small tributary of the Mogol River (Limpopo River System).

3.13


Table 2: Enzyme commission numbers (E.C. no) of the proteins separated by electrophoretic analysis, abbreviations used for loci resolved and buffers giving best results.

E.C. No Enzyme Locus Buffer

2.6.1.1 Aspartate aminotransferase AAT-1, -2 MF

1.1.1.1 Alcohol dehydrogenase ADH RW

2.7.4.3 Adenylate kinase AK TC

2.7.3.2 Creatine kinase CK RW

3.1.1.1 Esterase EST-1, -2 -3, -4 MF

- - - - General (unidentified) protein PROT-1, -2, -3, -4, -5 MF

5.3.1.9 Glucose-6-phosphate isomerase GPI-1 , -2 MF

1.2.1.12 Glyceraldehyde-3-phosphate dehydrogenase ' GAPDH RW

1.1.1.8 Glycerol-3-phosphate dehydrogenase G3PDH MF

1.1.1.14 L-Iditol dehydrogenase IDDH RW

1.1.1.42 Isocitrate dehydrogenase (NADP .) IDHP TC

1.1.1.27 Lactate dehydrogenase LDH-1, -2 TC,MF

1.1.1.37 Malate dehydrogenase MDH-1, -2 RW

1.1.1.40 Malic enzyme (NADP±) ME MF

5.3.1.8 Mannose-6-phosphate isomerase MPI MF

3.4.-.- Peptidase Dipeptidase PEPA MF Peptidase-S PEPS

5.4.2.2 Phoshoglucomutase PGM RW

1.1.1.44 Phosphogluconate dehydrogenase PGDH MF

1.15.1.1 Superoxide dismutase SOD RW

MF: continuous Tris, boric acid, EDTA buffer (pH 8.6) described by Marked and Faulhaber (1965).

RW: discontinuous Tris, citric acid (gel pH 8.7), lithium hydroxide, boric acid (tray pH 8.0) buffer system (Ridgway et al., 1970).

TC: continuous Tris, citric acid (pH 6.9) buffer system (Whitt, 1970).

3.14


Table 3: Relative mobilities, allele frequencies and G-test values of polymorphic loci and loci where differences were detected and values with standard errors for four B. brevipinnis populations.

Locus Allele

Eastern Escarpment Populations

Sable Sand

Waterberg Populations

Mogol Mogalakwena

EST-1 100 1.000 0.867 0.725 0.875 90 0.133 0.275 0.125

G-test 0.465 0.805 0.678

EST-4 100 1.000 1.000 1.000 00 1.000

GPI-1 100 1.000 0.283 1.000 1.000 90 0.717

G-test 3.913

IDHP 110 0.750 0.550 0.150 0.300 100 0.250 0.450 0.850 0.700

G-test 0.015 0.074 1.328 0.201

LDH-1 110 0.048 100 1.000 1.000 0.952 1.000

G-test 0.024

MDH-2 100 0.967 0.950 1.000 0.190 90 0.033 0.050 0.810

G-test 0.158 8.769 0.201

ME 110 1.000 1.000 - 100 - 0.786 1.000 90 0.214 -

G-test 0.002

PEPA 100 1.000 1.000 90 - 1.000 1.000

PGM 110 - 0.365 100 1.000 0.443 1.000 1.000 90 0.192

G-test 0.545

PGDH 100 1.000 1.000 90 1.000 1.000

PROT-4 100 1.000 1.000 1.000 00 1.000

SOD 100 1 .000 1.000 90 1.000 1 .000

H 0.0146 0.0620 0.0221 0.0462 Standard error ±0.0126 ±0.0295 ±0.0158 ±0.0210

3.15


Table 4: Genetic distances between populations calculated with the distance coe icients: (D = Nei, 1972, D° = Nei, 1978 and C = Cavelli-Sforza 81 Edwards, 1967 Chord Distance) for brevipinnis and ? . brevipinnis.

Population

Barbus brevipinnis

Sabie Sand

?Barbus brevipinnis

Mogalakwena Mogol

Sabie D D ' C

Sand D 0.028 D' 0.027 C 0.154

Mogalakwena D 0.240 0.269 D' 0.239 0.268 C 0.412 0.430

Mogol D 0.254 0.280 0.133 D' 0.255 0.279 0.131 C 0.420 0.439 0.334

3.16

Aliozynie differences between pop iations of

hubbyhead barb (Barbus an .plus Weber, 1897)

and Marico barb (B. g otebensis Steindachner,

1894).

CONTENTS:

ABSTRACT 4.1

INTRODUCTION 4.1


RESULTS 4.3

DISCUSSION 4.5

Figure 1: Map of the former Transvaal depicting the distribution of B. motebensis and B. anoplus with sampling sites. 4.11

Figure 2:

Phylogenetic trees obtained by using a) DISINAG and b) PAUP, showing the relationship between as well as within populations of B. motebensis and B. anoplus. 4.12

Table 1: Localities where B. anoplus (a, b) and B. motebensis (c, d) populations were collected. 4.13

Table 2: Enzyme commission numbers (E.C. no), proteins examined, abbreviations used for loci resolved and buffers giving best results. Loci nomenclature according to Schaklee et al. (1990). 4.14

Table 3: Relative mobilities (RM), allele frequencies, average heterozygosity (H), exact significance probability values (P) for polymorphic loci and loci where mobility differences were detected between B. motebensis and B. anoplus populations. 4.15

Table 4: Genetic distances between B. motebensis and B. anoplus populations calculated using ID (Nei, 1972), D' (Nei, 1978) and Dc (Cavalli-Sforza and Edwards, 1967). 4.17

Allozyme Differences Between Populations of Barbus anoplus and B. motebensis

BSTRACT

Starch gel-electrophoresis was used to assess genetic differences between

two morphologically similar barb species. Two population samples of each

species were analysed and polymorphism was detected in one or both

species, at 10 of the 30 protein coding loci examined. Relative mobility

differences of alleles among the four populations were found at 20 of these

loci (66.7%). It was concluded that the extent of the genetic differences

between the two species supports the present taxonomic status of these

species, which were previously thought to be synonymous. The genetic

differences between the species and populations are of conservation

importance and can be used to study possible migration routes and the

evolution of the species.

1NTR D CTIO

Barbus anoplus is the most widely distributed fish species south of the

Limpopo River and it is mostly limited to altitudes above 915m. Barbus

anoplus was initially described from the Buffels River (Gouritz System) in the

Cape (Jubb, 1968). Morphologically, this species resembles B. motebensis

occurring in some tributaries of the Limpopo River. According to Jubb (1968)

B. motebensis differs from B. anoplus in having a lower caudal peduncle

scale count and the breeding males of the former species exhibit numerous

conical tubercles on the snout, forehead and the lower jaw. Both Gaigher

(1969) and Groenewald (1958) experienced difficulties in separating B.

4.1


anoplus from B. motebensis and suggested that the two species are

synonymous. In the present study the genetic variation within and between

four geographically isolated populations were investigated to determine

whether B. anoplus and B. motebensis represent one or more species and

to what extent the various populations differ from each other.

TE

HAILS Ail D ET HODS R

Fifty-six B. anoplus specimens were collected from the Crocodile (Incomati

River system) and Vaal Rivers and 63 B. motebensis specimens were

sampled from the Mario° and Ohrigstad Rivers (Table 1). Figure 1 shows the

distribution of the two species in the former Transvaal and the sampling

points used in this study. Samples were analysed by starch

gel-electrophoresis as described by Engelbrecht and Van der Bank (1994).

Average heterozygosity (H) was calculated according to Nei (1978) and

exact probabilities were used to determine possible deviations of allele

frequencies from expected Hardy-Weinberg proportions (Elston and

Forthofer, 1977; Swofford and Selander, 1981). Different fixation indices

were used to analyse genetic differentiation between populations (Wright,

1978) using 10SYS-1 (Swofford and Selander, 1981): where F IT and F is are

the fixation indices of individuals relative to the total population and its

subpopulations respectively and FsT measures the amount of differentiation

among subpopulations relative to the limiting amount under complete

4.2


fixation. Genetic variance was also calculated for each level of hierarchy with

the W - DGHT78 procedure (Swofford and Selander, 1981), using the formula

of Wright (1978). According to Swofford and Selander (1981), this method

is similar to gene diversity analysis used by Nei (1973). The genetic

distances of Nei (1972, 1978), D (standard) and D' (unbiased) respectively

and the Cavalli-Sforza and Edwards' (1967) chord distances (Dc) were

calculated between populations.

Phylogenetic relationships were determined using the DOS G routine

(Swofford and Selander, 1981) with Dc values (phenetic approach) and a

cladogram (cladistic approach) was constructed by phylogenetic analysis

using parsimony (PAUP). The latter procedure uses an allelic frequency data

matrix which is transformed into a presence/absence matrix (an allele

present in sample = 1 and absent = 0). This program is guaranteed to find

the shortest (most parsimonious) tree (Swofford, 1985) and it was preferred

to FREQP RS (Swofford and Berlocher, 1987), because analysis using the

latter method produces a completely bifurcating tree that is confusing when

analysing only four populations to compare the two species.

ES U LTS

The 21 enzymes studied produced interpretable results at 30 protein coding

loci. The enzyme commission numbers, names of the proteins giving

interpretable results, locus abbreviations and buffers giving the best results

4.3


are presented in Table 2. Polymorphism was detected at 10 loci (33%) in the

four populations studied and mobility differences of alleles were present at

20 (67%) of the protein coding loci.

The relative allele mobilities at loci where differences between populations

occurred, allelic frequencies and exact significance probabilities for

polymorphic loci, as well as average heterozygosity (H) values and standard

errors are presented in Table 3. All four populations displayed identical allele

mobilities at ADH, AK, G PDH, ODDH, OD1-1', L 2 and PEP Allele

mobility differences separating the two species or the four populations from

one another was detected at the T-1, -2, Cry, EST-I - 4, GPI-I, -2, litfiD11-1-

1, -2, ia1PO, PGDH, PG DA, PROT-I - -5 and SOD protein coding loci.

Relatively low exact significance probabilities for alleles that deviated from

expected Hardy-Weinberg proportions were encountered at 60% of the

polymorphic loci studied (Table 3). Deviations from expected

Hardy-Weinberg proportions were evident at the l E protein coding locus for

all four populations studied; T-1, G 1-2 and Pr-'01--2 for the population

from the Vaal River; EST-I for the Crocodile River population; PEPS for the

Crocodile and Ohrigstad River populations (Table 3). Average heterozygosity

values for the Barbus populations studied ranged between 0.038 and 0.076

(Table 3). F-statistics mean values of -0.008, 0.849 and 0.850 were

calculated for Fps, Fir and FST respectively. The genetic variance values were

4.4


6.15, 8.25 and 2.10 for locality-species, locality-total and species-total

analysis respectively.

Genetic distance (D) between the B. anoplus the B. motebensis

populations studied averaged 0.614. Smaller D values (0.230 and 0.329)

were found between the B. anoplus and B. motebensis populations

respectively (Table 4). Values obtained by using various other coefficients

displayed a similar trend (Table 4). The phenetic tree (Fig. 2a) obtained by

using DISVV G, rooted at the midpoint of greatest patristic distance and

based on Dc values (Table 4), illustrates the genetic differences between the

barb populations studied and clearly shows the existence of two separate

groups, namely a chubbyhead (B. anoplus) and a tubercled barb group (B.

motebensis). The cladogram obtained using IPAUIP (Fig. 2b) is almost

identical to the grouping produced by the phenogram (Fig. 2a). This is

probably a result of the high genetic divergence between populations and

the relatively small influence of polymorphic gene frequencies on genetic

distances between these populations.

IDDSCUSSION

Deviations from expected Hardy-Weinberg proportions were encountered at

60% of the polymorphic loci studied (Table 3). Perfect Hardy-Weinberg

populations do not actually exist in nature and departures from

Hardy-Weinberg proportions may occur because of several factors such as

4.5


the Wahlund (1928) effect, natural selection, interbreeding and population

bottlenecks (Ferreira et al., 1984). In the present study the deviations from

expected Hardy-Weinberg proportions were mainly caused by a deficiency

of heterozygotes. A deficiency of heterozygotes can be the consequence of

selection against a heterozygote or a homozygote, which is a common

phenomenon within fish populations (Kirpicnikov, 1981). Barbus anoplus

and B. motebensis are mostly confined to the upper catchments of rivers

where natural and artificial barriers often subdivide the species into

numerous isolated populations. It is therefore possible that these deviations

from expected Hardy-Weinberg proportions may be the result of

interbreeding in small and isolated populations, causing a reduction of

heterozygotes (Chakraborty and Nei, 1977).

The H values obtained in the present study (0.038-0.079) are lower than

those found by Mulder (1989) for large Barbus species (0.052-0.216).

However, it compares favourably with the average of H value (0.051) given

by Nevo et al. (1984) for 183 species of fish and by Engelbrecht and Van der

Bank (1994) for small Barbus species. According to Berrebi et al. (1990) and

Agnese et al. (1990) small Barbus species are diploid while the large Barbus

species tend to be tetraploid and it is therefore reasonable to assume that

the relative lower t t values found in certain small Barbus species could be

associated with smaller numbers of active loci in diploids.

4.6


The fixation index (F ST), quantifies inbreeding due to population

subdivision or the reduction of heterozygosity of a subdivision due to genetic

drift (Lawson et al., 1989). The FST value (0.850) over all populations

suggests a great genetic differentiation between the populations and is also

comparable with FST values (0.609) found for isolated cave populations of

fish (Avise and Selander, 1972).

The genetic variance is similar to the gene diversity analysis of Nei (1973)

so that the variance of populations in terms of the total (8.25) gives an

indication of total genetic divergence. Most of this divergence is derived from

the variance of the populations compared with the species (6.15), which is

considerably larger than the variance for species compared with the total

variance (2.10). This is indicative of the relative large genetic differentiation

between the four populations and a relative small genetic differentiation

within the populations.

D values ranging between 0.230 and 0.798 where observed in the

present study among the four populations (Table 4) which compares well

with D values reported by Mulder (1989) between nine large Barbus species.

Similar ED values were also found by Agnese et al. (1990) between two

species of small barbs (0.128) and seven species of large barbs (0.086-

0.274). Berrebi et al. (1990) reported values of between 0.112 and 0.565 for

five small barbs. The former author obtained a D value between two

4.7


conspecific populations of large Barbus species of approximately 0.01. For

fish, Schaklee et al. (1982) found that D values between pairs of conspecific

populations ranged from 0.002-0.07 (average 0.05) and for congeneric

species it ranged from 0.03-0.61 (average 0.3). According to Grant and Stahl

(1988) the boundaries between taxonomic categories are not sharp but, in

general, the distances between conspecific populations are larger than 0.05

and average about 0.40 for congeneric species. The average value of D

between B. motebensis and B. anoplus (0.614) in the present study falls

within the upper range for congeneric species as discussed above,

supporting the present taxonomic status of separate species. It is likely that

sympatric B. anoplus and B. motebensis communities can occur and that it

could have caused some confusion concerning the specific status of the two

species. The genetic trees (Fig. 2a and b) also depict two groups of

genetically different barbs, namely a chubbyhead barb group (B. anoplus)

and a tubercled barb group (B. motebensis). The comparatively high ID value

(0.329) found between the two B. motebensis populations also falls within

the range for congeneric species, mainly as a result of relative mobility

differences of monomorphic (fixed) alleles at the EST-2, -3 GIPD-1, -2,

MID B1-1, -2 and PROT-5 protein coding loci (Table 3). The presence of these

biochemical markers should be investigated in relation to other

taxonomically associated barb species from southern Africa (e.g. B.

amatolicus and B. gurney!) to determine the taxonomic significance of these

differences. The unexpectedly high level of divergence between the two B.

4.8


anoplus populations examined (D=0.230) on the other hand is the result of

mobility differences at fixed alleles at the GPI-1, -2, MPO and PGODH protein

coding loci (Table 3). The taxonomic importance of geographically

subdivided B. anoplus populations have been detected by Barnard (1943),

who divided the chubbyhead barbs into two species, namely B. karkensis

from Natal and B. anoplus for the rest of its distribution. Barnard (1943) also

subdivided B. anoplus into three geographically isolated forms (Orange,

Olifants and Gouritz River Systems) substantiated by morphological

differences among the three forms. The genetic differences between the B.

anoplus populations found in the present study suggest that the

morphological subdivision by Barnard (1943) may be substantiated by a

more detailed study of the genetic differences between these geographically

subdivided B. anoplus populations.

These results also suggest that the dispersion and isolation of these fish

species into the different rivers of southern Africa have created ideal

conditions for speciation. These conditions would result in allopatric

speciation, which is a very common phenomenon in fish populations (Bush,

1975). A subdivided population structure will result in a faster rate of adaptive

morphological evolution and a founder effect in such populations will most

likely lead to genetic changes (Templeton, 1980). According to the latter

author, an adaptive divergence mode of speciation can be present where

populations are divided by intrinsic barriers, as with the present study. Since

4.9


the mutation process is random and selection always interacts to some

degree with genetic drift, ordinary micro-evolutionary processes lead to

adaptive divergence between isolated populations even if they inhabit

identical environments. However, the rate of adaptive divergence can be

greatly increased if the environments are also different. The large genetic

variation obtained between the B. anoplus and B. motebensis populations

could therefore be the result of such processes. The ecological isolation of

these species into small isolated populations make these species valuable,

mainly because the genetic variation between the populations can be used

to study micro-evolution, speciation and migration of fish in South Africa.

4.1


(1) cs)

U-

Map

of t

he fo

rme

r Tra

nsva

al d

epic

ting

the

dist

ribu

tion

of

B. m

ote

bens

is a

nd B.

ano

plu

s w

ith s

amp

ling

site

s.

4.11

(b)

MARICO (B. motebensis)

CROCODILE (B. anoplus)

VAAL (B. anoplus)

0 1.0

Character state changes PAIR

OHRIGSTAD (B. motebensis)

OHRIGSTAD (B. motebensis) (a)

MARICO (B. motebensis)

CROCODILE (B. anoplus)

VAAL (B. anoplus)

J

0 0.4 0.08 0.12 0.16 0.20 0.24 0.28 0.32 0.36

Distance from root

DESWAG


Figure 2: Phylogenetic trees obtained by using a) DISWAG and b) PAUP, showing the

relationship between as well as within populations of B. motebensis and B. anoplus.

4.12


Tab11-1 o LocaDMes where z ant.' .11us (a, L) and la

motel ensis (c, d) p. pull Uons were c , Hect d. .

Species Locality Lat. Long. River

31 B. anoplus Buffelskloofspruit a 24°47'S 30°30'E Crocodile River

25 B. anoplus Blesbokspruit b 26°11'S 28°23'E Vaal River

33 B. motebensis Ohrigstad River c 24°53'S 30°36'E Blyde River

30 B. motebensis Kaaloog se Loop d 25°47'S 26°24'E Marico River

a Tributary of the Crocodile River (Incomati River System).

b Tributary of the Vaal River (Orange River System).

Tributary of the Olifants River (Limpopo River System).

d Tributary of the Marico River (Limpopo River System).

4.13


Table 2: Enzyme Commission Numbers, proteins examined, abbreviations used for loci resolved and buffers giving best results. Loci nomenclature according to Schaklee et al. (1990).






3.1.1.1 Esterase EST-1, -2 , -3 MF


5.3.1.9 Glucose-6-phosphate isomerase GPI-1, -2 MF

1.2.1.12 Glyceraldehyde-3-phosphate dehydrogenase GAPDH RW



1.1.1.42 Isocitrate dehydrogenase (NADP+) IDHP TC

1.1.1.27 L-Lactate dehydrogenase LDH-1, -2 TC,MF


1.1.1.40 Malic enzyme (NADP+) ME MF


3.4.-.- Peptidase Dipeptidase PEPA MF Peptidase-S PEPS




MF: continuous Tris, boric acid, EDTA buffer (pH 8.6) described by Markert and Faulhaber (1965).

RW: discontinuous Tris, citric acid (gel pH 8.7), lithium hydroxide, boric acid (tray pH 8.0) buffer system (Ridgway et al., 1970).


4.14

Table 3: Reiative rnobiiities (RI I), allele frequencies, average heter zyi osity ( , ex ct significance probabilities values (P) for polymorphic loci and loci where mobility differences were detected betw en 11_3 motebensis and °plus populations.

Barbus motebensis Barbus anoplus

Locus Rilfl Ohrigstad Mario° Crocodile Vaal

AAT-1 100 1.000 1.000 90 1.000 0.652 80 0.348

P 0.004

AAT-2 100 1.000 90 1.000 1.000 00 - 1.000

CK 100 1.000 1.000 90 - 1.000 1.000

EST-1 100 0.952 1.000 90 1.000 1.000 - 80 0.048

P 0.049

EST-2 100 1.000 1.000 1.000 90 1.000

EST-3 100 1.000 1.000 90 1.000 00 1.000

G3PDH 100 0.970 1.000 1.000 0.900 90 0.030 0.100

P 1.000 1.000

GPI-1 100 1.000 90 0.030 1.000 80 .970 70 1.000

0.015

GPI-2 100 0.300 90 1.000 0.700 80 1.000 70 0.984 60 0.016

P 1.000 0.012

LDH-1 100 0.985 1.000 1.000 1.000 90 0.015

P 1.000

MDH-1 100 1.000 90 1.000 1.000 1.000


4.15


TA LE 3 - continued

Locus WI

Barbus motebensis

Ohrigstad Marico

Barbus anoplus

Crocodile Vaal

MDH-2 110 1.000 -- 100 - 1.000 1.000 1.000

ME 100 0.621 0.417 0.515 0.700 90 0.379 0.583 0.485 0.300

P 0.284 0.256 0.308 0.060

MPI 100 1.000 1.000 - 1.000 90 - 1.000 -

PEPS 100 0.561 0.650 0.581 0.900 90 0.439 0.350 0.419 0.100

P 0.078 1.000 0.139 1.000

PGDH 100 - 1.000 - 90 1.000 1.000 - 1.000

PGM 100 - - 0.100 90 1.000 1.000 1.000 0.900

P 1.000

PROT-1 100 1.000 1.000 - 1.000 90 - 1.000 -

PROT-2 100 0.136 - - 90 0.864 1.000 - - 80 - - 0.440 70 - - 1.000 0.560

P 1.000 0.420

PROT-3 100 1.000 1.000 - - 90 - 1.000 1.000

PROT-4 100 1.000 0.917 1.000 1.000 90 - 0.083 - -

P 1.000

PROT-5 100 - 1.000 1.000 1.000 00 1.000 - -

SOD 100 - 1.000 1.000 10 1.000 1.000 - -

H 0.044 0.038 0.044 0.079 a ± 0.024 ± 0.023 ± 0.023 ± 0.029

4.16


Table 4: Genetic distances between 0. motebensis and z. anoplus populations calculated using D ( ei, 1972), Du( Nei, 1978) and roc (Cavalli-Sforza nd Edwards, 1967).

Barbus motebensis Barbus anoplus

Ohrigstad Mario° Crocodile Vaal

Mario° River

D 0.329 -

D' 0.328 -

Dc 0.470 -

Crocodile River

D 0.798 0.660 -

D' 0.797 0.659 -

Dc 0.658 0.616 -

Vaal River

D 0.575 0.424 0.230 -

D' 0.574 0.423 0.228 -

Dc 0.596 0.530 0.416 -

4.17

Genetic relationships betw. n Mari= barb

(Barbus motebensis Steindach r, 1! 4) 9 'edtaH

barb (B. gum yi G nth r9 186 ) a d A atoL barb

(B. amatoiicus Skeito 199 ).

C NTENTS:

ABSTRACT 5.1

INTRODUCTION 5.1


RESULTS 5.6

DISCUSSION 5.7

Figure 1: Map of southern Africa depicting the distribution of B.

motebensis, ?B. motebensis, B. gurneyi and B. amatolicus

with sampling sites. 5.14

Figure 2: Phenetic tree obtained by using DHSPAN and D showing

phylogenetic relationship between B. motebensis, ?B.

motebensis, B. gumeyi and B. amatolicus. Bootstrap numbers

are listed at nodes. 5.15

Table 1: Enzyme commission numbers (E.C. no), proteins examined,

abbreviations used for loci resolved and buffers giving best

results. 5.16

Table 2: Relative mobilities, allele frequencies, average heterozygosity

(H), exact significance probability values (P) for polymorphic

loci and loci where mobility differences were detected

between four Barbus populations. 5.17

Table 3: Genetic distances calculated between populations of ?B.

motebensis, B. motebensis, B. gurneyi and B. amatolicus

using D (Nei, 1972), 13' (Nei, 1978) and Dc (Cavelli-Sforza &

Edwards, 1967). 5.19

Genetic Relationships Between Barbus motebensis, B. gumeyi and B. amatolicus.

A ,t, ST : A CT

Starch gel-electrophoresis was used to assess genetic differences between two

populations of B. motebensis and two other taxonomic related and

morphologically similar barb species. Ten of the 29 protein coding loci studied

were polymorphic. In spite of the small biometric and meristic differences

between the species studied, the extent of genetic differences suggests four

distinct species. The evidence in support of a new and endemic barb species

in the Ohrigstad River further adds to the current conservation importance of

the rivers of the north-eastern escarpment as a "hotspot" for endemic fish

species.

INTRO UCT1

Chubbyhead barbs are endemic to South Africa (Skelton, 1993) and comprise

the chubbyhead barb (B. anoplus), the redtail barb (B. gurney°, the Marico barb

(B. motebensis) and the Amatola barb (B. amatolicus). Common morphological

features between chubbyhead barbs (Cyprinidae) include a flexible primary

dorsal ray (D iii, 7); anal fin (A iii, 5); scales with numerous radiate striae; at

least three or four scales between the lateral line and pelvic fin; and no distinct

pelvic axil scale. Pigmentation in these barbs is usually simple, consisting of an

indefinite body stripe and a small spot on the base of the caudal fin.

Several systematic studies of the Cyprinidae have used tuberculation as

5 . 1


a primary character to differentiate species-groups, species and subspecies

(Lachner, 1967; Gibbs, 1957; Skelton, 1988). Therefore, the presence of

tubercles in B. gurneyi, B. motebensis and B. amatolicus clearly differentiates

these species from B. anoplus. The former two species also differ from B.

anoplus in having a lower caudal peduncle scale count and two pairs of barbels

(Jubb, 1968; Skelton, 1993). Barbus amatolicus, which occur sympatrically with

B. anoplus, remained undetected until recently because of the subtle

differences between the two species, but its inferior mouth (compared with a

terminal mouth), tubercles in males and one pair of well-developed barbels set

it apart from B. anoplus (Skelton, 1990). In the rivers of Mpumalanga and

Kwazulu-Natal, B. anoplus and B. motebensis are only found above an altitude

of 915m while B. gumeyi occurs in rivers at an altitude between 300 and 900m.

This difference in habitat preference may be the reason why B. anoplus and B.

gurneyi have not been collected from the same localities (Crass, 1964).

Gaigher (1969) and Groenewald (1958) could not separate B. anoplus

from B. motebensis and therefore suggested that the two species might be

synonymous. However, Engelbrecht and Van der Bank (submitteda) have

confirmed the taxonomic validity of B. anoplus and B. motebensis.

Morphological examination of several preserved specimens from a collection

at the Fisheries Institute (Lydenburg) as well as Skelton's (1993) distribution

maps suggest that the distribution of B. anoplus and B. motebensis may

5.2


overlap in the upper reaches of the Olifants and Crocodile Rivers (Limpopo

System, former Transvaal). It is therefore possible that sympatric communities

were interpreted as intraspecific variation in the past, causing the confusion

concerning the specific status of B. motebensis.

Engelbrecht and Van der Bank (submitted') also found that populations of

B. motebensis from the Ohrigstad River (former Transvaal) differed genetically

from specimens from the type locality in the Marico River and, therefore,

suggested that its genetic relationship with the other "tubercled" chubbyhead

barb species should be investigated further. The possibility of a new barb

species in the eastern Transvaal escarpment was initially suggested by Crass

(1960; 1964) who considered some barb specimens collected from a small

tributary of the Sabie River (which is in the close proximity of the Ohrigstad

River) comparable to B. gumeyi. According to Gilchrist and Thompson (1913),

Jubb (1967) and Skelton (1993) B. gurneyi occurs in the rivers of Kwazulu-

Natal below an altitude of 900m, leaving the specific status of ?B. motebensis

from the north-eastern Transvaal escarpment uncertain.

Because of these difficulties involved in the identification of the different

chubbyhead barb species, a reliable method independent of morphological

characteristics was needed to define the nature of these species. The

computation of genetic distances based on data from several loci has been

5.3


used to evaluate the taxonomic relationship between taxa (Berrebi et al., 1990;

Agnese et al., 1990; Engelbrecht and Van der Bank, 1994). In the present

study, the genetic variation within and between populations of "tubercled"

chubbyhead barbs (B. gumeyi, B. motebensis and B. amatolicus) were

investigated to determine phylogenetic relationships between species and to

examine their relationship with ?B. motebensis from the Ohrigstad River.

MATE i

LS AND METH •

I S i■

Eighty-three specimens were collected from four localities representing three

different species and the ?B. motebensis population from the Ohrigstad River.

Samples were analysed by starch gel-electrophoresis as described in

Engelbrecht and Van der Bank (1994). Figure 1 shows the distribution of B.

motebensis, B. gumeyi, B. amatolicus in southern Africa and the sampling sites

used in this study.

Average heterozygosity values (H) were calculated according to Nei (1978)

and exact probabilities were used to determine possible deviations of allele

frequencies from expected Hardy-Weinberg proportions (Elston and Forthofer,

1977; Swofford and Selander, 1981). Different fixation indices were used to

analyse genetic differentiation between populations (Wright, 1978) using

00SYS-1 (Swofford and Selander, 1981): where F IT and Fis are the fixation

indices of individuals compared with the total population and its subpopulations

5.4

Genetic Relationships Between Barbus motebensis, B. qumeyi and B. amatolicus.

respectively, and FsT measures the amount of differentiation among

subpopulations compared with the limiting amount under complete fixation.

Genetic variance was also calculated for each level of hierarchy with the

WRIGHT78 procedure (Swofford and Selander, 1981), using the formula of

Wright (1978). According to Swofford and Selander (1981), this method is

similar to gene diversity analysis used by Nei (1973). The genetic distances of

Nei (1972; 1978), D and D' respectively, and Cavalli-Sforza and Edwards'

(1967) chord distance (Dc) were calculated between populations.

Phylogenetic reconstructions were done by (1) a phenetic approach using

the D1SP N analysis (Ota, T., 1993. Pennsylvania State University), neighbor-

joining (NJ) method and Dc values to construct a phenogram (1000 bootstrap

replications were performed) and (2) a cladistic approach with phylogenetic

analysis using < parsimony (PAUP). The latter procedure uses an allelic

frequency data matrix transformed into a presence/absence matrix (an allele

present in sample=1 and absent=0). This program is guaranteed to find the

shortest (most parsimonious) tree (Swofford, 1985) and it was preferred to

F EQP RS (Swofford and Berlocher, 1987), because analysis using the latter

method produces a completely bifurcating tree that is confusing when analysing

only four populations.

5.5


ES U LTS

The 20 enzymes studied produced interpretable results at 29 protein coding loci

(Table 1). The enzyme commission numbers, names of the proteins giving

interpretable results, locus abbreviations and buffers giving the best results are

presented in Table 1. All four populations studied displayed identical isozyme

mobilities for T-1, ADIHI, K, GAPDH, ID H, L H-2, PEP , PROT-1 and

SOD. Polymorphism (no criteria) was detected at 10 of the protein coding loci

in at least one of the populations studied (Table 2). Mobility differences among

the four populations were present at 13 (43%) of the protein coding loci eA AT-2,

CK, EST-1, -2, -3 GPI-11, -2, MDH-1, -2, PD, PGM, P OT-3 and -5). The

relative mobilities for loci where allozyme differences between populations

occurred, allelic frequencies and exact significance probabilities of polymorphic

loci as well as the average heterozygosity (H) values and standard errors (a)

thereof are presented in Table 2.

Average heterozygosity values for the Barbus populations studied, based

on all 30 protein coding loci, ranged between 0.021 and 0.110 (Table 2). Exact

significance probabilities of polymorphic loci, calculated for the various

populations studied, indicates a deviation from expected Hardy-Weinberg

proportions ( 0.95) in six (35.3%) of the 17 instances (Table 2). At four of the

loci deviating from expected Hardy-Weinberg proportions, a heterozygote

deficiency was indicated, whereas a heterozygote excess was observed only

5.6


at the E protein coding locus for the B gurneyi, B. amotolicus and Ohrigstad

River populations.

Population differentiation was examined by calculating fixation indices Fis ,

FIT and FsT for each locus as well as the mean value across all loci. The mean

Fis, FIT and FsT values across all loci were 0.042, 0.878 and 0.802 respectively.

The genetic variance components (Wright, 1978), based on a hierarchy of one

group with four species, were 5.93, 8.02 and 2.1 for the species towards the

total, the species towards the group and the group towards the total

respectively.

Genetic distance (D) between the populations averaged 0.343. Values

obtained from various other coefficients displayed a similar trend (Table 3). The

phenetic tree (Fig. 2) obtained by DISPA 1, based on 13 values (Table 3),

illustrates the genetic differences between the barb populations studied. It

shows that B. gurneyi is genetically closest to B. amatolicus and that ?B.

motebensis is closer related to these two species than to B. motebensis (Marico

River). A cladogram closely resembling the grouping obtained in the above-

mentioned phenogram (Fig. 2) was obtained using P UP.

DOSCUSSION

Deviations of allele frequencies from expected Hardy-Weinberg proportions

5.7


(P<0.95) were present in 40% of the polymorphic loci in all the populations

studied. The allele frequencies at four of the five loci deviated from expected

Hardy-Weinberg proportions due to a deficiency of heterozygotes whereas an

excess of heterozygotes was observed at WilE for the B. gumeyi, B. amotolicus

and ?B. motebensis populations. Selection against heterozygotes or

homozygotes is a common phenomenon in fish populations (Kirpichnikov,

1981) and despite the fact that sample size in some populations was small, it

was concluded that there is no reason to question the genetic models used to

interpret the observed variation (Table 2). This assumption is supported by

comparable results for other Barbus species (Mulder, 1989; Engelbrecht and

Van der Bank, 1994).

According to Chakraborty and Nei (1977) a reduction in heterozygotes can

be the result of inbreeding, which is more often found in freshwater fish than in

sea fish because freshwater fish are often topographically subdivided into

smaller breeding populations (Utter et al., 1973). The most obvious factor

restricting the distribution of freshwater fish is the physical subdivision of river

systems. Rivers change over time in accordance with geological processes and

an actively eroding river may capture an adjacent drainage system, transferring

fauna between the systems. Other factors limiting the distribution of fish within

a river system include physical barriers such as waterfalls and dams, physio-

chemical tolerance (e.g. temperature) as well as behavioural and biological

5.8


factors. Considering that chubbyhead barbs are typically restricted to upper

catchment areas and are often isolated in the headstreams of the rivers,

inbreeding may be very relevant in some populations. Inbreeding in these

barbs may also be aggravated by further subdivision of small populations by

weirs and dams as well as the introduction of predatory fish which can reduce

their numbers significantly.

The fixation index (FsT), quantifies inbreeding due to population subdivision

or the reduction of heterozygosity of a subdivision due to genetic drift (Lawson

et al., 1989). The mean FsT value (0.802) for polymorphic loci in the

chubbyhead barbs studied was interpreted as large genetic differentiation

between the populations studied due to genetic drift. The inbreeding

coefficients Fis and FIT suggest a departure from random mating in the

populations (Lawson et al., 1989). Therefore, the relatively low mean Fis value

(0.042) is considered to be the result of a reduction of heterozygosity within the

populations studied and the relatively high mean FIT value (0.878) as indicative

of effective barriers to gene flow between populations.

Average heterozygosity values obtained in the present study are slightly

lower than those found by Mulder (1989) for large Barbus species, which may

be the result of small and/or isolated breeding populations sampled in the

present study. Species with small or isolated breeding populations tend to lose

5.9


alleles as a result of selection and inbreeding (Grant and Stahl, 1988).

However, these values are similar to the H values given by Nevo et al. (1984)

for 183 species of fish as well as the values obtained by Engelbrecht and Van

der Bank (1994) for small Barbus species. According to Berrebi et al. (1990)

and Agnese et al. (1990) small Barbus species are diploid whereas large

Barbus species tend to be tetraploid, and it is therefore reasonable to assume

that the relative lower t 1 values found in small barbs (compared to large barbs)

could be associated with the smaller number of active loci in diploids.

The significant allozyme differences observed between populations (Table

2) are reflected by D values that range between 0.252 and 0.398 (Table 3).

Unlike conspecific populations where polymorphic loci are largely accountable

for genetic differences, the extent of the genetic distances between the

populations studied is mainly the result of relative mobility differences at the

AAT-2, CGS, EST-1, -2, -3 GPD-1 , -2, MDH-1 , -2, MPH, PG , P OT-3 and -5

protein coding loci (Table 2). The D values compare well with values previously

estimated between congeneric Barbus species (Mulder, 1989; Agnese et al.,

1990; Berrebi et al., 1990; Engelbrecht and Van der Bank, 1994). For fish,

Schaklee et al. (1982) found that D values between congeneric species ranged

between 0.03 and 0.61 (average 0.3). According to Grant and Stahl (1988) the

boundaries between taxonomic categories are not sharp, but in general the

distances between congeneric species average 0.40. The average value of D

5.10


obtained in the present study among the four populations studied (0.343) thus

falls within the upper range suggested for congeneric species and ?B.

motebensis could therefore be viewed as a valid species instead of a

conspecific population of B. motebensis.

The genetic variance (Wright, 1978), based on a hierarchy of one group

with four species, showed that the variance contribution of species towards the

group (8.02) was greater than an estimate of the total variance (5.93), hence

the small value for the variance contribution of the group towards the total (2.1).

This is indicative of the large heterogeneity among the four populations studied,

which supports four congeneric species as suggested above.

Based on morphological characteristics, Skelton (1990) regarded B.

amatolicus most similar to B. gumeyi. According to this author the specific

status of these two species is not in doubt, because the former species differs

from the latter in having an inferior mouth (compared with a terminal mouth),

one pair of well-developed barbels (instead of two), different distribution

patterns of the tubercles, differences in breeding colours, a less curved lateral

line, two or three gill rakers (compared to six or seven) and a shorter first gill

arch. The importance of the genetic distance between ?B. motebensis and B.

motebensis is emphasized by the smaller genetic distance between two

morphologically distinct species (B. amatolicus and B. gurney!).

5.11


The genetic relationships between the populations (Fig. 2) correspond with

the geographical distribution of the populations studied, i.e. B. amatolicus and

B. motebensis are both genetically and geographically (Fig. 1) the most distant

of the four species. The observed genetic distances are most likely directly

linked with the migration routes used by a common ancestor, resulting in the

colonization of new and different environments, followed by isolation and

speciation as described by Lowe-McConnell (1959). It was proposed that the

rivers in southern Africa were colonized by means of several southward

invasions of fishes from further north in Africa (Gaigher and Pott, 1973; Skelton,

1993) and that B. motebensis populated the Marico River during the late

Pliocene via a linkage between the Limpopo and the Okavango/Upper-Zambezi

Rivers.

The ancestor of ?B. motebensis, B. gumeyi and B. amatolicus could have

colonized the rivers of the eastern escarpment via a major paleodrainage

system, that crossed the present course of the Olifants River through the

Mogalakwena River to join the Limpopo River, combined with stream capturing

events along the Great Escarpment induced by the massive uplift and new

erosive potential in the eastern part of the country during the late Pliocene

(Partridge and Maud, 1987). The presence of various other endemic species

in the upper catchments of the north-eastern escarpment region (Blyde,

Incomati and Pongolo Rivers), i.e. B. brevipinnis, B. treurensis, Chiloglanis

5.12

Genetic Relationships Between Barbus motebensis, B. gurneyi and B. amatolicus.

anoterus, C. bifurcus, C. emarginatus and Varicorhinus nelspruitensis (Skelton,

1993), indicate that a great deal of evolution and speciation occurred within this

region. An appropriate description for this mode of speciation is adaptive

divergence where the creation of an extrinsic isolation barrier is followed by

independent micro-evolution. A subdivided population structure will result in a

faster rate of adaptive morphological evolution and a founder effect in such

populations will most likely lead to genetic changes (Templeton, 1980). Since

the mutational process is random and selection always interacts to some

degree with genetic drift, ordinary micro-evolutionary processes lead to

adaptive divergence between isolated populations even if they inhabit identical

environments, although the rate of adaptive divergence can be greatly

increased if the environments are also different.

Based on the results of the present study and the specific status of B.

motebensis, B. gurneyi and B. amatolicus, which is in no doubt, ?Barbus

motebensis (Ohrigstad River population) represents a new and endemic

species. This study has also emphasised the importance of biochemical

methods to identify cryptic species. Although Bruton (1989) points out that

many populations that are currently recognized as species may be no more

than ecophenotypes of one or another homeorhetic state, the present and other

studies have shown that morphological similarity does not necessarily imply

little genetic variation (Grant and Cherry, 1985; Scholl, 1973).

5.13


O

c3- cB E

O

O z

0.)

cci

O

O O E

n.

ci)

O O

O C 0

0 LT.

C

17)

Q

O -o as

L .

C a) ui

4:1) Cl)

O N 2 0 fa. O. E 2 2

LL

5.14


72

100

?B. motebensis

B. motebensis

B. amatolicus

B. gurneyi

Figure 2: Phenetic tree obtained by using DHSPAN and ID showing phylogenetic

relationship between B. motebensis, ?B. motebensis, B. gurneyi and B.

amatolicus. Bootstrap numbers are listed at nodes.

5.15


Table 1: Enzyme commission numbers (E. C. no), proteins examined, abbreviations used for loci resolved and ',suffers giving best results.






3.1.1.1 Esterase EST-1, -2 , -3 MF


5.3.1.9 Glucose-6-phosphate isomerase GPI-1, -2 MF

1.2.1.12 Glyceraldehyde-3-phosphate dehydrogenase GAPDH RW



1.1.1.27 L-Lactate dehydrogenase LDH-1, -2 TC,MF


1.1.1.40 Malic enzyme (NADP + ) ME MF


3.4.-.- Peptidase Dipeptidase PEPA MF

Peptidase-S PEPS




MF: continuous Tris, boric acid, EDTA buffer (pH 8.6) described by Marked and Faulhaber (1965). RW: discontinuous Tris, citric acid (gel pH 8.7), lithium hydroxide, boric acid (tray pH 8.0) buffer

system (Ridgway et al., 1970). TC: continuous Tris, citric acid (pH 6.9) buffer system (Whitt, 1970).

5.16

Table 2: r=elative mobilities, all& frequencies, verage heterozygosity (H), exact significa ce probability values (F) for polymorphic loci and loci where mobility differences were detected between four t arbus populations.

Locus Allele ? B. motebensis B. motebensis B. amatolicus B. gurney!

AAT-2 100 1.000 1.000 1.000 00 1.000

CK 100 1.000 1.000 1.000 90 - 0.100 - 80 - 0.900 -

P 1.000

EST-1 100 1.000 1.000 1.000 90 1.000

EST-2 100 1.000 00 1.000 1.000 1.000

EST-3 100 1.000 00 1.000 1.000 1.000

GPD 100 0.970 1.000 0.100 1.000 90 0.030 0.500 80 - 0.400

P 1.000 1.000

GPI-1 100 1.000 1.000 1.000 90 1.000

GPI-2 100 1.000 90 1.000 1.000 80 1.000

LDH-1 100 0.985 1.000 1.000 1.000 90 0.015

1.000

100 1.000 70 1.000 20 1.000 10 1.000


5.17


"11-,., LE 2: CeNTNUED

Locus Allele ? B. motebensis B. motebensis B. amatolicus B. gurney!

JiDH-2 110 1.000 1.000 1.000 100 - 1.000 -

ME 100 0.621 0.417 0.300 0.700 90 0.379 0.583 0.400 0.300 70 - 0.300

P 0.284 0.706 0.582 0.480

MPI 100 1.000 1.000 - 90 - 1.000 1.000

PEPS 100 0.561 0.650 1.000 1.000 90 0.439 0.350 -

P 0.078 1.000

PGM 100 - - 0.800 90 - 0.200 1.000 80 1.000 1.000 -

P 1.000

PROT-2 100 0.136 - 90 0.864 1.000 0.700 1.000 80 - 0.300

P 1.000 0.480

PROT-3 100 1.000 1.000 0.800 - 90 - 0.200 1.000

P 1.000

PROT-4 100 1.000 0.917 0.200 1.000 90 - 0.083 0.800 -

P 1.000 0.009

PROT-5 100 - 1.000 0.200 1.000 00 1.000 0.800 ' -

P 0.//1

H 0.044 0.037 0.110 0.015 a ± 0.024 ± 0.023 ± 0.040 ± 0.015

5.18


Table 3. Genetic distances between ? . motebensis, motebensis, gurneyi and . amatolicus populations calculated using D (Nei, 1972), D' (Nei, 1978) and Dc (Cavelli-Sforza & Edwards, 1967).

. motebensis 1. motebensis matolicus

. motebensis

D

D'

Dc

0.343

0 .342

0.478

B. amatolicus

D 0.375 0.398

D' 0.368 0.390

Dc

gumeyi

0.501 0 .506

D 0.296 0.395 0.252

D' 0.295 0.393 0.245

Dc 0.453 0.509 0.422

5.19

Genetic relationships between sever sp cies

withi the chap • byh d d godc os a groups of

osces 9 ro i m ida ).

CO TE TS

ABSTRACT 6.1

INTRODUCTION 6.2


RESULTS 6.5

DISCUSSION 6.7

Figure 1: Phylogenetic tree, using =PAN and Nei's (1978) values,

depicting relationships within and between the goldie,

chubbyhead and tubercle barb groups of minnows studied.

Bootstrap numbers are listed at nodes and population

designations are presented in Table 1. 6.18

Table 1: Sample size, species and localities where populations of

chubbyhead and goldie barbs were collected. Species listed

according to existing taxonomy. 6.19

Table 2: Enzyme commission numbers (E. C. No.), proteins examined,

abbreviations used for loci resolved and buffers giving best

results. 6.20

Table 3: Relative mobilities (Rilli), allele frequencies, average

heterozygosity (H), exact significance probability values ( ) for

polymorphic loci and loci where mobility differences were

detected between chubbyhead, tubercle and goldie barb

population. See Table 1 for abbreviations of populations. 6.21

Table 4: Genetic distance values (Nei, 1972) between populations of

chubbyhead, tubercle and goldie barbs. See Table 1 for

abbreviations of populations. 6.23

Genetic relationships between seven species within the chubbyhead and goldie barb groups of minnows

ABSTRACT

The chubbyhead and goldie barbs are the most widely distributed groups of

minnows in the cooler rivers of southern Africa, south of the Limpopo River. The

morphological similarity between species within these two groups led to

uncertainty regarding the taxonomy, distribution and conservation status of the

different species. Consequently, conservation bodies experienced difficulties to

describe the variation within these two groups through the existing taxonomy and

to determine the conservation status of the different species. To contribute

towards solving this problem, starch gel-electrophoresis was used to assess the

genetic differences between populations and species within these two groups of

minnows.

The result of the present study suggests that a third group of barbs exist,

splitting the present chubbyhead group of barbs into a tubercle and a

chubbyhead group. Furthermore, the chubbyhead barbs can be subdivided into

at least two groups consisting of a eastern CapeNaal River group and a north-

eastern escarpment group (Crocodile and Tugela Rivers). The present study

also suggests that the goldie barbs consist of a B. brevipinnis group from the

north-eastern escarpment (Sabie and Pongola Rivers), a goldie barb group from

the rivers of the Waterberg (north-western area of high topographical elevation

in the catchment of the Limpopo River), and a goldie barb group from the

eastern interior (Ohrigstad, Crocodile and Tugela Rivers). The genetic distances

between three species of tubercle barbs (B. motebensis, B. gumeyi and B.

6.1


amatolicus) and a population of tubercle barbs from the Ohrigstad River (mean:

0.329) suggest that the latter population may represent a fourth species of

tubercle barb endemic to the Steelpoort and Blyde Rivers. Besides the

taxonomic and biogeographic implications of this study, it also provides

evidence that B. neefi was probably recently translocated between catchments.

BBB DUCTBON

Describing existing diversity of organisms in terms of their spatial and temporal

distribution and abundance is essential for the conservation of biotic diversity.

In southern Africa, researchers have been mainly involved in studies

surrounding the classification and distribution of freshwater fish species based

on their morphological characteristics. (Gilchrist and Thompson, 1913; Barnard

1938, 1943; Groenewald, 1958; Crass, 1960, 1964; Jubb, 1967; Gaigher, 1969,

1973, 1976; Kleynhans, 1983; Skelton, 1993). The need for phylogenetic

studies of freshwater fish species is highlighted by an extensive history of

uncertainty regarding the taxonomy of the different species of goldie and

chubbyhead barbs in these studies. Because of these uncertainties, it is not

always possible for conservation bodies to determine whether a species within

these two groups is common, rare or threatened. For example, Skelton et al.

(1995) lists the conservation status of B. motebensis as "status not known

sufficiently".

The chubbyhead barbs are endemic to South Africa (Skelton, 1993) and

6.2


according to present taxonomy, they comprise the chubbyhead barb (Barbus

anoplus), the redtail barb (B. gurney!), the Marico barb (B. motebensis) and the

Amatola barb (B. amatolicus). Tubercles in the males of the latter three species

set them apart from B. anoplus (Skelton, 1993). Chubbyhead barbs are widely

distributed in the headwaters of many rivers south of the Limpopo River.

Pigmentation in the chubbyhead barbs is usually simple, consisting of an

indefinite body stripe and a small spot on the base of the caudal fin. This group

of minnows are of ecological and scientific interest because they are frequently

the only fish species in many of these rivers and are often found in small

isolated stretches of river above waterfalls.

The "goldie" barbs are also a distinct southern African group of minnows and

comprise the goldie barb (B. pallidus), the shortfin barb (B. brevipinnis) and the

sidespot barb (B. neefi). Goldie barbs are relatively small barbs and

pigmentation usually consist of spots, dashes or a single stripe along the

midbody. The goldie barbs are limited to the headwaters of rivers and

occasionally occur with chubbyhead barbs.

The morphological similarity of the species, the difficulties encountered

separating them from each other and attempts to decide their phylogenetic

relationships showed some inconsistencies in terms of their present taxonomic

status. Therefore, a reliable method independent of morphological

characteristics was required to define the species within these groups. The

6.3


computation of genetic distances based on data from several loci has been

successfully used to evaluate the taxonomic relationship between Cyprinidae

(Berrebi et al., 1990; Agnese et al., 1990; Cook et al., 1992; Alves and Coelho,

1994; Engelbrecht and Van der Bank, 1994; Coelho et al., 1995; Karakousis et

al., 1995; Machordom et al., 1995). In the present study the phylogenetic

relationships between seven species (17 populations) were investigated to

determine their taxonomic relationships.

MATERIALS AND METHOL S

Three hundred and thirty Barbus specimens were collected from 17 localities

within South Africa (Table 1). Samples were analysed by starch

gel-electrophoresis as described by Engelbrecht and Van der Bank (1994).

Average heterozygosity (H) was calculated according to Nei (1978) and exact

probabilities were used to determine possible deviations of allele frequencies

from expected Hardy-Weinberg proportions (Elston and Forthofer, 1977;

Swofford and Selander, 1981). Genetic variances were calculated for each level

of hierarchy with the WRDGHT78 procedure (Swofford and Selander, 1981).

According to Swofford and Selander (1981), this method is similar to gene

diversity analysis (Nei, 1973). The genetic distances of Nei (1972, 1978), D and

D' respectively were calculated between populations. The D value was used

for comparisons with other studies. However, D' is a more appropriate measure

for small sample sizes and was used for determining phylogenetic relationships.

6.4


D1SP N (Copyright: Tatsua Ota, 1993, Pennsylvania State University, USA)

was used to construct a phylogenetic tree from D ° values, using neighbour-

joining and bootstrap tests (1000 replications). Phylogenetic relationships were

also determined, using a phenetic approach from D ° values, with the CASTE

and =WAG routines (Swofford and Selander, 1981), and a cladistic approach

using F EQPA (.•1) S (Swofford and Berlocher, 1987) which produces a

completely bifurcating tree.

ESULTS

Twenty-one proteins were studied which produced interpretable results at 30

protein coding loci, of which 53% displayed polymorphism. The enzyme

commission numbers of the proteins, locus abbreviations and buffers giving the

best results are presented in Table 2. Five of the loci ( DH, AK, GAPDH, DDDH

and ILDH-2) displayed monomorphic gel banding patterns. Mobility differences

of alleles between populations studied were present at 22 (73%) of the protein

coding loci studied. Products of the

T-2, EST-3 and P

T-5 protein coding A •

loci migrated cathodally. Products of the SOD protein coding locus in the four

tubercle barb populations (TU1-4 in Table 1) also migrated cathodally whereas

it migrated anodally in all of the other populations studied.

The relative allele mobilities at loci where differences between populations

occurred, allelic frequencies and exact significance probabilities for polymorphic

loci, as well as average heterozygosity (H) values and standard errors thereof

6.5


are presented in Table 3. For polymorphic loci, low exact significance

probabilities of alleles which deviated from expected Hardy-Weinberg

proportions were encountered in 48%, 58% and 100% of the cases within the

chubbyhead, tubercle and goldie barb populations studied respectively (Table

3). A deficiency of heterozygotes was indicated in 87% of the cases where

deviations from expected Hardy-Weinberg proportions occurred. As indicated

in Table 3, an excess of heterozygotes was only present at the ME protein

coding locus for three of the chubbyhead (CH1, CH2, and CH5) and tubercle

barb (TU1, TU3 and TU4) populations respectively, and at the ODHP protein

coding locus in two of the goldie barb populations (G5 and G6). Average

heterozygosity values for all the populations studied, based on 30 protein

coding loci, ranged between 0.015 and 0.110 (Table 3).

Values of ID° differed from ID values only at the second decimal. Since these

values were almost similar, reference is made to D values to compare the

results of the present study with those of other authors. The mean ID values

between the chubbyhead - tubercle, chubbyhead - goldie and tubercle - goldie

barb groups were 0.706 (range: 0.424-1.052), 0.633 (range: 0.449-0.907) and

1.006 (range: 0.770-1.247) respectively. The mean ID values between species

and/or populations within these three groups were 0.221 (range: 0.111-0.404),

0.328 (range: 0.242-0.380) and 0.182 (range: 0-0.293) for the chubbyhead,

tubercle and goldie barb groups respectively (Table 4).

6.6


The phylogenetic tree (Fig. 1) obtained by using DOSPAN and DU values

(Table 4) illustrates the genetic relationships between the barb populations

studied and depicted the existence of three main groups consisting of a

chubbyhead, a tubercle and a goldie barb group. The present study found

genetic distances of the same magnitude between three species of tubercle

barbs (B. motebensis, B. gurneyi and B. amatolicus) and a population of

tubercle barbs from the Ohrigstad River (?B. motebensis) and is depicted as a

separate species in Fig. 1. The chubbyhead barbs are subdivided into at least

two groups consisting of a group from the Vaal River (CH5) and rivers of the

eastern Cape (CH3 and CH4), and a group from the Crocodile (CH2) and

Tugela Rivers (CH1). The clustering also suggests that the goldie barbs consist

of a B. brevipinnis group from the north-eastern escarpment (G3 and G4), a

goldie barb group from the Waterberg (Mogol and Mogalakwena Rivers), and

a goldie barb group from the Ohrigstad (G5), Crocodile (G6) and Tugela Rivers

(G7). This grouping is almost identical to those obtained by using FREQP A RS,

CLUSTER and MS G with DU values. A hierarchy based on these groupings

(three main groups with nine specie-groups) gave genetic variance (Wright,

1978) values of 2.86, 11.29 and 8.43 for population to species-groups,

population to total and species-groups to total analysis respectively.

DISCUSSI

Allelic frequencies (Table 3) deviated from expected Hardy-Weinberg

proportions at a relative large number of the polymorphic loci studied (48%,

6.7


58% and 100% for chubbyhead, tubercle and goldie barb groups respectively).

Departures from Hardy-Weinberg proportions may occur because of several

factors such as the Wahlund (1928) effect, natural selection, interbreeding and

population bottlenecks (Ferreira et al.,1984). In the present study the deviations

of allele frequencies from expected Hardy-Weinberg proportions were mainly

caused by a deficiency of heterozygotes. A deficiency of heterozygotes can be

the consequence of selection against a heterozygote or a homozygote, which

is a common phenomenon within fish populations (Kirpicnikov, 1981).

Chubbyhead, tubercle and goldie barbs are mostly confined to the upper

catchments of rivers where natural and artificial barriers often subdivide the

species into numerous isolated populations. It is therefore possible that the

observed deviations of allelic frequencies from expected Hardy-Weinberg

proportions may be the result of interbreeding in small and isolated populations,

causing a reduction of heterozygotes (Chakraborty and Nei, 1977).

The H values obtained in the present study (range: 0.015-0.110) are slightly

lower than those reported by Mulder (1989) for large Barbus species (range:

0.052-0.216). However, it is similar to the H values given by Nevo et al. (1984)

for 183 species of fish, Alves and Coelho (1994) for a cyprinid species and by

Engelbrecht and Van der Bank (1994, 1996, submitted') for small Barbus

species. According to Berrebi et al. (1990) and Agnese et al. (1990) small

Barbus species are diploid while the large Barbus species tend to be tetraploid

and it is therefore reasonable to assume that the relative lower H values found

•


in small Barbus species could be associated with smaller numbers of active loci

in diploids. The highest heterozygosity values in the present study were

observed in populations sampled from the eastern Cape (TU4, CH4 and CH5),

which may suggest a possible relation between genetic diversity and this

specific habitat. This may also be related to extensive breeding (Pirie Trout

hatchery) and stocking of minnows in these rivers as trout fodder (Skelton, pers.

comm). In contrast to these, the low heterozygosity values obtained (i.e. TU3

and G1) may be the result of environmental degradation. This assumption is

supported by the fact that some of these populations were heavily infested with

parasites, showing possible stress. Alves and Coelho (1994) emphasized that

habitat degradation will inevitably lead to the reduction of intraspecies genetic

diversity.

The phylogenetic tree (Fig. 1) obtained in the present study give plausible

groupings that could contribute towards understanding the taxonomy of the

chubbyhead and goldie barb groups (three main groups and nine specie-

groups). Because the taxonomy of the species has not yet been properly

settled, this phylogenetic clustering of the different populations was used to

deduce species-groups. Based on these three main groups and nine species-

groups, it was found that most of the total genetic variance (11.29) is derived

from the species-groups towards the total (8.43), which is much larger that the

variance for populations in terms of the species-groups (2.86). This shows the

relative large genetic differentiation between species-groups and much less

differentiation between populations within these species-groups. This

6.9


delineation (Fig. 1) is therefore viewed as representative of the genetic

relationship between the populations studied. This delineation is also supported

by distribution patterns and morphological characteristics of the populations

within these species-groups (see discussion below).

Phyllogenetic subdivision of chubbyheads into a chubbyhead group and

tubenclle barb group

The most important division in Fig. 1 is the split between the chubbyhead group

and the tubercle barb group which were previously thought to be closely related

(Skelton, 1993) or synonymous (Groenewald, 1958; Gaigher, 1969, 1973,

1976). The allele mobility differences at the .A T-1 (see Fig. 2 in Appendix), -2,

Clcc, EST-1, GDH, P OT-1 - -3 and SOD protein coding loci separated the

tubercle barb group from the chubbyhead barb group (Table 3). The D value

between these two groups averaged 0.706 (range: 0.424-1.052). For fish,

Shaklee et al. (1982) found that D values between pairs of conspecific

populations ranged from 0.002-0.07 (average 0.05) and for congeneric species

it ranged from 0.03-0.61 (average 0.3). According to Grant and Stahl (1988) the

boundaries between taxonomic categories are not sharp. Usually the distances

between conspecific populations are larger than 0.05 and average about 0.40

for congeneric species. The average value of D in the present study between

chubbyhead and tubercle barbs (0.706) falls within the upper range for

congeneric species as discussed above, supporting the present taxonomic

status of separate species. Because of evolutionary age differences between

6.10


taxa, generalised average D values between congeneric species are not

necessarily reliable. However, comparisons with closely related taxa could be

useful.

According to Thorpe and Sole-Cava (1994) and Karakousis et al. (1995)

genetic divergence can be related to evolutionary time. Therefore the large

genetic differences found in the present study between the different species

and populations of chubbyhead and tubercle barbs suggest that these

populations have been isolated from each other for millions of years. In view of

these large genetic differences and the overlapping distribution patterns of

these two groups, the possibility that these two groups have their origin in

different lineages and that their morphological similarity is the result of

convergence should not be excluded.

Several systematic studies of the Cyprinidae which have used tuberculation

as a primary character to differentiate species-groups, species and subspecies

(Gibbs, 1957; Lachner, 1967; Skelton, 1988). The presence of tubercles in the

tubercle barbs would substantiate a split between the chubbyhead and tubercle

barbs. Therefore, the presence of tubercles in the tubercle barb group (B.

gurney!, B. motebensis and B. amatolicus) clearly differentiates these species

from the chubbyhead barb (B. anoplus). The former two species (B. gurney! and

B. motebensis) also differ from B. anoplus in having a lower caudal peduncle

scale count and two pairs of barbels (Jubb, 1968; Skelton, 1993). Barbus

6.11


amatolicus, which occur sympatrically with B. anoplus, remained undetected

until recently because of the subtle differences between the two species, but its

inferior mouth (compared with a terminal mouth), tubercles in males and one

pair of well-developed barbels set it apart from B. anoplus (Skelton, 1990). The

phylogenetic differences between B. anoplus and B. motebensis are discussed

in more detail by Engelbrecht and Van der Bank (submitteda).

Phyiogenetic subdivision of the tubercle barb group of minnows

Engelbrecht and Van der Bank (1996) discussed the phylogenetic relationships

between three species of tubercle barbs (B. motebensis, B. gurneyi and B.

amatolicus) and a population of tubercle barbs from the Ohrigstad River, and

showed that the latter population may represent a fourth species of tubercle

barb endemic to the Steelpoort and Blyde Rivers. The comparatively high D

value (0.327) calculated between the B. motebensis population (TU2) and the

tubercle barb population from the Ohrigstad River (TU1), which were previously

classified as B. motebensis, is similar to

values obtained between three I)

species of tubercle barbs and falls within the range predicted for congeneric

species. This division is mainly caused by fixed allele differences at the

following monomorphic loci:

T-2, EST-2, -3, GPi-1, -2 (see Fig. 3 in AA

Appendix), DH-1, -2 and PROT-5 (Table 3).

The present study has shown that the tubercle barb population from the

Ohrigstad River (TU1) is phylogenetically closest to the B. gurneyi (TU3)

6.12


population (Figure 2). Morphological similarities between the tubercle barb from

the Ohrigstad River (TU1) and B. gumeyi (TU3), which divide them from B.

motebensis (TU2), is the presence of only two to three scales between the

complete lateral line and the ventral fin in comparison with four scales and an

incomplete lateral line in the latter species. However, the D value (0.282)

between the tubercle barb population from the Ohrigstad River (TU1) and B.

gurneyi (TU3) also falls within the range for congeneric species and is mainly

the result of mobility differences at the EST-I, -3,

Dll-I-1,liiiiP0 and IPG^Y1 protein m

coding loci. The first reference to this tubercle barb species in the north-eastern

escarpment was made by Crass (1960, 1964), who considered barb specimens

collected from a small tributary of the Sabie River (which is in the proximity of

the Ohrigstad River) comparable to B. gurneyi.

Phylogenetic subdivision of the chubbyhead barb group of minnows

Based on the result of the present study (Fig. 1) the existing chubbyhead group

of barbs are subdivided into at least two groups: B. anoplus from the Vaal River

and eastern Cape rivers, and ?B. anoplus from the Crocodile and Tugela Rivers

(north-eastern escarpment). The mean value of

(0.274) between these two [I)

groups falls within the range for congeneric species and is mainly supported by

relative mobility differences of alleles at the GR-2, PGDFI (see Fig. 4 in

Appendix), PROT-1 and -2 protein coding loci. This subdivision agrees largely

with Barnard's (1943) suggestion to subdivide the chubbyhead barbs into two

species namely B. karkensis from Natal, and B. anoplus from the rest of its

6.13


distribution. The most obvious morphological difference found in the present

study between these two groups is the presence of an incomplete lateral line

in the Vaal River/eastern Cape group and a complete lateral line in the north-

eastern escarpment group. The latter group also lack the distinctive

"chubbyhead" profile. These differences are discussed in more detail by

Barnard (1943) and Gaigher (1969, 1973, 1976).

Phyiogenetic subdivision of the goldie barb group of rninnows

The goldie barbs are depicted in Fig. 1 as a group separate from the

chubbyhead and tubercled barbs. This is supported by relative mobility

differences of alleles at the LIDF0-1 (see Fig. 1 in Appendix), « E, PGDH, PROT-

1 and SOID protein coding loci. Although Skelton (1993) divided the goldie barbs

into three species, he mentioned that the taxonomy of the goldie barbs has not

yet been properly settled. According to this division the goldie barbs consist of

B. pallidus which has a divided distribution in the coastal streams of eastern

Cape, and highveld tributaries of the Vaal, Crocodile, Steelpoort, Pongola and

Tugela Rivers; B. neefi with a divided distribution in the Steelpoort River

(Limpopo River System), and Upper Zambezi, Kafue Rivers and rivers of

southern Zaire; B. brevipinnis with a divided distribution in the Sabie River

(Incomati River System), and rivers from the Waterberg. Although the latter

distribution was not referred to in the publication mentioned above, material

from this locality was referred to as B. brevipinnis by the same author.

6.14


Based on the result of the present study (Fig. 1), the goldie barbs were

divided into three groups that differ slightly from these groupings. Group 1

consists of the Sabie, Sand and Pongola River populations (G1, G2 and G8).

It is believed that this group represents the species B. brevipinnis because G1

was sampled close to its type locality. It is interesting that the Pongola River

population (G8) was classified as B. pallidus during sampling (Table 1).

Group 2 consists of the Ohrigstad, Crocodile and Tugela River populations

(G5, G6 and G7), which suggests that the B. pallidus and B. neefi populations

studied are conspecific populations and not congeneric species. However, B.

pallidus and B. neefi were not sampled from the vicinity of their type localities

for the present study (eastern Cape coastal streams and headwaters of the

Zambezi River respectively). These localities are geographically completely

separated from the populations studied and may represent two distinct species-

groups. This conclusion must be confirmed by further studies. The D value

between populations G5 and G6 in the present study was zero, which is unlikely

for two congeneric species completely isolated from each other in different river

systems (Crocodile River in the Incomati System and Ohrigstad River in the

Limpopo River System). In comparison, the D value between these two

populations and G7 were 0.125 and 0.128 respectively. Agnese et al. (1990),

Machordom et al. (1995) and Cook et al. (1992) obtained D values ranging from

0.001-0.01 between conspecific populations of cyprinids (including Barbus).

Therefore, it is believed that the low D value obtained between these two goldie

6.15


barbs (G5 and G6) show conspecific population variation and that these fish

could have been accidentally translocated from the Lydenburg Fisheries Station

to these localities, which would explain the genetic similarity between the two

populations. This hypothesis is supported by the fact that Gaigher (1969, 1973)

did not sample goldie barbs during their surveys in the Crocodile River (G6),

and that B. neefi (G5) was collected for the first time in the Ohrigstad River

during the mid-sixties (Jubb, 1967, 1968).

Group 3 consists of the Mogol and Mogalakwena River populations (G3 and

G4). Engelbrecht and Van der Bank (1994) suggested that this group may

represent a new species and that it may be limited to the rivers of the

Waterberg. Although the subdivision of group 1 and 2 are supported by a

relative low bootstrap value (38), this subdivision is also supported by

morphological differences between specimens from populations G5 (group 2)

and G1 (group 1). These differences include a shorter head with neurocranium

more curved, narrower ethmoid and infraorbital bone in the former species

(Skelton, pers. comm.).

The present study contributed towards understanding the phylogenetic

relationships, taxonomy and distribution (artificial and natural) within the

chubbyhead and goldie barb groups of minnows, which has troubled

researchers for decades (Barnard, 1938, 1943; Groenewald, 1958; Crass,

1960, 1964; Gaigher, 1969, 1973, 1976; Skelton, 1993). This study is yet

another illustration that molecular methods can be very useful for the study of

6.16


(conspecific) populations and morphologically very similar species. The relative

large D values between most of the populations studied suggests that most of

these populations have been isolated from each other for considerable time in

habitats which differ and that genetic divergence within and between species

have taken place. The results of this study also suggest that the distribution of

some species is more limited than previously thought (i.e. B. motebensis and

B. brevipinnis) and their status may be rare. However, the most important

conclusion drawn from these results is that the species-concept is a poor

measure of ecological important diversity because species are phylogenetically

not always clear cut, but could contain a continuum of genetically-unique

populations. It is known that conservation priorities are often driven by the

presence of rare and endangered species. These priorities often ignore

genetically-unique populations, which could be rare or even endangered.

Therefore, conservation priorities should perhaps not be bound by species and

species diversity, but should also strive to conserve genetic diversity within

common and often ecologically important populations such as within the

chubbyhead, tubercle and goldie barbs.

6.17

YCJ =

Iii ■1

),- U 14 , at .0., gt VI E co CO co 1.1.1 03 .-I ad = Ce 03 CC 0 < = < = < 0 CO U CO i••• CO

V

V co

te•■=.==.3

r-

U

U 1.1

U

U

U

co V

7


a) u) .c a) O

CD C a) 4—

CZ

..;:a)

02 c - (L)

(1) E a_ YE c 2 c-

E2 ca -(7; 2 8 moo

0 a) a La a) z

"C:3

U)- .

O r' Ct) C CD

C

N. 4— o IA

eL

z a) C

O 9.2 < E.2 co CL a) co

c . o

cm a C C c c -(7) co ,D) = 7) (L)- -CO

_c c 0 _a .2

j3 76 c a) cf.. :0 o

0— CI) ca

6.18


Table 1: Sat ple size, species and localities where populations of chubbyhead annd goldie rbs were collected. Species listed ccordi.r g to existing taxonomy.

N Species Locality Latitude Longitude Abbreviation

31 10 25 10 10

33 30 10 10

30 30 21 20

10 10 30 30

CHUBIBYHEAD BAR IS

B. anoplus Buffelskloofspruit A 24°47'S 30°30'E CH2 B. anoplus Ngagagane B 27°59'S 29°52'E CH1 B. anoplus Blesbokspruit C 26°11'S 28°23'E CH5 B. anoplus Gatbergvlei ° 31°15'S 28°04'E CH3 B. anoplus Bloukrans River E 33°19'S 26°33'E CH4

B. motebensis Ohrigstad River F 24°53'S 30°36'E TU1 B. motebensis Kaaloog se Loop ° 25 ° 47'S 26°24'E TU2 B. gurneyi Msinduzi H 29°38'S 30°25'E TU3 B. amatolicus Xuka River' - 31 °28'S 27°51'E TU4

GOLDIE BARBS

B. brevipinnis Marite River' 24°47'S 31°05'E 01 B. brevipinnis Sand River " 24°09'S 31 °02'E G2 B. brevipinnis Sterk River '' 24°32'S 28°31'E G4 B. brevipinnis Grootspruit I' 24°29'S 27°51'E G3

B. pallidus Manzaan River N 27°38'S 30°53'E G8 B. pallidus Ngagagane B 27°59'S 29°52'E G7 B. pallidus Buffelskloofspruit A 24°47'S 30°30'E G6 B. neefi Ohrigstad River F 24°53'S 30°36'E G5

A B c D E F G H 1 , K L M N

Tributary of the Crocodile River (Inkomati River System). Tributary of the Buffels River (Tugela River System). Tributary of the Vaal River (Orange River System). Tributary of Inxu River (Mzimvubu River System). Tributary of the Cowie River (Cowie River System). Tributary of the Olifants River (Limpopo River System). Tributary of the Marico River (Limpopo River System). Tributary of the Mgeni River (Mgeni River System). Tributary of the Xuka River (Mbashe River System). Type Locality (Jubb, 1966), Tributary of Sabie River (Incomati River System). Tributary of Sabie River (Incomati River System). Tributary of Mogalakwena River (Limpopo River System). Tributary of Mogol River (Limpopo River System). Tributary of the Pongola River (Pongola River System).

6.19

Genetic relationships between seven species within the chubbvhead and goldie barb groups of minnows

Table 2: Enzyme commission numbers (E. C. no), proteins exami ed, abbreviations used for loci resolved and buffers giving best results.


2.6.1.1 Aspartate aminotransferase AAT-1, -2

1.1.1.1 Alcohol dehydrogenase ADH

2.7.4.3 Adenylate kinase AK

2.7.3.2 Creatine kinase CK

3.1.1.1 Esterase EST-1, -2 , -3

- - - - General (unidentified) protein PROT-1, -2, -3, -4, -5

5.3.1.9 Glucose-6-phosphate isomerase GPI-1, -2

1.2.1.12 Glyceraldehyde-3-phosphate dehydrogenase GAPDH

1.1.1.8 Glycerol-3-phosphate dehydrogenase G3PDH

1.1.1.14 L-Iditol dehydrogenase IDDH

1.1.1.42 Isocitrate dehydrogenase (NADP+) IDHP

1.1.1.27 L-Lactate dehydrogenase LDH-1, -2

1.1.1.37 Malate dehydrogenase MDH-1, -2

1.1.1.40 Malic enzyme (NADP+) ME

5.3.1.8 Mannose-6-phosphate isomerase MPI

3.4.-.- Peptidase Dipeptidase PEPA

Peptidase-S PEPS

5.4.2.2 Phoshoglucomutase PGM

1.1.1.44 Phosphogluconate dehydrogenase PGDH

1.15.1.1 Superoxide dismutase SOD

MF

RW

TC

RW

MF

MF

MF

RW

MF

RW

TC

TC,MF

RW

MF

MF

ME

RW

ME

RW

MF: continuous Tris, boric acid, EDTA buffer (pH 8.6) described by Markert and Faulhaber (1965).

RW: discontinuous Tris, citric acid (gel pH 8.7), lithium hydroxide, boric acid (tray pH 8.0) buffer system (Ridgway et a/., 1970).


6.20

Genetic relationships between seven species within the chubbvhead and qoldie barb groups of minnows

Table 3: Relative mobilities (RM), allele frequencies, average heterozygosity (H), exact significance probability values (P) for polymorphic loci and loci where mobility differences were detected between chubbyhead, tubercle and goldie barb population. See Table I for abbreviations of populations.

RM CH1

CHUBBYHEAD BARBS TUBERCLED BARBS GOLDIE BARBS

CH2 CH3 CH4 CH5 TU1 TU2 TU3 TU4 01 G2 G3 G4 05 06 G7 G8

AAT-1 100 1.000 1.000 1.000 1.000 90 1.000 1.000 1.000 1.000 0.652 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 80 0.348

0.004 AAT-2 100 1.000 1.000 1.000 90 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 00 1.000 1.000 CK 100 1.000 1.000 1.000 - 90 1.000 1.000 1.000 1.000 1.000 0.100 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 80 0.900

1.000 EST-1 100 1.000 0.952 1.000 0.200 1.000 1.000 1.000 1.000 - 1.000 90 1.000 1.000 1.000 0.883 0.917 - 80 0.048 0.800 1.000 0.881 0.117 0.083 0.400 70 0.119 - 0.600

0.049 0.009 0.225 0.001 0.004 0.002 EST-2 100 1.000 1.000 1.000 1.000 1.000 1.000 0.867 0.725 1.000 1.000 1.000 1.000 1.000 90 1.000 1.000 1.000 1.000 0.133 0.275 -

P 0.002 0.017 EST-3 100 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 90 1.000 80 1.000 1.000 1.000 1.000 G3PDH 100 1.000 1.000 0.600 1.000 0.900 0.969 1.000 1.0000 0.100 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 90 0.100 0.031 0.500 80 0.400 0.400

1.000 1.000 0.016 1.000 GPI-1 100 1.000 - 90 0.030 1.000 1.000 1.000 1.000 0.283 1.000 1.000 1.000 1.000 1.000 0.100 80 1.000 0.100 1.000 0.970 0.900 70 1.000 0.900 0.717

1.000 0.015 0.025 1.000 GPI-2 100 0.300 1.000 0.800 90 1.000 1.000 0.700 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 0.200 1.000 80 0.900 - 1.000 70 0.984 60 0.016 50 0.100 -

P 1.000 1.000 0.012 0.009 IDHP 100 0.750 0.550 0.300 0.300 0.700 0.800 0.500 0.700 90 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 0.250 0.450 0.700 0.700 0.300 0.200 0.500 0.300

P 0.000 0.000 0.003 0.000 0.000* 0.560* 0.007 0.003 LDH1 100 1.000 1.000 1.000 1.000 1.000 0.985 1.000 1.000 1.000 90 0.015 80 1.000 1.000 0.950 1.000 1.000 1.000 1.000 1.000 70 0.050

1.000 0.026 MDH1 100 1.000 90 1.000 80 1.000 70 1.000 1.000 1.000 1.000 1.000 - 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000

6.21

Genetic relationships between seven species within the chubbvhead and goldie barb groups of minnows

Table 3. - Continued. CHUBBYHEAD BARBS TUBERCULED BARBS GOLDIE BARBS

RM CH1 CH2 CH3 CH4 CH5 TU1 TU2 TU3 TU4 01 G2 G3 G4 G5 G6 G7 013

MDH-2 100 1.000 1.000 1.000 90 1.000 1.000 1.000 1.000 1.000 1.000 - 80 1.000 0.967 0.917 0.190 - 70 0.033 0.083 0.810 0.933 0.967 0.650 0.550 60 0.067 0.033 0.350 0.450

P 0.017 0.004 0.007 0.001 0.017 0.022 0.015 ME 100 0.300 0.515 1.000 0.600 0.700 0.621 0.417 0.700 0.300 90 0.700 0.485 0.400 0.300 0.379 0.583 0.300 0.400 80 1.000 70 0.300 1.000 1.000 1.000 1.000 1.000 60 1.000 0.868 50 0.132

P 0.480* 0.308* 0.002 0.060* 0.284* 0.706 0.480* 0.582* 0.247 MPI 100 1.000 1.000 1.000 1.000 1.000 90 1.000 1.000 1.000 1.000 1.000 1.000 1.000 80 1.000 1.000 1.000 1.000 1.000 PEPA 100 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 90 1.000 1.000 PEPS 100 0.800 0.581 1.000 0.700 0.900 0.561 0.650 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 90 0.200 0.419 0.300 0.100 0.439 0.350

P 1.000 0.139 0.003 1.000 0.078 1.000 PGDH 100 1.000 1.000 90 1.000 1.000 80 1.000 1.000 1.000 1.000 1.000 1.000 1.000 70 1.000 1.000 60 1.000 1.000 1.000 1.000 PGM 100 0.100 0.800 1.000 0.466 1.000 1.000 1.000 1.000 1.000 1.000 90 0.150 0.100 1.000 0.200 0.155 80 1.000 1.000 0.545 0.750 0.900 1.000 1.000 70 0.379 60 0.455 -

P 1.000 1.000 1.000 1.000 0.709 PROT-1 100 - 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 90 1.000 1.000 1.000 1.000 1.000 80 1.000 1.000 70 1.000 1.000 PROT-2 100 0.136 90 0.864 1.000 1.000 0.700 80 0.450 0.440 - 0.300 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 70 1.000 1.000 60 1.000 0.550 0.560

1.000 0.420 1.000 0.480 PROT-3 100 1.000 1.000 1.000 0.800 90 1.000 1.000 1.000 1.000 1.000 0.200 80 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000

1.000 PROT-4 100 1.000 1.000 1.000 1.000 1.000 1.000 0.917 1.000 0.200 90 0.083 0.800 80 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000

P 1.000 0.009 PROT-5 100 - 1.000 1.000 1.000 1.000 1.000 0.200 1.000 1.000 1.000 1.000 1.000 1.000 1.000 90 1.000 1.000 1.000 0.800 1.000

0.111 SOD 100 1.000 1.000 1.000 1.000 1.000 - 90 1.000 1.000 1.000 1.000 1.000 1.000 80 1.000 1.000 - 70 1.000 1.000 1.000 1.000 -

0.032 0.038 0.040 0.074 0.079 0.046 0.037 0.015 0.110 0.015 0.064 0.032 0.040 0.025 0.018 0.062 0.032 0.019 0.023 0.024 0.031 0.029 0.024 0.023 0.015 0.037 0.013 0.030 0.020 0.020 0.016 0.012 0.030 0.022

• = excess of heterozygots

6.22


Table 4: Genetic distance values (Nei, 1972) between pop uliations of

chubbyhead, tubercle and goildie barbs. See Table 1 for

abbreviations of populations.

Pop

Chubbbyhead barbs Tubercle barbs Goldie barbs CHI CH2 CH3 CH4 CH5 TUI TU2 TU3 TU4 GI G2 G3 G4 G5 G6 G7

CH2

CH3

CH4

CH5

TU1

TU2

TU3

TU4

G 1

G2

G3

G4

05

G6

G7

G8

0.189

0.404

0.303

0.319

0.483

0.725

0.797

0.882

0.669

0.643

0.715

0.907

0.800

0.812

0.768

0.722

0.223

0.210

0.186

0.723

0.595

0.965

1.052

0.616

0.542

0.662

0.773

0.669

0.680

0.638

0.603

0.120

0.141

0.764

0.465

0.787

0.745

0.611

0.526

0.620

0.774

0.668

0.678

0.691

0.598

0.111

0.587

0.472

0.778

0.818

0.542

0.514

0.584

0.594

0.540

0.551

0.545

0.529

0.574

0.424

0.731

0.747

0.530

0.499

0.572

0.681

0.582

0.592

0.572

0.514

0.327

0.282

0.356

1.195

1.165

1.167

1.247

1.089

1.099

1.144

1.180

0.378

0.380

0.899

0.906

0.914

0.849

0.818

0.827

0.923

0.889

0.242

1.162

1.162

1.079

1.115

1.038

1.055

0.904

1.152

0.936

0.984

0.926

1.004

0.770

0.779

0.900

0.927

0.028

0.240

0.256

0.177

0.179

0.197

0.060

0.269

0.278

0.211

0.214

0.231

0.087

0.142

0.278

0.283

0.293

0.189

0.221

0.227

0.231

0.194

0.000

0.125

0.110

0.128

0.112 0.128

6.23

A key bas al ri blochemic 0 g netic d ta g

hokgy distrg t RI to rent \'/

phA gene& rWat s • wide— r Lops uOtHril

the chubbyhe d = rid ie b rows of

hl co Vso

CONTENTS

ABSTRACT 7.1

INTRODUCTION 7.1


RESULTS AND DISCUSSION 7.2

Table 1: Key based on morphology and distribution to

phylogenetic related species-groups within the

chubbyhead and goldie barb groups of minnows. 7.6

Key Based on Biochemical Genetic Data, Morphology and Distribution

ABSTRACT

Because of morphological similarities between species within the chubbyhead

and goldie barb groups, the taxonomies of these species have not yet been

properly settled. Starch gel-electrophoresis was used to assess the genetic

differences between populations and species within these two groups of

minnows. The results showed that phylogenetic related species-groups obtained

in the present study differed slightly from existing classification, but

corresponded with distribution patterns, pigmentation and morphological

characters. A key based on above characteristics is presented.

INT 0DUCT1t N

The chubbyhead barbs are endemic to South Africa (Skelton, 1993) and

according to present taxonomy comprise the chubbyhead barb (Barbus

anoplus), the redtail barb (B. gurney!), the Marico barb (B. motebensis) and the

Amatola barb (B. amatolicus). Tubercles in the males of the latter three species

set them apart from B. anoplus (Skelton, 1993). The "goldie" barbs are also a

distinct southern African group of minnows and comprise the goldie barb (B.

pallidus), the shortfin barb (B. brevipinnis) and the sidespot barb (B. neefi).

However, the morphological similarity between the different species within the

goldie and chubbyhead barb groups of minnows generated a great deal of

uncertainty concerning their classification and distribution patterns. Because of

these uncertainties, it is not always possible for conservation bodies to

determine whether a species is common, rare or threatened. Consequently,

7.1


Skelton et al. (1995) lists the conservation status of B. motebensis as "status not

known sufficiently". The aim of this paper is to address some of these

uncertainties by linking the phylogenetic relationships between the populations

studied with more useable external morphological characteristics, distribution

patterns and pigmentation.

TERDALS AND ETHODS

Morphometric and meristic measurements were taken from 311 specimens from

14 populations within the goldie and chubbyhead barb groups of minnows

(Chapter 2). Gene products of 30 protein coding loci for a total of 330 individuals

from 17 populations (Table 1 in Chapter 6) within the goldie and chubbyhead

barb groups of minnows were studied by means of gel-electrophoresis

(Engelbrecht and Van der Bank, submitted b). The genetic distance ( °) of Nei

(1978) were calculated between populations because it is a more appropriate

measure for small sample sizes and it was used for determining phylogenetic

relationships. DOSPAN (Copyright: Tatsua Ota, 1993, Pennsylvania State

University, USA) was used to construct a phylogenetic (Fig. 1 in Chapter 6) tree

from ° values, using neighbour-joining and bootstrap tests (1000 replications).

RESULTS AN DSCE. SSI

•

The phylogenetic tree (Figure 1 in Chapter 6) illustrates the genetic relationships

between the barb populations studied and depicts the existence of three main

groups consisting of a chubbyhead, a tubercle and a goldie barb group.

7.2


Engelbrecht and Van der Bank (1996) discussed the phylogenetic relationships

between three species of tubercle barbs (B. motebensis (TU1 and TU2), B.

gumeyi (TU3) and B. amatolicus (TU4)). They concluded that ?B. motebensis

from the Ohrigstad River (TU1) represent a fourth species of tubercle barb

endemic to the Steelpoort and Blyde Rivers (Fig. 1). The presence of two to

three scales between the complete lateral line and the pectoral fin in this

population separates it from B. motebensis (TU2). The morphological similarity

between ?B. motebensis and B. gumeyi was first noticed by Crass (1960, 1964).

The chubbyhead barb group are subdivided into at least two groups

consisting of a Vaal River/eastern Cape group (CH3, CH4 and CH5 in Chapter

6), and a north-eastern escarpment group (Crocodile and Tugela Rivers; CH1

and CH2 in Chapter 6). This phylogenetic subdivision between these two

species-groups largely agree with the morphological differences noted by

Barnard (1943) and Gaigher (1969, 1973, 1976). Gaigher (1969, 1976) found

that these two species-groups occur sympatrically in the Pongola River around

Chrissiesmeer (watershed of Vaal and Pongola Rivers).

The goldie barbs studied consist of a B. brevipinnis group from the Sabie,

Sand and Pongola Rivers (G1, G2 and G8 in Chapter 6), a goldie barb group

from the Mogol and Mogalakwena Rivers (G3 and G4 in Chapter 6), and a goldie

barb group from the Ohrigstad, Crocodile and Tugela Rivers (G5, G6 and G7 in

Chapter 6). Skelton (1993) depicts the distributions of B. pallidus and B.neefi as

7.3


three geographically subdivided areas, overlapping in the third area of

distribution (central area of distribution). However, the results of the present

study show that the B. pallidus and B.neefi populations are phylogenetically

closely related in the central area mentioned above. Although the study did not

include samples of B. pallidus (southern area), Skelton (pers. comm.) found that

B. neefi from the Ohrigstad River (central area), have a few characteristics

which set it apart from B. brevipinnis and B. pallidus (southern area), e.g. shorter

head with neurocranium more curved, narrower ethmoid and infraorbital bones.

Therefore, B. pallidus (southern area) and B. neefi (northern area) could

represent a fourth and fifth species-group. This view is strengthened by the more

stubby appearance of B. neefi (northern area) in relation to the B. pallidus/neefi

complex (central area) showed in Greenwood (1962) and Jubb (1967).

Not only did the genetic interpretation of isozymes and allozymes prove to be

a useful tool to study the taxonomy of cryptic species (i.e., goldie and

chubbyhead barbs), but fixed allele mobility differences between conspecific

populations demonstrated that these populations have evolved independently

(Engelbrecht and Van der Bank, 1994, 1996, submitteda, submitted b ). For

example, fixed allele, morphological and distributional differences between B.

motebensis from the Marico River (TU2) and ?B. motebensis from the Ohrigstad

River (TU1), suggest that these populations represent two distinct species with

limited distribution patterns. Although some goldie and chubbyhead barb species

are perceived to be quite common (i.e., B. anoplus), the above-mentioned

7.4


results suggested that many populations may be genetically unique and could

be restricted to very small catchments. These minnows are often the only fish

present in the upper reaches of many southern African rivers, adapted to local

conditions over millions of years. Therefore, these populations are of

conservation and scientific importance, not only as relics representing the

evolution of the species, but also as an integral part of a specialized ecosystem.

The above-mentioned phylogenetic related species-groups clearly

corresponded with external morphological characters, pigmentation and

distribution patterns, which are useful to identify these species-groups. Since

biochemical keys are impractical to use when access to specialised laboratories

is denied, a simplified key to these species-groups is presented (Table 1).

7.5

Fig. I a

Fig. 1 b

Fig. l c

Fig. I d

Key based on biochemical genetic data, morphology and distribution

Table 1: Key based on morphology and distribution to phylogenetic rel ted species-groups within the chubbyhead and goidie barb groups of minnows.

la Body of one or both sexes with prominent or clearly defined markings,

usually eight dorsal branched rays, less than 32 lateral line scales II

lb Body of one or both sexes without prominent or clearly defined markings,

usually seven dorsal branched rays, more than 31 lateral line scales III

it Goldie bath group

Ha Specimens normally exhibit a dark unbroken line

above the lateral line, sometimes consisting of

numerous small dashes and spots along the midbody (Fig.

1 a), headwaters of the Waterberg rivers (Fig. 1 b; G3

and G4 in Chapter 6) brevipinnis*

lib Specimens normally exhibit a dark broken line above

the lateral line, consisting of three to six dashes along

the midbody (Fig. 1c), limited to the headwaters of the

Sabie and Pongola Rivers (Fig. 1 d; G1, G2 and G8

in Chapter 6) lB. br vipinnis Jubb, 1966*

7.6

Fig. le

Fig. I f

Fig. 2a

Fig. 2b


TabBe 1: - continued

IIc Variable number of large diffuse spots in a linear

sequence along the midbody and wavy parallel lines

between each row of scales (Fig. le). Headwaters in

Mpumalanga, Kwazulu-Natal, Gauteng and Free State (Fig.

lf; G5, G6 and G7 in Chapter 6) rgeefi*

Barbus pallidus from the coastal rivers of the eastern Cape (dark

area in Fig. 2a) and B. neefi from the Upper Zambezi, Kafue and

southern Zaire systems in Zambia and Zaire (light area in Fig. 2b)

was not sampled. The distributions of B. pallidus and B.neefi shown

by Skelton (1993) overlap in the middle of Mpumalanga and

Kwazulu-Natal (dark area in Fig. 2b). However, in this area they are

phylogenetically closely related (see IIc). Skelton (pers. comm.)

found that Barbus neefi from the Ohrigstad River (former

Transvaal), have a few characteristics that set it apart from B. pallidus and B.

brevipinnis, e.g. shorter head with neurocranium more curved than the others,

narrower ethmoid and infraorbital bones. Consequently, B. pallidus from the

coastal rivers of the eastern Cape (dark area in Fig. 2a) could represent a fourth

species-group. Greenwood (1962) and Jubb (1967) show B. neefi from the

Zambezi River (light area in Fig. 2b) as a more stubby fish than ?B. neefi

collected from Mpumalanga and Kwazulu-Natal (dark area in Fig. 2b). These two

groups are geographically isolated from each other (light and dark areas in Fig.

2b), and may differ phylogenetically from one another. Barbus neefi (light area in

Fig. 2b) could represent a fifth species-group.

7.7

Fig. 3b

Fig. 3c

Fig. 3d


Table 1: - Continued

Ill Chubbyhead barb group

IIIa Ripe males without prominent (visible) white tubercles on

head, sixteen caudal peduncle scales IV

1111) Ripe males with prominent (visible) white tubercles on

head, twelve to fourteen caudal peduncle scales V

IV Chubbyhead barbs

IVa Incomplete lateral line, single pair of weakly

developed barbels, typical chubbyhead profile

(Fig. 3a), limited to the Orange River and rivers

of the eastern Cape (Fig. 3b; CH3, CH4 and CH5

la, ■ Chapter 6) an plus M. Weber, 1897*

IVb Complete lateral line, single pair of well

developed barbels, absence of typical

chubbyhead profile (Fig. 3c), limited to the

headwaters of Incomati, Pongola and Tugela River

Systems (Fig. 3d; CH1 and CH2 in Chapter 6).

anoplus*

Fig. 4a


Table 1: - Continued

V TaAberde barbs

Va Lateral line incomplete, four rows of scales between

lateral line and ventral fin, two pairs of barbels (Fig.

4a), headwaters of Crocodile and Marico Rivers -

Limpopo River system (Fig 4b; TU2 in Chapter

6) motebensis Steindachner, 1984*

Fig. 4b

Vb Lateral line complete, single pair of well developed

barbels, inferior mouth, limited to the headwaters of the

Kei and Mbashe Rivers (eastern Cape) (Fig. 4c; TU4 in

Fig. 1 in Chapter 6) amatolicus Skelton, 1990* Fig. 4c

Vc Lateral line complete, two pairs of barbels and

two or three rows of scales between lateral line

and pelvic fin (Fig. 4d) VI Fig. 4d

VIa Ripe males with orange fins, headwaters of the

Umtamvuna River northwards to the Amatikulu River

Kwazulu-Natal (Fig. 5a; TU3 in Chapter

6) fr gumeyi Gunther, 1968*

7.9


VIb Ripe males golden yellow, headwaters of the Blyde,

Steelpoort and Sabie Rivers (Fig. 5b; TU1 in Chapter 6)

?f';. motebensis* Fig. 5b

* = Possibly an undescribed species.

* = Phylogenetic grouping agree with Skelton's (1993) classification.

7.10

CONTE TS

INTRODUCTION 8.1

TAXONOMY 8.2

DISTRIBUTION AND BIOGEOGRAPHY 8.6

CONSERVATION OF THE GOLDIE AND CHUBBYHEAD BARBS 8.9

CONCLUSIONS 8.12

SUMMARY

ONTRODUCTa0M

Systematic and biogeographical distribution of species in time and space are

essential for the determination, and therefore maintenance of biotic diversity. To

this extent, researchers from conservation organisations, the Albany Museum

and the J. L. B. Smith Institute of Ichthyology have been involved for many years

in studies surrounding the classification and distribution of freshwater fish

species in southern Africa (Skelton et al., 1995). However, the morphological

similarity (Chapter 2) between the different species within the goldie and

chubbyhead barb groups of minnows generated a great deal of uncertainty

concerning their classification and distribution patterns. Because of these

uncertainties, it is not always possible for conservation bodies to decide whether

a species is common, rare or threatened. For example, Skelton et al. (1995) list

the conservation status of B. motebensis as "status not known sufficiently".

The primary objective of the present study was to focus on the phylogenetic

relationships within and between different species of goldie and chubbyhead

barbs to broaden the knowledge-base of their systematics and biogeography.

In Chapter 1, nine questions were raised which outlines some major

uncertainties regarding the classification and distribution patterns within the

goldie and chubbyhead barbs. The objectives of this study were set to address

these questions and the populations chosen for this study were accordingly

selected.

SUMMARY

Gene products of 30 protein coding loci in a total of 330 individuals from 17

populations within the goldie and chubbyhead groups of barbs were studied by

means of gel-electrophoresis. Phylogenetic relationships between populations

were used to explain the genealogy of different species-groups (Chapters 3 to

6). It was found that these phylogenetic related species-groups corresponded

with distribution patterns and morphological characters, which were useful to

identify these species-groups (Chapter 7). The results of the present study are

hereby reviewed to establish to what extent the six objectives of the study were

met. The most important contributions of the present study towards the

systematics, distribution and conservation status of these minnows include the

following:

T ONO Y

Based on the phylogenetic relationships between 17 goldie and chubbyhead

barb populations it was concluded that:

1. The chubbyhead group, as defined by Skelton (1993), could

phylogenetically be subdivided into a tubercle and a chubbyhead barb

group (Chapter 4). These two groups were previously thought to be

closely related or synonymous (Groenewald, 1958; Gaigher, 1969, 1973,

1976). Morphological characteristics such as the presence of tubercles

in the former group (B. gurney!, B. motebensis and B. amatolicus) can

serve to differentiate between these species and the chubbyhead barbs

(B. anoplus). Several systematic studies of the Cyprinidae have used

SUMMARY

tuberculation as a primary character to differentiate species-groups,

species and subspecies (Gibbs, 1957; Lachner, 1967; Skelton, 1988),

which supports this split between the chubbyhead and tubercle barb

groups. Concerning objective 3 (Chapter 1), the present study shows

beyond doubt that B. anoplus and B. motebensis are not synonymous

and that the tubercles in B. motebensis are of taxonomic importance.

With reference to objective 6 (Chapter 1), the present study found

genetic distances of the same magnitude between three distinct species

of tubercle barbs (B. motebensis, B. gurneyi and B. amatolicus) and a

population of tubercle barbs from the Ohrigstad River (?8. motebensis),

which suggest that the latter population represents a fourth species of

tubercle barb endemic to the Steelpoort and Blyde Rivers. The most

conspicuous morphological differences between the ?B. motebensis

population from the Ohrigstad River and the B. motebensis population

from the Marico River's (a type locality) are the presence of four scales

between the origin of the ventral fins and the incomplete lateral line in

the latter population compared to three scales between the origin of the

ventral fins and the complete lateral line in the Ohrigstad River

population (Chapter 5 and 6).

Barbus anoplus could phylogenetically be subdivided into at least two

groups (Chapter 6), namely: a group from the Vaal River (Orange River

System) and eastern Cape rivers and a group from the Crocodile and

Tugela Rivers (eastern escarpment). The most apparent morphological

8.3

SUMMARY

difference found in the present study between these two groups is the

presence of an incomplete lateral line in the Vaal River/eastern Cape

group in contrast to a complete lateral line in the eastern escarpment

group. The latter group also lacks the distinctive "chubbyhead" profile.

The results of the present study agree largely with the morphological

differences noted by Barnard (1943) and Gaigher (1969, 1973, 1976)

between populations of B. anoplus. Therefore, the present study

suggests that these morphological differences within the distribution

range of B. anoplus may be more than intraspecific variations and could

include interspecific variations which may be of taxonomic importance

(objective 5, Chapter 1).

4. In the present study the goldie barb populations from the former

Transvaal and Kwazulu-Natal could be divided into three groups (refer

to objective 1 and 4, Chapter 1):

Group 1: This group is delineated as the goldie barb populations from the

Sabie, Sand and Pongola Rivers and represent the species B.

brevipinnis. The Pongola River population was classified as B.

pallidus during sampling (Table 6.1). The latter locality represents

a new goldie barb distribution record (Coke, pers. comm.) and a

new species for Kwazulu-Natal. Barbus brevipinnis has a similar

distribution pattern to two other endemic species (e.g.

Varicorhinus nelspruitensis and Chiloglanis anoterus). This may

be an indication that these fish species have evolved

SUMMARY

simultaneously in the Pongola and Sabie Rivers, and that these

rivers were historically linked.

Group 2: This group comprises the goldie barb populations from the Mogol

and Mogalakwena Rivers (rivers of the Waterberg). In response to

objective 1 (Chapter 1), Engelbrecht and Van der Bank (1994)

have suggested that this group may represent a new species

(Chapter 3).

Group 3: Based on the genetic similarity between populations of B. pallidus

and B. neefi from the Ohrigstad, Crocodile and Tugela Rivers,

these three populations represent a single species-group. This

result suggests that B. neefi and B. pallidus within the former

Transvaal and Kwazulu-Natal represent conspecific population

variation and not congeneric species (refer to objective 2, Chapter

1).

For the present study, B. pallidus and B. neefi were not sampled from the

geographical area where the type specimens were collected (eastern Cape

coastal streams and upper reaches of the Zambezi and Kafue Rivers

respectively). These localities are geographically completely separated from the

populations studied and may represent two distinct species-groups. This

hypothesis is supported by the fact that Barbus neefi from the Ohrigstad River

has a few characteristics (i.e. shorter head with neurocranium more curved,

narrower ethmoid and infraorbital bones) that set it apart from B. pallidus and B.

8.5

SUMMARY

brevipinnis (Skelton, pers. comm.). However, this must be confirmed by further

studies.

DDST OBUTDON AND BDOGEOGRAPHY

The above-mentioned phylogenetic related species-groups corresponded with

distribution patterns and therefore an explanation for this inter- and intraspecific

variation in relation to existing theories is presented. Most of the theories

explaining the present distribution patterns of the fish fauna of southern Africa

suggest a central African origin (Bell-Cross and Minshull, 1965; Gaigher and

Pott, 1973; Kleynhans, 1984; Engelbrecht and Van der Bank, 1994). However,

Skelton et al. (1995) divide the fish fauna of southern Africa into a Zambezian

and a southern temperate fauna based on distinct historical derivations of these

faunas. Skelton (1993) also found evidence of closest relatives to some fish

species from the southern temperate fauna on other continents. For example,

the closest relative to Labeo umbratus was found in Asia, which suggest that it

could stem from a time before the separation of India 120 million years ago and

traces the early origins of the redfins minnows (Pseudobarbus species) back to

the early drainage history of the Cape Fold Mountains and high Drakensberg.

This would imply that a southern temperate fish fauna has evolved in the south.

According to Thorpe and Sole-Cava (1994) genetic divergence is related to

evolutionary time, therefore the large genetic differences found in the present

study between the different populations of chubbyhead, tubercle and goldie

8.6

SUMMARY

barbs suggest that these populations have been isolated from each other for

periods that do not conform to a recent invasion from the north. Karakousis et

al. (1995) suggested that a constant mutation rate of k=20x10 6 for Barbus

species conformed to geological events. If this constant mutation rate is

accepted for the groups of minnows studied, the tubercle barb species and B.

anoplus may have diverged from a common ancestor as distant as 19.08x10 6

years ago. Based on this theory, species of goldie barbs within the former

Transvaal have colonised these rivers more recently (between 2.2 and 5.6x10 6

years ago). Fixed allele mobility differences between conspecific populations

studied (Chapters 3 to 7) suggest genetic isolation of these populations for

periods which would conform to an independent and ancient lineage of a

southern temperate fish fauna as suggested by Skelton (1993) and Skelton et

al. (1995). The present study indicates that chubbyhead barbs must have been

present in the south for millions of years, before some geomorphic changes in

the present landscape occurred, which explains why chubbyhead barbs are

often found above waterfalls and why they are frequently the only species in the

headwaters of rivers. This would also explain why chubbyhead barbs are widely

distributed within South Africa but are absent from some major rivers in the

former Transvaal and the rest of the continent to the north thereof. The

presence of tubercles in some of the barb species studied and the red fins in B.

gurneyi (redtail barb) may suggest a phylogenetic relationship with the redfin

minnows (Pseudobarbus species), which is a further indication of an

independent and ancient lineage of a southern temperate fish fauna.

SUMMARY

Although the chubbyhead and goldie barbs are widely distributed within

temperate rivers south of the Limpopo River, many of the species within these

groups have a very restricted distribution (e.g. B. brevipinnis). This restricted

distribution, supported by the relative large genetic distances between the

populations studied, suggested that little or no gene flow has occurred recently

between these populations. Therefore it is unlikely that these fish populations

have been involved in a massive recent invasion from the north. Most likely,

recent movement of the chubbyhead and goldie barbs were limited to stream

capture events between rivers of the advancing watersheds, e.g. the eastern

escarpment. The present distribution pattern of B. anoplus could be partly

explained by the advancing watersheds of the eastern escarpment which

captured the upper reaches of the Orange River.

A goldie barb population (B. pallidus) collected from the Crocodile River

(Incomati River System) were genetically identical to a population (B. neefi)

collected from the Ohrigstad River (Limpopo River System). This is not likely to

occur in populations of two different species that have evolved independently

and which were completely isolated from each other in different river systems.

A possible explanation for this occurrence is that these two goldie barb

populations were accidentally translocated from the Lydenburg Fisheries station

to these localities. This theory is supported by the fact that Gaigher (1969, 1973)

did not collect goldie barbs during their surveys in the Crocodile River (?B.

pallidus), and that the goldie barb in the Ohrigstad River (?B. neefi) was

8.8

SUMMARY

collected for the first time in this river during the mid-sixtees (Jubb, 1967, 1968).

This theory exposes the possibility that many similar accidental translocations

of goldie barbs could have taken place between catchments and would explain

why B. neefi and B. pallidus, collected from two different river systems are

genetically almost identical. However, the full extent of such accidental

translocations cannot be determined without a more detailed study. Such

translocations can cause interspecific competition and genetic contamination

between conspecific populations that could lead to the loss of genetic

characteristics in some populations.

C SERVATDO OF THE GOLIDDE AND CHUBBYHEAD BARBS

Several authors have identified areas of high topographical elevation where

conditions are most favourable for sustaining relict populations of fish through

variable climatic periods, for example, the north-eastern escarpment was

identified as "hotspots" for endemic fish species richness (Kleynhans, 1984;

Skelton et al., 1995). The present study also found taxonomic significant genetic

divergence between populations of goldie and chubbyhead barbs within these

mountain areas, thus adding to its importance for conservation in accordance

with the primary goals of the IUCN. It is important to note that the water derived

from the mountainous areas is not only important for the conservation of the

local fish communities but are also essential for maintenance of the fish

communities downstream.

SUMMARY

Conserving selected streams or segments of rivers in its natural state,

comparable to the "Wild and Scenic Rivers Act" of the USA (Doyle et al., 1977),

could contribute towards the conservation of endemic fish species richness in

these areas. However, conserving the wide spectrum of endemic fish species

richness in the mountainous areas of southern Africa with its genetic diversity

component by means of reserves would be extremely difficult or impossible.

Therefore, focusing on the broader issues of holistic river catchment

conservation that embraces multiple usage of the resource whilst maintaining

the ecological functioning, would be necessary to conserve these areas. This

view was also expressed by Skelton et al. (1995).

Because the goldie, tubercle and chubbyhead barb populations are mostly

confined to small perennial streams in the upper-catchments of rivers, they are

mainly threatened by afforestation, invasive exotic plant species, mining, and the

introduction of exotic fish species. Afforestation can cause a marked decrease

in runoff during the dry season, which could reduce the number and size of small

perennial streams (Bosch and Smith, 1989). This situation can be aggravated

by invasive exotic plants such as wattles (Acacia meamsii, A. dealbata and A.

melanoxylon). These exotics can cause increased erosion, siltation and a loss

of riverine marginal vegetation, loss of pool-riffle sequences and loss of diversity

in water-depth classes. Cambray (1983) has noted that siltation affects B.

anoplus eggs. In addition forestry and invasive exotic plants could be an

important factor causing acidification of freshwater, which could increase

8.10

SUMMARY

aluminium ions that are of particular significance because of its toxicity (Mason,

1991). Exotics may also affect the timing and nutritional value of the primary

food base on which many riverine organisms are dependant (Dallas and Day,

1993). Sound forestry practices, i.e. maintenance of riparian buffer zones,

important wetlands as well as invasive exotic plant control could greatly reduce

these impacts on the ecosystem and should form an integral part of catchment

management.

The water quality of several small perennial streams in the upper-catchments

of rivers in Mpumalanga and Kwazulu-Natal, which are suitable for chubbyhead

and goldie barbs, are affected by drainage water from active and abandoned

mines. These water quality changes include increased total dissolved solids,

total suspended solids, trace metals and acidification (acid mine drainage). The

streams immediately below these mines are mostly devoid of life and reaches

further away seldom support any number of fish. Pollution control, proper

rehabilitation and comprehensive environmental impact assessments before

commencement of mining could reduce the risk of mining to the survival of

endemic fish fauna.

The habitat of goldie, tubercle and chubbyhead minnows are often stocked

with exotic piscivorous fish species. These piscivorous fish species have

decimated many endemic fish populations and is one of the most important

factors threatening the endemic fish species in the mountainous areas of

.11

SUMMARY

southern Africa (De Moor and Bruton, 1988). Therefore, to conserve the

endemic fish species richness of southern Africa, it would be important not to

expand the present distribution range of exotic species such as trout.

Exotic parasites, introduced with exotic fish species, e.g. lchthyopthirius

multifilis (white spot) and Bothtiocephales acheillognathiy (fish tape worm), were

affecting the goldie and tubercle barb populations studied, which could pose a

serious threat to their survival. In comparison with the survey of De Moor and

Bruton (1988), the present study suggests that the distribution of B.

acheillognathiy has increased and that it is also affecting two additional endemic

species (B. brevipinnis and B. gumeyi from the Sand and Mgeni Rivers

respectively). Sixty percent of the B. brevipinnis females collected in the present

study from the Pongola River was completely sterile because large numbers of

metacecaria of an unidentified digenetic trematode completely displaced their

ovaries. The rapid spread and the potential threat of these parasites to the

southern African endemic fish population stresses the importance of stringent

import control to prevent further introductions of exotic parasites.

CONCLUS00 S

The present study proved that genetic interpretation of isozyme patterns can be

a useful tool to study the phylogenetic relationships of cryptic and morphological

similar species (Engelbrecht and Van der Bank, 1994, 1996, submitteda,

submitted b, submittedc). These results contributed towards the understanding

.112

SUMMARY

and description of the systematics, genetic diversity, distribution patterns and

conservation status of the freshwater fishes of southern Africa. In view of these

results, the objectives of the present study were successfully met. Besides the

present study reaching its primary goals which were expanded as the study

evolved (see Chapter 1), it also showed taxonomic significant variation between

conspecific populations of minnows which could be the result of their ancient

lineage and independent evolution. Therefore, this study suggests that minnow

species that are often perceived to be quite common, may comprise many

genetically unique populations which could be restricted to very small

catchments. These minnows are often the only species present in the upper

reaches of many southern African rivers, adapted to local conditions over

millions of years. Thus they are of conservation and scientific importance, not

only as relic populations representing the evolution of the species, but also as

an integral part of a small and specialized ecosystems. Although the present

study could not possibly address all the questions relating to the systematics,

distribution and conservation status of these minnows, it could be regarded as

an important milestone which contributed towards the conservation and

knowledge-base of these species.

The present study shows the importance of molecular methods to study the

phylogenetic relationships between conspecific populations of cryptic species

in relation to their distribution patterns and morphology, and its application to

conservation. In a comparison of several modern and traditional molecular

8.13

SUMMARY

methodologies, Park and Moran (1995) rated the use of protein electrophoresis

for population genetic studies as excellent. Allozyme electrophoresis remained

the method of choice for Naish and Skibinski (in press) to characterize tilapia

species because the method is informative and technically simple. However,

they foresee that simplified DNA methods may prove to be more useful in many

laboratories in the future.

In retrospect, the results of the present study largely verify the taxonomic

importance of morphological differences between chubbyhead and goldie barb

species, which may have been overlooked as intraspecific variation in the past

(Chapter 7). The morphometric and meristic characterization of fish populations

and strains, confirmed by recent genetic approaches (Teugels, in press),

showed that the results of these two approaches can support each other.

Therefore it is believed that an in-depth morphological study of the chubbyhead

and goldie barbs, supported by selected biochemical analysis, could lead to a

simple key that is practical and useful for the identification the different species.

8.14

EFE r•\ E CES

References

Agnese, J. F. P., Berrebi, P., Leveque, C. and Guegan, J. -F. 1990. Two

lineages, diploid and tetraploid demonstrated in African species of Barbus

(Osteichtyes, Cyprinidae). Aquat. Living Resour. 3: 305-311.

Alves, M. J. and Coelho, M. M. 1994. Genetic variation and population

subdivision of the endangered Iberian cyprinid Chondrostoma lusitanicum.

J. Fish Biol. 44: 627-636.

Avise, J. C. and Selander, R. K. 1972. Evolutionary genetics of cave-dwelling

fishes of the genus Astyanyx. Evolution 26: 1-19.

Barnard, K. H. 1938. Notes on the species of Barbus from the Cape Province,

with descriptions of new species. Ann. Mag. Nat. Hist. 11(2): 80-88.

Barnard, K. H. 1943. Revision of the indigenous freshwater fishes of the South

Western Cape region. Ann. S. Afr. Mus. 36(2): 101-262.

Bell-Cross, G. and Minshull, J. L. 1988. The Fishes of Zimbabwe. National

Museums and Monuments of Zimbabwe, Harare.

Berrebi, P., Leveque, C., Cattaneo-Berrebi, G., Agnese, J. -F., Guegan, J. -F.

and Machordon, A. 1990. Diploid and tetraploid African Barbus

(Osteichtyes, Cyprinidae): on the coding of differential gene expression.

Aquat. Living Resour. 3: 313-323.

Bosch, J. M. and Smith, R. E. 1989. The effect of afforestation of indigenous

scrub forest with Eucalyptus on streamflow from a small catchment in the

Transvaal, South Africa. S. Afr. For. J. 150: 7-17.

Boulenger, G. A. 1909-1916. Catalogue of the freshwater fishes of Africa, 1-4.

British Museum (Natural History), London.

9.1

References

Bruton, M. N. 1989. The Ecological significance of alternative life-history styles.

In Alternative Life-history Styles of Animals (Bruton, M. N., ed.), pp. 503-

553. Kluwer Academic Publishers, Dordrecht, Netherlands.

Bush, G. L. 1975. Modes of speciation. Ann. Rev. Ecol. Syst. 6: 339-364.

Cambray, J. A. 1983. Early development and larval behaviour of the minnow,

Barbus anoplus (Pisces: Cyprinidae). S. Afr. J. Zoo!. 18: 331-336.

Carvalho, G. R. and Pitcher, T. J. 1995. Molecular Genetics in Fisheries.

Chapman & Hall, London.

Cavalli-Sforza, L. L. and Edwards A. W. F. 1967. Phylogenetic analysis: models

and estimation procedures. Evolution 21: 550-570.

Chakraborty, R. and Nei, M. 1977. Bottleneck effects on average heterozygosity

and genetic distance with the stepwise mutation model. Evolution 31: 347-

356.

Chakraborty, R., Haag, M., Ryman, N. and Stahl, G. 1982. Hierarchical gene

diversity analysis and its application to brown trout population data.

Hereditas 97: 27-22.

Coelho, M. M., Brito, R. M., Pacheco, T. R., Figueiredo, D. and Pires, A. M.

1995. Genetic variation and divergence of Leuciscus pyrenaicus and L.

carolitertii (Pisces, Cyprinidae). J. Fish Biol. 47: 243-258.

Cook, J. A., Bestgen, K. R., Propst, D. L. and Yates, T. L. 1992. Allozymic

divergence and systematics of the Rio Grande silvery minnow,

Hybognathus amarus (Teleostei: Cyprinidae). Copeia 1992: 36-44.

Crass, R. S. 1960. Notes on the freshwater fishes of Natal with the description

9.2

References

of four new species. Ann. Natal Mus. 14(3): 405-458.

Crass, R. S. 1964. Freshwater Fishes of Natal. Shuter and Shooter,

Pietermaritzburg, Natal.

Dallas, H. F. and Day, J. A. 1993. The effect of water quality variables on riverine

ecosystems: a review. TT61/93, Water Research Commission, Pretoria.

De Moor, I. J. and Bruton, M. N. 1988. Atlas of alien and translocated indigenous

aquatic animals in southern Africa. South African National Scientific

Programmes Report no. 144. Foundation for Research and Development,

Pretoria.

Doyle, R. E., Chadwick, D. H., Rudloe, J., Boyer, A., Kennedy, D. S. and

Kauffman, J. M. 1977. Rivers wild and pure: a priceless legacy. National

Geographic 153: 1-35.

Elston, R. C. and Forthofer, R. 1977. Testing for Hardy-Weinberg equilibrium in

small samples. Biometrics 27: 873-881.

Engelbrecht, J. S. and Van der Bank, F. H. 1994. Isozyme and allozyme

differences in four shortfin barb (Barbus brevipinnis Jubb, 1966)

populations with reference to an undescribed species from the Transvaal.

Biochem. Syst. Ecol. 22(3): 269-276.

Engelbrecht, J. S. and Van der Bank, F. H. 1996. Genetic relationships between

Marico barb (Barbus motebensis Steindachner, 1894), Redtail barb (B.

gumeyi Gunther, 1868) and Amatola barb (B. amatolicus Skelton, 1990).

Anim. Genet. 27: 77-84.

Engelbrecht, J. S. and Van der Bank, F. H. Allozyme differences between

9.3

References

populations of chubbyhead barb (Barbus anoplus Weber, 1987) and

Marico barb (B. motebensis Steindachner, 1894). Water SA (submitteda).

Engelbrecht, J. S. and Van der Bank, F. H. Genetic relationships between seven

species within the chubbyhead and goldie barb groups of minnows

(Cyprinidae). J. Afr. Zool. (submitted b).

Engelbrecht, J. S. and Van der Bank, F. H. A key based on biochemical genetic

data, morphology and distribution to identify species-groups within the

chubbyhead and goldie barb groups of minnows. J. Afr. Zool. (submittedc).

Ferreira, J. T., Grant, W. S. and Avtalion, R. R. 1984. Workshop on fish genetics.

CSIR special publication, Pretoria.

Gaigher, I. G. 1969. Aspekte met betrekking tot die ekologie, geografie en

taksonomie van varswatervisse in die Limpopo- en

Incomatiriviersisteem. Ph.D Thesis, Rand Afrikaans University,

Johannesburg (in Afrikaans).

Gaigher, I. G. 1973. The habitat preferences of fishes from the Limpopo River

system, Transvaal and Mozambique. Koedoe, 16: 103-116.

Gaigher, I. G. 1976. Geographical variation and synonymy of the minnows

Barbus motebensis Steindachner, 1868 and B. anoplus M. Weber,

1897. Cape Nature Conservation Reports. 171-178.

Gaigher, I. G. and Pott, R. 1973. Distribution of fishes in southern Africa. S. Afr.

J. Sci. 69: 25-27.

Gibbs, R. H. 1957. Cyprinid fishes of the subgenus Cyprinella of Notropis. I.

Systematic status of the subgenus Cyprinella, with a key to the species

9.4

References

exclusive of the lutrensis-ornatus complex. Copeia 3: 185-195.

Gilchrist, J. D. F. and Thompson, W. W. 1913. The freshwater fish of South

Africa. Ann. S. A. Mus. 11: 321-579.

Grant, W. S., Cherry, M. I. and Lombard, A. T. 1984. A cryptic species of

Mytilus (Mollusca: Bivalvia) on the west coast of South Africa. S. Afr.

J. Mar. Sci. 2: 149-162.

Grant, W. S. and Cherry M. I. 1985. Mytilus galloprovincialis Lmk. in southern

Africa. J. Exp. Mar. B. 90: 179-191.

Grant, W. S. and Stahl, G. 1988. Description of electrophoretic loci in Atlantic

cod Gadus morhua, and comparison with Pacific cod Gadus

macrocephalus. Hereditas 108: 27-36.

Greenacre, M. J. 1987. The geometric interpretation of correspondence analysis.

J. Am.Stat. A. 82: 437-447.

Greenacre, M. J. 1990. SIFT C version 2. Personal Computer Software for

Correspondence Analysis. UNISA, Pretoria.

Greenwood, P. H. 1962. A new species of Barbus (Pisces, Cyprinidae) from the

upper Zambezi River, Rhodesia. Rev. Zool. Bot. Afr. 65: 3-4, 211-216.

Groenewald, A. A. V. J. 1958. A revision of the genera Barbus and Varicorhinus

(Pisces, Cyprinidae) in the Transvaal. Ann. Transvaal Mus. 23(3):

263-330.

Harris, H. and Hopkinson, D. A. 1976. Handbook of Enzyme Electrophoresis

in Human Genetics. North-Holland, Amsterdam.

Harrison, A. D. 1959. General statement on South African Hydrobiological

9.5

References

Regions. NIWR, CSIR Report No. 1 Project 6.8H, Pretoria.

Hillis, M. D. and Moritz, C. 1990. Molecular systematics. Sinauer Associates,

Sunderland, Massachusetts.

Jubb, R. A. 1966. A new species of Barbus (Pisces, Cyprinidae) from the Sabie

River, North-Eastern Transvaal, Annals of the Cape Provincial

Museums 5: 157-160.

Jubb, R. A. 1967. Freshwater fishes of southern Africa. A. A. Balkema, Cape

Town, South Africa.

Jubb, R. A. 1968. The Barbus and Varicorhinus species (Pisces, Cyprinidae) in

the Transvaal. Ann. Transvaal Mus. 26: 79-97.

Jaccard, P. 1928. Die statistisch-floristische Methode als Grundlage der

Pflanzensoziologie. In Abderhalden. Handb. Biol. Arbeitsmeth. 11:

165-202.

Karakousis, Y., Machordom, A., Doadrio, I. and Economidis, P. S. 1995.

Phylogenetic relationships of Barbus peloponnesius Valenciennes,

1842 (Osteichthyes: Cyprinidae) from Greece and other species of

Barbus as revealed by allozyme electrophoresis. Biochem. Syst. Ecol.

23(4): 365-375.

Kirpichnikov, V. S. 1981. Genetic bases of fish selection. Springer - Verlang, New

York.

Kleynhans, C. J. 1983. A checklist of the fish species of the Mogol and Palala

Rivers (Limpopo system) of the Transvaal. J. Limnol. Soc. sth. Afr. 9:

29-32.

9.6

References

Kleynhans, C. J. 1984. Die verspreiding en status van sekere seldsame vissoorte

van die Transvaal en die ekologie van sommige spesies. Ph.D thesis,

Pretoria University, Pretoria (in Afrikaans).

Lachner, E. A. 1967. Systematics, distribution, and evolution of the chub

Nocomis (Cyprinidae) in the southwestern Ohio River Basin, with the

description of a new species. Copeia 3: 557-580.

Lawson, R., Kofron, C. P. and Dessauer, H. C. 1989. Allozyme variation in

natural populations of Nile crocodile. Am. Zoolog. 29: 863-871.

Lowe-McConnel, R. H. 1959. Breeding behaviour patterns and ecological

differences between Tilapia species and their significance for evolution

within the genus Tilapia (Pisces: Cichlidae). Proc. Zool. Soc. Lond.

132: 1-30.

Machordom, A., Doadrio, I. and Berrebi, P. 1995. Phylogeny and evolution of the

genus Barbus in the Iberian Peninsula as revealed by allozyme

electrophoresis. J. Fish Biol. 47: 211-239.

Markert, C. L. and Faulhaber, I. 1965. Lactate dehydrogenase isozyme patterns

of fish. J. Exp. Zool. 159: 319-332.

Mason, C. F. 1991. Biology of Freshwater Pollution. Longman Scientific &

Technical. Essex England.

Mulder, P. F. S. 1989. Die ontwikkeling van 'n genetiese databank vir die

ekonomies belangrike Barbus spesies in Suid Afrika. Ph.D Thesis.

Rand Afrikaans University, Johannesburg (In Afrikaans).

Naish, K. and Skibinski, D. Review of techniques applied to the characterization

9.7

References

of Tilapiine species and populations. Workshop on the characterization

of Ghanaian tilapia genetic resources for use in fisheries and

aquaculture. Institute of Aquatic Biology, Acara, Ghana (in press).

Nei, M. 1972. Genetic distance between populations. Am. Nat. 106: 238-292.

Nei, M. 1973. Analysis of gene diversity in subdivided populations. Proc. Natn.

Acad. Sci. U.S.A. 70: 3321-3323.

Nei, M.1975. Molecular Population Genetics and Evolution. North Holland,

Amsterdam.

Nei, M. 1978. Estimation of average heterozygosity and genetic distance from

a small number of individuals. Genetics 89: 583-590.

Nei, M. and Roychoudhury, A. K. 1974. Sampling variances of heterozygosity

and genetic distance. Genetics 76: 379-390.

Nevo, E., Beiles, A. and Ben-Shlomo, R. 1984. The evolutionary significance of

genetic diversity, ecological demographic and life history correlates.

Evolutionary dynamics of genetic diversity. Lecture Notes in

Biomathematics 53: 13-213.

Park, L. K. and Moran, P. 1995. Developments in molecular genetic techniques

in fisheries. In Molecular genetics in fisheries (Ed. by Carvalho, G. R.

and Pitcher, T. J.), pp 1-22. Chapman and Hall.

Partridge, T. C. and Maud, R. R. 1987. Geomorphic evolution of southern Africa

since the Mesozoic. S. Afr. J. Geol. 90(2): 179-208.

Raven, P. H. and Wilson, E. 0. 1992. A fifty-year plan for biodiversity surveys

Science 258: 1099-1100.

References

Ridgway, G., Sherburne, S. W. and Lewis, R. D. 1970. Polymorphism in the

esterases of Atlantic herring. Trans. Am. Fish. Soc. 99: 2-15.

Schaklee, J. B., Allendorf, F. W., Morizot, D. C. and Whitt, G. S. 1990. Gene

nomenclature for protein-coding loci in fish. Trans. Am. Fish. Soc. 119:

2-15.

Schaklee, J. B., Tamaru, C. S. and Waples, R. S. 1982. Speciation and

evolution of marine fishes studied by electrophoretic analysis of

proteins. Pacific Sci. 36: 141-157.

Scholl, A. 1973. Biochemical evolution of the Genus Xiphophorus (Poeciliidae,

Teleostei). In Genetics and mutagenesis of fish (Ed. by SchrOder, J.

H.), pp 277-299. Springer-verlang, Berlin.

Skelton, P. H. 1988. A taxonomic revision of the redfin minnows (Pisces,

Cyprinidae) from southern Africa. Ann. Cape Prov. Mus. (Nat. Hist.)

16(10): 201-207.

Skelton, P. H. 1990. A new Barbus minnow (Pisces, Cyprinidae) from the eastern

Cape Province, South Africa. S. A. Jour. Zoo!. 25(3): 188-193.

Skelton, P. H. 1993. A Complete Guide to the Freshwater Fishes of Southern

Africa. Southern Book Publishers (PTY) Ltd. Halfway House. South

Africa.

Skelton, P. H., Cambray, J. A., Lombaard, A. and Benn, G. A. 1995. Patterns of

distribution and conservation status of freshwater fishes in South

Africa. S. Afr. J. Zoo!. 30(3): 71-81.

Sokal, R. R. and Rohlf, J. H. 1969. Biometry. W. H. Freeman, San Francisco,

9.9

References

CA.

Swofford, D. L. 1985. PAUP: Phylogenetic analysis using parsimony. User's

manual. Illinois Natural History Survey, Champaign, Illinois, 61820,

U.S.A.

Swofford, D. L. and Berlocher, S. H. 1987. Inferring evolutionary trees from gene

frequency data under the principle of maximum parsimony. Syst. Zool.

36: 293-325.

Swofford, D. L. and Selander, R. B. 1981. MOSYS-1 : A FORTRAN program for

the comprehensive analysis of electrophoretic data in population

genetics and systematics. J. Hered. 72: 281-83.

Templeton, A. R. 1980. Modes of speciation and inferences based on genetic

distances. Evolution 34(4): 719-729.

Teugels, G. G. Morphometric characterization of populations and strains of

Oreochromis niloticus, Sarotherodon melanotheron (Cichlidae), Clarias

gariepinus, C. anguillaris (Clahidae) and Chrysichthys nigrodigitatus

(Claroteidae). Workshop on the characterization of Ghanaian tilapia

genetic resources for use in fisheries and aquaculture. Institute of

Aquatic Biology, Acara, Ghana (in press).

Thorpe, J. P. and Sole-Cava, A. M. 1994. The use of allozyme electrophoresis

in invertebrate systematics. Zool. Scr. 23: 3-18.

Utter, F. M., Allendorf, F. W. and Hodgins, H. 0. 1973. Genetic variability and

relationships in Pacific Salmon and related trout based on protein

variations. Syst. Zool. 22: 257-70.

9.10

References

Van der Bank, F. H., Grobler, J. P. and Du Preez, H. H. 1992. A comparative

biochemical genetic study of three populations of domesticated and

wild African catfish. Comp. Biochem. Physiol. 101B(3): 387-390.

Wahlund, S. 1928. The combination of populations and the appearance of

correlation examined from the standpoint of the study of heredity.

Hereditas 11:65-106.

Whitt, G. S. 1970. Developmental genetics of the lactate dehydrogenase

isozymes of fish. J. Exp. Zool. 175: 1-35.

Wright, S. 1978. Evolution and the genetics of populations, Vol 4. Variability

within and among natural populations. University of Chicago, Chicago.

9.11

10.1

Appendix

Rel

latta

ve m

ob

ahity

of L

-Lac

tate

deh

ydro

gen

ase

Od

es

Chu

bby h

ead

bar b

s G

o➢di

e ba

r bs

Appendix

10.2

Appendix

~r

Rel

ativ

e m

obi

lity

o f G

lltne

ose-

6-p

hosp

hate

iso

mer

ase

alle

les

Tu

berc

le b

arbs

0

t'

t)

PL

Appendix

Rel

ativ

e m

ob

ility

of P

hosp

hogi

lue o

nate

deh

ydr

ogen

ase

alle

les

bA

sA.

Riv

er/s

outh

eas

tern

Cap

e gr

oup

en

.ta

No

rth

east

ern

ese

arp

m

twee

n th

e no

rth

east

er

10.4

phyilogenetic relationships etween morphs logically

Documents