Ecological studies on crater lakes in West Cameroon Fishes of Barombi Mbo

Abstract

In the West Cameroonian crater lake, Barombi Mbo, and its inflowing stream there are 17 species of fishes, of which 12, including the 11 cichlids, appear to be endemic. We give a systematic account of the endemics, including four new species and two new genera, Konia and Myaka. The ecology of the lake is described in relation to the feeding biology of the fishes, studied by underwater observation in the lake and the examination of stomach contents. The cichlids in Barombi Mbo probably evolved there from two or three ancestral populations and now show clear ecological separation in their feeding and breeding. The problem of speciation within the lake is of particular interest because of the apparent absence of physical barriers to account for the genetic isolation of incipient species.

Full text

J.
Zool.,
Lond.
(1972)
167,
41-95
Ecological studies on crater lakes
in
West Cameroon
Fishes
of
Barombi Mbo
ETHELWYNN
TREWAVAS
British
Museum (Natural
History),
Cromwell Road, London
S.
W.7
J.
GREEN
AND
SARAH
A.
CORBET
Westjield College (University
of
London), London N. W.3
(Accepted
9
November
1971)
(With
1
plate and 29 figures in the text)
In the West Cameroonian crater lake, Barombi Mbo, and its inflowing stream there are
17
species of fishes, of which
12,
including the
11
cichlids, appear to be endemic. We give a
systematic account of the endemics, including four new species and two new genera,
Konia
and
Myaka.
The ecology of the lake is described in relation to the feeding biology of the
fishes, studied by underwater observation in the lake and the examination of stomach con-
tents. The cichlids in Barombi Mbo probably evolved there from two
or
three ancestral
populations and now show clear ecological separation in their feeding and breeding. The
problem of speciation within the lake is
of
particular interest because of the apparent
absence of physical barriers to account for the genetic isolation of incipient species.
Contents
Introduction
................
Systematics of the fishes (E.T.)
............
Nomenclature
................
Measurements
................
List of species
................
Descriptions of species
..........
..
Ecology of the lake
(J.G.
and
S.A.C.)
........
..
Feeding habits and ecological relationships
....
..
Predators of the fishes
..........
..
......
Fish fauna of the surrounding streams
........
Cichlids of lower zones of rivers of Cameroon
....
*.
Eye lens proteins
..............
Evolution
of
the cichlids in Barombi Mbo
(E.T.,
J.G.
and
S.A.C.)
.
,
References
..................
Relationships of the fishes in Barombi Mbo (E.T.)
Introduction
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
*.
..
..
..
..
..
..
..
..
..
..
Page
41
42
43
43
44
44
76
78
83
85
85
86
88
89
93
Barombi Mbo
is
a
crater lake in West Cameroon, about 35 miles north-north-east
of
Mount Cameroon at 9’22’E and 4’38”. Barombi
Mbo
(from the Barombi word mbo,
pronounced mbu, meaning
a
deep lake) is one of several names for the lake, which
is
also
41

42
E. TREWAVAS,
J.
GREEN AND
S.
A.
CORBET
known as Barombi Lake, or Kumba Lake from its proximity to the town of Kumba.
Tomczek named it Elefanten-see in
1883.
The lake is clear and lies in a small forested crater. The Kake Gorge, to the south-east,
carries the outflow which flows via the Kumba River to join the Mungo River. The
physical geography and geology of the area around the lake have been described by Gkze
(1
943).
Trewavas
(1962)
reviewed the collections of fishes made by previous expeditions and
described several new species. In March and April 1970 we visited the lake and in this
paper we describe the new species of fishes that we found there, review those
already known, and give an account of the ecological and evolutionary relationships of
the fishes in Barombi Mbo and its inflowing stream.
Borornbi
Boathouse
t. Kake
FIG.
1.
Bathymetric map of Barombi Mbo (after Hassert,
1912).
The isobaths are given in metres.
Systematics
of
the fishes
ETHELWYNN
TREWAVAS
Our plans for work on Barombi Mbo envisaged ecological enquiries based on a fairly
well-defined taxonomic situation. In the event there were
so
many additions and correc-
tions to the earlier account of the fish fauna (Trewavas,
1962)
that the fish taxonomist
was kept occupied full-time on her speciality.

ECOLOGICAL STUDIES
ON
CRATER LAKES
IN
WEST CAMEROON
43
Briefly, the non-cichlid taxonomy was in the main confirmed, with the addition of
a
second species of
Clarias,
the widespread
C.
walkeri,
to the lake fauna and two other
cyprinodonts to that of the feeder stream. Instead of the five cichlid species recognized in
1962, we found that there were 11, all endemic.
In the first place,
Pungu
maclareni,
recorded (as
Barombia)
in 1962 as from Lake
Barombi Kotto, was found by Thys van den Audenaerde to have been wrongly assigned
to that lake and on his personal information the record was transferred in
“The Red Book”
(Trewavas in Fisher
et al.,
1969) to Barombi Mbo, where it proves to be abundant.
Our own work on the lake revealed other misapprehensions in the 1962 report. The single
specimen listed by me as a possible older individual
of
Tilapia eisentrauti
proves to belong
to a distinct, unnamed species. The small fishes with enlarged lateral line pores placed by
me as young of
Stomatepia mariae
are members of a separate species of
Stomatepia
and
a
third species of the same genus was introduced to
us
by the Barombis by the name of
“mongo”. Lastly it became apparent that we were confusing two species under the name
of
T.
linnellii
and that the second one had already been described by Holly (1930) as
T.
caroli,
a name that I had wrongly placed in the synonymy of T.
linnellii.
The distinction
between these two was the only one that gave
us
any trouble in the field, and the fishermen
themselves distinguished
T.
caroli
(fissi) only in the mature stage.
In this account
I
give full descriptions of the new species, define some new genera and
recapitulate the diagnostic characters of the previously known endemic species, adding
any new information about them. A critical account of the non-endemic species is given
in a separate paper (Trewavas, in prep.).
Nomenclature
The name
Sarotherodon
Ruppell, 1852, has been used as a subgenus of
Tilapia
by
Regan
(1920) and recently (1968a) by Thys. Accumulating knowledge, not least of the cichlids of
Barombi Mbo, points to its being worthy of full generic rank and
I
so use it here. This and
the recognition of new genera result in the name
Tilapia
disappearing from Barombi Mbo.
Regrettably, the name
Barombia
proves to be preoccupied* and
I
have replaced it by
Pungu,
the Barombi name.
Measurements
The abbreviation SL is the standard length of taxonomic practice, from tip of snout
to
base of caudal fin, excluding any projecting part of the lower jaw. The depth of the pre-
orbital (=lacrimal) bone is measured from the middle of its orbital rim to its oral edge,
along
a
line that continues the radius of the eye. Since the two edges of the bone are not
parallel the measurement is subject to individual bias, but some of the variation is real
and may exist between right and left sides. It is also affected to some extent by allometry,
but very little within the size-ranges of our cichlids. The interorbital width is measured
between points of dividers pressed against the skin of the tops of the orbits and is little
more than the least width of the skull roof between the eyes.
*
I am very grateful
to
Mr
G.
P.
Whitley for pointing this
out
to me.

44
E. TREWAVAS,
J.
GREEN AND
S.
A.
CORBET
List
of
species of.jish from Barombi Mbo and its inflowing stream
Species Barombi Systematics Ecology
name Page Page
C
y
p r i n i d a e
Clariidae
Barbus batesii
Boulenger
Clarias walkeri
Gun ther
Clarias maclareni
Trewavas
Cyprinodontidae
Aphyosemion oeseri
(Ah1 in Schmidt)
(inflow stream only)
Epiplatys sexfasciatus
Gill
Procatopus similis
Ah1
(inflow stream only)
Cichlidae
Sarotherodon lohbergeri
(Holly)
S.
steinbachi
(Trewavas)
S.
linnellii
(Lonnberg)
S.
caroli
(Holly)
Myaka myaka
sp.
n.
Konia eisentrauti
(Trewavas)
K.
dikume
sp.
n.
Pungu maclareni
(Trewavas)
Stomatepia mariae
(Holly)
S.
pindu
sp.
n.
S.
mongo
sp.
n.
suh, kimbundu
nyongo
nyongo
longo katta
lenge
leka keppe
kululu
kippC, unga
fissi, unga
myaka myaka
konye
dikume
PuWu
nsess
pindu
mongo
44
45
45
47
47
47
48
50
52
54
59
61
63
66
68
70
72
78
78
78
80
80
80
80
81
81
81
81
82
82
82
83
83
Of the
17
species now known from the lake and its inflowing stream,
12,
includingall
the
11
cichlids, appear to be endemic. Only
Barbus batesii, Clarias walkeri, Procatopus
similis, Aphyosemion oeseri
and
Epiplatys sexfasciatus
are known from elsewhere.
Description
of
species
CYPRINIDAE
Barbus batesii
Boulenger,
1903: 25,
pl. iii fig.
2; 191 1
:
43,
fig.
24;
Trewavas,
1962: 152;
Trewavas, in
Barbus linnellii
Lonnberg,
1904
:
138.
Barombi names
:
suh, kimbundu (young).
prep*

ECOLOGICAL STUDIES
ON
CRATER LAKES
IN
WEST CAMEROON
CLARIIDAE
The clariids of Barombi Mbo belong to the subgenus
Clarioides
David of
Clarias,
having all the bones
of
the side of the head present and fewer than 20 gill-rakers on the
lower part of the first arch. There are two species in the lake and (young only) its feeder
stream.
Clarias walkeri
45
(Fig. 3(4)
Gunther, 1896: 274, pl. xiv fig. B (Ogowe); Boulenger, 1911,
2:
248, fig. 206 (Ogowe,
Gabon, Corisco
Is.,
Benito
R.,
Ja
R.,
Cameroon); Holly, 1927: 198
(R.
Sanaga and
Yaoundk); Trewavas, 1962: 156, fig. 2 (stream near Barombi Kotto); Thys van den
Audenaerde, 1967: 51 (Fernando
Poo).
Clarias poensis
Boulenger, 1908: 1078, fig, 252; Boulenger, 191
1,
2:
244, fig.
203.
Barombi name
:
nyongo (part).
Clarias maclareni
(Figs 2 and 3(b))
C.
maclareni (part.,
syntypes only) Trewavas, 1962: 157, fig. 3 (lower sketches).
Barombi name: nyongo (part).
IIN
-v
\x
--&
FIG.
2.
Clarias maclareni,
a specimen
of
160
mm
SL.
Dotted lines show the positions
of
the postorbital and
suprapreopercular bones. The head is viewed from the left and
a
little above. Above, suprabranchial trees.
C. walkeri
is the common
Clarias
of Cameroun and is reported from the Ogowe to the
northwestern Cameroons, River Logone and Fernando
Poo.
Records from farther west
are not well supported (e.g. a doubtful report in Trewavas
&
Irvine, 1947) and Daget
&
Iltis (1965) did not find it in the Ivory Coast. A single specimen in the British Museum
(Natural History) from Sierra Leone
so
labelled is indeed very much like
C.
walkeri,
although the suprabranchial tree of epibranchial I1 is less developed than in
C.
walkeri
of
the same size.
C.
maclareni
is peculiar to Barombi Mbo, its only close relation being the form from
Lake Soden to be described by Griffith (in prep.). Other West African clariids in the
British Museum (Natural History) with the bones
of
the side of the head loosely connected
have small eyes and less reduced suprabranchial trees than
C. maclareni.
The fishermen did not distinguish between
C. walkeri
and
C.
maclareni,
but the species
are fairly easily distinguished at sight by the size
of
the eye, smaller in
C.
walkeri.
Deter-
mination can be confirmed by examining the suprabranchial trees and the bones of the

46
E.
TREWAVAS,
J.
GREEN AND
S.
A. CORBET
FIG.
3.
(a)
Clarias walkeri,
holotype of
151
mm
SL,
shape of premaxillary and vomerine tooth-patches, supra-
branchial trees of left side and postorbital and suprapreopercular bones of right side. (b) The same for the
holo-
type of
C.
maclareni,
a specimen of
SL
222
nun.
From Trewavas,
1962.
e, Eye; p, postorbital;
s,
suprapreopercular.
side
of
the head, especially if specimens of the same size are compared (Table
I
and Figs
2
and
3).
These two features remain the most reliable distinguishing characters.
The vomerine teeth form
a
continuous band at all sizes examined and in
C.
maclareni
the band is narrower in the middle. The width proves to be variable in
C.
walkeri,
so that
the width of the band is not
a
very reliable distinguishing feature. The nature of the union
of the bones of the side of the head usually provides
a
contrast between the two species
within the lake basin, but in some
C.
walkeri
from the River Wowe and the River Kumba
the postorbital and suprapreopercular meet tangentially or are separated by fibrous tissue.
The gill-raker count in
C.
maclareni,
9-13
on the lower part of the first arch, also provides
no contrast with
C.
walkeri.
TABLE
I
Distinctions between
Clarias walkeri
and
C.
maclareni
C.
walkeri
N=15
(9
from Mbo,
C. maclareni
N=15 6
from elsewhere)
Standard length
(mm)
114.5-222.5 107-3 19
Diameter
of
eye
As
%
length of head
As
%
interocular width
Suprabranchial trees
9.5-1
1.8
6.5-9.5
20.0-250 13.0-19.5
Relatively simple Elaborate
Union of postorbital and suprapreopercular Tangential or not quite meeting Sutural
or
tangential

ECOLOGICAL STUDIES ON CRATER LAKES
IN
WEST CAMEROON
41
The poor development of the suprabranchial trees in
C.
maclareni
may be related to
lacustrine life, where there
is
no need to have recourse
to
air-breathing (cf. the endemic
clariids of Lake Malawi, Greenwood, 1961) and perhaps the bigger eyes are more effic-
ient aids to the pursuit of mid-water prey like dikume and myaka myaka (see
p.
80)
than
the barbels with their organs of touch and chemical sense on which most clariids seem to
place greater reliance. Unfortunately we did not note the species from whose stomachs
these frequenters of deeper waters were recorded.
One specimen of
C.
maclareni
had no eyes, and in another the right eye was gone. In
both cases the skin had healed over the eye sockets, but there was a pit showing that the
socket was as big as that for a normal eye.
No
doubt some predator had removed the
missing eyes.
CYPRINODONTID
AE
Aphyosemion oeseri
Panchax oeseri
Ah1 in Schmidt, 1928: 165, fig.
on
p. 166 (Fernando
Poo).
Panchax vexillifer
Meinken, 1929
:
255, fig. (“Westkuste Afrikas”).,
Aphyosemion santa-isabellae
Scheel, 19683: 332, fig. 1 (Fernando
Poo).
Aphyosemion santaisabellae;
Foersch, 1971
:
20,
4
figs.
Aphyosemion oeseri;
Trewavas, in prep.
No
Barombi name known.
Epiplatys sexfasciatus
Gill,
1862:
136 (Gaboon River); Trewavas, 1962: 159 (Lake Kotto and Barombi Mbo);
Haplochilus infrafmciatus
Giinther, 1866
:
3 13 (Old Calabar)
;
Lonnberg, 1903
:
39
Barombi name: longo katta.
Scheel, 1968a: 383, figs; Trewavas, in prep.
(River at Sanye, near Lake Kotto).
Procatopus similis
Ahl, 1927: 79 (“Logobaba, Kamerun”); Trewavas, in prep.
Barombi name
:
lenge.
CICHLIDAE
Structure
of
the ovaries
We found eggs and young in the mouths of parents of only one species,
Surotherodon
linnellii,
in which the mother was the brooding parent in all of the eight instances.
In two cichlid species we found no certainly ripe eggs,
Myaka myaka
and
Konia dikume,
but in all the others
(S.
caroli
from previously collected specimens only) we found one or
more females with large, green, oval eggs nearly ready to be shed.
Many other ovaries were
in
the stage described
as
“starting”, in which distinct oocytes
could be seen with the dissecting microscope. The oocytes ranged from minute, colourless
transparent ones to cream-coloured or yellow spheres from 0.5 to 3.0 mm diameter. In
some there was a fairly smooth gradation in sizes, but usually there were two or three
size-groups, possibly representing two or three future spawning-cycles.

48
E.
TREWAVAS,
J.
GREEN AND
S.
A. CORBET
As well as the oocytes there were amorphous bodies coloured bright rust-red or yellow.
As these are generally fewer, smaller or absent in ripe ovaries we think they may be nutrient
bodies developed in some of the follicles at the expense of the oocyte. They are apparently
the same as bodies described by Bretschneider
&
DuyvenC de Wit (1947) and Hoar (1957)
in
Rhodeus
as “preovulation corpora lutea” or “corpora atretica”. As there
is
no evidence
for a hormonal function nor any close analogy with the corpora lutea of mammals we do
not use the former term, and although the latter may be more appropriate too little is
known about the development of the bodies in our species for us to feel happy in using it.
We refer to them as “amorphous bodies”.
The surface of eggs from the mouth of
S.
linnellii
bears sparsely distributed bunches of
filaments, like those figured for
S.
galilaeus
by Kraft
&
Peters (1963). Examination with
the electron scanning microscope of ripe ovarian eggs of
S.
lohbergeri, St. pindu
and
Pungu
revealed
no
filaments, but no conclusion can be drawn from this until discharged eggs can
be compared.
TABLE
I1
Vertebral numbers and long diameter
of
largest ovarian eggs in Cichlidae
of
Barombi Mbo
Vertebrae
27
28 29
30
31
Diameter
of
eggs
Sarotherodon steinbachi
781
S.
lohbergeri
4
S.
linnellii
15
S. caroli
31
K.
eisentrauti
5
K.
dikume
5
Pungu maclareni
7
Myaka myaka
16
Stomatepia mariae
16
S.
pindu
92
S.
mongo
7
3.6
4.5
5-53
5-53
4
?
4
?
4
4
4.5-5
.
.. .
Testes
Although we handled many ripe and ripening females, we found no males with very
large testes such as
I
have seen for instance in the males of mouth-brooding species in
Lake Malawi.
I
have followed this observation with no quantitative or histological data,
but it seems worth while recording it in view of the fact that Heinrich (1967: 724-725)
records otherwise unpublished data of Peters showing that in
Sarotherodon melanotheron
(“heudeloti”)
and
S.
galilaeus
the weight of the testis is relatively lower than in
S.
niloticus
and
S.
mossambicus,
one of a number
of
characters distinguishing these two pairs of species
from each other.
Saro therodon lohbergeri
(Figs
4
and
5)
Tilapia lohbergeri
Holly, 1930:
203,
pl. ii fig. lO;Trewavas, 1962: 171,fig. 7. Holotype:
89+23 mm, Vienna Mus. 13951, coll. Haberer, “Cameroun”.

ECOLOGICAL
STUDIES
ON
CRATER LAKES
IN
WEST CAMEROON
49
FIG.
4.
Surotherodon lohbergeri,
a specimen of
SL
162
mni.
FIG.
5.
Surotherodon lohbergeri,
inner and outer jaw teeth; lower pharyngeal bone and isolated anterior and
posterior teeth; first gill-arch and oral view
of
two inner and two outer gill-rakers; the inner rakers have a subsidiary
tubercle. In part from Trewavas,
1962.

50
E. TREWAVAS,
J.
GREEN AND
S.
A. CORBET
Barombi name: leka keppe.
This species is easily recognized in life and when freshly caught by the prominent black
band extending from the opercular spot to the caudal peduncle. When preserved there is
also an upper band between the dorsal fin and the upper lateral line. At all times it is
distinguished from
K. eisentrauti
and
S.
steinbachi
by the acute snout and the terminal
mouth with rather thick lips. The young of about
30
mm
SL
have pearly iridescent bodies
with no stripe (but the stripe may appear in preserved young) and their pelvic fins and the
anterior edge of the anal are orange to vermilion.
Closer examination reveals the characteristic movable teeth of the jaws, in four to seven
rows in adults and already three or four rows in young of 27-34mm. The shafts are
slender and the expanded crowns are spoon-shaped with a lateral notch in the outer row,
tricuspid in the inner rows. This dentition corresponds to the method of feeding described
on page 80 and has evolved several times in African Cichlidae-in
Tilapia mariae,
in the
subspecies of
S.
nilotica
in Lake Rudolf, in
Neotilapia
(Lake Tanganyika),
Petrochromis
and
Petrotilapia.
These are unrelated to
S.
Iohbergeri
and to each other; in some, but not
in
S.
lohbergeri,
the outer teeth may become tricuspid, either in the adult or throughout
life. In most of them the pharyngeal teeth are fine, unicuspid and close-set
on
broad tooth
plates, as they are in
S.
lohbergeri
(Fig. 5).
The intestine of a fish
of
86 mm SL measures 410 mm, 4.8 times the standard length.
A
fish of 78 mm SL contained green ovarian eggs of 4.5 mm long diameter, 13 in the
right ovary, 27 in the left. Two of 73.5 mm and 83 mm had eggs of 4.0-4.2 mm; in the
73.5 mm fish there was a total of 46 eggs, 23 in each ovary. Many ovaries contained yellow
and rust-red amorphous bodies as well as small oocytes of two sizes, the larger cream-
coloured and up to 2.5 mm diameter. At 3.0-3.5
mm
the oocytes were already turning
green.
We observed
no
mating or parental behaviour, but eggs of this size must surely be in-
cubated in the mouth. This is supported by the evidence of Dr Pasqual (quoted by
Trewavas, 1962) that in aquaria they are mouth-brooders; he had the impression that
both parents carried the larvae in this way. Biparental care is consonant with the lack of
external differences between the sexes.
The species
in
the River Kumba
Two young Tilapias of 24.5 and
30
mm SL were caught by Dr
R.
H.
L. Disney in narrow-
meshed traps in the River Kumba at Buea Road bridge, just below the town of Kumba.
These agreed very well with young
S.
lohbergeri
in colour, having in life no black markings
(no tilapia-mark), but vermilion pelvic fins, anterior edge of anal and lower edge of
caudal; there were three rows of teeth in the jaws and the numbers of dorsal rays (XVI
11 and XVII 10) and gill-rakers (respectively 14 and 16 on the lower part
of
the first arch)
are within the ranges of
S.
lohbergeri.
The significance of this find
is
discussed on page
85.
Saro
therodon
steinbachi
(Figs 6 and 7)
Tilapia steinbachi
Trewavas, 1962: 178, fig. 10,
11.
Holotype and paratype:
BMNH
1961.10.18.40-41, respectively 113 and 90 mm SL,
collected by Professor Eisentraut’s expeditions.

ECOLOGICAL STUDIES
ON
CRATER LAKES
IN
WEST CAMEROON
51
FIG.
6.
Sarotherodon
steinbachi,
holotype.
c-.
/
FIG.
7.
Sarothevodon steinbachi.
Inner and outer jaw teeth and lower pharyngeal bone with two
of
its teeth;
first gill-arch and oral view
of
two inner and two outer gill-rakers. In part from Trewavas,
1962.

52
E.
TREWAVAS,
J.
GREEN AND
S.
A.
CORBET
Barombi name: kululu.
When alive or freshly caught this species had
no
black band along the body or at most
a
very vague one, but this vague band is present in most preserved specimens. There is no
tilapia-mark at the sizes seen, except
in
the smallest,
a
fish
of 39.5
mm
SL,
in
which it is
grey and rather ill-defined.
We
found no sexual differentiation in colour.
Kululu is recognisable by the very small mouth and rather abruptly decurved snout
profile, the Roman-nose-like appearance being produced by the projecting premaxillary
pedicels,
so
placed
in
using the upper jaw to rake bottom deposits into the mouth.
Further diagnostic features are the very small teeth of the jaws, usually in three, rarely
two or four series; the short gill-rakers, 18-23
on
the lower part of the first arch; and the
massive pharyngeal bones, whose length (including the anterior blade) is nearly half that
of the head (45.0-49.5
x)
in adults. The condition of the slender pharyngeal teeth of the
type as described in 1962 proves to be characteristic, the complete brown-tipped ones
being interspersed with shorter teeth whose brown tips are worn off. The cause of this,
deduced from the stomach contents of the types, is confirmed by the observations of
Professor Green and Dr Corbet (p.
81).
In the young fish of
39.5
mm this wear had
not started.
The intestine, coiled seven to nine times, measures 350-460 mm in two fishes of 75 and
77 mm
SL,
4.7-6.0 times the standard length.
Green ovarian eggs in a fish
of
84 mm SL measured 3.6
x
3.0
mm. With them were
small white oocytes but no red and yellow amorphous bodies; they were probably nearly
ready to be shed. The eggs are already green when the
long
diameter is 2.6 mm and the
ovaries then contain several coloured amorphous bodies.
We have no information on mating or parental behaviour.
Sarotherodon linnellii
(
Lonnberg) and
Sarotherodon caroli
(Holly)
Barombi names
:
kippC, unga and fissi, unga, respectively.
In 1962
I
regarded these as one species, but experience at the lakeside, attention to the
opinions of the Barombi fishermen and subsequent analysis of about
40
specimens
of
each have persuaded me that there are two and that the two scientific names already
available apply to them respectively. They are distinguished as follows.
Sarotherodon linnellii
(Figs
8,
9(a)-(c) and
1
I(b))
Tilapia (Gephyrochromis) linnellii
Lonnberg,
1903
:
42.
Tilapia linnellii;
Boulenger,
19
I5
:
159,
fig. 104.
Tilapia linnellii
(in part); Trewavas, 1962: 174, figs
8,
9.
(1)
An
intense black tilapia-mark present on the dorsal fin in the young from the
stage when they are still in the maternal mouth until they reach a SL
of
about 100 mm,
after which it becomes vaguer in outline and fainter; it may be absent altogether
above
SL
about 120 mm or may still be recognisable at
SL
150
mm.
(2) Dorsal spines
(XIV) XV
or
XVT,
mode
XV.
Modal formula
XV
12 (Table
111).
(3)
Gill-rakers
on
lower part of anterior arch
15-1
8, mode
I7
(Table
111).

53
ECOLOGICAL STUDIES ON CRATER LAKES
IN
WEST
CAMEROON
FIG.
8.
Sarorheroclon linnellii,
one
of
the syntypes, a kippk,
(from
Boulenger,
1915,
fig.
104).
@-
Y
1
mrn
P
am
-
1
crn
FIG.
9.
Lower pharyngeal bones of, (a)
S.
linnellii
of
SL
150
mm and,
(b)
S.
caroli
of
SL
143
mm
with
a
posterior
tooth (p)
of
each and anterior (a) and middle (m) teeth
of
S.
linne//ii.
(c) Outer and inner jaw teeth
of
male and
female
S.
linnellii.

54
E. TREWAVAS,
J.
GREEN AND
S.
A. CORBET
(4) Diameter of eye in young of 60-91 mm
SL
28-31
(5)
Blade of lower pharyngeal bone
in
young of 64-83 mm
SL
0.8-1
-8
times median
length of dentigerous area, in bigger fishes more than twice as long.
(6)
Teeth of jaws and pharynx very small, but always a little bigger than in
S.
caroli
of the same size.
(7)
Breeding males green.
of
length of head.
TABLE
111
Frequencies
of
meristic numbers in
Surotherodon
linnellii
and
S.
caroli
S.
linnellii
S.
curoli
Dorsal formula
XIV
12
xv
I1
XVI
10
xv
12
XVI
11
XVI
12
Dorsal spines
XIV
xv
XVI
26
27
28
8
9
10
11
29
30
15
16
17
18
19
20
Dorsal totals
Soft
anal rays
Vertebrae
Gill-rakers on lower part
of
arch
1
13
1
23
3
-
I
36
4
15
26
-
1
17
13
2
15
-
3
13
21.5
3.5
-
-
-
3
1
13
17
4
-
16
22
4
30
4
-
9
16
4
3
1
-
-
2
(juveniles)
11.5
13.5
10
Surotherodon caroli
(Figs 9(b),
10
and
1
I(a))
Tilapia curoli
Holly, 1930: 204, pl.
ii
fig.
11.
Tilapia
linnellii
(in part, not of Lonnberg); Trewavas, 1962: 174.
(1)
No
tilapia-mark
on
dorsal fin, at least at
SL
62 mm and over.
(2)
Dorsal
spines
XV
or
XVI,
mode
XVI.
Modal formula
XVT
1
I.

ECOLOGICAL
STUDIES
ON
CRATER LAKES
IN
WEST CAMEROON
55
(3)
Gill-rakers on lower part
of
anterior arch 18-20 (rarely 17 in juveniles), mode 19
(Table 111).
(4) Diameter of eye in young of 62-99 mm
SL
21.5-27.6
(5) Blade of lower pharyngeal bone always more than twice the median length
of
the
dentigerous area, at least at
SL
62 mm and over.
(6) Teeth of jaws and pharynx even smaller than those of
S.
linnellii
of
the same size.
(7) Breeding (or spent?) males grey with black or dirty grey smudges.
of length of head.
Teeth
of'juws
In both species the young have bicuspid outer and tricuspid inner teeth, and females
retain the cusps in the inner and in all but a few posterior teeth
of
the outermost series.
But
in
sexually active males the outer teeth become replaced by unicuspids and in some
of the larger males some or all
of
the inner teeth are unicuspid, the anterior as large as
the outer.
In
S.
linnellii
the unicuspid replacement teeth are usually fewer than the bicuspids
of
a
female of the same size, as the following figures show.
Upper outer teeth in 10 females
of
SL
135-150 mm: 90-1 16.
Upper outer teeth in
10
males of
SL
139-163 mm: 70-86.
The change to unicuspid teeth was less complete in the males of
S.
curoli
at
our disposal
than in those of
S.
linnellii.
In most specimens, at least some outer teeth retained a shoulder
representing the minor cusp, and the numbers were not reduced. Thus
of
seven females
of
SL
95-145 mm, one had only
80
teeth in the outer series of the upper jaw, but some
gaps; the others had 100-124;
of
15 males, 6 had 80-96 and the rest
100-1
18.
In
S.
caroli
too we have none with unicuspid inner teeth.
In both species the teeth are small, but they are smaller in
S.
curoli
and slightly more
numerous.
Lower phurjwzgeul bone
(Fig. 9)
In 1962 I recorded that although the blade
of
the lower pharyngeal bone
in
the adult
(94-148 mm
SL)
is 2.4-2.8 times the median length of the dentigerous area,
in
the young
of
50-59 mm these lengths are about equal.
The first indication that we were dealing with two species came from young of an
intermediate size, 62-85 mm. Some of these had no tilapia-mark and in them the blade
of the pharyngeal was of the same relative length as in the adult. This pointed to the
taxonomic importance of the tilapia-mark in small individuals
of
these species and we
then found different modal numbers of gill-rakers and fin rays associated with the presence
or absence of the tilapia-mark and used these, especially the gill-raker counts, to separate
adults in which the tilapia-mark was no longer a reliable guide.
The pharyngeal proportions distinguishing the two species are
:
blade/dentigerous area at SL 64-95 mm
in
S.
linnellii:
0.8-1.8
[lo]
in
S.
caroli:
2.1-3.0 [8]
in
S.
linnellii:
2.0-2.5 [12]
in
S.
curoli:
2.3-24
[I21
at
SL
102-147 mm
[In square brackets are the numbers of specimens used.]

56
E.
TREWAVAS,
J.
GREEN AND
S.
A. CORBET
FIG.
10.
Surotherodon
curoli,
holotype, (from
Holly,
1930,
pi.
ii,
fig.
11).
/
/
FIG.
11.
Gill-rakers
of
first arch in, (a)
S.
curoli
of
SL
130
mm and, (b)
S.
linnellii
of
135
mm.
A
broken line
shows the outline of the
gill.
(i) Inner (posterior) view
of
an upper part
of
ceratobranchial.
(0)
Oral view
of
a
lower portion:
f,
Line
of
attachment of gill-filaments.
This difference is more marked in young than in adults.
In
both species the pharyngeal
teeth are very slender and the posterior ones, are densely crowded, but
in
specimens
of
the
same size they are smaller and denser in
S.
caroli
and the triangular area of the bone has
a great space bare
of
teeth (Fig. 9(b)).
These differences are too subtle to be used alone for diagnosis but are confirmatory
when used with the more obvious ones.
Functionally the higher number of gill-rakers and the longer pharyngeal blade in
S.
caroli
are parts
of
the means by which the capacity of the pharynx
is
enlarged. The addi-

ECOLOGlCAL STUDIES ON CRATER LAKES
IN
WEST CAMEROON
57
tional gill-rakers do not make
a
more efficient filter because they are spaced along
a
longer
arch. They are in fact shorter than those of
S.
linnellii
(Fig. 11) and as Greenwood (1953)
showed for
S.
esculenta
filtering is
a
function to which tilapiine gill-rakers are poorly
adapted. The rakers of all the arches interlock to form a fairly solid floor to the pharynx
when the arches are adducted. When the pharynx is expanded water and the fine plankton
that forms most of the food of these species (see p.
81)
can pass freely between the arches
and the minute food organisms are captured in mucus, helped perhaps by the micro-
branchiospines that are present as usual along three of the arches.
In
addition to allowing a large volume
of
water to be passed through it, it may be that
the functional significance of the enlarged pharynx is the retention
of
the young in the
maternal mouth to
a
larger size than in most species.
T
make this suggestion by analogy
with
S.
lidole
of Lake Malawi, which in its large head and long pharyngeal blade parallels
S.
caroli
so
closely, and which harbours young to
a
total length of
40-60
mm.
Baromhi names
The Barombis used three names for these two species: unga, fissi and kippC. Individuals
in breeding colours they distinguished as fissi
(S.
caroli)
or kippC
(S.
linnellii);
and juve-
niles of both species, which look very much alike except for the tilapia-mark, they classified
together as unga.
The name kippC was used for fishes usually with a large amount of brassy yellow or
yellow-green in the lower half
of
the head and body and with the body tapering away
from the back of the head to the tail. Some
of
these had intense black areas ventrally from
the tip of the lower jaw backwards to the anal fin; in others this area remained brassy.
Both shape and colour were taken into account in applying the name kippC. Most of these,
and
all
with the most brilliant colour, proved to be males in some stage
of
sexual activity
with at least the outer teeth unicuspid. But
a
few were females, one with eggs in the
mouth; these had
a
more silvery general colour with a little greenish iridescence on
the preorbital, and it was mainly the shape that caused them to be classed as kippC.
Only the largest of these “female kippk” had
a
few of the outer teeth unicuspid. We
are not sure whether the fishermen regarded kippC
as
a phase
of
unga or as a separate
species.
The male kippi, when preserved, were
a
good match for the type of
S.
linnellii
and in
their meristic characters and pharyngeal dentition we could find no reason to regard them
as specifically distinct from the silver-grey female unga, including those brooding eggs or
young. Confirmation of this (and of the variability of the tilapia-mark at breeding sizes)
comes from Dr Corbet’s observation of
a
green kippC with a tilapia-mark courting
a
grey
fish without
a
tilapia-mark.
One “young kipp6” of
SL
105
mm proved, however, to have the meristic characters of
S.
caroli;
its testes were still small.
On dissection most of the kippC proved to have little food in the gut and much fat
among the viscera.
The name fissi began to be used more frequently towards the end of the second week
of
our visit and was then applied to males of about
I30
mm SL upwards, mainly caught in
nets, grey in colour with irregular dark grey or black smudges on the lower parts
of
the
head and body and with the snout and dorsum black. We were not sure whether the testes
were ripening or recently spent. These fishes had the characters of
S.
caroli.

58
E.
TREWAVAS,
J.
GREEN AND
S.
A.
CORBET
Although we offered special inducement, no one was able to find a female fissi for
us.
The females of
S.
caroli
in our collection were all juveniles, not recognised by the Barombis
as fissi.
On returning to the British Museum (Natural History)
I
found that in Maclaren’s
collection made in September, 1948, there are three females of
S.
curoli,
one of them,
141 mm in
SL,
with ovarian eggs 5.5 mm in long diameter.
Breeding
The data in the preceding paragraphs suggest that the form of isolation that now keeps
these two extremely similar species apart is
a
difference
in
the time (and place?) of breed-
ing, possibly reinforced by
a
different breeding colouration in the male. This need not
involve a rigid annual breeding season, but simply
a
non-synchronisation of breeding and
feeding cycles between the two species. A ripe female
S.
linnellii
was collected by Dr
C.
A.
Wright in November 1963, and although specimens of
S.
caroli
were present in his collec-
tion too their gonads were in an earlier phase.
In both species right and left gonads are about equal in size; the eggs are large, 5.0-
5.5
mm in long diameter, and oval.
In
S.
linriellii
they are green (colour unknown in
S.
caroli).
In
the right ovary of S.
curoli
of 141 mm
SL
there were 29 ripe eggs, in that of a
S.
linnellii
of 140 mm 35; two
S.
linnellii
of 150 and 155.5 mm SL had respectively 68 and
62
ripe eggs in the right ovary. These figures are not enough to establish
a
specific differ-
ence in fecundity.
The surface of the egg bears bunches of minute filaments similar to those figured for
S.
galilaeus
by Kraft
&
Peters (1963, fig. 26(e)).
Our collection includes eight brooding
S.
linnellii
of
SL
I
18-163 mm, all females, three
of them with young in the mouth, the rest with developing eggs. The young measured
13.5-13.7 mm in
SL
and 17 mm in total length and already had
a
large black tilapia-mark.
The ovaries of brooding females contained opaque, cream-coloured oocytes of 0.7-2.0 mm
and smaller translucent oocytes, as well as rust-red and yellow bodies of irregular shape.
Fat-bodies among the viscera were unusually large. We deduce that two or more breeding
cycles would have followed the one just in its last phase.
As recorded on page
81,
unga exhibiting territorial and courting behaviour were
common near the shore and we deduce that whereas adult
S.
limwllii
feed in mid-water
offshore, they come inshore
to
mate and brood the young.
The largest
S.
caroli
in our collections is a male of
SL
148.5 mm; we have females
nearly as long. Of
S.
linnellii
the largest mature male is one of the syntypes, 163
mm
in
SL,
and we have females up
to
the same size. In addition we had
a
large unga of SL 185 mm.
This was
a
silver-grey fish with
a
little green iridescence
on
the preorbital and the back
and with fins of a watery yellow-green colour.
It
was
in
good condition with food in the
gut and much fat among the viscera. It was apparently male, but the genital papilla was
very small and the gonads were as thin as in a juvenile.
This has its parallel among the tilapias of Lake Malawi, where similar large, fat speci-
mens of
S.
squatnipinnis
were given a special name,
rigongo,
and on examination proved to
have minute inactive gonads.
This largest of the unga had all the outer teeth of the jaws bicuspid, evidence that re-
placement by unicuspid teeth in the active males is a function of breeding condition and
not
of
mere size.

ECOLOGICAL STUDIES
ON
CRATER LAKES
IN
WEST CAMEROON
59
Myaka
gen.
n.
Type-species
:
Myaka myaka
sp.
n.
The tautonymous generic and specific names together constitute the Barombi name for
for this species. We were corrected if we referred to it as “myaka”. The
“y”
is pronounced
as a consonant.
Cichlid fishes of small size with 28-29 vertebrae (mode 29), the mesethmoid free from
the vomer, with the facet for the upper pharyngeals weakly developed, not including the
basioccipital; scales cycloid, with concentric circuli. Teeth of the jaws few, the inner
minute and tricuspid, outer with the minor cusp reduced or absent; teeth of the pharynx
also reduced in number and size; lower pharyngeal bone with
long
blade and restricted
dentigerous area. Mouth at an angle of about
50”
with the horizontal, the lower jaw
usually slightly projecting; gill-rakers short, 19-23 on lower part of first arch; intestine
about
1
times the
SL.
Gonads paired.
Relationship to
S.
linnellii
and
S.
caroli
is suggested by the shape of the lower pharyn-
geal (cf. Figs 9 and 13).
A
single species, confined to Barombi Mbo.
Mjiaka
myaka
sp.
n.
(Figs 12 and 13)
Holotype a mature
3
of 65+34 mm. BMNH 1971.10.20.34.
Paratypes
15
specimens of 44.5-66.5 mm
SL.
BMNH 1971 JO.20.35-49.
Barombi name: myakamyaka.
Proportions
as
SL
caudal peduncle 14-17 (1.05-1.3 times its own depth).
Proportions
as
o/;l
length
of
head
Length of snout 27-33, diameter of eye 23-29, depth of preorbital 17.5-22 (usually
19.0-21.5), interorbital width 26-31, length of lower jaw 33-39, usually 35-37 (33 is the
percentage of a very long head, 39 of a short one), length of premaxillary pedicels 22-26,
length of lower pharyngeal bone 33.3-37.5, its width 24.5-26.5.
Pharyngeal teeth very weak, confined to a short posterior area whose length is contained
3.5-5.3 times in the length
of
the blade.
Teeth of jaws very small, in two series, the inner tricuspid, outer either simple or with a
very small minor cusp; when simple either truncate,
so
probably simple by wear, or
conical or near conical apparently by replacement; anterior teeth directed forwards, with
a gap at the symphysis; 24-42 teeth in outer series of upper jaw.
Gill-rakers short, (3-6)+(0-1)+(19-23) on the anterior arch. Microbranchiospines
present on outer sides of 2nd, 3rd and 4th arches.
Scales 29-37 in the lateral line series, 4-5 from origin
of
dorsal fin to lateral line, 16-1 9
around caudal peduncle. When, as in several, there are more than 31 or 32 in the
1.1.
series this is because the rows above and including the lateral line are irregular.
Dorsal with 25-27 rays, usually 26 or 27. Formulae
XV
10
(f.2),
XIV
12 (f.3),
XV
11
(f.5) or
XV
12 (f.7).
Depth of body 32.0-35.5; length of head 37.0-44.0; length
of
pectoral
fin
27.0-32.5,
of

60
E.
TREWAVAS,
J.
GREEN
AND
S.
A. CORBET
FIG.
12.
Myaka
rq~aka,
holotype.
\
\
\
/
/
/
,,
,
/
,
/
-
-
----_-__-
FIG.
13.
Myaka rnyaka,
dentition
of
jaws, lower pharyngeal bone and gill-rakers
of
first
arch, with oral view
of
two inner and two outer rakers
from
the upper part
of
the ceratobranchial.
Anal
111
8
(f.]),
111
9
(f.18)
or
Ill
10
(f.7).
Pelvics reaching anus or genital papilla.
Colour
watery grey or greenish grey, usually with yellow pelvic
fins.
No
tilapia-mark.
Size.
Known
only
at the size-range
of
the paratypes.
Breeding
The holotype was the only male seen by
us
with advanced gonads. The testis was
1.7
mm wide, but its anterior third was narrower. It was feeding and there were fat-bodies

ECOLOGICAL STUDIES
ON
CRATER
LAKES
1N
WEST CAMEROON
61
among the viscera; there was more black pigment about the head and body than in any
other.
In
the females the ovaries were all very short, at the hind end of the abdomen, and
contained very few oocytes and some amorphous bodies. In a fish of 60 mm
SL,
examined
when preserved, the right ovary contained a single, not quite spherical, orange-coloured
body that may have been a nearly ripe egg; two similar bodies were in the left ovary and
both ovaries contained a few much smaller oocytes.
I
think it possible that
in
this species the eggs are smaller and fewer than
in
other cich-
lids of Barombi Mbo. The fact that in several there was little material
in
the gut, the
viscera were packed with fat and the genital papilla was swollen suggests that they were
preparing to breed or recently spent. If this interpretation is correct
Myaka
mj,aka
is a
species of low individual fecundity. Nothing is known about its parental habits.
Konia
gen.
n.
Type species
:
Tilapia eisentrauti
Trewavas
=
Konia eisentrauti.
Konia
from the Barombi name of the type species-konye.
Resembling
Tilapia
and
Sarotherodon
in having the apophysis on the base of the skull
for the upper pharyngeals formed from the parasphenoid alone a ventral apophysis
on
the
3rd vertebra, cycloid scales with concentric circuli and in the outer bicuspid and inner
tricuspid teeth of the jaws; but differing from them
in
the narrower interorbital space
(23.5-28.3
%
length of head), the shorter intestine (1
1-1
4
times the
SL
of the fish) corres-
ponding to a more carnivorous diet, and the absence of a “tilapia-mark” at the junction
of spinous and soft parts of the dorsal fin.
It resembles the genus
Surotherodon
in having the mesethmoid free from the vomer, in
the large eggs
(4
mm diameter in the type species) presumably brooded in the mouth
of
one or both parents, fine pharyngeal teeth, the lower
on
a bone whose anterior blade
is
0.7-1.5 times the length of the toothed area, and gill-rakers numbering
14-1
7
on
the lower
part of the first arch.
Vertebrae 29.
Within
Sarotherodon
it resembles none
so
much as
S.
lohbergeri,
its companion in
Barombi Mbo, but this, in addition to having a
Sarotherodon-like
interorbital width and
a long intestine, has a very distinct dentition.
Konia eisentraut
i
(Figs
14
and
15)
Tilapia eisentrauti
Trewavas, 1962
:
168,
fig. 5A, 6A.
T.
(Pelmatolapia) eisentrauti;
Thys,
1968
:
xxviii.
Barombi name
:
konye.
The “dubious specimen” whose description was appended to that
of
this species in
1962 is now known to be a dikume
(K. dikume).
K. eisentrauti
is recognised in the field by the evenly decurved profile of the snout
descending
to
a nearly horizontal mouth and by its colour-pattern. This consists of an
upper series of black blotches parallel to the dorsal outline and a black band of uneven
width extending from the opercular spot to the anterior part of the caudal peduncle; at
the posterior end of the peduncle is a vertical blotch meeting its fellow over the top. The

62
E.
TREWAVAS,
J.
GREEN AND
S.
A.
CORBET
FIG.
14.
Koni~i
ei.trntrtiiiti,
a paratype
of
SL
64.5 nini.
_----
.
FIG.
15.
Konh
eiser~/rau/i,
(a)
inner
and
(b)
outer jaw teeth and lower pharyngeal bone
of
a
paratype
of
SL
62
mm
and gill-rakers
of
first
arch in mother specimen.
Both
inner and
out
gill-rakers have a subsidiary tubercle.
In part, from Trewavas, 1962.
fins are colourless (in the smallest the soft dorsal had faint pinkish stripes); there is
no
tilapia-mark.
Larger individuals have a dark grey snout and back and the upper series of blotches
is
no
longer visible. The colour-pattern is then similar to that of leka keppe
(S.
lohbergeri),
but the shape of the snout
is
distinctive.
As
in
leka, individuals of both sexes with maturing
gonads often have the branchiostegal membrane, isthmus and
in
some chest and lower
jaw black. This seems to be related
to
gonad condition and
not
to size.
Further distinguishing characters are the three regular rows of teeth (two in the
35.5
mm
fish, a few teeth
of
a fourth
in
the lower jaw
of
an
86
mm one), the outer having less
slender shafts than in
S.
lohbergeri;
the shape of the lower pharyngeal bone (Fig.
15),
in
which the blade is
0.7-0.95
of
the median length
of
the toothed area; and the narrow
interorbital space,
24-27
of the length of head.

ECOLOGICAL STUDIES
ON
CRATER
LAKES
IN
WEST
CAMEROON
63
In
a specimen of 66 mm
SL
the intestine measured 86 mm 1.43 times the
SL.
Our specimens ranged from 35.5 to 86 mm
SL.
The 86 mm fish was the only male with
advanced testes; these were about 2.3 mm wide and looped to fit their increased length
into the body-cavity the viscera were packed with ribbons of fat.
The largest ovarian eggs seen by us were in
a
fish of 54 mm
SL;
they were green, nearly
spherical, with a long diameter of
4
mm. The abdomen was nearly filled with these eggs,
of which there were seven in the right ovary, nine
in
the left.
In
the interstices between
them were a few very small translucent oocytes. Two other females contained green eggs
of 3.6-3.7 mm, one a fish of 59.6 mm SL, the other 82 mm. The latter had a total of 58
eggs, 33 in the right ovary, 25 in the left. We deduce that the size of the eggs is fairly
constant and the number depends on the size of the mother.
The majority of our specimens had very narrow testes, or ovaries
in
the stage at which
small cream-coloured and smaller translucent oocytes are mixed with red and yellow
amorphous bodies. We identified no very young specimens (but perhaps because the mesh
of the traps was too wide) and we think that the species was just entering upon a breeding-
cycle. Ripe eggs are of
a
size at which mouth-brooding is the probable mode of parental
care, but we found no brooding partents.
Konia dikume
sp.
n.
(Figs 16 and 17)
Tilapia eisentrairti
(“dubious part”) Trewavas, 1962: 169, fig.
5B,
69,
c
and d.
Holotype BMNH 1971.10.20.50, 104.5 mm
in
SL.
Paratypes
BMNH
1971.10.20.51-67,
Barombi name: dikume.
62-1
11.5
mm
SL
(17 specimens).
Distinguishing characters
In
contrast to
K.
eisentrauti
the snout is acute, the mouth at an angle of about 20”-30”
with the horizontal, the lower jaw with a slight mental process, its front teeth often pro-
jecting beyond those of the upper jaw. When freshly caught no dark markings are to be
seen
on
the body and it can always be recognized by the blood that swells the capillaries,
especially at the bases of the fins, and oozes over the surface, staining its silvery colour.
Preserved, there appear two to five dark blotches between upper lateral line and dorsal
fin and in some
a
vague dark stripe at the level
of
the horizontal myoseptum.
The teeth are usually in only two series and the lower pharyngeal bone is very slender,
with a relatively longer blade than in
K. eisentrauti.
Description
of
the holotypes and paratypes
Proportions as
SL
Depth of body 36.5-40.0; length of head 37.5-40.7; length
of
pectoral fin 32.0-36.0;
last dorsal spine 12.5-16.5; third anal spine 12.0-15.5 (from a little shorter to
a
little longer
than last dorsal); length of caudal peduncle 13.5-16.5 (0.9-1.2 times its
own
depth).
Proportions as
%
length
of
head
Length of snout 31.5-35.0; diameter of eye 27.3-31.8 at SL
of
88-1 12 mm, 31.2-33.3
at SL 62-71 mm; depth
of
preorbital 19.0-23.0; interorbital width 23.5-28.3; length
of
lower jaw 33.3-36.7, of premaxillary pedicels 20.0-25.0.

64
E.
TREWAVAS,
J.
GREEN AND
S.
A. CORBET
FIG.
16.
Konia
rlikurnr.
holotype.
\
FIG.
17.
Koniu
rlikunze,
two lateral teeth
of
the lower jaw, outer teeth from near midline
of
upper and lower
jaws, an inner tooth, lower pharyngeal bone and outer gill-rakers
of
first arch in
a
specimen
of
SL 109
mm.
A
section
of
the arch is drawn in oral view
to
show that each inner raker
has
a
subsidiary tubercle on the ridge
of
the arch. In part from Trewavas.
1962.

ECOLOGICAL
STUDIES
ON
CRATER LAKES
IN
WEST CAMEROON
65
Length
of
pharyngeal bone 29.6-30.7, its width 26.0-26.7, the blade up to
14
times the
median length of the toothed area; in young of 62-71 mm
SL
blade equal to dentigerous
area
in
length and whole bone only 26.3
%
length of head.
Teeth of jaws in two rows, occasionally with
a
few teeth forming a third row anteriorly;
45-66 in outer row of upper jaw; outer bicuspid, the two cusps separated by
a
terminal
narrow notch (Fig. 17), in adult 2-4 at end of row with minor cusp reduced or absent;
inner tricuspid. Teeth of lower pharyngeal bone very small and slender, not crowded,
unicuspid or with minor cusp
a
mere shoulder.
Gill-rakers on first arch (3-4)+
1
+(14-16), usually
14-1
5
on lower part. Microbranchio-
spines present
on
outer sides of 2nd, 3rd and 4th arches.
Intestine about
11
times the
SL
(a specimen of
SL
101
mm).
Scales
on
cheek in two series, usually expanding to three on the anterior part; in lateral
lines series 31
in
14
specimens of the
15
counted,
30
in the other; 3a-4 between origin
of
dorsal and lateral line, 16 around caudal peduncle.
Dorsal
XV
I I
(f.13),
XV
12 (f.2) or
XVI
I1
(f.
I).
Anal
I11
9-10.
Caudal truncate or very slightly emarginate when spread.
Culour.
Silvery when freshly caught, with fins tinged watery yellow and snout dark.
When preserved, two to five dark blotches between dorsum and upper lateral line and a
vague dark band at level
of
horizontal myoseptum. At 60-70 mm
SL
preserved specimens
with faint traces of 8-10 vertical bars on upper half of body. Dorsal lappets black-tipped.
No
tilapia-mark.
Most of the many specimens brought to us were of the same size, about 90-1
10
mm
SL,
but we saw a few of 60-70 mm. Although absence
of
smaller ones may be explained by
the mesh
of
the traps and nets used in the deeper waters, the upper size-limit probably
represents that of the species, since larger
S.
Iinnellii
and
S.
caroli
were caught with the
same gear. The narrower-meshed traps set near or at the shore caught few
if
any of this
species.
No
ripe individuals were encountered; the males had narrow testes and several females
had ovaries at the stage where the largest oocytes were cream-coloured spheres about
1
mm
in
diameter. In addition there were smaller oocytes and yellow and red amorphous
bodies.
Dikume probably spends much of its time at depths of
20m
or more. The freshly-
caught fish is suffused with a thin film
of
bright red blood, the pectoral axil
is
a red blotch
from the swollen and oozing capillaries, the gills are deep red, with gas-bubbles enmeshed
among the filaments, the swim-bladder is swollen and all the blood-vessels of the viscera
and the lining of the gill-cover are conspicuous and swollen. When the gill-cover of the
preserved fish is lifted
a
perfect impression of the first gill is often seen in the layer
of
clotted blood lining the operculum.
The occasional
S.
linnellii
or
S.
caruli
caught with dikume shows no such phenomenon,
but they are said to be caught in the upper part of the net.
Pungu
nom.
n.
Buromhia
Trewavas, 1962: 184 (preoccupied in Insecta, see p. 43).

66
E.
TREWAVAS,
J.
GREEN AND
S.
Pungu maclareni
(Plate
I
and Figs
18
and
Baronibia maclareni
Trewavas, 1962: 184, fig.
14,
15
Barombi name
:
pungu.
347, pl.
31
fig. e.
A.
CORBET
9)
Trewavas
in
Fisher
et
al.,
1969:
Holotype:
8
75+17 mm,
BMNH
1959.8.18.177. coll. P.I.R. Maclaren (originally
wrongly labelled as from Barombi Kotto).
The golden-yellow colour
with
black blotches is well shown in Plate
I.
The black mark-
ings, although frequently quite irregular, are often arranged along a mid-lateral horizontal
band and an upper band parallel to the dorsal outline, like those of
K.
eisentrauti
and
S.
lohbergeri
(and many other African cichlids). There is endless variety both in the
amount of black and in its arrangement. The contrasting gold
that
makes this such a
conspicuous little fish develops as the fish matures, replacing the more usual silvery back-
ground of the young. There is no tilapia-mark.
Even without the distinctive colour pungu would be instantly recognisable by its small
terminal mouth and thick lips.
Our specimens ranged from
35
to 100 mm
SL.
From them something can be learnt
of
the development of the remarkable dentition, of which that of the holotype (Fig.
18)
is
typical. The inner teeth have three subequal cusps at all ages, except that from SL 90 to
100
mm two to four teeth in the middle of the lower jaw are larger than the others and
have their lateral cusps reduced or absent. There are two inner rows in the lower jaw, one,
or one with an incomplete second row, in the upper. The 35 mm fish has only three or
four inner teeth in each jaw. The lower outer teeth do not increase in number, being seven
or eight from
SL
35 mm to
100
mm (one 60 mm fish has only five), but the upper outer
increase from 12 at
SL
35 mm to 23 at 100 mm. At 35 mm all the outer teeth are notched
(except some with the brown crowns worn
off).
At 53 mm only the two middle upper and
six of the eight lower are without trace of a minor cusp. At
100
mm all the outer teeth
are simple, including one partly erupted and not yet
in
use, which shows that at least some
of the teeth are simple by replacement, not by wear.
The lower jaw is stout and broad and the jaw-muscles are highly developed, a consider-
able superficial layer of the A2 part of the adductor mandibulae inserting musculously on
the coronoid process (Fig. 19). The proportions of the lower pharyngeal bone and the
nature of its teeth are as in the type (Fig.
18).
Microbranchiospines are absent in smaller specimens, but at 65 mm they have developed
along the 4th arch and parts of the 2nd and 3rd;
in
a 95 nim fish there are none on the
2nd arch, a short series (and a corresponding groove) on the 3rd and 4th;
in
the 100 mm
fish there is an interrupted series
on
the 2nd arch, complete series on the 3rd and 4th.
The intestine in two specimens of 66 and 70 mm
SL
measured 2.6-2.8 times
SL.
The ranges of meristic characters for
I1
specimens are as follows:
Gill-rakers (2-3)
t(0-1)+(13-16)
on the first arch; 13 only
in
the smallest specimen, 14
Scales 29 or 30 in the lateral line series, 14-16 around the caudal peduncle.
Dorsal XV
10
(f.4), XV1
10
(f.5),
XVI
11
(f.1)
or XVlI
10
(f.1). Anal
111
8-9.
Vertebrae: 29 (seven specimens).
The largest ovarian eggs were
in
a
fish of 65 mm SL and measured 4
Y
3 mm, six in the
in
only one.

J.
Zool.,
Lond.
(1972)
167
PLATE
1.
Pmgu
maclareni.

ECOLOGICAL STUDIES ON CRATER LAKES
IN
WEST CAMEROON
67
-
1
mm
U
1
mm
FIG.
18.
Pungii
maclareni,
anterior and lateral views of lips and teeth in the holotype, an outer lower tooth in
a juvenile
of
SL
35
mm, lower pharyngeal bone
of
the holotype and gill-rakers of another specimen. In part from
Trewavas,
1962.
FIG.
19.
Pungu
maclareni,
dissection
of
jaws and their muscles and isolated lower jaw; the upper jaw has slipped
forward a little; normally the teeth meet as in Fig.
18.
A1
and
A2,
parts of adductor mandibulae muscle.
art, “Articular”; d, dentary; iop, interopercular bone; lig, ligaments; m.gen, geniohyoid muscle;
m,
maxilla;
pal, palatine; pmx, premaxilla; pop, preopercular; q, quadrate; r.art, retroarticular
or
“angular”;
s,
anterior tooth
and part of lower jaw of left side; uet, unerupted tooth (probably suppressed right anterior tooth).

68
E.
TREWAVAS,
J.
GREEN
AND
S.
A.
CORBET
right ovary, nine in the left. They were green and lapwing-egg-shaped, with contrast
between the widths of the two ends,
so
were probably nearly ready to be shed. In the same
ovaries were small white oocytes of
0.5
mm diameter and some amorphous masses.
We have no information on mating or parental behaviour.
Genus
Stomatepia
Trewavas, 1962
:
18
1
;
type species
Paratilapia mariae
Holly, 1930
=
Stornatepia mariae.
Cichlid fishes with 27-29 vertebrae (mode 28), the third with a pair of apophyses united
below, for the attachment of the swim-bladder and the pharyngeal muscles; the apophysis
for the upper pharyngeal bones formed by the parasphenoid alone; scales cycloid with
concentric circuli; inner teeth of the jaws tricuspid at all ages, outer bicuspid in the young,
the small minor cusp becoming minute or obsolete in the adult; lower pharyngeal bone
with blade from approximately equal in length to the dentigerous area to nearly twice as
long; pharyngeal teeth slender, but not crowded, the lower posterior with minor cusp
respresented by a shoulder; snout long, cleft of mouth oblique (at 40-50" with the hori-
zontal); gill-rakers
15-21
on lower part of the first arch, slender, the upper slightly ex-
panded distally; a narrow interorbital region (16.0-24.5
"/,
of length of head). The
mesethmoid is free from the vomer.
Intestine shorter than
SL
of fish.
Eggs large
(4-5
mm in long diameter) and olive-green.
Three species, restricted to Barombi Mbo.
Key to species of
Stornatepia
lb. Depth
of
body 29.5-35.5%
SL;
upper
jaw 28.5-35.2% length
of
head
. . .. ..
2
la.
Depth
of
body 25-30.5%
of
SL;
upper jaw 26.8-28.80/, length
of
head
.
.
. . .
.
S.
inongo
2a. Lateral line cavities
of
the head enlarged; colour-pattern,
if
not masked by black, a series
of
dark blotches along the side;
a
small, vague tilapia-mark present
or
absent in the
young: usually no micro-branchiospines
.
. .
.
..
.. .. .. .
.
S.pindu
2b. Lateral-line cavities
of
head not enlarged; colour-pattern
a
continuous dark band along
the side;
a
strong, clear-rimmed tilapia-mark in the young, reduced or absent in adults:
microbranchiopsines present
.
.
..
..
..
..
I.
..
..
.
. S.
mariae
The proportions and meristic characters of the species are summarized in Tables
IV
and
V.
Stomatepia mariae
(Figs 20, 21 and 22)
Paratilapia mariae
Holly, 1930
:
206, pl. ii fig. 12.
Stomatepia mariae
(part.)
;
Trewavas, I962
:
182, fig. 13 (but
not
13(a)
and
(b)).
Barombi name
:
nsess.
Holotype: of
109
mm. Vienna Mus. 13950.
From Tables
IV
and
V
and Figs 22, 23 and 26 it can be seen that mess has usually a
shorter pectoral fin than pindu and a slightly wider interorbital region than either of the
other two species. The range and modal number of the gill-rakers are higher than in
S.
pindu
and
S.
mongo.
The modal number
of
dorsal rays is the same as that in
S.
mongo
and
one higher than in
S.
pindu.

ECOLOGICAL STUDIES ON CRATER LAKES IN WEST CAMEROON
69
FIG.
20.
Stornarepiu
rnariue,
lower pharyngeal bone and first gill-arch
of
a
specimen
of
SL
92
mm.
(a) Outer jaw
teeth and one inner
of
the same.
(b),
(c) Outer and inner jaw teeth
of
two specimens
of
S.
pindu,
respectively,
58
and
45.5
mm in
SL.
From Trewavas,
1962.
Microbranchiospines are present on the outer sides of the 2nd, 3rd and 4th arches.
At all sizes it is distinguished from
S.
pindu
by the normal, smaller size of the lateral
line cavities and by the colour-pattern.
The outer teeth have a small minor cusp up to about 35 mm
SL.
Above this size they
are unicuspid or (a few) with a shoulder representing the minor cusp. The inner remain
tricuspid with the middle cusp dominant. There are 40-60 outer teeth in the upper jaw
from
50
mm
SL
upwards, fewer in the young.
We found
no
black individuals. The colour is silvery grey with some brassy iridescence;
a dark grey or black continuous horizontal band extends along the body, usually ending
at the anterior end of the caudal peduncle, at the caudal end of which is a dark vertical
blotch. Dorsal, caudal and pectoral fins are watery yellow, orange yellow in mature males;
pelvic and anal fins may be dusky. In the young there
is
a
large black tilapia-mark; this
dwindles with age, but a trace may remain in the adult (Fig.
22).
The largest ovarian eggs were found in fishes of
75-80
mm
SL.
They were oval, green
and about
4
mm in long diameter. In a 75 mm specimen each ovary contained
12
ripe
eggs and they took up
so
much space in the abdomen that one ovary, the left, was almost
entirely anterior to the other. In another
(80
mm) the eggs, though already green, were
not
so
large (about 3 mm) and the ovaries were side by side.
A
third specimen (79.5 mm),
also with green eggs of 2.9 mm, had
14
in one ovary; the other ovary was smaller.
Ripe or ripening males were bigger, 91.5-109 mm
SL.
All specimens with ripe
or
ripening gonads had large fat-bodies among the viscera.
The smallest specimen caught was captured in
a
narrow-meshed trap set among the
We saw no mouth-brooding fish but we suppose that such large eggs must be cared for
shore vegetation. It measured 19-5 mm and had a conspicuous tilapia-mark.
in this way.

70
E. TREWAVAS,
J.
GREEN AND
S.
A.
CORBET
~ ~~
1
crn
FIG.
21.
Lower jaw and preoperculum from, left,
Sromutepiupindu
of
SL
88
mm, right,
S.
rnariue
of
SL
107 mm,
as examples of lateral line bones
of
head, in which the canals and bony openings are enlarged in
S.
pindu.
The canal
in the “articular”
of
S.pindu
is
so
much enlarged that the retroarticular is reduced in comparison with that of
S.
mariae,
and the posterior opening in the dentary is extended to involve the articular too. In the drawing the
foreshortened coronoid process is omitted.
Stomatepia pindu
sp.
n.
(Figs 20(b), (c), 21, 23 and 26)
Stomatepia mariae
(part., nec Holly); Trewavas, 1962: 182, figs 12, 13(a) and (b).
Barombi name
:
pindu.
Holotype:
8
81.5+20 mm collected in March 1970,
BMNH
1971.10.20.1. Paratypes:
The meristic characters and proportions are set out in Tables
IV
and
V.
The teeth are in two series in the young, a third is gradually added and (in one)
a
few
of
a fourth. From about 40 mm
SL
upwards the outer are unicuspid or with the minor cusp
reduced to a shoulder, but the inner remain tricuspid, with the middle cusp wider and
longer than the lateral; 40-60 in the outer series
of
the upper jaw at and above 40 mm
SL.
21 specimens, 32.5+8
to
91+21 mm,
BMNH
1971.10.20.2-22.
Lower pharyngeal as in
S.
mariae,
with blade slightly longer than toothed area.

ECOLOGICAL STUDIES ON CRATER LAKES
IN
WEST CAMEROON
71
FIG.
22.
Sromatepia mariae
of
SL
97
111111.
FIG.
23.
Stomatepia pindu,
holotype.
Pindu is deeper-bodied than mongo and usually than mess
(S.
muriae).
The depth
is
35
%
SL
or more in 10 of 18 in which it was measured, whereas in
S.
rnuriue
only three of
15
were so deep and these were young fish less than
40
mm in
SL.
Pindu has a longer pectoral
fin
than the other two species and in mature fish the dorsal
and anal are often produced beyond the base of the caudal.
In contrast to
S.
muriue,
there are usually no microbranchiospines. Among 13 specimens
examined for this feature they were found in only one and this had incomplete series
on
the outer sides of arches 2 and
3,
none on the fourth.
At all sizes pindu
is
easily distinguished from
S.
rnuriue
by the enlarged lateral-line cavities
of the head (Figs 21 and 23), and the tubules of the lateral line itself are more prominent

72
E. TREWAVAS,
J.
GREEN AND
S.
A. CORBET
too. The colour-pattern is distinctive. Many individuals are totally black, including the
fins and in some even the genital papilla. But the black may be restricted to the dorsal side,
passing into uneven blotches on the lower parts of the body and fins. Others are grey with
a series of dark grey or black blotches along the side (never the continuous band of
nsess) and an upper series of blotches above the lateral line (see Trewavas, 1962, fig. 12).
In both black and grey individuals there is often a characteristic crimson iridescence on the
operculum. One fish when alive had a shining green ground-colour instead of the usual
grey and its pelvics were orange.
The most advanced male examined was an all-black fish of 75 mm
SL.
A female of
68.5 mm had both ovaries full of large (about 4
x
3 mm), oval, green eggs, eight in the right
ovary, seven in the left. This fish had the black pigment in large irregular blotches, darker
dorsally, greyer on the lower half of the body. A female of 91 mm with ovaries “starting”
was all black except grey zones on caudal and anal fins. It seems, therefore, that the all-
black pigmentation has no simple relation to sex and breeding phase, and indeed individuals
have been observed to change colour.
The three smallest individuals preserved were pale, two of them with the mid-lateral
series of blotches, the third
(SL
44
mm) with no dark pattern; and the eight young of the
Eisentraut collection (Trewavas, 1962), 41-60 mm in
SL,
were all, when preserved, grey
with the lateral blotch pattern. Some
of
the latter, up to
50
mm in
SL,
had
a
small tilapia-
mark. This is less intense than in
S.
mariae
of the same size and does not persist into the
adult.
Stomatepia mongo
sp.
n.
(Figs 24 to 26)
Barombi name
:
mongo.
Holotype:
$2
93-5+19.5 mm, with ripe ovarian eggs, collected 24.1V.1970. BMNH
Paratypes: 10 of 79-100 mm
SL,
collected March-April 1970. BMNH 1971.10.20.24-33.
Meristic characters and proportions are set out in Tables IV and
V.
This is a very well characterized species, contrasting with its congeners in the shallow
body and long snout. Except the depth of body, most of its proportions are like those of
the other two species, but the snout and preorbital bone are a little longer. The lower jaw
bears the same relation to the length of head, but the upper jaw
is
smaller, as shown by
the measurement of both the pedicels and the oral edge (Table IV).
As
in the other species the inner teeth are tricuspid. The outer have two very unequal
cusps or the minor cusp may be reduced to a shoulder or (some teeth of one fish) obsolete.
There are 40-64 outer in the upper jaw.
The pharyngeal bone is narrower than in the other two species and its dentigerous area
occupies a smaller proportion of the length (Fig. 25).
The lateral-line cavities of the head are somewhat enlarged, but less
so
than in pindu.
The modal number
of
gill-rakers is lower than in either of the others. Microbranchio-
spines may be present or absent.
The colour of the body is usually dark grey, counter-shaded, sometimes darker along
the side, but without a well-defined pattern of band or spots. Often the pelvic and anal fins
are tinged with salmon-pink and the dorsal lappets may be pink or (in males) orange,
tipped with black. Some individuals are wholly black as in pindu. There
is
no tilapia-mark.
1971.10.20.23.

ECOLOGICAL STUDIES
ON
CRATER LAKES
IN
WEST
CAMEROON
13
FIG.
24.
Stomatepia
mongo,
holotype.
-
1
rnrn
*
c
--
1
crn
I
1
rnrn
FIG.
25.
Stomatepia
mongo,
from a fish
of
92
mrn. Lower pharyngeal bone and isolated anterior and posterior
teeth; left first gill-arch and enlarged oral view
of
an upper part of the ceratobranchial; and typical inner and outer
teeth from the jaws.
Many of the specimens examined had much fat among the viscera, and gonads starting
to swell, but one female with cream-coloured oocytes had
no
fat.
The holotype had ripe, olive green ovarian eggs of long diameter
4.5-5.0
mm as well as
some white oocytes of
1
mm. The right ovary contained
18
eggs, the left about
13.
In
this
fish the lower lip
is
very thick at the symphysis.

74
35
30
E
E
-
25-
e
,o
20-
._
c
-
u
Q
-
15-
c
0
c
+
3
E. TREWAVAS,
J.
GREEN AND S.
A.
CORBET
I
I
I
1
I
I
I
I
0.0
0
A-
0
AoAo
0
OCb
-
OOA'
A
AA
-
O:OoO
AA
00
0
-
0
0
0
00
-
0
10-
.o
-
0.
5-
-
I
I
I
I
I
I
I
I
30
40
50
60
70
80
90
I00
110
FIG.
26.
TABLE
IV
Proportions
in
the species
of
Stomatepia
Proportions as
%
SL
Depth
of
body
Length
of
head
Length
of
pectoral
Longest dorsal spine
3rd anal spine
Length
of
caudal peduncle
Caudal peduncle:
lengthldepth
mariae
SL
29-109
Proportions as
%
head
Length
of
snout
Diameter
of
eye
below
70
mm
SL
70
mm or more
Depth
of
preorbital
Interorbital width
Length
of
lower jaw
Length
of
premaxillary pedicel
(at
SL
40
mm
+)
Length
of
upper jaw
29.5-35.4
39.143.7
28.0-3 3.3
11.0-15.4
10'0-14.4
13.3-18'0
pindu mongo
32.5-91 70-101
304-37.0 25.3-30.6
38.6-44.0 38'4-41.4
32.2-37.0 27'5-32'0
12'3-15.0 12.0-14'0
12.0-15.0 12.4-14.2
12.8-15.2 11'0-13'5
1.0-1.3 0.95-1.2 0.9-1'1
30.5-40.0
27'8-33'5
24.5-29.4
18.5-24.0
18.0-24.5
38.243.5
22.2-26'6
28.0-32.5
31.0-36.6
25.0-31 '4
24.0-28.0
17.8-24.5
16.0-20.6
36.5-42.2
22'6-27.0
28.5-35.2
34.0-39.0
-
23.0-28.5
20.0-25.2
16'0-18.5
3 7.6-40.7
19'3-23.4
26'8-28.8
mongo.

ECOLOGICAL STUDIES
ON
CRATER LAKES
IN
WEST CAMEROON
TABLE
V
Frequencies
of
meristic numbers in the species
of
Stomatepia
(for vertebrae
see
Table
I1
on
p.
48)
75
S.
mariae
S.
pindu
S.
mongo
Dorsal
formula
XI11 11
XIV 10
XIV 11
xv
10
xv
11
XIV 12
xv
12
XVI 11
Dorsal Spines
XI11
XIV
xv
XVI
Dorsal totals
24
25
26
27
Soft anal rays
8
9
10
Gill-rakers
15
16
17
18
19
20
21
-
1
3
1
8
6
10
9
1
4
14
-
-
12
7
-
3.5
8
5.5
3
1
3
23
2
2
1
26
4
4
25
2
5
22
2
3
9
12
1
2
-
-
1
2
6
4
1
-
-
1
12
1
4
10
1
10
2
We found no mouth-brooding specimens, but probably the eggs are cared for in this
way. The species was very rarely caught during our time on the lake, but several specimens
turned up in the middle of April in traps. The fishermen were making special efforts
to
find them for
us
and in March said that we might expect to find breeding individuals in
April. This proved
to
be the case and it is possible that we were seeing the beginning
of
an
inshore breeding migration.

16
E. TREWAVAS,
J.
GREEN AND
S.
A.
CORBET
Ecology
of
the
lake
J.
GREEN
AND
SARAH A.
CORBET
Barombi Mbo is roughly circular and about
2.5
km across, with one major outflow
through the Kake Gorge at the south-eastern corner and several small inflows, some
of
them seasonal. The only permanent inflow is near the Barombi village to the north-west.
The bottom
of
the lake descends steeply all around to a maximum depth of
11
1
m near
the middle. Our visit coincided with the end of the dry season, when the water level was
probably at its lowest. From markings on rocks and tree trunks projecting out of the water
we estimated that the water level fluctuated through about
1
m during the course of a
year.
The rim
of
the crater is heavily forested, and projects up well above the surface of the
lake, providing considerable shelter,
so
that the surface of the lake is generally calm.
On
only two occasions during five weeks did we observe winds strong enough to produce
waves that broke offshore as white horses. The lake is clear,
so
that underwater observa-
tion of fish
is
easy.
The surface temperature of the lake in March and April was
29"-30°C.
We did not have
apparatus to measure temperatures in the deep water, but we did suspend a thermometer
at a depth of
4
m and then read it in
situ.
When the surface temperature was
29.5"C
the
temperature at
4
m was
28°C.
The conductivity of the water was low. Measurements made both inshore and in the
middle
of
the lake gave a value of
39
pmho/cm/2O0C in March and April.
The clarity
of
the water and the low conductivity indicate that Barombi Mbo is not a
very productive lake. Vertical hauls with phytoplankton nets through
a
depth of
100
metres yielded very little plant material. Most of the phytoplankton seems
to
be too small
to be retained by the
180
mesh nets. A high proportion of the stomach contents of the
main phytoplankton-eating fishes (unga-Surotherodon linnellii and
S.
curoli) consisted of
unicellular green algae with
a
diameter of about
10
pm or less. Small dinoflagellates were
also present in most
of
these fishes, but were absent from the net samples.
The zooplankton of the open water is also relatively sparse (Green,
1972).
For instance
the total number
of
cyclopoids under one square metre is only about one tenth of the num-
ber found in Lake Mutanda, the least productive of three lakes
in
the Kigezi district of
Uganda (Green,
1965).
The most striking feature of the zooplankton of Barombi Mbo is
the total absence of Cladocera.
The bottom
of
the lake in the deeper regions appears to consist of fine sand and fine
organic debris. Our only means
of
sampling at depth was to allow the plankton nets to
drag along the bottom. Such samples yielded very few animals, and were remarkable in
lacking chironomid larvae.
Around the shore of the lake, plant material falling into the water from the surrounding
forest makes an important contribution to the food supply. The sandy bottom was thickly
covered with fallen leaves in some places, and almost clear
of
them in others. Where the
bottom was overlain
by
dead leaves the invertebrate fauna was more obvious.
On
the
leaves the tubes of chironomid larvae were abundant, and ancylid limpets (Ferrissiu sp.)
occurred more rarely. Among the leaves were the larvae of caenid and baetid mayflies, as
well as ostracods and dragonfly larvae. Small stones provided shelter for a similar group

ECOLOGICAL
STUDIES
ON
CRATER
LAKES
IN
WEST
CAMEROON
of invertebrates including mayfly larvae, ostracods and hydracarines, together with chiro-
nomid larvae. In addition, untidy silken webs on stones were found to contain hydro-
psychid larvae, or sometimes the larvae of the mayfly
Povilla adusta,
which was more
typically found in burrows in wood. Peritrichs lived epibiotically on mayfly larvae and
chironomid larvae. Epibiotic rotifers were found on the ancylid gastropod and on mayfly
larvae.
The endemic sponge
Corvospongilla thysi
(Brien) covered large areas on the outer sur-
faces of rocks from near the surface to a depth of at least
3
or 4 m. Another species.
Corvospongilla bohmii
(Hilgendorf), was found in crevices and hollows in rocks and wood
over
a
similar depth range. The rocks and boulders, and the sponges on them, bore a
thin coat of aufwuchs and debris derived from the faeces of shore fishes and from
material that settled
in
the water. Fishes were most abundant in these rockier parts
of
the shore.
The trunks of trees that had fallen into the lake formed an important habitat for inverte-
brates. The burrowing larvae of
Povilla adusta
were very abundant in suitably old branches
and logs. The crevices resulting from their tunnelling harboured numerous other inverte-
brates, including the larvae of mayflies, caddis, zygopterans and chironomids. Patches of
Corvospongilla thysi
were occasionally found on submerged wood, though it was very much
less common on wood than on rock.
Corvospongilla bohmii
was fairly common lining the
cavities in wood. Like the rocks the submerged logs and branches carried a thin covering
of aufwuchs and debris, and as on the rocks this formed the food of fishes. A crab was
once seen breaking up wood as if to extract the invertebrates living in it.
In relatively unshaded regions of the shore there were
a
few patches of rooted water
plants. The least common was
Potamogeton octandrus
which formed loose mats about
5
cm thick on the bottom in water about
0.5-1
m deep. Waterlilies grew in about
0.5-1
m
of water and had remarkably few animals, either invertebrates or fishes, associated with
them, though their leaves were scored with the oviposition marks of coenagriid dragon-
flies. Large patches of
Najas pectinata
grew in water
0.5-4.5
m deep. In the shallow parts
of its range these plants reached the surface, and the tangle of stems and leaves floating
there formed an oviposition site for anisopteran and zygopteran dragonflies. In the
deeper part of its range the tops of the plants were about
1
m below the surface of the
water. The leaves and stems of
Najas
were densely clothed with epibionts, including
gelatinous masses of the blue-green alga
Gloeotrichia,
the diatom
Gomphonema,
the green
algae
Spirogyra, Bulbochaete
and
Mougeotia,
flosculariacean rotifers and the peritrich
Cothurnia.
Among the leaves of
Najas
were zygopteran larvae, mayfly larvae, ostracods
and numerous shrimps
(Caridina).
We did not see fishes feeding on these free-living in-
vertebrates, but large shoals of leka keppe
(S.
lohbergeri)
used to graze the epibionts from
the leaves and stems of
Najas.
Animal and plant debris falling onto the surface of the lake contributed to the diets of
several species of fishes.
Stomatepia mariae, Barbus batesii
and young unga frequently
took floating terrestrial insects, and
Epiplatys sexfasciatus
fed almost entirely on these and
on emerging mayflies, also taken at the surface. Fruits and seeds, which were often seen
floating, were found as the main contents in the stomachs of some
Konia eisentrauti,
Stomatepia pindu
and
Barbus batesii.
Shoals of
Sarotherodon steinbachi
were often seen
dimpling at the surface, and cottony plant hairs, a characteristic component of the flotsam,
were often found in their stomachs.
I7

78
E.
TREWAVAS,
J.
GREEN AND
S.
A. CORBET
Feeding habits and ecological relationships
The information in this section is derived from two sources: the examination
or
stomach
contents, and observations made underwater in the natural habitat of the fishes. Using
a
face mask and snorkel, it was possible to observe ten species in the clear water around the
edge of Barombi Mbo. Eight of these species were identifiable in the field, but we were
unable to distinguish the two species of unga under field conditions. Five other species
(Konia dikume, Myaka myaka, Stornatepia
mongo
and the two species of
Clarias)
were not
seen in shallow water, or in dives down
to
depths of
5
or
6
m; their feeding habits are
known only from the stomach contents. Two other species
(Procatopus similis
and
Aphyo-
seinion oeseri)
were found only in the small stream flowing into the northern side of the
lake, where no underwater observations could be made.
The main food found
in
the stomachs of most of the species is shown in the trophic
spectrum (Fig.
27)
(see Darnell,
1961).
In our study the item occupying the greatest
volume in a stomach was classified as the main contents, and then any other items were
listed. Thus the relative importance of each food is expressed in two ways; as the percentage
of fishes with a given item as the main contents; and as the percentage of fishes in which
a
given item occurred, either as main or as subsidiary contents. Both percentages are shown
in the trophic spectrum.
Barbus batcsii
Suh and kimbundu
Suh
is the Barombi name for large
Barbus
and the name kimbundu is applied to smaller
specimens with an
SL
up to about
150
mm.
Barbus
is caught in both the lake and the
tributary stream, and we are not certain that all the fishes we dissected came from the
lake itself. We were able to watch both large and small
Barbus
swimming near the shore,
the smaller ones in water
0.5-1
m deep and the larger ones usually in depths greater than
1
m. We did not see the larger fishes feeding. They swam strongly to and fro, alone or in
small groups. Other fishes seemed to avoid them. Small specimens were observed in shallow
water vigorously turning over small stones with their snouts, and snapping at the place
where the stone had stood. This seems to be an effective technique for catching insect
larvae. Several kimbundu collected to feed, together with some
Konia eisentrauti,
around
a footprint where the sandy bottom had been disturbed.
In
Barbus
the stomach is not distinct, and whole guts were examined.
A
high proportion
of the fishes contained both mayfly larvae and chironomid larvae. Other arthropod prey
included caddis larvae, decapod crustaceans and terrestrial insects. The proportion con-
taining terrestrial insects may be unnaturally high because some of the suh were caught
on hooks baited with grasshoppers. Organic debris, sand grains and sponge spicules were
probably picked up from the bottom when insect larvae were taken. Some vegetable matter
was also found in the guts; one fish had been eating seeds.
Clarias walkeri
and
C.
maclareni
These two species were not separated in our dissections. Our specimens came from
both the tributary stream, where they were caught by women using hand nets, and the
lake. In the lake nyongo appeared to have a wide depth distribution; some were caught
in traps set close inshore, and some must have entered deeper water, because they had
eaten
Chaoborus
larvae, which we did not find in water less than 20m deep. We never
saw nyongo underwater.
Nyongo

FIG.
27.
Trophic spectrum illustrating the results
of
analyses
of
stomach contents
of
the fishes
of
Barombi Mbo.
The depths
of
the black blocks show the percentages
of
stomachs with each
food
as
the main contents, and the
depths
of
the hatched blocks show the percentages
of
stomachs in which each
food
was present.

80
E. TREWAVAS,
J.
GREEN
AND
S.
A. CORBET
The stomachs contained
a
wide range of food. Insect larvae (mayflies, chironomids and
caddis) formed the main food of small and middle-sized individuals. Some of the larger
specimens (up to 440mm SL) had eaten fish, including one
Konia dikurne,
one
S.
stein-
bachi,
and one
Myaka
myaka.
Epiplatys
sexfasciatus
Longo katta
This species swam in small groups close to the surface in inshore waters, but we did not
see it feeding. Sometimes the fishes lay just below the surface of the water, but they were
also observed keeping station about
20
cm below the surface. The stomach contents were
remarkably uniform, with over
89
%
of the main contents consisting of terrestrial insects
including beetles, lepidopterans, caddis and dipteran flies. The other main item was
mayflies in the process of emerging. It seemed as if these fishes took nearly all their food
at or very close to the surface.
Procatopus
similis
Lenge
P.
similis
was found only in the small stream near the Barombi village. The stomach
contents consisted entirely of insects, and the majority of these were terrestrial forms, includ-
ing a high proportion of ants. In spite of its upwardly directed mouth lenge also fed from
the bottom; about one third of the stomachs contained mayfly larvae or chironomid larvae.
Sarotherodon lohbergeri
Leka keppe
S.
lohbergeri
was the most abundant fish in shallow water. The main substance in all
the stomachs that we examined was organic debris. This was accompanied by other items
such as rhizopods
(Arcella, Euglypha
and
Dijlugia),
bdelloid rotifers, and sedentary
diatoms, which indicate that the detritus was picked up from the surface of a plant or
a
stone. This was supported by field observations. The adults of this species opened their
mouths very wide and applied them to the surface of a rock, as though sucking material
in. This apparently dislodged any loose organic debris. Young specimens, which have
silvery bodies and orange fins, fed by closing their mouths against the surface of the rock
and then giving a sideways twist as though wrenching something from the surface. The
silver flash produced by this movement in sunlight was a characteristic feature in water
less than a metre deep all round the lake. This species also fed by removing the rich auf-
wuchs from
Najaspectinata.
The fish turned
on
its side,
so
that the width
of
the mouth lay
along the axis of the stem.
As
far as we could see the tissues of the plant were not eaten,
but debris on the surface was removed together with filamentous algae or other aufwuchs.
S.
Iohbergeri
has also been observed in shoals at the surface. The upper lip was pro-
truded through the surface film, and small movements were made as if ingesting material
from the surface film. These movements may also have been respiratory, but the presence
of fine cotton-like plant hairs in the stomachs of some specimens indicated that they had
been feeding at the surface. The surface film of the inshore waters was often littered with
various plant products, such as seeds, hairs, petals and fallen stamens. The two species
which seemed to utilize this source of food to the greatest extent were the present one and
S.
steinbachi.
Sarotherodon steinbachi
Kululu
This was one of the easier species to recognise in the field, with
a
generally pale colour,
sometimes with a vague stripe, and
a
characteristic profile. We saw a single specimen with

ECOLOGICAL STUDIES
ON
CRATER LAKES
IN
WEST CAMEROON
81
a completely black caudal fin among a flock of normally pigmented individuals. The main
item in the stomachs was organic debris.
A
high proportion of the fishes contained sand
grains. This agreed with our field observations of feeding.
S.
steinbachi
was often seen
feeding in shallow sandy areas. The fish would take in a mouthful
of
sand, and then move
the material around in the mouth and spit out the sand, following this with a swallowing
movement. After this another mouthful of sand would be taken in. We also observed this
species collecting organic debris from the surface
of
rocks and sponges. The movement
appeared to be much more gentle than the corresponding movements of
S.
lohbergeri
feeding on the same rock. Shoals at the surface were also observed, apparently ingesting
material in a manner similar to that described for
S.
lohbergeri.
This would account for
the presence of cotton-like plant hairs and adult insects in the stomachs
of
a few indivi-
duals.
Sarotherodon linnellii
and
S.
caroli
We did not distinguish between these two species in the field or in our early dissections
of
the stomachs, but dissections of
14
fishes which were positively identified revealed no
apparent differences in their stomach contents. The adults of both species ate phyto-
plankton, but the young included a high proportion of animals in their diet. Specimens
with a standard length of 20-70 mm contained mayfly larvae and various terrestrial insects,
which were consistently absent from the stomachs of larger specimens. The young unga
were abundant close inshore, darting about and feeding in an opportunistic manner,
investigating objects at the surface or on the bottom. It was a common experience to find
these small fishes removing pieces of dead skin from one’s feet and attacking scars on
one’s legs.
Large unga were observed in very shallow water (0.5-1 m deep), often behaving in a
territorial manner, and indulging in mouth fighting. Large fishes seen in deeper water,
down to about 5 m, were sometimes in groups of
up
to
six
or eight individuals.
Myaka myaka
Myakamyaka
This small silvery fish seemed to live in the middle of the lake. We did not see it in
shallow water, and only once saw it alive. It was occasionally caught in deep traps and
sometimes in nets set far out from the shore. The stomachs contained very fine organic
debris and phytoplankton. The organic debris may have been taken from the bottom or
from suspension near the bottom, and much of it seemed to be derived from phytoplankton.
About a third of the stomachs contained
Chaoborus
larvae, which indicates that
Mj)aka
myaka
feeds in relatively deep water.
Konia eisentrauti
Konye
K.
eisentrauti
was not always easy to distinguish from
S.
lohbergeri
in the field.
A
good
lateral profile view was essential. It was not
a
very abundant species, and our observations
on
it were rather few. This species included more darkened plant tissues in its diet than
did
S.
lohbergeri,
and it included a wider range of animals. Several specimens had mayfly
larvae as the main contents, and one specimen had eaten large cichlid eggs. The main
impression from the stomach contents was that
K.
eisentrauti
fed in regions with abundant
decaying leaves, catching small invertebrates that it found there. This species normally
occurred in small groups, of two to eight individuals. When it fed,
K.
eistentrauti
seemed
to watch a particular object and snap it up, and then spit out some of the debris that had
Unga, fissi, kippe

82
E.
TREWAVAS,
J.
GREEN
AND
S.
A.
CORBET
been taken in with it. The feeding behaviour resembled that of
S.
steinbachi;
but the snap-
ping movement appeared to be directed at some object in the sand (possibly a mayfly
larva), whereas the feeding of
S.
steinbachi
seemed to be a sorting out of organic material
from the sand.
Konia dikume
Dikume
K. dikume
was abundant in the lake, but we never saw it underwater, probably because
it lived in deep water. The fishermen caught
K.
dikuine
in the deeper parts of gill nets set
well away from the shore, or in baskets set deep, and the fishes were always dead when
they were brought in. Their stomach contents provided further evidence of a deepwater
habitat. The presence of organic debris in most of the stomachs indicated that these
fishes often fed near the bottom, where they may also have picked up the sand grains and
sponge spicules that were found in some of them.
Chaoborus
larvae, which formed the
main contents in many of them, were sometimes mixed with organic debris, and sometimes
very clean, as though they had been caught at a time when they were swimming in the
plankton. (In plankton hauls taken by day, some of them including material from the
bottom,
Cliaoborus
larvae were not found in water less than
20
m
deep.)
Pungu maclareni
Pungu
P.
maclareni
was abundant in the shallow water near the shore, and was easily recognized
underwater by its very thick lips and its pattern of black patches on a background of rich
golden yellow. The larger individuals were seen in groups of about ten to thirty feeding
over submerged logs or patches of sponge on rocks, usually
1-2
m below the water surface.
In the shallower part of its range,
P.
maclareni
often fed at rocks in company with
S.
lohbergeri,
but on the deeper rocks,
2-3
m below the surface, it fed alone. Typically
P.
maclareni
oriented itself head down or on its side as it fed at rock or at wood, moving
steadily over the surface taking repeated rasping bites at it. When feeding on sponges it
applied its open mouth firmly and briefly to the surface, apparently scraping
off
chunks
with its prominent teeth. Probably the rough surfaces of the sponge and
of
the submerged
wood were rich in detritus and epibiotic organisms that had settled there, and, as well as
groups of sponge spicules, the stomachs often contained organic debris, sand grains and
benthic and epilithic diatoms. But their diet sometimes included insect larvae, particularly
larvae of chironomids and mayflies, which suggested that these fish could vary their feeding
habits when the opportunity arose. Further evidence for this was provided by the observa-
tion of a
P.
maclareni
following a crab which was using its pincers to prise bits
off
a
log
tunnelled by larvae
of
Povilla.
When the crab removed a piece of wood the fish moved in
and picked up something from under the wood as the crab lifted
it.
Some of the stomachs
contained higher plant material, and two of them were packed with the tissues of fruits
which must have fallen from a tree into the lake.
Although the larger fish at rocks normally fed only on patches of sponge, a group
of
about ten small
P.
maclareni
(20-30
mm
SL)
was seen over bare rock in
0.5
m of water.
Instead of working regularly over the surface, as the larger fishes did, these young ones
would pause as though watching, and then snap at particular objects on the rock.
Stomatepia mariae
Nsess
S.
mariae
lived rather deeper than
S.
pindu,
in water about
0.5-1
m deep. They usually
swam alone or
in
groups of two or three, but we have seen larger shoals of up to about
20

ECOLOGICAL
STUDIES ON
CRATER LAKES IN WEST CAMEROON
83
fish. When seen underwater they were usually lurking in some sheltered place or cruising
slowly around. They were rarely found feeding. Although plant tissues and organic debris
formed part of their diet, their stomach contents showed that they are predators. Like
S.
pindu,
they often took mayfly larvae and other aquatic arthropods. Unlike
S.
pindu,
they frequently visited the surface to take terrestrial insects floating there, and they some-
times preyed on fishes.
Stomatepia pindu
Pindu
S.
pindu
was a common species inshore wherever rocks or submerged logs provided
crevices into which the fishes could retreat. They were usually found, in twos and threes,
in shallow water, up to about
0.5
m deep, but they have been seen as deep as
3
m. The
species was usually recognizable underwater by the dark colouring of even very small
fishes; but they could change colour rapidly. When they were pale, sometimes with a row
of dark lateral patches,
S.
pindu
could be distinguished from
S.
mariae
by their humped
backs and pointed faces.
S.
pindu
is a predator, probably snapping up small aquatic arth-
ropods individually. Those with food in the stomach contained mayfly larvae, caddis
larvae, chironomid larvae and decapod crustaceans as well as smaller quantities of organic
debris, dark plant tissues and sponge spicules which may have been taken up accidentally
together with the prey. Fourteen of the
32
stomachs were empty. We saw adult
S.
pindu
feeding only twice. On one occasion, during the daytime, the fish hovered for some minutes
as though watching
a
leaf lying on the bottom; then it moved on and seemed to study two
more leaves; and at the fourth leaf, after a period of watching, it suddenly darted down
and took up a mouthful of material from under the leaf. It spat out debris and made
chewing and swallowing movements. On the other occasion
S.
pindu
was seen by torch-
light feeding at night. It applied its mouth to a rock and twisted its body as though picking
something up. Small
S.
pindu
about
15
mm in
SL
were seen feeding in a similar way,
watching and snapping at individual objects on the surface of a rock.
As
a predator,
S.
pindu
probably spends less time eating than does a scavenger such as
Sarotherodon steinbachi,
but it is surprising that such a common fish was
so
rarely seen to
feed. Possibly it fed in a place or at a time outside the range of our underwater observations,
which were confined to well-lit regions in full daylight.
Stoniatepia mongo
Mongo
S.
mongo
seemed to live rather deep in the lake. The fishermen caught very few, and these
all came from basket traps specially set in several metres of water. Of the nine whose
stomachs were examined, six had quite empty guts and the other three contained very little.
One contained mayfly larvae and woody tissues; one contained woody tissues only; and
one contained scraps of arthropod cuticle. Evidently
S.
mongo,
with its large mouth and
short gut, takes invertebrate prey.
It
is
not clear whether
S.
mongo
is a rare fish, or simply lives too deep to be caught by
the Barombis’ basket traps, which are usually set in shallow water.
Predators
of
the fishes
Two of the species of
fish
that we examined had eaten other fishes. We were unable to
identify the fish eaten by
Stomatepia mariae,
but those eaten by
CIarias
included
S.

84
E.
TREWAVAS,
J.
GREEN AND
S.
A.
CORBET
steinbachi,
K.
dikume and
Mjiaka
nyaka. It is probable that Barbus also takes fishes, but
we did not find any in the few large specimens that we examined.
Both of the two species of snakes found in the lake (Grayia smythii and Natris anascopus)
have been recorded as eating fishes. The Barombis said that G. srnythii was often caught
in fish traps. One specimen that they showed us was nearly
2
m long.
No
crocodiles or turtles were seen in the lake, and the Barombis said that there were
none.
The birds around the lake included the following species. The nomenclature follows
Mackworth-Praed
&
Grant (1970).
Senegal finfoot-Podica senegalensis
Little grebe-Podiceps ruji’collis
Long-tailed cormorant-Phalacrocoras africanus
Darter-Anhinga ruja
Grey heron-Ardea cinerea
Pied kingfisher-Ceryle
rudis
Malachite kingfisher-Corythornis cristata
Shining-blue kingfisher-A lcedo quadribrachys
The Long-tailed cormorant and the Darter are probably the most serious predators,
although their populations were not large. We estimate that there were no more than 15
to
20
Long-tailed cormorants and about four Darters on the whole lake. The Pied king-
fisher may also be a serious predator, being the commonest of the kingfishers. This species
has been reported to confine its attacks to fish less than
10
cm in length (Daget,
1954).
A
specimen
of
this kingfisher and a Little grebe were found dead in fish traps which they
may have entered in pursuit of fishes.
The Black kite (Milvus migrans) was frequently seen patrolling the lake, and once was
seen to swoop and remove a fish from the water. This was probably a dead fish thrown
out from the nets by the Barombis. The Palm-nut vulture (Gj’pohierax angolensis) was
also frequently seen over the lake, but we did not see it take any fishes, although Brown
&
Amadon
(1968)
state that it will sometimes snatch live fishes from the water surface.
We did not see any otters in the lake, but an Otter shrew (Potamogale velox) was caught
in one of the Barombi gill nets. Its stomach was full of broken-up crabs.
Probably the most serious predator
on
the fishes in the lake is man. The Barombi
tribesmen have the sole fishing rights over the lake, and rely for their livelihood on the
fishes. Each day they catch about
300
to
500
large unga, which with smaller numbers of
Clarias,
Barbus and mos (the prawn Macrobrachium vollenhovenii) are sold to “middle
men” who carry them to Kumba market. The smaller fishes, such as
P.
maclareni,
S.
pindu
and
S.
lohbergeri are kept and eaten by the villagers. It is difficult to estimate the numbers
removed for this purpose, but
if
each person in the village eats five small fishes a day the
total number eaten per day may be as high as 1000. The fishing methods of the Barombis
on this lake effectively exploit fishes of particular sizes in particular habitats, but the
smallest (and largest?) fishes are not caught, and the deeper waters are hardly fished at
all.
There are on the lake about five large gill nets, each about
100
m long and
2
m deep,
with a stretched mesh of about
5
cm. These nets are set at the surface some distance from
the shore in water about
30-60
m deep. They are left out all the time (except when they

ECOLOGICAL STUDIES ON CRATER LAKES IN WEST CAMEROON
85
are taken up to be washed every four weeks or
so)
and are examined every morning by
the fishermen. These nets catch unga and, in the lower part of the net,
K.
dikume.
The mesh
is too large to catch
Myaka myaka,
the other pelagic species.
A
few shorter gill nets of smaller mesh are set close to the shore, tied to overhanging
branches, in water 0.5-2 m deep.
Most of the smaller fishes and some of the marketable unga are caught in basket traps
woven from strips of the rhachis of the leaves of the raffia palm (Fig. 29). Most of these
traps are about
1
m long and 35 cm in diameter, with meshes about
1
cm wide. Smaller
traps with finer meshes are sometimes made for special purposes. The large traps are very
common all around the edge of the lake. They are usually set on the bottom in shallow
water
so
that the top of the trap just breaks the surface of the water. Sometimes a V-shaped
fence is constructed to guide the fishes into the trap. These traps are set without bait, and
are left in position for many days at a time. Each fisherman has trapping rights on a
particular stretch of the shore and examines
his
traps daily. The traps catch large unga,
which have apparently come inshore to breed, and large numbers of the smaller fishes:
S.
lohbergeri,
P.
maclareni,
K.
eisentrauti,
S.
pindu,
S.
mariae,
young
Clarias
and the larger
specimens of
E.
sexfasciatus.
Sometimes traps of this kind are set deep on
a
cane rope at
depths of 2-3 m or more. It was in deep traps that
S.
mongo,
young
K.
dikume
and sexually
mature
Myaka myaka
were caught.
Other methods of fishing are used near the Barombi village in the lake and in the tri-
butary stream. Boys fish with hook and line for
Clarias
and
Barbus,
and baited hooks are
sometimes set in the lake, tied
to
overhanging vegetation. Women use hand nets of about
2
cm mesh netting
on
an oval wood frame to scoop for
Clarias
in
the muddy pools
of
the
stream. Closely woven circular baskets are used by women and children to dip
P.
similis
and
E.
sexfasciatus
from the village stream.
Relationships
of
the fishes
of
Barombi
Mbo
ETHELWYNN
TREWAVAS
Fish fauna
of the
surrounding
streams
The streams in the neighbourhood of Barombi Mbo are tributaries of the Mungo River,
with the exception of the River Bille, an upper tributary of the Meme.
A
score
of
species
were caught in these streams by Dr Disney and his staff, using narrow-mesh basket traps
and hand nets.
None of the cichlid species of the lake was found either in the feeder stream or in any
of
the streams of the district, with the exception of two juveniles of
S.
Zohbergeri,
caught in
the part of the Kumba below the town. We understand that there is no barrier to upstream
migration of cichlids between the Kake-Kumba confluence and the place where these were
caught and it is probable that their parents were washed down from the lake into the
Kake and
so
entered the Kumba.
The only cichlids caught in the streams or the main stream of the Mungo were
Hemi-
chromis
fasciatus, Chromidotilapia
sp.,
C.
guntheri
and
Tilapia camerunensis,
none
of
which is even congeneric with the cichlids of the lake.
Of the six species, all non-cichlids, caught in the feeder stream of Barombi Mbo,
five

86
E.
TREWAVAS,
J.
GREEN AND
S.
A.
CORBET
were also present
in
other streams, but only four in the lake.
Of
the two species of
Proca-
topus
caught in the district the one,
P.
siinilis,
living in the feeder stream was found else-
where only in the River Kumba of the Mungo tributaries and in the River Bille of the
Meme system. These are the two streams nearest to the lake, respectively on its east and
west sides, but not communicating with it directly. Of the two species of
Barbus
with
parallel-striate scales, only one,
B. batesii,
was caught in the River Kumba, and this is the
one present in Barombi Mbo. The only lake-endemic in Barombi Mbo’s feeder stream was
Clarias maclareni
(young).
Clarias walkeri
was caught outside the lake basin in the Rivers
Kake, Kumba and Wowe and in the Kotto area, where we found no
C.
maclareni.
The
Clarias
from Lisoni described by me in 1962 proved to be the representative
of
a form
closely related to
C. maclareni,
but not identical (Griffith, Green
&
Corbet, in prep.).
In conclusion, we have from the river faunas confirmation of the endemism of the
11
species of cichlids of Barombi Mbo, and no contradiction of the supposed endemism of
C.
maclareni,
although its relationship with the form of Lake Soden is confirmed. The
cichlid fauna of the upper Mungo and Meme systems provides no species that could be
considered a source-species for the Mbo endemics. We must look elsewhere for the rela-
tionships of these.
Cichlids
of
lower
zones
of
rivers
of
Cameroon
The volcanoes that form the characteristic landscape of West Cameroon are believed to
be late Quaternary in date and Mount Cameroon
is
still active. Their conical shape and
little-weathered craters support this view (Gkze, 1943) and the freshwater fish-fauna of
Fernando
Po0
(Thys, 1967) testifies to the recent unity
of
this volcanic island with the
area. The Mungo and Meme are now relatively short rivers. Probably before the volcanic
activity rejuvenated their tributaries the cichlids of the lower zone lived farther inland as
they do in the Sanaga, and one or more of them might have got into the crater lake, per-
haps before all its clan had migrated to the more suitable lowland waters.
We must therefore take into account all the cichlids known from the Mungo and Meme
systems including lakes Kotto and Mboandong.
Two
genera can be excluded as a source
for the Mbo cichlids,
Hemichromis
because
of
the participation of the basioccipital in the
apophysis for the upper pharyngeals and the structure of its scales,
Chromidotilapia
because of the “hanging pad” in the pharynx and associated feeding mechanism.
Tilupiu
(Coptodon) kottae
Lonnberg,
T.
(Pelmatolapia) mariae
Boulenger and
Sarotherodon
galilaeus
(Linn.) inhabit the lakes assocated with the lower Mungo and Meme. In the
brackish water of rivers to the west and south, and probably also of the Mungo and
Meme, is
Sarotherodon melanotheron
Riippell or its subspecies
S.
m.
nigripinnis
(Guichenot in Dumkril, 1859). Inland, in Rivers Nyong, Ja and Ogowe is the related
S.
mvogoi
(Thys).
The extreme trophic adaptions of the Barombi Mbo cichlids must be ignored in looking
for relationships, since we have looked in vain for equivalents that are not obvious cases
of convergence or at most parallelism.
Groupings among the lake species that suggest multiple origin are
:
(1)
Sarotherodon steinbachi.
(2)
S.
linnellii,
S.
caroli,
S.
lohbergeri
and
Myaka myaka.
(3)
Konia eisentrauti,
giving rise to
K.
dikume
and
Stomatepia
(three species).
(4)
Puiigu maclareni.

ECOLOGICAL STUDIES ON CRATER LAKES
IN
WEST CAMEROON
As a source group for the first three the genus
Surotherodon
has the requirements of
a
mesethmoid free from the vomer (see Trewavas, 1972), a long buccopharynx, 14-24 gill-
rakers on the lower part of the first arch and mouth-brooding still in the “experimental”
or flexible stage (Kraft
&
Peters, 1963; Fishelson, 1966; Heinrich, 1967).
(1)
S.
steinbuchi,
as suggested in 1962, is probably related to
S.
gulilueus.
The evidence
is, (a) the range of vertebral numbers is as in
S.
gulilueus
and higher than in the other
Barombi Mbo species (see Table 11), (b) the head is shorter, 33.0-36.5
%
SL,
cf. 35.5-48.0
in group 2 and 35.0-40.0 in group 3 and (c) the mouth is smaller; its ratio (lower jaw
26-31
%
length of head) although nearly the same as in
S.
lohbergeri
is measured against
a smaller head. Seventeen dorsal spines, its modal number, were not found in any other
Mbo species except in one specimen of
S.
lohbergeri,
but it is rare too in
S.
gulilueus.
It
differs from
S.
gulilueus
in
its
small size as well as in its specialized pharyngeal dentition,
which could be derived from that of
S.
gulilueus
or from
S.
lohbergeri.
(2)
S.
linnellii
and
S.
curoli
show in their dentition of jaws and pharynx and in the long
snout and preorbital bone resemblance to
S.
melanotheron,
S.
mvogoi
and
S.
schwebischi,
but the massive head is not paralleled by any of these.
S.
linnellii
(and perhaps the others)
differs from
S.
melanotheron
and resembles
S.
schwebischi
in being a maternal mouth-
brooder.
S.
lohbergeri
is included with these because of its long head, but the length is not more
than it may be in
S.
melanotheron.
Alternatively it may be related to the third group, as
suggested by the similarity of its colour-pattern to that of
K.
eisentrauti.
Myuku
myaku
is one of the strangest of the cichlids in the lake. It is the smallest, and
has a much less robust appearance than any other. Its long head and big mouth give it the
appearance of a miniature
S.
linnellii,
but its interorbital width is usually less than even in
a small
S.
linnellii.
It has reduced both the size and number of the teeth
of
jaws and
pharynx; its scales are small, thin and irregular in arrangement and numbers. Its ovaries
are very small. In spite
of
manifest differences
I
suggest that it is most nearly related to
S.
linnellii
and
S.
curoli,
mainly because the long narrow pharyngeal bone is derivable
from that of
S.
curoli.
If
so,
the shortness
of
the intestine is probably secondary, another
reduction character.
(3) All the species in the third group have a rather narrow interorbital roof of the skull
and a relatively short intestine, The interorbital roof can be narrow in
Tilupiu
but not
in
Surotherodon.
The shortness
of
the intestine may be secondary, but this and other features
(relatively short pharyngeal blade, diversity of trophic adaptations) suggest that these may
have been derived from a
proto-Sarotherodon,
a tilapiine group with bi- and tri-cuspid
teeth, a short epibranchial series
of
gill-rakers and a rather long ceratobranchial series,
still feeding largely on invertebrates, having taken steps towards mouth-brooding.
a7
All further specializations have evolved in the lake.
(4)
Thys (1971
:
167) has suggested that
Pungu
is related to
Tilupiu kottue
and
T.
guineen-
sis,
but has not yet given his reasons.
In
favour of this is the fact that its ethmoid cartilage
is very well ossified, the mesethmoid as well as the lateral ethmoid meeting the dorso-
lateral wing
of
the vomer on each side in
a
brief suture.
In
this it resembles
T. zillii,
T.
kottue, T. camerunensis
and some individuals of
T.
guineensis
as well as
T. muriue
and
differs from all the other species of Barombi Mbo and all species of
Surotherodon
that
I
have examined. It differs from all species of
Tilupia,
including
T.
mariue,
in the large eggs,
probably mouth-brooded, and from all except
T.
muriae
and
T. brevimunus
in the higher

88
E.
TREWAVAS,
J.
GREEN AND
S.
A. CORBET
numbers of gill-rakers. Further, the species of
Tilapia
have typically a deep cheek with
three
or
four horizontal rows of scales on it (although
T.
kottae
may have only two),
whereas
Pungu
has two oblique rows as in
Sarotherodon
and the other Mbo cichlids. Also,
Pungu
has 29 vertebrae and the modal number in
T.
mariae
and most species of
“Coptodon”
is
28
(of this group only
T.
rendalli
has a mode of
29).
Its relationship presents a baffling problem, since its extreme specialization carries it far
from any possible ancestor. Little is known of the variation in ethmoid ossification, the
study of which involves
a
disfiguring of museum specimens justifiable only to a limited
extent. Regan (1920) distinguished the subgenus
Coptodon
from subgenus
Tilapia
on this
character, but the type species of
Tilapia,
T.
sparronanii,
also has an ethmovomerine union.
Preliminary studies suggest that it has a certain taxonomic value and that
T.
rendalli
is
an
exceptional
Tilapia,
in this respect paralleling
Sarotherodon.
In
Pungu
does the well-ossified ethmoid indicate a close relationship with
Tilqia,
perhaps
T.
mariae,
or is it an atavistic convergence, part of the general strengthening of
the snout
?
Finally, we cannot find outside the lake more than three possible source-species for the
Mbo cichlids and we can postulate a fourth (the
“proto-Sarotherodon”).
If this fourth
really existed it could have given rise to groups
(1)
and
(2)
as well as to the third. The
separate origin of
Pungu
is still open to question.
Eye
lens
proteins
Acting
on
the advice of my colleague
Dr
C.
A. Wright,
I
collected lenses from the eyes
of a number of cichlids in Cameroon and Ghana. They were dissected from the fresh fish
as soon as possible after death and kept chilled until they were delivered at the Museum.
For this possibility
I
have to thank the members of the Helminthiasis Research Unit,
Professor
D.
W. Ewer of Ghana University and the proprietor
of
the village store at
Kuntanasi, near Bosumtwi, for the use of refrigerators. The lenses were transported in a
chilled thermos flask.
At the British Museum (Natural History) Mr Ross processed them to produce electro-
phoretic protein patterns. Without knowing either the geographical or zoological relation-
ships of the species Mr Ross arranged the patterns in three groups. The first comprised
eight species of Barombi Mbo, the only ones from that lake that gave successful results.
The division between the other two groups seems to me to be more arbitrary. Together
they include
Chromidotilapia loennbergi
(Lake Kotto),
Tilapia guineensis
(near Accra),
T.
discolor
(Lake Bosumtwi),
T.
muriae(Lake
Kotto),
Sarotheron melunotheron
(two lagoons
in Ghana),
S.
multifasciatus
(Bosumtwi) and
S.
galilueus
(Kotto and Nungua reservoir
near Accra).
The Barombi Mbo pattern is of nine evenly spaced bands, becoming gradually fainter
from the cathode end towards the anode, giving very similar patterns for
K.
eisentrauti,
K.
dikume, St. pindu
and
Pungu.
That of
S.
lohbergeri
is
similar but fainter. In
S.
steinbachi,
Myaka myaka
and
S.
linnellii
the band immediately at the cathode side of the starting line
is stronger than the other three cathode bars and there
is
a
blank at the anode side until
three or four fairly well defined bars show up at the anode end. These three species, and
especially
S.
linnellii,
are the only ones that show some resemblance to any of the outside
patterns, namely to that of
S.
galilaeus,
in which, however, the densest cathode bar is farther

ECOLOGICAL STUDIES
ON
CRATER LAKES
IN
WEST
CAMEROON
89
from the starting line. In all the other external cichlids tested there are two strong bands
on the cathode side.
Unfortunately we do not know how to interpret these results. Have they anthing to do
with relationship or does the chemistry of the environment influence the lens proteins?
For relationship we have the close similarity of the patterns of
T.
guineensis
and
T.
discolor,
whose macrostructure also bespeaks close relationship, but whose environments, respec-
tively a brackish lagoon and a soda lake, are dissimilar. But the patterns of the unrelated
S.
melanotheron
and
C.
loennbergi
are also very similar, to each other and to
T.
guineensis,
the one from the same lagoon as
T.
guineensis,
the other from Kotto.
The only legitimate conclusion is that there is something that causes the eye lens proteins
of the cichlids of Barombi Mbo to be alike and to differ from those of other cichlids in
which they are known, but whether it is genetic similarity or an environmental factor we
do not know.
Photographs of the electrophoretic patterns and Mr
Ross’s
account of his methods are
available for reference in the Experimental Taxonomy Section of the British Museum
(Natural History).
Evolution
of
the
cichlids
in
Barombi
Mbo
ETHELWYNN
TREWAVAS,
J.
GREEN
AND
SARAH
A.
CORBET
The
11
species of cichlids in Barombi Mbo thus appear to be endemic and without very
close relatives elsewhere. Their evolution within the lake probably began with the initial
invasion of the lake by ancestral cichlids followed
by
their subsequent adaptive radiation
which involved, firstly, the initiation of reproductive isolation between units of the
ancestral populations, and, secondly, the progressive divergence of the newly-formed
species that resulted in their avoidance of competition.
The initial invasions of Barombi Mbo may have been by two to four species (see p.
86).
This isolated lake, a crater situated on top of a high volcanic hill, must have been difficult
of access for fishes throughout its history, and the first to colonize it will have found them-
selves in an environment which, though presumably different in major respects from their
original home, offered food and space enough to permit
a
rapid increase in population.
It is evident from work on man-made lakes that in such circumstances fish species may
change their habits relatively quickly. Petr (1967) has recorded that the schilbeids of the
River Volta,
Schilbe mystus
and
Eutropius niloticus,
which in the river were generalized
feeders on insects and small invertebrates, have in the Volta Lake become specialists on
Povillu
larvae and are never to be found far from the clumps of rotting wood in which
these insects burrow. Loiselle (in press) describes the ready adaptation of
Leptotilapiu
irvinei,
previously known only from the Volta rapids, to the rocky places in the new lake.
Both these changes in habit are such as might eventually result in changed structure.
If
the first-colonizing cichlids underwent changes of this kind on entering the lake it is
scarcely surprising that we can now see no close similarity to the species elsewhere
(if
they
still exist) that provided the pioneers.
Changes
of
this kind would affect the whole species in its new environment; to explain
the early stages of speciation there we must envisage some physical or biological barrier

90
E. TREWAVAS,
J.
GREEN AND
S.
A. CORBET
within the lake capable of producing genetic isolation between the populations that are
incipient species. Fryer’s (1959) view that incipient species of cichlids in Lake Malawi
were kept apart by alternating rocky headlands and sandy bays is not applicable here.
None of the species in Barombi
Mbo
seems to be confined to either rocks or sand, and
there are no major discontinuities in the shore-line. Nor can we invoke the separation
of
basins within the lake by a former low water level, as Greenwood (1951) proposed for
Lake Victoria, because the crater of Barombi Mbo is evenly conical. Currents in the lake
are gentle, but it is possible that there may at some time have been currents near the out-
flow strong enough to prevent the passage of fishes from one side to the other, but only at
that point.
Since no major physical barrier exists to account for the initial separation of the popu-
lations that became species, the possibility of speciation in the absence of such
a
barrier
deserves very serious consideration.
The divergence of the newly-formed species
is
paralleled in other organisms: Ayala
(1 970) has given examples of experiments demonstrating selection for avoidance of compe-
tition in sibling species of
Drosophila
resulting in the establishment of greater differences
between populations
of
such species living together than between the same two living
apart, and Brown
&
Wilson (1956) have given instances of this latter phenomenon in wild
populations. There is some evidence for selection for the avoidance of competition in
Barombi Mbo: our study showed that most of the cichlids are clearly separated from one
another in their distribution through space or time, or in their feeding habits.
There is a clear spatial separation between those species that live and feed inshore and
those that feed in the deeper water of the open lake, coming inshore only to breed. Within
this group of inshore species there is some overlap in diet, but each species selects a main
course different from the others.
The main fish foods in the lake are phytoplankton, organic debris, aquatic invertebrates,
fishes and adult insects (Fig.
28).
The two inshore species of
Sarotherodon
both feed to a
large extent on organic debris, but
S.
Zohbergeri
collects it, together with aufwuchs, from
the surfaces of rocks and plants, while
S.
steinbuchi
collects it mainly from sandy areas.
Pungu maclareni
is
the only species in the lake (and perhaps the only freshwater fish) that
feeds largely on chunks of living sponge. Its prominent teeth and thick lips may be adapta-
tions to this unusual diet. Insect larvae support a group of predatory inshore fishes which
differ from one another in the depths or times at which they feed, the size of their prey, and
the relative importance of adult insects in their diets. Young individuals of unga
(Saro-
therodon ZinneZlii
and/or
S.
curoli)
often feed
very
close inshore in water too shallow for
other fishes, taking flotsam as
well
as bottom aufwuchs and insect larvae.
Konia eisentrauti
feeds in water about
1-2
m deep, snapping up small invertebrates and sometimes taking
the eggs of other fishes.
Stomutepia mariae
takes larger prey than
K.
eisentrauti,
including
small fishes, and frequently takes adult insects from the surface of the water.
Stomatepiu
pindu
hunts
in
a similar way to
K.
eisentrauti,
but seems to feed mainly in twilight or at
night and
so
probably encounters a different range of prey. The cyprinodont
Epiplatys
sexfasciutus,
feeding at the surface close inshore, makes use of
a
food source scarcely
tapped by the cichlids.
The cichlids in deeper water are silver grey and without markings when alive in contrast
to the inshore species which are blotched with black. The significance of such colouration
in the camouflage of open-water fishes has been discussed by Denton
&
Nicol(1965). Those

ECOLOGICAL STUDIES
ON
CRATER LAKES
IN
WEST CAMEROON
91
~-
I
Epiplatys
I
FIG.
28.
A
food web for the fishes of Barombi Mbo based on the examination of stomach contents. Wide
arrows: more than
30%
of fishes have the food item as main contents; narrow arrows: fewer than
30%
have.
individuals found in breeding condition inshore
(of
Surotherodon linnellii,
S.
curoli
and
Myuka myuka)
showed darker markings resembling those of inshore species. Phytoplank-
ton forms the sole food of adult
S.
linnellii
and
S.
caroli,
both of which swim a few feet
below the surface in mid-lake.
Myuku myaka
probably feeds on dead phytoplankton and
other organic debris close to the bottom.
No
fishes seem to prey directly on the copepods
and rotifers in the open water, but these must support the
Chuoborus
larvae which con-
stitute the chief food for
Koniu dikume.
From this survey it
is
evident that most species of cichlids in the lake differ
in
their diets
so
much that no two species are likely to come into direct competition for a scarce item
of food. The sibling species
Surotherodon linnellii
and
S.
caroli
provide an interesting
exception. They seem to swim at similar depths and they share a diet of phytoplankton
;
so
that if the resource were scarce one might expect severe competition.
But
Lowe-
McConnell(l969), who found three phytoplankton-feeding species of Tilupiu coexisting in
Lake Malawi, has suggested that for such fishes food may not be a limiting resource. More
important for cichlids may be the limited space for breeding sites around the shore. If
so,
competition between these sibling species would be reduced by
a
difference in breeding
seasons. We have some evidence that
S.
linnellii
and
S.
caroli
do breed at different seasons
We suggest, then, that the stages in the evolution of the cichlids in Barombi Mbo were
these. The lake was colonized by two to four cichlid species,
(a)
an
S.gulilueus-like
form,
(P.
58).

92
E.
TREWAVAS,
J.
GREEN AND
S.
A. CORBET
(b) a form ancestral also to
S.
m.
nigripinnis and
S.
mvogoi, (c) a Konia eisentrauti-like
form and (d) the ancestor of
Pungu,
feeding respectively on (a) material sifted from fine
sand, (b) phytoplankton, detritus and aufwuchs, and (c and d?) invertebrates. The sand-
sifter, (a) became more exclusively adapted to that habit; the second form, (b), became
divided
in
response to increased opportunity and competition (especially for limited
aufwuchs) into a specialized aufwuchs-feeder
(S.
lohbergeri) and two phytoplankton-
feeders; and the invertebrate-feeders, (c), diverged to produce
Koniu
with its diet
of
bottom
invertebrates, and the species of Stonqatepia which avoided competition by feeding at
night
(S.
pindu) or in deeper water
(S.
mongo),
or by including in their diet waterlogged
insects from the surface and even small fishes
(S.
mariae). Driven from the crowded inshore
rocks to the deeper water near the middle of the lake,
K.
dikuine exploited the dense
populations of Chaoborus, and Myaka myuka (from group b?) fed on detritus near the
bottom. The question whether
Pungu
also diverged from group (c) or was highly modified
from
a
separate ancestor has been discussed on p.
87-88.
FIG.
29.
A
basket
trap,
110
cm
long,
from Barombi
Mbo.
This impressive ecological separation has clear adaptive advantages for
a
dense popula-
tion of fishes exploiting the meagre resources of an oligotrophic lake; but the nature
of
the
factors that initiated speciation within the lake, in the apparent absence of geographical
barriers, remains open to speculation.
Comparison
with
other African lakes
The situation in Barombi Mbo has a double interest from its resemblance
in
miniature
to
that in the great Rift Valley lakes, particularly Tanganyika. The scale, both temporal
and spatial, is immensely greater in Tanganyika, but this lake has the Malagarazi entering
it on the east and an outlet to the Congo on the west, the faunas of the two rivers being
related although they have been separated long enough for a degree of divergence to take
place. Across these once continuous rivers lies Lake Tanganyika, with a rich fish-fauna
consisting mainly of endemic species, especially cichlids, for most of which no obvious
source species still exists outside. It has evidently been derived from
a
very old fauna (old
in
terms of cichlid time-scale), which may have inhabited the huge shallow lake that is
believed to have occupied the central Congo basin. What has happed once in Barombi
Mbo has happened again and again in the Great Lakes, giving a multi-tier radiation with
convergent adaptations. Fryer has expressed this idea in a diagram
(1959:
446).
Even in
Barombi Mbo we may be seeing the beginning of a second tier in the three species of
Stomatepia and in Sarotherodon
linnellii
and
S.
caroli.

ECOLOGICAL STUDIES ON CRATER LAKES IN WEST CAMEROON 93
Our visit to Cameroon was financed by grants from the Royal Society, the Central Research
Fund
of
the University of London and the Godman Exploration Fund. We are particularly
indebted to Dr Brian Duke, Mr Peter Moore and Dr Henry Disney, of the Helminthiasis Research
Unit at Kumba, for their hospitality and for helping us with accommodation and transport, and
in many other ways. Dr Disney’s enthusiastic cooperation was especially valuable and much
appreciated. We are grateful to our colleague Dr John Griffith who, while holding a Churchill
Fellowship, gave us valuable collaboration in the field.
Introduction by the Helminthiasis Research Unit ensured us a friendly reception by the
Barombi people at the village of Barombi Mbo, who managed to combine a very helpful attitude
with a firm reminder that the lake was their property. We welcomed this as a good augury for
the conservation of this unique ecosystem. In particular, we thank Joseph Ndokpe Sangwa, who
shared his knowledge of the local fishes and told us how to spell their Barombi names. We pay
tribute to the memory of the chief, Martin Malleh, who died in
1971.
One of us
(E.T.)
is grateful for the facilities provided by the British Museum (Natural History),
where Dr C.
A.
Wright and Mr G. C.
Ross
carried out the electrophoresis of the eye-lens proteins,
Mr R. H. Harris and Mr
C.
G.
Ogden assisted in the preparation and examination of fish eggs
with the scanning electron microscope and Mr G. Howes in radiography for vertebral counts.
Figures
4,
12,
16,
and
22-24
were drawn by Sharon Chambers and Figs
6
and
14
and Plate I by
Denys Ovenden with their well-known skill. We are grateful
to
Messrs George Rainbird Ltd and
the International Union for the Conservation of Nature and Natural Resources for permission to
use Denys Ovenden’s colour picture
of
Pungu
from
“The Red
Book-
Wild
Life
in Danger”.
The
painting used is a slightly modified copy by Mr Ovenden of that picture, which was based on
a
colour photograph by Dr Thys van den Audenaerde.
Our thanks are due to Dr
P.
Brien, Institut Zoologique, Universitk Libre de Bruxelles, for
identification of sponges, and to the Director
of
Kew Gardens for arranging for the identification
of the aquatic plants.
We are indebted to Mrs Margaret Clarke for her unfailing patience in typing the manuscript.
REFERENCES
Ahl,
E.
(1927). Neue afrikanische Fische der Familien Anabantidae und Cypronidontidae.
Sber. Ges. naturf.
Ayala,
F.
J.
(1970). Competition, coexistence and evolution. In
Essays
in
evolution andgenetics
in
honor
of
Theodosius
Boulenger,
G.
A. (1903). On the fishes collected by Mr.
G.
L. Bates in Southern Cameroon.
Proc.
zool.
Soc.
Lond.
Boulenger,
G.
A. (1908). A revision of the African silurid fishes of the subfamily Clariinae.
Proc.
zool.
Soc.
Lond.
Boulenger, G. A. (191 1-1916).
Catalogue of the fresh-water fishes
of
Africa
in
the British Museum (Nut. Hist.)
Bretschneider, L. H.
&
Duyvene de Wit, J.
J.
(1947).
Sexual endocrinology
of
non-mamnmlinn vertebrates.
New
Brown, L.
&
Amadon, D. (1968).
Eagles, hawks and falcons
of
the world.
London: Country Life Books.
Brown, W. L.
&
Wilson, E.
0.
(1956). Character displacement.
Syst.
Zool.
5:
49-64.
Daget,
J.
(1954). Note sur le rtgime alimentaire du cCryle pie.
Notes afr.
62:
55-57.
Daget,
J.
&
Iltis, A. (1965). Poissons de Cbte d‘Ivoire (eaux douces et saumatres).
Meni. Inst. fr. Afr. noire
No. 74:
Darnell,
R.
M. (1961). Trophic spectrum
of
an estuarine community based on studies of Lake Pontchartrain,
Denton, E.
J.
&
Nicol,
J.
A.
C. (1965). Studies on reflexion of light from silvery surfaces of fishes, with special
Freunde Bed.
1927: 76-81.
Dobzhansky:
121-158. Amsterdam: North Holland Publishing Co.
1903
(1):
21-29.
1907: 1062-1097.
1
and 2, 1911;
3,
1915;
4,
1916. London: Trustees of the British Museum.
Y
ork
:
Elsevier.
1-385.
Louisiana.
Ecology
42:
553-568.
reference to the bleak,
Alburnus alburnus.
J.
mar.
biol.
Ass.
U.K.
45:
683-703.

94
Dumeril,
A.
(1859). Reptiles et poissons de 1’Afrique occidentale.
Archs
MIIS.
not.
Hist. Paris
10:
137-268.
Fishelson, L. (1966). Untersuchungen zur vergleichenden Entwicklungsgeschichte der Gattung
Tilapia
(Cichlidae,
Fisher,
J.,
Simon, N.
&
Vincent, J. (1969).
The Red Book: Wildlife
in
Danger.
London: Collins.
Foersch, W. (1971).
Aphyosemion snntoisnbellae
Schcel 1968.
Aquar.-u. Terrar.
Z.
24:
20-23.
Fryer,
G.
(1959). Some aspects of evolution in Lake Nyasa.
Evolution, Lancaster, Pa.
13:
440-451.
Gtze, B. (1943). Gtographie physique et gCologie du Cameroun occidentale.
Mem.
Mus.
Paris
17:
1-272.
Gill, T. (1862). On the West African genus
Hemichromis
and description of new species in the museums of the
Green,
J.
(1965). Zooplankton of Lakes Mutanda, Bunyoni and Mulehe.
Proc. zool.
Soc.
Lond.
144:
383-402.
Green, J. (1972). Ecological stuides on crater lakes in West Cameroon. The zooplankton of Barombi Mbo, Mboan-
Greenwood,
P.
H. (1951). Evolution of the African cichlid fishes; the
Huplorhroniis
species-flock in Lake Victoria.
Greenwood, P. H. (1953). Feeding mechanism of the cichlid fish
Tilapia
escirlenta
Graham.
Nalure, Lond.
172:
207.
Greenwood,
P.
H. (1961).
A
revision of the genus
Dinotopterus
Blgr. (Pisces, Clariidae) with notes on the compara-
tive anatomy of the suprabranchial organs in the Clariidae.
Bull.
Er.
Mus.
(Not.
His?.)
7:
215-241.
Griffith, J., Green,
J.
&
Corbet,
S.
A.
(in preparation).
Ecological studies
on
crater lakes
in
West
Cameroon.
Fishes
of
Lake Soden.
Gunther,
A.
(1866).
Catalogue
offishes
it7 ?he collec?ion of
the
Briti.sh Mu.reiin7
6.
London: Trustees of the British
Museum.
Gunther, A. (1896). Report on a collection of reptiles and fishes made by Miss M. H. Kingsley during her travels
on
the Ogowe River and in Old Calabar.
Ann. Mag. not. Hiit.
(6)
17:
261-285.
Hassert,
K.
(1912). Seenstudien in Nord-Kamerun.
Z.
Ges. Erdk.
Bed.
1912:
7-41, 135-144, 203-216.
Heinrich, W.
(I
967). Untersuchungen zum Sexualverhalten in der Gattung
Tilapia
(Cichlidae, Teleostei) und bei
Hoar, W.
S.
(1957). Gonads and reproduction. In
Thephysiology
offishes,
1:
287-321. Brown, M. E. (Ed.) New
Holly,
M.
(1927). Siluriden, Cyprinodontiden und Cypriniden aus Kamerun.
Sher.
Akad. Wiss. Wien
136:
195-203.
Holly,
M.
(1930). Synopsis der Susswasserfische Kameruns.
Sber. Akad.
Wiss.
Wien
139:
195-281.
Kraft, A. von
&
Peters, H.
M.
(1963). Vergleichende Studien uber die Oogenese in der Gattung
Tilapia
(Cichlidae,
Lioselle, P. V. (in press).
Thefishes
of
Volta Lake.
Lonnberg,
E.
(1895). Notes on fishes collected
in
the Cameroons by Mr
Y.
Sjostedt.
0fver.s.
K.
VetenskAkad.
Lonnberg,
E.
(1903). On
a
collection of fishes from the Cameroons containing new species.
Ann. Mag. nat. His/.
Lonnberg,
E.
(1904). On some fishes from the lakes of the Cameroon Mountain.
Ann.
Mag.
not.
Hist.
(7)
13:
Lowe-McConnell, R.
H.
(1969). Speciation
in
tropical fresh-water fishes.
Eiol.
J.
Linrr.
SOC.
1:
51-75.
Mackworth-Praed, C. W.
&
Grant, C.
H.
B.
(1970).
Birds
of
West Cen?ra/and Western Africa.
London: Longman.
Meinken,
H.
(1929). uber eine neue Panchax-Art
(Pancha.u vexillifer
nsp.). BI.
Aquar. Terrerienk.
40:
255-257.
Petr, T. (1967). Food-preferences of the commercial fishes of the Volta Lake.
Rep. Tech.
Ghana
Volta Basin Res.
Regan,
C.
T. (1920). The classification of the fishes of the family Cichlidae.
1.
The Tanganiyika genera.
Ann. Mug.
Ruppell,
E.
(1
852).
Verzeichnis der in detn Museum der Senclietzbergischerz naturforschenden Gesellschaft Samnz-
Scheel,
J.
(1968~).
Rivulins ofthe Old Workd.
Jersey City, N.J.
:
T.F.H. Publications Inc.
Scheel,
J.
(19686). A new species of
Aphyosemion
(Cyprinodontidae, Pisces) from Fernando
Poo.
Revue Zool.
Bor.
Schmidt,
H.
(1928). uber zwei neuimportierte Fische.
Aquarium, Berl.
1928:
165-166.
Thys van den Audenaerde,
D.
F.
E.
(1967). The freshwater fishes of Fernando
Poo.
Verk.
K.
Vlaam.
Akad. Wet.
29
E.
TREWAVAS, J. GREEN AND
S.
A. CORBET
Teleostei).
2001.
J6. (Anat.)
83:
571-656.
Academy and Smithsonian Institution.
Proc. Acad.
nut.
Sci.
Philad.
1862:
136-142.
dong, Lake Kotto and Lake Soden.
J.
Zoo/.,
Lond.
166:
283-301.
Nature, Lond.
167:
19-20.
Artbastarden.
Z. Tierpsychol.
24:
684-754.
York: Academic Press.
Teleostei).
Z. Zellforsch. mikrosk. Anat.
61
:
434-485.
Forh.
52:
179-195.
(7)
12:
37-46.
135-1 39.
Proj.
No.
10:
1-22.
nat. Hist.
(9)
5:
33-53.
lungen,
pt. 4: 1-40. Frankfurt-am-Main.
afr.
78:
332-342.
NO. 100: 1-167.

ECOLOGICAL STUDIES ON CRATER LAKES IN WEST CAMEROON
95
Thys van den Audenaerde, D.
F.
E. (1968). An annotated bibliography
of
Tilapia
(Pisces, Cichlidae).
Dociimn
zool.
Thys van den Audenaerde, D.
F.
E. (1971). Some new data concerning the Tilapia-species of the subgenus
Coptodon
Trewavas, E. (1962). Fishes of the crater-lakes
of
the Northwestern Cameroons.
Bonn.
zool.
Beitr.
13: 146-192.
Trewavas, E. (1969). Freshwater fishes
of
Africa. In
The Red Book: Wildlife in Danger:
344-350.
Fisher, J., Simon,
N.
&
Vincent
J.
(Eds.) London: Collins.
Trewavas,
E.
(1972). On the cichlid fishes
of
the genus
Pelmatochromis
with proposal of a new genus for
E.
congicus;
on the relationship between
Pelmatochromis
and
Tilapia
and the recognition
of
Sarotherodon
as
a
distinct genus.
Bull. Br.
Mus.
(Zool.)
Mus.
r. Afr. centr.
no.
14:
1-406.
(Pisces, Cichlidae).
Revue
2001.
Bot.
afr.
84:
160-168.
Trewavas, E. (in preparation).
Taxonomic notes
on
the fishes
of
West Cameroon.
Trewavas,
E.
&
Irvine,
F.
R. (1947). Freshwater fishes.
In
The fishes and fisheries
of
the Gold Coast:
221-282.
Irvine,
F.
R. (Ed). London: Crown Agents
for
the Colonies.