Content uploaded by M. Clara P. Amorim
Author content
All content in this area was uploaded by M. Clara P. Amorim on Mar 16, 2018
Content may be subject to copyright.
Differences in sounds made by courting males of three
closely related Lake Malawi cichlid species
M. C. P. AMORIM*†,M.E.KNIGHT‡, Y. STRATOUDAKIS§
AND G. F. TURNER{
*Unidade de Investigac¸a
˜o em Eco-Etologia, ISPA, Rua Jardim do Tabaco 34,
1149-041 Lisboa, Portugal, ‡School of Biological Sciences, University of
Southampton, SO16 7PX, U.K., §Instituto de Investigac¸a
˜o das Pescas
e do Mar (INIAP/IPIMAR), 1449-006 Lisboa, Portugal and
{Department of Biological Sciences, University of Hull,
HU6 7RX, U.K.
(Received 10 November 2003, Accepted 29 July 2004)
Courtship sounds made by three sympatric cichlid species, Pseudotropheus zebra,P. callainos
and an undescribed species known as P. ‘zebra gold’ were recorded and compared to investigate
the potential role of acoustic signals in mate choice. Sounds were emitted during ‘quiver’ and
‘circle’ components of the male courtship display and consisted of rapidly repeated pulse units.
Some sound variables differed significantly among species with P. callainos generally being
separated from the other two species. This species produced sounds with higher peak frequency
(for a given length) and lower number of pulses than P. ‘zebra gold’ and higher pulse durations
than P. zebra. In addition, standard length was inversely related to peak frequency in both
P. ‘zebra gold’ and P. callainos (this relation was not tested in P. zebra due to the small sample
size). These differences might indicate different regimes of intraspecific sexual selection among
the three species. #2004 The Fisheries Society of the British Isles
Key words: bioacoustics; Cichlidae; mate choice; Pseudotropheus; reproductive isolation;
sound production.
INTRODUCTION
It is thought that there are between 450 and 650 species of cichlids in Lake
Malawi. They represent one of the fastest and most spectacular examples of
speciation and adaptive radiation known (Turner, 1999). The conspicuous
differences in male colours of closely related species (Ribbink et al., 1983;
Konings, 2002) have led to the suggestion that speciation may be driven by
sexual selection, based largely on divergent female preferences for male colour
(Dominey, 1984; Turner, 1994). Seehausen & van Alphen (1998) demonstrated
that assortative mating among closely related Lake Victorian cichlid species
breaks down under monochromatic light, suggesting that visual cues alone may
be responsible for reproductive isolation, in these species at least. Later studies
†Author to whom correspondence should be addressed. Tel.: þ351 218811700; fax: þ351 218860954;
email: amorim@ispa.pt
Journal of Fish Biology (2004) 65, 1358–1371
doi:10.1111/j.1095-8649.2004.00535.x,availableonline at http://www.blackwell-synergy.com
1358
#2004 The Fisheries Society of the British Isles
of Malawi cichlids, however, have failed to replicate this result (Jordan et al.,
2003; N.J. Barson, M.E. Knight & G.F. Turner, unpubl. data).
Males of many cichlid species are known to make sounds during courtship
(Lobel, 2001; Amorim et al., 2003). Acoustic species recognition has been
demonstrated in pomacentrids (Myrberg et al., 1978; Spanier, 1979), and
suggested for several other fishes, including cichlids (Lobel, 1998). Could
acoustic cues be involved in species recognition among closely related species
of Lake Malawi cichlids? If so, this might suggest that investigations of the
establishment and maintenance of reproductive isolation among cichlids should
consider acoustic as well as other non-visual cues.
To investigate interspecific differences between closely related species, court-
ship calls of three species of Pseudotropheus that co-occur at Nkhata Bay, on
the western shore of Lake Malawi, were recorded and analysed. These are
members of the most species-rich subgenus within the most species-rich group
of Malawi cichlids, the mbuna, which are small largely herbivorous fishes found
in shallow rocky habitats (Konings, 2002).
METHODS
STUDY SPECIES
The three focal species chosen were members of the Pseudotropheus zebra complex,
now considered to belong to the subgenus Maylandia Meyer & Foerster. The name
Metriaclima Stauffer, Bowers, Kellogg & Mckaye is sometimes used as a generic name
for the group, although this is now considered a junior synonym of Maylandia (Conde
´&
Ge
´ry, 1999). The three taxa differ strikingly in male courtship colours but are similar in
other morphological traits. Male P. zebra (Boulenger) are pale blue with black vertical
bars, males of the undescribed species P. ‘zebra gold’ (Ribbink et al., 1983) have a very
similar pattern of brown bars on a yellow background, while P. callainos Stauffer & Hert
males are uniform pale blue. Pseudotropheus callainos females are also distinctively
coloured (blue or white), but those of P. zebra and P. ‘zebra gold’ are generally cryptic
and easily confused, both by human observers and by courting males (Knight & Turner,
1999). Analysis of population allele frequencies of microsatellite DNA loci indicates that
these species are reproductively isolated (van Oppen et al., 1998) and they mate assorta-
tively in the laboratory, indicating that reproductive isolation can be maintained by
direct mate choice alone (Knight et al., 1998). The three species are considered to be
probably at least very closely related if not sister species (Allender et al., 2003).
EXPERIMENTAL PROCEDURE
To minimize external noise, trials were conducted in a heavily insulated room. All
tanks were fitted with internal power filters, 250 W heaters (25C) and 36 W overhead
fluorescent lights (photoperiod of 12L : 12D). Tanks were given weekly one-third water
changes with tap water (pH 80–83). All fishes were fed daily on a mixture of flake and,
pea and shrimp mix (Fohrman, 2002). The fishes used were all first generation laboratory
stock, bred from parents collected at Nkhata Bay, Malawi (11360N; 34170E).
Three aquaria (150 45 60 cm) with a common acoustic environment (i.e. with
similar background noise) were used. Each aquarium was divided into three 50 cm long
compartments by two opaque, removable partitions. Each tank housed a single species.
Seven or eight P. ‘zebra gold’, P. zebra or P. callainos females were kept permanently in
each central compartment. Each end compartment held a single male. A terracotta pot
provided a refuge and prospective spawning site for each male. After introduction, males
were left visually isolated to acclimatize for a minimum of 24 h. Once recording was
LAKE MALAWI CICHLID COURTSHIP SOUNDS 1359
#2004 The Fisheries Society of the British Isles, Journal of Fish Biology 2004, 65, 1358–1371
complete, a male was removed, weighed (wet mass, M) and measured (standard length,
L
S
), returned to a stock tank and replaced with another male of the same species. A total
of 12 P. ‘zebra gold’, 14 P. zebra and 14 P. callainos adult males were used. Recordings
for the three species were carried out within 2 months.
DATA COLLECTION
At the start of a recording period, all electrical appliances were switched off, apart
from the tank lights. One of the opaque partitions was then removed, allowing one male
free access to the females in the central compartment. The fishes were allowed to
acclimatize to the altered conditions for 5 min before recording began. Sounds were
recorded with a High Tech 94 SSQ hydrophone (sensitivity of 165 dB re 1 V mPa
1
,
flat frequency response up to 6 kHz 1 dB) placed near the entrance of the male’s
spawning site, and with a Sony TCD-D8 DAT recorder. Sound production was moni-
tored by one researcher listening with headphones who communicated any sound pro-
duction with hand signals to the other researcher who recorded behavioural data. Thus
the association between sound and behaviour could be recorded. Recorded sounds could
be attributed to the subject males because their intensity varied with distance from the
hydrophone and were consistently associated with particular courtship displays. Male
behaviour was recorded for 20 min. Males that did not court on their first trial were
re-tested up to three times before being returned to stock tanks. Female aggression or
response to male courtship was also recorded.
Male courtship behaviour in these species consists of a series of distinctive and
relatively invariant movements that are not always displayed in a fixed order (Baerends
& Baerends van Roon, 1950; Vodegel, 1978): quiver, male positioned laterally in front of
female, body trembling; dart, male makes exaggerated but rapid 180turns displaying
opposite flanks in quick succession; lead swim, male attempts to lead the female to his
spawning site, swimming in front of her with exaggerated movements (‘fluttering’) of the
caudal and dorsal fins; circle, if male is successful in attracting the female to the spawning
site the pair may then follow each other around head to tail in tight circles. During
‘circle’; the male again often quivers. Spawning may subsequently commence.
Aggressive behaviour typically consists of chasing (towards sub-dominant individuals),
which may involve physical contact and biting, or lateral and frontal display. Lateral
display is similar to the courtship quiver and frontal display involves charging head on to
the opponent with all fins fully extended and flared opercula (Baerends & Baerends van
Roon, 1950; Vodegel, 1978).
SOUND ANALYSIS
Sounds were digitized at 22 kHz (16 bit resolution) and analysed with Canary 1.2.4 for
Macintosh. Only sounds that showed a clear structure and were recorded close to the
hydrophone and outside the terracotta pot were considered for analysis. Typically the
sounds with sufficient quality for analysis were produced at approximately one fish body
length (L
S
) from the hydrophone, sometimes up to two body lengths. The sounds
recorded were made up of repeated pulses (see Fig. 1). The following sound variables
were measured: sound duration, time elapsed from the start of the first pulse to the end of
the last pulse in ms; number of pulses; pulse period, average peak to peak interval
between consecutive pulse units in the entire sound in ms; pulse duration, average
duration of five pulses in a sound in ms; peak frequency, frequency component with
the highest energy in the entire sound in Hz. Temporal variables were measured from
oscillograms. As background noise may confound the beginning and the end of pulses,
the following criteria were used to measure pulse duration. Sound (pulse) onset was
measured from the point of a rapid rise in positive (or negative) energy, and sound
termination was determined at the zero crossing of the last cycle where sound signal
decayed into the background. As it is sometimes difficult to determine sound limits
perfectly, especially the sound termination since the amplitude is decaying, a further
criterion was used. As the rapid oscillation in the sound waveform is much more
1360 M. C. P. AMORIM
ET AL.
#2004 The Fisheries Society of the British Isles, Journal of Fish Biology 2004, 65, 1358–1371
homogenous than the background noise, which is usually irregular, pulse limits were
considered when a more homogenous sound wave stopped and joined with irregular
noise (see Fig. 1). It should be noted that the pulse structure of sounds recorded in small
tanks may also be affected by reverberations in the tank (Akamatsu et al., 2002). Peak
frequency was measured from the power spectra (filter bandwidth of 3497 Hz). High-
pass filters (up to 120 Hz but usually 60 Hz) were used in oscillograms when needed to
remove low-frequency background noise, but peak frequencies were always measured
from unfiltered sounds.
2·0
1·0
20
0
–20
5·0
0·0
–5·0
200
200
400
400
600
600
9·8 ms
9·8 ms
10·4 ms
11·2 ms
600
600
800
800
1000
1000
1200
1200
1400
1400
1600
1600
Frequency (kHz)Frequency (kHz)
2·0
1·0
3400 3600
Time (ms)
3800 4000
3400 3600 3800 4000
Frequency (kHz) Amplitude (µPa)Amplitude (µPa)
20
0
–20
Amplitude (µPa)
2·0
(a)
(b)
(c)
1·0
FIG. 1. Sonograms (top) and oscillograms (bottom) of typical courtship sounds produced by males of
(a) Pseudotropheus ‘zebra gold’, (b) P. zebra and (c) P. callainos. For each species is depicted one or
two pulses of sound and their durations. For P. ‘zebra gold’ a portion of the oscillograms showing
two pulses is expanded to illustrate differences between the background noise and the sound pulse.
Oscillograms were filtered with a (b) 60 and (a), (c) 120 Hz high-pass filters. The y-axis of the
oscillograms show relative amplitudes.
LAKE MALAWI CICHLID COURTSHIP SOUNDS 1361
#2004 The Fisheries Society of the British Isles, Journal of Fish Biology 2004, 65, 1358–1371
DATA ANALYSIS
Preliminary experiments to check for sound production by these three species pro-
duced good quality recordings for one P. ‘zebra gold’ male and two P. zebra males, but
male size was not registered at the time. Data from these recordings are included on the
general description of sounds and in the comparison among species in cases where the
L
S
variable was not considered important. A w
2
-test was used to test for independence
of behaviour and sound production.
Differences among species in the five sound variables measured were tested with linear
mixed models (LMMs). The LMMs provide an alternative to nested ANOVA for
unbalanced data sets (variable number of fish recorded within species and variable
number of sounds analysed within fish in the present study) and permit the incorporation
of information from continuous independent variables (fish L
S
in this case). It is required
that the random effects (expected response for each fish centred about the expected
response for the species in this case) are normally distributed with a zero mean and a
constant variance and that the within-group errors [residual deviation of each sound
observation from both the fitted fixed (species) and random (individual fish) effect in this
case] have a zero mean and constant variance and are independent of the random effects.
Pinheiro & Bates (2000) give an outline of the statistical principles and McRoberts et al.
(1998) provide an ecological application of mixed-effects models. Exploratory analysis on
the residuals of the fitted models [distribution of standardized residuals against the
grouping factor (i.e. the random effect) and against fitted values, separately for each
level of the classification factor (i.e. the fixed effect)] revealed that the above assumptions
were adequately met (in the case of sound duration and number of pulses after logarith-
mic transformation), according to the inspection criteria described by Pinheiro & Bates
(2000).
In the first part of the analysis, an LMM was fitted to each sound variable using
species (three level classification factor) as the fixed effect and individual fish nested
within species as the random effect (nine, five and eight level grouping factor for P. ‘zebra
gold’, P. zebra and P. callainos, respectively). Evidence of any differences between species
was assessed by referring to the Wald statistic (a multivariate generalization of the
t-statistic) divided by the fixed-effect d.f. to an F-distribution on 2 and 19 d.f.
(the fixed-effect degrees of freedom and the degrees of freedom for estimating the
within-group residual variance respectively). In the results, the Wald test is reported as
the equivalent F-value on the d.f. Significant differences were investigated further by
pair-wise comparisons between species using t-tests on 19 d.f. with Bonferroni adjustment
for multiple comparisons.
The effect of fish L
S
was not considered in this first analysis, as these data were not
available for all fishes (particularly for P. zebra). Fish L
S
only had a significant effect on
peak frequency, so in the second part of the analysis, LMMs were fitted to the sound
peak frequency observations from the species P. ‘zebra gold’ and P. callainos for all
individuals with a known L
S
(eight fish for each species). In the second analysis, the
model considered the fixed effect of species, L
S
and their interaction, while the random
effect consisted of the individuals nested within species. The LMMs were fitted in R
(open source software) using the library nlme (Pinheiro & Bates, 2000).
Differences between males of the same species were tested with one-way ANOVA for
each sound variable.
RESULTS
SOUND PRODUCTION
A total of 638 sounds were detected over the course of the experiment: 212
from 11 P. ‘zebra gold’, 152 from seven P. zebra and 274 from 10 P. callainos
males. Twelve males neither attempted courtship nor produced any sound
(Table I). Table I also shows information on L
S
and mass as well as the number
1362 M. C. P. AMORIM
ET AL.
#2004 The Fisheries Society of the British Isles, Journal of Fish Biology 2004, 65, 1358–1371
of sounds produced per male per session by species. Sound was produced by
every male that attempted to court a female. The sounds emitted by the males
of these three species generally had peak frequencies <720 Hz, pulse durations
of c. 9 to 12 ms, pulse period of 60–70 ms and sound duration varying between
500 and 700 ms (Table II and Fig. 2). Example sonograms and oscillograms of
sounds produced by each species are shown in Fig. 1.
ASSOCIATION WITH BEHAVIOUR
Courtship behaviour and sound production were significantly related
(w
2
, d.f. ¼4, P<0001); the great majority of sounds being produced during
the male ‘quiver’ and ‘circling’ displays (Table III). Sound production was also
detected on one occasion while a P. zebra female performed a lateral display to
a second female revealing that sound production is not restricted to males in
this species.
COMPARISON AMONG SPECIES
Variation within each sound variable was generally similar across species
(Fig. 2) and, with few exceptions, observations within species were symmetric-
ally distributed about the mean (in the case of sound duration and number of
pulses after ln transformation). None of the five sound variables differed
significantly among males of the same species (one-way ANOVA, P>005).
The results of the LMMs showed significant differences among species in pulse
duration (Wald test, d.f. ¼2 and 19, P¼0021) and peak frequency (Wald test,
d.f. ¼2 and 19, P¼0004). Pair-wise comparisons (t-tests after Bonferroni
correction for multiple comparisons), however, demonstrated that for both
sound variables only one comparison between species was significant at the
5%level. Pseudotropheus callainos was the species most clearly distinguished
based on sound properties, having a significantly higher peak frequency than
P. ‘zebra gold’ and a significantly higher pulse duration than P. zebra.
TABLE I. Number of males of Pseudotropheus ‘zebra gold’, P. zebra and P. callainos that
were tested, recorded, and with sounds analysed. The mean S.E. (range in parentheses)
number of sounds emitted per male per recording sessions and standard length of males
with sounds analysed is also shown
P. ‘zebra gold’ P. zebra P. callainos
Number of males tested 12 14 14
Number of males that produced sounds 11 7 10
Number of males with sounds analysed 8 4 8
Number of sounds per male per
recording session
1925 432 2533 789 2491 729
Male L
S
(mm) 105309
(1007–1096)
99923
(960–1065)
98328
(868–1108)
Male mass (g) 4206 263
(3341–5608)
3921 350
(3193–4724)
3242 308
(2226–4633)
LAKE MALAWI CICHLID COURTSHIP SOUNDS 1363
#2004 The Fisheries Society of the British Isles, Journal of Fish Biology 2004, 65, 1358–1371
TABLE II. Description of sounds made during courtship by Pseudotropheus ‘zebra gold’, P. zebra and P. callainos (present study) and by other
cichlids. Values are means with range in parentheses
Species
Sound
type
Sound
duration (ms)
Number of
pulses
Pulse
duration (ms)
Pulse
period (ms)
Peak
frequency (Hz)
Motor
pattern
Sample
size (n) and sex
P. ‘zebra gold’ (1) – 6676
(3104–10224)
109
(40–199)
104
(89–113)
633
(516–770)
4356
(3887–4798)
Quivering
Circle
9
M
P. zebra (1) – 5320
(3835–7855)
91
(70–141)
97
(75–114)
656
(604–742)
4604
(3993–5091)
Quivering
Circle
6
M
P. callainos (1) – 4763
(3494–6523)
80
(57–114)
115
(96–134)
668
(501–789)
5174
(4249–6199)
Quivering
Circle
8
M
Haplochromis burtoni
(Gu
¨nther) (2, 3)
Br-r-r 370 (230–1410) 13 (5–37) – – 320 (120–630)
(80–1000)
Quivering 40
M
Hemichromis bimaculatus
Gill (4)
– – – – 10–15 pulses s
1
i.e. 71–110 ms
–
<100
Quivering ?
M
Herotilapia multispinosa
(Gu
¨nther) (5)
Growl 200 (108–1600) – 5 (3–10) 121 (76–316)
a
150–350
(100–600)
Quivering
b
20
Mainly M,
but also F
Oreochromis mossambicus
(Peters) (6; also see 7–8)
– 712 (100–2834) 17 (4–60) 12 (9–15) 44 (22–137) 354
(207–524)
Mainly tail
wagging
19
M
1364 M. C. P. AMORIM
ET AL.
#2004 The Fisheries Society of the British Isles, Journal of Fish Biology 2004, 65, 1358–1371
Simochromis babaulti
Pellegrin (3)
Br-r-r 590 (240–1470) 26 (8–40) 20 (10–20) – 193 (125–500)
(50–800)
Quivering 28
M
Simochromis diagramma
(Gu
¨nther) (3, 9)
c
Br-r-r 1100 (500–2000) 26 (16–37) 100 (100–100) 33 200 (100–200)
(50–500)
Quivering
b
1 (3)
M
Tropheus brichardi
Nelissen & Thys (3)
Br-r-r 830 (250–2350) 32 (9–92) 20 (10–30) – 620 (800–900)
(200–2000)
Quivering 7
M
Tropheus duboisi
Marlier (3)
Br-r-r 460 (70–1470) 19 (3–33) 10 (10–10) – 407 (125–1250)
(63–5000)
Quivering 36
M
Tropheus moorii
Boulenger (2, 3)
Br-r-r 400 (150–900) 20 (5–35) 20 (6–70) – 840 (50–1250)
(25–3150)
Quivering 46 (3)
M
Copadichromus
conophorus
Stauffer, LoVullo &
MacKaye (10)
– 181 (78–654) 10 (5–30) 67 (3–10) 180 (14–22)
d
471 (372–594) Quivering fin
fluttering
27
M
Tramitichromis cf.
intermedius
(Trewavas) (10)
– 199 (120–304) 9 (6–12) 62 (3–11) 215 (17–33)
d
388 (305–480) Quivering fin
fluttering
20
M
(1), Present study; (2, 3), Nelissen (1977, 1978); (4), Rowland (1978); (5), Brown & Marshall (1978); (6), Amorim et al. (2003); (7), Konstantinova et al. (1979); (8), Marshall (1971);
(9), Nelissen (1975); (10), Lobel (1998).
a
Interpulse distance;
b
also produced during agonistic displays;
c
data presented includes the range found in more than one study;
d
determined by dividing call duration by number of pulses.
LAKE MALAWI CICHLID COURTSHIP SOUNDS 1365
#2004 The Fisheries Society of the British Isles, Journal of Fish Biology 2004, 65, 1358–1371
P. ‘zebra
gold’
P. zebra P. callainos
6
8
10
12
14
16
18
a,b a
b
Species
(d)
2
3
4
5
6
7
8
Sound duration (ms)Pulse duration (ms)
(a)
P. ‘zebra
gold’
P. zebra P. callainos
300
400
500
600
700
a
a,b
b
(e)
Peak frequency (Hz) Number of pulses
0
1
2
3
4
a
b
(b)
Pulse period (ms)
P. ‘zebra
gold’
P. zebra P. callainos
50
100
150
(c)
FIG. 2. Distribution of observations by species for the five sound variables (a) sound duration, (b) number of pulses, (c) pulse period, (d) pulse duration and (e) peak
frequency measured (data pooled for all sounds of individuals from the same species; sound duration and number of pulses are transformed to their natural
logarithms). The boxes in the plots correspond to the first, second and third quartile of observations respectively, the whiskers to values that would correspond to
the 99%range of a normally distributed variable, while the individual points (*) represent outliers. Significant differences among species (P<005) are indicated
by different letters (t-tests for pair-wise comparisons with Bonferroni correction for multiple comparisons). Significant differences shown for the number of pulses
result from a comparison between Pseudotropheus ‘zebra gold’ and P. callainos only.
1366 M. C. P. AMORIM
ET AL.
#2004 The Fisheries Society of the British Isles, Journal of Fish Biology 2004, 65, 1358–1371
Although the comparisons among species were non-significant for the other
three sound variables (Wald test, d.f. ¼2 and 19, P¼0229 for sound duration;
P¼0090 for number of pulses; P¼0268 for pulse period), there are some
indications that P. callainos generally produced sounds of lower duration and
with a fewer number of pulses (Fig. 2). Repeating the analysis only for P. ‘zebra
gold’ and P. callainos (the two species with higher number of observations) led
to a significant difference in the number of pulses between the two species
(Wald test, d.f. ¼1 and 15, P¼0046) but not in sound duration. If P. callainos
is compared with P. ‘zebra gold’ and P. zebra (which behave in a similar fashion
for both sound duration and number of pulses), however, differences in the
number of pulses become more evident (Wald test, d.f. ¼1 and 20, P¼0021)
and marginally non-significant for sound duration (Wald test, d.f. ¼1 and 20,
P¼0083). It is therefore likely that significant differences among species may
also exist in other sound variables (in particular in sound duration and perhaps
pulse period), but there is insufficient power to detect them in the present
analysis (where sounds from only four P. zebra were used).
The above analyses did not consider the effect of fish L
S
on the sound
variables. Spearman rank correlations were calculated between mean sound
variables and L
S
for P. ‘zebra gold’ and P. callainos. Despite the small sample
size (n¼8), the results indicated a highly significant negative correlation
between peak frequency and P. callainos L
S
(r¼090, n¼8, P¼0002).
Based on the above results, the LMM for peak frequency was repeated con-
sidering only P. callainos and P. ‘zebra gold’ and including the effect of L
S
and
its interaction with species. The exclusion of P. zebra and the inclusion of L
S
as
an explanatory variable did not alter the previous results, leading to the accentu-
ation of the differences in peak frequency between the two species (Wald test,
d.f. ¼1 and 12, P<0001) and indicating a highly significant negative effect of
L
S
(Wald test, d.f. ¼1 and 12, P¼0009) that was common in the two species
(i.e. non-significant interaction term: Wald test, d.f. ¼1 and 12, P¼0380). The
resulting model demonstrating the significant differences in the peak frequency
of the two species and the negative effect of fish L
S
is shown in Fig. 3, where the
mean peak frequency values for each individual are also superimposed.
TABLE III. Sound production and associated male behaviour (Q, quiver; LS, lead swim;
D, dart; C, circle; Ch, chase; none, sound produced with no obvious associated
behaviour) in Pseudotropheus ‘zebra gold’, P. zebra and P. callainos (n¼4478)
Q* Q %Q** LS* LS %LS** D* D %D** C* C %C** Ch* Ch %Ch** *none
P. ‘zebra
gold’
185 106 636 7 191 35 0 246 0 10 43 188 0 783 0 10
P. zebra 119 35 773 3 96 30 0 114 0 14 19 424 0 348 0 16
P. callainos 235 38 861 20 358 53 7 205 33 3 30 91 1 1262 018
*, Sound produced simultaneously, otherwise behaviour recorded but no associated sound detected.
**, Proportion of the total number of any behaviour observed for each species where sounds were
simultaneously recorded.
LAKE MALAWI CICHLID COURTSHIP SOUNDS 1367
#2004 The Fisheries Society of the British Isles, Journal of Fish Biology 2004, 65, 1358–1371
DISCUSSION
Males of three species of the P. zebra complex produced low frequency pulsed
sounds while quivering and circling. The mean values of the recorded sound
variables lay within the overall ranges expected from other studies of African
mouth-brooding cichlids (Table II), although the sounds were longer in overall
duration that those reported for other Malawi cichlids (Lobel, 1998).
The analysis demonstrated statistically significant differences in several prop-
erties of the sounds produced by these three closely related species. Although small
sample sizes and lack of L
S
data (particularly for P. zebra where reliable sound
data were only available in four fish of registered length) reduced the power and
scope of the analysis considering all species, the LMM results generally sepa-
rated P. callainos from the other two species. Pseudotropheus callainos produced
sounds with higher peak frequency and lower number of pulses than P. ‘zebra
gold’, and of higher pulse durations than P. zebra. Marginally non-significant
results also suggest that P. callainos sounds may have lower sound durations
than P. ‘zebra gold’ and P. zebra. In addition, the results demonstrated that
fish size (L
S
) is inversely related to peak sound frequency in P. ‘zebra gold’ and
P. callainos, thus differences in fish size should be taken into account when
sound frequency properties between species are compared.
Lobel (1998) showed statistically significant differences in pulse rates and
durations for two sympatric Malawian cichlids and proposed that courtship
sounds could play a role in mate choice and species recognition. He also
suggested that it would be worth investigating the possibility that differentiation
of acoustic signals might play a role in speciation. Lobel (1998) worked on
species which are not currently believed to be particularly closely related, and
are placed in different genera. Thus in this study three closely related sympatric
species were examined. Although the results are consistent with Lobel’s (1998)
g
g
g
gg
g
gg
80 90 100 110 120
350
400
450
500
550
600
650
LS (mm)
Mean peak frequency (Hz)
c
c
c
c
c
c
c
c
FIG. 3. Mean individual peak frequency observations in relation to fish standard length for Pseudotropheus
‘zebra gold’ (g) and P. callainos (c). Lines indicate LMM fit as a function of length and species (—,
P. ‘zebra gold’; ---, P. callainos) from final analysis on two species.
1368 M. C. P. AMORIM
ET AL.
#2004 The Fisheries Society of the British Isles, Journal of Fish Biology 2004, 65, 1358–1371
prediction, suggesting differences between closely related species in the nature of
their courtship calls, it remains to be demonstrated that females of the present
study species can actually differentiate among conspecific and heterospecific
males on the basis of these differences in courtship calls. Temporal discrimin-
ation has not been investigated in hearing generalists (fishes that do not possess
accessory hearing structures to enhance hearing ability, in contrast with hearing
specialists; Ladich & Bass, 2003) but the plainfin midshipman fish Porichthys
notatus Girard (Batrachoididae) can perceive pulse intervals as small as 10 ms
(McKibben & Bass, 2001) and frequency discrimination ability in hearing
generalists is generally slightly >10%difference (Fay & Simmons, 1999) and
sometimes comparable to specialists (McKibben & Bass, 1999). Such evidence
suggests that cichlids may be able to perceive the differences found among
species in sound duration, number of pulses and in peak frequency, but prob-
ably not in pulse duration (Wysocki & Ladich, 2003).
Temporal characteristics and pulse grouping patterns are believed to play a
fundamental role in species recognition in fishes (Winn, 1964; Spanier, 1979;
Crawford et al., 1997), although as far as is known, no one has demonstrated
that courtship sounds were effective in species separation in any fish species.
Equally, peak frequency is also important in animal acoustic communication
and is commonly used in individual recognition and assessment in fishes
(Myrberg & Riggio, 1985; Ladich et al., 1992) and other animals (Davies &
Halliday, 1978), since it may provide information on the size of an individual.
The acoustical variables that are less likely to be affected by fish morphology
(e.g. body size) or environmental constraints (e.g. ambient noise and reflections)
are more reliable and therefore better candidates for species recognition. As
pulse duration may be affected by the signal-to noise ratio, it is suggested that
the number of pulses in a sound and perhaps sound duration are good candi-
dates for specific discrimination among the studied species. Although peak
frequency is dependent on body size highly significant differences were found
between P. ‘zebra gold’ and P. callainos for a given length, suggesting it is also a
good candidate for species discrimination.
The dramatic differences in male colours among closely related species has led to
an interest in the possibility that sexual selection acting on male breeding dress may
have driven speciation in cichlids from Lake Malawi and elsewhere. Although the
present findings are still preliminary, attention should be given to the possible role
of divergent sexual selection acting on other sensory modalities, including acoustic
and perhaps also olfactory stimuli produced during courtship.
MCPA thanks FCT, Portugal (grant Praxis XXI/BPD/11806/97 and pluriannual
programme UI&D 331/94) for financial support and A.D. Hawkins for advice during
the early stages of this work. MEK was supported by NERC grant NER/A/S/1998/
00011. We are especially grateful to R. Fryer for statistical help and advice. R. Robinson
provided helpful comments on earlier versions of the manuscript.
References
Akamatsu, T., Okumura, T., Novarini, N. & Yan, H. Y. (2002). Empirical refinements
applicable to the recording of fish sounds in small tanks. The Journal of the
Acoustical Society of America 112, 3073–3082.
LAKE MALAWI CICHLID COURTSHIP SOUNDS 1369
#2004 The Fisheries Society of the British Isles, Journal of Fish Biology 2004, 65, 1358–1371
Allender, C. J., Seehausen, O., Knight, M. E., Turner, G. F. & Maclean, N. (2003).
Divergent selection during speciation of Lake Malawi cichlid fishes inferred from
parallel radiations in nuptial coloration. Proceedings of the National Academy of
Sciences, USA 100, 14074–14079.
Amorim, M. C. P., Fonseca, P. J. & Almada, V. C. (2003). Sound production during courtship
and spawning of the cichlid Oreochromis mossambicus: male-female and male-male
interactions. Journal of Fish Biology 62, 658–672. doi: 10.1046/j.0022-1112.2003.00054.x
Baerends, G. P. & Baerends van Roon, J. M. (1950). An introduction to the study of the
ethology of cichlid fishes. Behaviour 1(Suppl.), 1–235.
Brown, D. H. & Marshall, J. A. (1978). Reproductive behaviour of the rainbow cichlid,
Herotilapia multispinosa (Pisces, Cichlidae). Behaviour 67, 299–322.
Conde
´,B.&Ge
´ry, J. (1999). Maylandia Meyer et Foerster, 1984, un nom ge
´ne
´rique
disponible (Teleostei, Perciformes, Cichlidae). Revue Franc¸ aise d’Aquariologie et de
Herpetologie 26, 21–22.
Crawford, J. D., Cook, A. P. & Heberlein, A. S. (1997). Bioacoustic behavior of African
fishes (Mormyridae): potential cues for species and individual recognition in
Pollimyrus.The Journal of the Acoustical Society of America 102, 1200–1212.
Davies, N. B. & Halliday, T. R. (1978). Deep croaks and fighting assessment in toads
Bufo bufo.Nature 274, 683–685.
Dominey, W. J. (1984). Effects of sexual selection and life history on speciation: species
flocks in African cichlid fish and Hawaiian Drosophila.InEvolution of Fish Species
Flocks (Echelle, A. A. & Kornfield, I., eds), pp. 231–249. Orono, ME: University of
Maine Press.
Fay, R. R. & Simmons, A. M. (1999). The sense of hearing in fishes and amphibians. In
Comparative Hearing: Fish and Amphibians (Fay, R. R. & Popper, A. N., eds),
pp. 269–318. New York: Springer.
Fohrman, K. (2002). Food and feeding. In Enjoying Cichlids, 2nd edn (Konings, A., ed.),
pp. 34–37. El Paso: Cichlid Press.
Jordan, R., Kellogg, K., Juanes, F. & Stauffer, J., Jr. (2003). Evaluation of female mate
choice cues in a group of Lake Malawi mbuna (Cichlidae). Copeia 2003, 181–186.
Knight, M. E. & Turner, G. F. (1999). Reproductive isolation among closely related Lake
Malawi cichlids: can males recognise conspecific females by visual cues? Animal
Behaviour 58, 761–768.
Knight, M. E., Turner, G. F., Rico, C., van Oppen, M. J. H. & Hewitt, G. M. (1998).
Microsatellite paternity analysis on captive Lake Malawi cichlids supports
reproductive isolation by direct mate choice. Molecular Ecology 7, 1605–1610.
Konings, A. (2002). Cichlid Fishes in their Natural Habitat, 3rd edn. El Paso: Cichlid Press.
Konstantinova, N. G., Nikol’skii, I. D. & Termen, L. S. (1979). Acoustic and electric
activity of Tilapia mossambica (Cichlidae: Pisces). Moscow University Biological
Science Bulletin 34, 35–39.
Ladich, F. & Bass, A. H. (2003). Underwater sound generation and acoustic reception in
fishes with some notes on frogs. In Sensory Processing in Aquatic Environments
(Collin, S. P. & Marshall, N. J., eds), pp. 173–193. New York: Springer.
Ladich, F., Brittinger, W. & Kratochvil, H. (1992). Significance of agonistic vocalization
in the croaking gourami (Trichopsis vittatus, Teleostei). Ethology 90, 307–314.
Lobel, P. S. (1998). Possible species specific courtship sounds by two sympatric cichlid
fishes in Lake Malawi, Africa. Environmental Biology of Fishes 52, 443–452.
Lobel, P. S. (2001). Acoustic behavior of cichlid fishes. Journal of Aquaculture and Aquatic
Sciences 9, 167–186.
Marshall, J. A. (1971). Sound production by Tilapia mossambica (Pisces: Cichlidae).
American Zoologist 11, 632.
McKibben, J. R. & Bass, A. H. (1999). Peripheral encoding of behaviorally relevant
acoustic signals in a vocal fish: single tones. Journal of Comparative Physiology A
184, 563–576.
McKibben, J. R. & Bass, A. H. (2001). Effects of temporal envelope modulation on
acoustic signal recognition in a vocal fish, the plainfin midshipman. The Journal of
the Acoustical Society of America 109, 2934–2943.
1370 M. C. P. AMORIM
ET AL.
#2004 The Fisheries Society of the British Isles, Journal of Fish Biology 2004, 65, 1358–1371
McRoberts, R. E., Brooks, R. T. & Rogers, L. L. (1998). Using nonlinear mixed effects
models to estimate size-age relationships for black bears. Canadian Journal of
Zoology 76, 1098–1106.
Myrberg, A. A., Jr. & Riggio, J. R. (1985). Acoustically mediated individual recognition
by a coral reef fish (Pomacentrus partitus). Animal Behaviour 33, 411–416.
Myrberg, A. A., Jr., Spanier, E. & Ha, S. J. (1978). Temporal patterning in acoustical
communication. In Contrasts in Behaviour (Reese, E. S. & Lighter, F. J., eds),
pp 137–179. New York: John Wiley & Sons.
Nelissen, M. (1975). Sound production by Simochromis diagramma (Gu
¨nther) (Pisces,
Cichlidae). Acta Zoologica et Pathologica Antverpiensia 61, 19–24.
Nelissen, M. (1977). Sound production by Haplochromis burtoni (Gu
¨nther) and Tropheus
moorii Boulenger (Pisces, Cichlidae). Annales de la Societe Royale Zoologique de
Belgique 106, 155–166.
Nelissen, M. (1978). Sound production by some Tanganyikan cichlid fishes and a hypothesis
for the evolution of the R communication mechanisms. Behaviour 64, 137–147.
van Oppen, M. J. H., Turner, G. F., Rico, C., Deutsch, J. C., Robinson, R. L., Genner, M. J.
& Hewitt, G. M. (1998). Biodiversity and evolution in rock-dwelling cichlid fishes
from Lake Malawi: evidence from microsatellite markers. Molecular Ecology 7,
991–1001.
Pinheiro, J. C. & Bates, D. M. (2000). Mixed-effects Models in S and S-PLUS. New York:
Springer.
Ribbink, A. J., Marsh, B. A., Marsh, A. C., Ribbink, A. C. & Sharp, B. J. (1983).
A preliminary survey of the cichlid fishes of rocky habitats in Lake Malawi.
South African Journal of Zoology 18, 149–310.
Rowland, W. J. (1978). Sound production and associated behavior in the jewel fish,
Hemichromis bimaculatus.Behaviour 64, 125–136.
Seehausen, O. & van Alphen, J. J. M. (1998). The effect of male coloration on female mate
choice in closely related Lake Victoria cichlids (Haplochromis nyererei complex).
Behavioral Ecology and Sociobiology 42, 1–8. doi: 10.1007/s002650050405
Spanier, E. (1979). Aspects of species recognition by sound in four species of damselfishes,
genus Eupomacentrus (Pisces: Pomacentridae). Zeitschrift fu
¨r Tierpsychologie 51,
301–316.
Turner, G. F. (1994). Speciation mechanisms in Lake Malawi cichlids: a critical review. In
Speciation in Ancient lakes (Martens, K., Goddeeris, B. & Coulter, G., eds), pp.
139–160. Stuttgart: Schweizerbart.
Turner, G. F. (1999). Explosive speciation of African cichlid fishes. In Evolution of
Biological Diversity (Magurran, A. E. & May, R. M., eds), pp. 217–229. Oxford:
Oxford University Press.
Vodegel, N. (1978). A causal analysis of the behaviour of Pseudotropheus zebra (Boulenger)
(Pisces: Cichlidae). PhD Thesis. University of Groningen, The Netherlands.
Winn, H. E. (1964). The biological significance of fish sounds. In Marine Bio-acoustics,
Vol. 1 (Tavolga, W. N., ed.), pp. 213–230. Oxford: Pergamon Press.
Wysocki, L. E. & Ladich, F. (2003). The representation of conspecifics sounds in the
auditory brainstem of teleost fishes. The Journal of Experimental Biology 206,
2229–2240.
LAKE MALAWI CICHLID COURTSHIP SOUNDS 1371
#2004 The Fisheries Society of the British Isles, Journal of Fish Biology 2004, 65, 1358–1371
