Content uploaded by Terry Burke
Author content
All content in this area was uploaded by Terry Burke
Content may be subject to copyright.
Abstract The buffalo weavers, Bubalornis spp., are
unique amongst birds in possessing a phalloid organ, a
phallus-like structure anterior to the cloaca. We studied
the red-billed buffalo weaver Bubalornis niger, to deter-
mine whether the phalloid organ has evolved in response
to sperm competition. The phalloid organ was signifi-
cantly longer in males that were resident at nests than in
non-resident males, and among resident males was sig-
nificantly longer in those males with a harem than in
those without. Red-billed buffalo weavers bred colonial-
ly and had either a cooperatively polygynandrous (usual-
ly two unrelated males and several females) or a polygy-
nous (one male and several females) mating system. Co-
operative polygynandry provided females with the op-
portunity to copulate with more than one male and pater-
nity analyses using DNA fingerprinting revealed that
63% of 16 multiple-offspring broods, comprising 43 off-
spring, had multiple sires, which included both nest-
owning males and extra-group males. Sperm competition
was therefore intense. Observations and experiments
with buffalo weavers in captivity revealed that the phal-
loid organ was not intromittent during copulation, but
functioned as a stimulatory organ which necessitated
protracted copulation in order to induce male ‘orgasm’
and ejaculation, a feature apparently unique to this spe-
cies.
Keywords Phalloid organ · Orgasm · Sperm
competition · Bubalornis niger · Penis
Introduction
The red-billed (Bubalornis niger) and white-billed (B.
albirostris) buffalo weavers have a unique phallus-like
appendage or phalloid organ whose function has excited
speculation for over 150 years (Lesson 1831; Hartert
1927; Sushkin 1927; Hoesch 1952; Crook 1958;
Birkhead et al. 1993a). The phalloid organ is non-erec-
tile, has no sperm duct and comprises only connective
tissue; it is not homologous with any penis found in oth-
er bird species (see King 1981). The phalloid organ lies
immediately anterior to the cloaca and is enlarged in the
male, but vestigial in the female (Bentz 1983). Because
of this, the phalloid organ is likely to function in repro-
duction (Lesson 1831; Sushkin 1927; Hoesch 1952;
Birkhead et al. 1993a), although whether it is intromit-
tent during copulation is not known (Birkhead et al.
1993a).
Anecdotal reports indicate that, compared with most
other passerine birds, buffalo weavers copulate in an un-
usual manner. Passerines typically mount once per ejacu-
lation and for only a few seconds (Birkhead and Møller
1992). In contrast, buffalo weavers mount more than
once per ejaculation and mounting attempts are pro-
longed, lasting several minutes (Hoesch 1952; Birkhead
et al. 1993a; Winterbottom et al. 1999). The presence of
the phalloid organ may be related to this protracted cop-
ulatory behaviour. If the phalloid organ functions in re-
production, it is likely to have evolved by sexual selec-
tion (Darwin 1871; Andersson 1994), possibly as an ad-
aptation to sperm competition (Birkhead and Møller
1992). Sperm competition is an important component of
sexual selection, and is responsible for a wide range of
behaviours and morphological features in birds and
Communicated by J. Höglund
Supplementary material: Electronic supplementary material to this
paper can be obtained by using the Springer LINK server located
at http://dx.doi.org/10.1007/s 002650100384
M. Winterbottom · T.R. Birkhead (
✉
)
Department of Animal and Plant Sciences,
The University of Sheffield, Sheffield, S10 2TN, UK
e-mail: t.r.birkhead@sheffield.ac.uk
Tel.: +44-114-2224622, Fax: +44-114-2220002
T. Burke
Department of Biology, University of Leicester,
Leicester, LE1 7RH, UK
Present address:
T. Burke, Department of Animal and Plant Sciences,
The University of Sheffield, Sheffield, S10 2TN, UK
Behav Ecol Sociobiol (2001) 50:474–482
DOI 10.1007/s002650100384
ORIGINAL ARTICLE
M. Winterbottom · T. Burke · T. R. Birkhead
The phalloid organ, orgasm and sperm competition in a
polygynandrous bird: the red-billed buffalo weaver (
Bubalornis niger
)
Received: 17 April 2001 / Revised: 14 May 2001 / Accepted: 15 May 2001 / Published online: 5 July 2001
© Springer-Verlag 2001
475
throughout the animal kingdom (Smith 1984; Birkhead
and Møller 1992, 1998). The opportunity for females to
copulate with more than one male determines the intensi-
ty of sperm competition and is often determined by a
species’ mating system (Birkhead and Møller 1992;
Davies 1992).
Previous observations indicated that the red-billed
buffalo weaver may be polygynous (Crook 1958; Kemp
and Kemp 1974). Jensen and Clinning (1983) also noted
that more than one male may be resident at a single nest
and that such males were not aggressive to each other.
This indicates that they have either a polyandrous mating
system or that they are cooperative, as in the closely re-
lated sparrow weavers (Philetarius socius, Plocepasser
mahali and Pseudonigrita arnaudi) (Maclean 1973;
Collias and Collias 1978; Bennun 1994). Either way, the
close proximity of more than one male may increase a
female’s opportunity to copulate with multiple males,
and the intensity of sperm competition.
The aims of this study were to (1) establish the inten-
sity of sperm competition in the red-billed buffalo weav-
er (hereafter buffalo weaver), (2) establish the role of the
phalloid organ in reproduction and (3) consider how the
phalloid organ may have evolved in response to sperm
competition.
Methods
The study areas and study species
Fieldwork was conducted during the summers of 1994, 1995 and
1996 on farms and in National Parks in Namibia. The buffalo
weaver’s range stretches across southern Africa, through regions
of semi-arid savannah and dry bush-veld (Maclean 1973, 1993).
Buffalo weavers breed in response to rainfall, which occurs sea-
sonally during the austral summer (Maclean 1973, 1993). During
the present study, a drought limited the amount of behavioural da-
ta that could be collected.
The buffalo weaver is sexually dimorphic, both in terms of size
[in this study: mean male (±SE) mass=84.4±0.48 g, n=109; mean
female mass=71.4±0.66 g, n=67], and plumage (males have black
plumage and a coral-red bill; females have brown plumage and an
orange-grey bill; Maclean 1993). Breeding appears to be colonial:
each nest contains several breeding chambers, each inhabited by a
different female, and two or more such compound nests are often
built together (see Figs. S1–3 (Electronic Supplementary Material
= ESM)). Incubation lasts 14 days and nestlings typically spend
20 days in the nest (Kemp and Kemp 1974; Maclean 1993).
Social interactions and cooperative behaviour
Buffalo weavers were captured using mist nets and given unique
combinations of colour rings. The length of the phalloid organ was
measured with callipers to the nearest 0.1 mm, from the cloacal
opening to its distal tip. Individuals were observed at breeding col-
onies and their interactions with other colony members monitored.
This was less successful than anticipated because many birds were
nomadic during the breeding season, visiting colonies for only a
few days at a time: only 27 (15%) of 182 marked birds were ever
observed again after re-release. Behavioural data were collected
during 1996. Three days were spent at each colony and observa-
tions comprised 48 10-min bouts each day from dawn to dusk:
24 h in total. The number of females associated with each com-
pound nest was estimated as the number of reproductively active
breeding chambers in each nest. Resident males were identified by
the amount of time they spent at the compound nest. To confirm
that all resident males had been ringed, the proportion of time
spent at the compound nest by unringed males was also recorded.
To establish whether males acted cooperatively to defend and
maintain a compound nest and to feed offspring, we recorded (1)
the rate at which each male added sticks or grass to the compound
nest, (2) the number of times each male fed offspring in each
breeding chamber and (3) the proportion of aggressive interactions
between a resident and a non-resident male, during which the oth-
er resident male was closer to the attacker.
Are cooperative males related
and how is paternity shared between males?
Genetic relatedness and paternity were assessed using multilocus
DNA fingerprinting (Burke and Bruford 1987; Bruford et al.
1998). Upon capture, approximately 100 µl of blood was collect-
ed, under licence, from adults and chicks (of at least 8 days old)
by venipuncture of the metatarsal vein and stored, at ambient tem-
perature, in 1.0 ml of sterile Queen’s lysis buffer (QLB; Seutin et
al. 1991). When chicks were found dead, samples of brain were
removed and stored in QLB (see Lifjeld and Robertson 1992). Al-
though QLB is reputed to be effective at preserving DNA in hot
climates, DNA degradation occurred in about 30% of samples.
Fingerprinting analyses were carried out at the Department of
Biology, University of Leicester, according to the methods of
Burke and Bruford (1987) and Bruford et al. (1998), and summari-
sed briefly as follows. DNA was purified from blood samples by
phenol-chloroform extraction, ethanol-precipitated and re-dis-
solved in sterile distilled water (SDW). Because DNA fingerprint-
ing depends on genomic DNA being cut only at restriction sites,
degraded samples, identified by agarose gel electrophoresis, were
not used for fingerprinting. A total of 7 µg of purified DNA, quan-
tified by fluorometry, was digested overnight at 37°C using
20 units of restriction endonuclease HinfI (Life Science Technolo-
gies). The enzyme was then removed by phenol-chloroform ex-
traction and the DNA ethanol-precipitated and redissolved in
SDW. A total of 5 µg of digested DNA was run through a 30-cm-
long 1% (w/v) TBE-buffered agarose gel at 45 V for 2,480 volt-
hours. The DNA in the gel was then depurinated, denatured and
Southern-blotted onto a Hybond-Nfp (Amersham) nylon mem-
brane. The membrane was air-dried, and the DNA cross-linked on-
to the membrane using ultraviolet illumination. To visualise the
fingerprints, the DNA on the membrane was probed separately
with radioactively labelled probes 33.6 and 33.15 (Jeffreys et al.
1985). Fuji RX X-ray film was then exposed to the membrane at
–70°C and autoradiographs of the final fingerprints were pro-
duced. Individuals to be compared were run on the same gels to
facilitate comparison. The mean±SD number of resolved bands
was 16.2±3.2 (n=60) for probe 33.6 and 19.1±5.4 (n=79) for
33.15. Bands were considered to be shared between two individu-
als if they were separated by less than 0.5 mm and they were less
than twofold different in intensity (Birkhead et al. 1990). A segre-
gation analysis (Burke and Bruford 1987) to investigate the effects
of linkage and so test hypotheses of relatedness based on Mende-
lian expectations (Birkhead et al. 1990) could not be conducted
because brood size was invariably too small.
Paternity was assessed as follows. For broods from compound
nests at which mothers and putative fathers had been blood-sam-
pled, novel bands were used to assign paternity. Such assignment
was confirmed by examining band-sharing coefficients. Band
sharing (s) between two individuals (a and b) was calculated using
the formula s=2N
ab
/(N
a
+N
b
), where N
ab
=number of shared bands
between individuals a and b, N
a
=number of bands present in the
fingerprint of individual a, and N
b
=number of bands present in the
fingerprint of individual b (Wetton et al. 1987). After accounting
for all the maternal bands in a chick’s fingerprint, the remaining,
paternally derived bands were compared with the fingerprints of
each of the putative fathers resident at that compound nest. If a
male’s fingerprint accounted for all paternally derived bands, or
all but one paternally derived bands (attributed to mutation) in a
chick’s fingerprint, he was assigned paternity. In all cases, this was
confirmed by the male and offspring having a high band-sharing
coefficient. If none of the resident males fulfilled these criteria, a
male not resident at that nest was assumed to be the father. Mean
band-sharing coefficients for first-order relatives, with 95% confi-
dence limits (CLs), were obtained from the 14 father-offspring
pairs determined by this novel band analysis.
For broods for which only the putative fathers and not the
mothers were blood-sampled, band-sharing coefficients alone
were used to assign paternity. Band-sharing coefficients were cal-
culated between each offspring and each of its putative fathers.
Any putative father that had a band-sharing coefficient with a
chick lying within the 95% CLs for father-offspring comparisons,
as described above, was assigned paternity. If none of the resident
males was assigned paternity in this way a male not resident at
that compound nest was assumed to be the father.
Relatedness between males resident at the same compound
nest was assigned as follows (see Bruford et al. 1998; Jamieson et
al. 1994). In most cases, a single band-sharing value between
males was calculated for probes 33.6 and 33.15 combined. In
some cases where we tested the paternity of candidate fathers, on-
ly 33.15 was used, but all decisions were made using CLs ob-
tained using the same probe(s). Band-sharing coefficients were
then calculated between ten independent pairs of randomly chosen
males that had been captured on different farms and were there-
fore unlikely to be related. Mean band-sharing coefficients were
calculated with 95% CLs for these unrelated birds as were 95%
CLs for band-sharing coefficients between first-order relatives.
CLs for half-siblings were obtained from comparisons of those
offspring within nests that were determined to have the same
mother but different fathers. Males were assumed to be relatives if
their band-sharing coefficients lay above both the upper CL for
unrelated birds and the lower CL for related birds. Not all the co-
alition pairs had offspring that were blood-sampled and there were
no cases in which we had to assign paternity to closely related
males. All CLs were calculated from angular-transformed data
(Sokal and Rohlf 1995).
The role of the phalloid organ
Attempts to observe and record copulations between marked indi-
viduals in the wild and establish the role of the phalloid organ
proved almost impossible because birds typically flew up to
500 m from the nest prior to copulating in dense vegetation. We
therefore captured 13 male and 6 female buffalo weavers, and un-
der licence brought them to Europe and maintained them in aviar-
ies at the Max-Planck Institute in Radolfzell, Germany. All obser-
vations of copulation refer to captive birds unless otherwise stated.
The birds settled into captivity and were maintained on an ad lib-
itum soft-bill diet, and some individuals built nests and copulated.
Even in captivity, observing the position of the phalloid organ dur-
ing copulation was not possible, even after removing the feathers
surrounding the phalloid organ and by using ‘close-up’ video.
Therefore, to establish whether the phalloid organ was intromittent
during copulation, two approaches were used. First, a piece of
card was fixed transversely to the end of the phalloid organ to pre-
vent intromission during copulation. Second, males were allowed
to copulate with a model female buffalo weaver fitted with a false
cloaca (see Pellatt and Birkhead 1994) to assess whether the phal-
loid organ was inserted into the model’s cloaca during copulation.
To help establish how the presence of the phalloid organ was relat-
ed to the unusual copulatory behaviour of the buffalo weaver, the
following aspects of copulatory behaviour were recorded: (1) the
number of mounts, (2) the duration of mounts which involved
ejaculation, (3) the duration of the whole series of mounts before
ejaculation and (4) the position of the male relative to the female
during mounting.
Where data were not normally distributed, non-parametric sta-
tistical tests were used (Siegel and Castellan 1988). Means±1 SE
are given unless otherwise stated. In captivity, 3 of 13 males per-
476
formed non-forced (solicited) copulations with live females and
also copulated with model females. Eight males performed forced
copulations with live females. We compared differences in copula-
tion behaviour between males and when significant differences oc-
curred we report these. If males did not differ significantly, we
treated copulations as independent data points in statistical tests,
although these should obviously be treated with caution.
Results
The phalloid organ
The phalloid organ (Fig. 1) was significantly longer
in males (15.7±0.29 mm, n=109) than females (6.1±
0.19 mm, n=68; Mann-Whitney U-test: U=1733,
P<0.001) (Figs. S4, S5 ESM), as noted by previous au-
thors (e.g. Bentz 1983). Among males, resident birds had
a significantly longer phalloid organ (17.4±0.53 mm,
n=24) than non-resident males (15.2±0.32 mm, n=85;
U=586, P<0.001), and among resident males those with
Fig. 1 A male buffalo weaver, showing the phalloid organ (inset
detail)
477
a harem of females had a significantly longer phalloid
organ (17.9±0.66 mm, n=7) than those without (16.4±
0.78 mm, n=17; U=20, P=0.012). However, relative
phalloid organ length was not significantly correlated
with harem size (r
s
=0.355, n=14, P=0.200).
Social interactions
Nest colonies comprised 5.7±1.10 nests (n=24). Each
nest was a compound structure containing a mean of
6.5±0.80 breeding chambers (n=16 colonies). The mean
number of reproductively active nests per nest colony
was 1.7±0.37 (n=13) and the mean number of reproduc-
tively active breeding chambers per active compound
nest in each colony was 4.6±0.68 (n=13 colonies).
At each of five colonies where males were ringed and
behavioural data were collected, only one compound
nest was reproductively active. At two of these colonies,
only one ringed male was frequently observed at the
compound nest (one-male-nests; Fig. S3 ESM). At these
two colonies, the single males were present 48% and
61% of the time, and unringed males were rarely if ever
recorded (0.1% and 0% of the time). In two other colo-
nies, two ringed males were frequently observed at the
compound nest (two-male-nests; Fig. S6 ESM). At one
of these, one male was present for 81% of the time while
the other was present for 23% of the time. At the second
colony, one male spent 55% of the time present, and the
second male spent just 13% of the time there. At both
colonies, unringed males were observed less frequently
(0.6% of the time at both colonies). At one colony, four
ringed males were frequently observed at the compound
nest, although two of these birds were observed more
frequently (male 1: 51%, male 2: 39% of the time) than
the others (male 3: 19% and male 4: 11% of the time).
Overall, two of the five compound nests appeared to be
one-male nests and three appeared to be two-male nests.
Less detailed observations at five other nests in the wild
indicated that each was a two-male nest, suggesting that
two-male-nests may be more prevalent than our detailed
data suggest.
Do males cooperate to rear offspring?
At two-male-nests, both males were observed building at
their compound nest (Table 1). This is consistent with
males either owning different breeding chambers in the
same compound nest, or sharing breeding chambers and
behaving cooperatively to rear the offspring in those
chambers. Two types of observation suggested that
males behaved cooperatively: (1) both resident males
participated in nest defence against other males (ob-
served on 41 occasions), but were never seen to attack
each other, and (2) at the one colony where we observed
chick-feeding, both males carried food for chicks into
the same breeding chambers (Table 2).
How is paternity shared between males?
Due to degradation of some DNA samples, fingerprints
were obtained for only four colonies of compound nests.
In all four, only one compound nest was reproductively
active. Fingerprints were obtained from 25 broods, com-
prising 55 offspring. Blood samples were obtained from
the mother and all the associated resident males for six
broods, comprising 16 offspring. These six broods origi-
nated from one compound nest. Three of the six broods
(50.0%) were fathered by more than one male (Table 3).
In one of these three, paternity was shared between resi-
dent males (Table 3), and in the other two broods, pater-
nity was shared between a resident male and a non-resi-
dent male. Among these six broods, resident males
gained paternity of 14 of 16 offspring (87.5%) and non-
resident males gained paternity of 2 of 16 offspring
(12.5%).
Examination of band-sharing coefficients for families
in which only the fathers had been blood-sampled re-
vealed the paternity of 39 offspring from 19 broods
(Table 4). Of these 19 broods, 6 (32%) were fathered
Table 1 Mean number of occasions per 10 minutes that each resi-
dent male built at the compound nest
Colony Male Building rate
Harry Big Dam BW 0.056
GB 0.333
Claratol Kaffee GY 0.667
PW 2.933
Du Preez 96 GG 0.840
OO 0.167
Table 2 Mean number of feeds per 10 minutes by resident males
at three breeding chambers at a single compound nest
Male Feeding rate at each chamber
123
GPW 0.043 0.070 0.208
GRG 0.021 0.014 0.224
Table 3 Assignment of paternity using novel-band analysis for
families in which mothers and putative fathers were sampled in
six nest chambers at a single compound nest
Brood Number Number of offspring fathered by
of offspring
Resident Resident Non-residents
male 1 male 2
L4 2 1 0 1
B1 3 3 0 0
L5 1 0 1 0
F2 3 0 3 0
L1 3 0 2 1
L3 4 2 2 0
478
solely by resident males, and 8 (42%) were fathered
solely by non-resident males. Seven of the 14 broods
containing more than one offspring (50%) were fathered
by more than one male; two of these broods were fa-
thered solely by the resident males and five broods had
at least one chick fathered by a non-resident male. Over-
all, resident males gained paternity of 22 of 39 offspring
(56%) and non-resident males gained paternity of 17 of
39 offspring (44%).
Combining the results of both analyses, resident
males gained paternity of 36 of 55 offspring (65%) and
non-resident males gained paternity of 19 of 55 offspring
(35%). In 17 of 25 broods (68%), some or all offspring
were fathered by one or more resident males. In 5 of
these 17 broods (29%), both resident males fathered
some offspring. In 15 of 25 broods (60%), some or all
offspring were fathered by one or more non-resident
males, and in 8 of these 15 broods (53%), non-resident
males fathered all the offspring. In three broods, both
offspring were fathered by extra-group males; we did not
attempt to determine if these pairs of chicks were full- or
half-siblings. Therefore, overall, in 10 of 16 broods
(63%) with two or more offspring, more than one male
sired offspring, indicating that sperm competition was
intense.
Are cooperative males related?
The frequency distribution of band-sharing coefficients
for the ten cooperative pairs of males is shown in Fig. 2
with 95% CLs superimposed from unrelated pairs of
males and from first-order relatives (father-offspring).
One band-sharing coefficient for one pair of males lay
above the upper 95% CL for unrelated birds and above
the lower 95% CL for first-order relatives. This indicates
that this pair of males was probably related, as either
brothers or father and son. Therefore, in only one of ten
cooperative male pairs (10%) were the two males appar-
ently closely related. However, we cannot rule out the
possibility that in some cases, including this one, the
males were more distantly related (e.g. half-sibs).
Males might be able to recognise their natal nest-
mates, and males in coalitions may therefore consist of
full- or half-siblings cooperating to obtain inclusive fit-
ness benefits. If all coalitions arose in this way, then the
distribution of kinship among coalitions should be the
same as that among chicks within nests. Here, we found
that males in coalitions had significantly lower band-
sharing similarities (mean±SD=0.32±0.10) than expect-
ed (0.56±0.12) under this scenario (taking only combina-
tions where both probes were used and a mean value
of similarity between chicks for each brood, t=–5.35,
df=25, P=0.00001). Additionally, we can estimate 95%
CLs for band-sharing between half-siblings based on
comparisons between offspring that we conclude, from
their assignment to different males, to be the result of
multiple paternity (Tables 3, 4; 0.26–0.57, n=11 for both
probes combined, taking only one randomly selected pair
of half-sib chicks per brood, or two separate pairs in two
broods of four). Comparison with the band-sharing val-
ues for cooperating males (Fig. 2) indicates that three of
Table 4 Assignment of paternity from band-sharing coefficients
between putative fathers and offspring in families in which moth-
ers were not sampled
Compound Brood Number of Resident Resident Non-
nest offspring male 1 male 2 resident
males
Dirk 95 1 2 2 0 0
22 1 1 0
33 1 1 1
Janbeh
a
12 2 – 0
22 2 – 0
31 0 – 1
42 2 – 0
52 1 – 1
Du Preez 96 L6 3 2 1 0
B2 4 2 1 1
b
F1 3 0 2 1
L2 3 1 0 2
Claratol
Dam 96
a
11 0 – 1
21 0 – 1
32 0 – 2
42 0 – 2
52 0 – 2
61 0 – 1
71 0 – 1
a
At these compound nests, there was only a single resident male:
the other compound nests had two resident males
b
For this individual, the band-sharing coefficient fell between the
confidence limits for unrelated and related birds: it was classified
as unrelated because the band-sharing coefficient lay closer to that
for unrelated birds, but this conclusion must be treated with cau-
tion
Fig. 2 Frequency distribution of band-sharing coefficients be-
tween ten pairs of cooperative males (band sharing data obtained
from probes 33.6 and 33.15 combined). The solid vertical line rep-
resents the lower confidence limit for band-sharing between full-
sibs, the dashed vertical line represents the upper confidence limit
for band-sharing between unrelated pairs of males, and the bracket
indicates the 95% confidence range for band-sharing between
half-siblings
ten were below the lower CL for half-sibs. Furthermore,
there was no difference between the similarities of males
in coalitions and those of ten unrelated pairs of males
(t=0.89, df=18, P=0.81), though this test is of low power.
We therefore conclude that at least 30% of cooperating
males were completely unrelated, and that the data are
consistent with 90% of the males in coalitions being un-
related.
Copulation and the function of the phalloid organ
A total of eight copulations was observed among wild
(unidentified) birds in the field. They occurred at a mean
distance of 280±55.0 m (range: 75–500 m) from the nest
where the birds originated. The mean number of mount-
ing attempts per copulation event was 5.8±1.63 (range:
1–14) and the mean duration of mounting attempts was
8.1±1.84 s (range: 1–16). The only copulation that took
place less than 100 m from the nest was interrupted by
another male (see also Birkhead et al. 1993a). It was im-
possible to determine how the phalloid organ was em-
ployed during copulations observed in the wild.
In captivity, the following copulations were observed:
(1) 57 solicited, unforced copulations with live females,
(2) 118 forced copulations (69 heterosexual copulations
and 49 homosexual copulations between two males) and
(3) 34 copulations with a model female. Figure 3 shows
copulation between a male and a live female in captivity
(see also video clip 1 (ESM) which shows a male pre-
copulatory display and video clip 2 (ESM) which shows
copulation with a live female).
The phalloid organ did not appear to be intromittent
during unforced copulations with live females for three
reasons. First, when intromission was prevented by at-
taching a piece of cardboard to the phalloid organ of one
male, he was still able to inseminate the female success-
fully. This indicates that insemination does not require
intromission of the phalloid organ. Second, in none of 34
copulations with a model female was a piece of sponge,
which had been placed in the neck of the false cloaca,
displaced. Third, in 89.4±5.4% (n=4 males) of occasions
after males copulated with a live female, the phalloid or-
gan neither appeared damp nor with matted feathers,
suggesting that it had not been inserted. In contrast,
when a model phalloid organ was inserted into the clo-
aca of live females, it subsequently appeared damp on
nine of ten occasions.
In captivity, all three reproductively active males ex-
perienced a form of ‘orgasm’ during copulations with
live and model females. During the ‘orgasm’, the wing
beat slowed to a quiver, the whole body shook and the
feet clenched hold of the female, the muscles apparently
in spasm. The effect of the orgasm was to pull the female
against the male at the point of ejaculation. Only after
copulations with the model in which males performed
orgasmic behaviour (n=34) was semen found in the en-
trance to the false cloaca. This suggests that orgasm is
prerequisite to ejaculation. Such orgasmic behaviour ap-
479
pears to be unique amongst birds. As the phalloid organ
is also unique to the buffalo weaver, it possibly functions
as a stimulatory organ facilitating orgasm and ejacula-
tion. The unusual copulatory behaviour of the male buf-
falo weaver may also provide stimulation to the male’s
phalloid organ to promote orgasm and ejaculation. Two
lines of observational evidence suggested this was the
case.
Fig. 3 Three frames from a video sequence of a pair of buffalo
weavers copulating in captivity, showing the extreme postures
adopted by both sexes (see also video clips 1 (ESM) and 2 (ESM))
1. Copulations which involved orgasm and ejaculation
comprised multiple mounts before orgasm occurred
(solicited copulations with live females: mean number
of mounts=24.8±3.76, n=57, range=2–159; solicited
copulations with the model female: mean number of
mounts=23.1±3.60, n=34, range=1–103. For neither
model nor live females were there any significant dif-
ferences between males; Kruskal-Wallis tests). By
contrast, forced heterosexual and homosexual copula-
tions which did not involve ejaculation only ever in-
volved a single mount (n=116). Sequences of mount-
ing attempts during solicited copulations with live fe-
males lasted 11.67±1.67 min (n=57) before ejacula-
tion occurred. Sequences of mounting attempts with
the model female lasted 31.5±5.99 min (n=30) before
ejaculation occurred (again, for both model and live
females there were no significant differences between
males; Kruskal-Wallis tests). In solicited copulations
with live females, the mean duration of those mounts
which involved ejaculation (mean duration=30.1±
2.91 s, n=25) was significantly greater than the dura-
tion of other mounts (mean duration=4.3±0.68 s,
n=25; Wilcoxon signed-rank test: z=–4.372, n=25,
P<0.001). In copulations with the model female, the
mean duration of those mounts which involved ejacu-
lation (mean duration=29.6±4.32 s, n=21; no differ-
ences between males) was significantly greater than
the duration of other mounts (mean duration=12.0±
1.58 s, n=21; no differences between males (Wilcox-
on signed-rank test: z=–3.389, n=21, P<0.001). These
observations suggest that multiple mounts and mounts
of long duration are prerequisite to orgasm and ejacu-
lation.
2. During the majority of copulations which involved
ejaculation, males adopted an unusual position, re-
peatedly leaning backwards, away from the female in
72% (n=57) of solicited copulations with live fe-
males and in 88% (n=34) of copulations with the
model female. Differences between males copulating
with live females were non-significant (χ
2
-tests), but
with model females there was a significant difference
between males (χ
2
=7.97, 2 df, P=0.02), mainly be-
cause copulations by a subordinate male were often
disrupted by other males. Males adopted the leaning-
back position less often during copulations which did
not involve ejaculation [forced copulation with fe-
male: males leaned back in 41% (n=68) of copula-
tions; forced copulation with male: males leaned
back in 31% (n=48) of copulations; no differences
between males]. Leaning away from the female may
help the male gain full stimulation along the length
of the phalloid organ to promote orgasm and ejacula-
tion. During all solicited copulations with live fe-
males (n=57), the female was pulled hard against the
male during orgasm and ejaculation. Because the
phalloid organ is not intromittent during copulation,
the female’s body would be pulled directly against
the phalloid organ and, in so doing, may provide it
with further stimulation.
To confirm whether such copulatory behaviour could
stimulate male orgasm and ejaculation, phalloid organs
were stimulated artificially. To simulate being rubbed re-
peatedly against the female’s body wall, the phalloid or-
gans of eight males were rubbed repeatedly for 25 min.
This was the average duration of the series of copulatory
mounts performed by males before ejaculation occurred.
Then, to simulate the female being pulled against the
phalloid organ, pressure was applied to the base and top
of the phalloid organ of each male. After rubbing the
phalloid organ, pressure on the base of the organ caused
some signs of orgasm (the wings extended and the feet
clenched) in six of the eight males. Although rubbing
stimulated signs of orgasm, it was not possible to stimu-
late ejaculation. However, in another experiment, three
different males were caught immediately after they had
copulated and ejaculated with live females, and their
phalloid organs were manipulated. On application of
pressure to the top of the phalloid organ, their wings ex-
tended, the birds shuddered slightly, and the feet
clenched. Then, when pressure was applied to the base
of the phalloid organ, the wings extended completely
and quivered, the whole body shuddered, the feet
clenched and the males ejaculated. This was repeated on
two separate occasions with each of three males. These
observations suggest that orgasm and ejaculation were
caused by stimulation received by the male’s phalloid or-
gan during copulation. Two stimuli were required for
ejaculation to occur: (1) repeated rubbing of the phalloid
organ for a prolonged period and (2) pressure upon a
stimulatory region at the base of the phalloid organ.
Discussion
The red-billed buffalo weaver had an unusual social or-
ganisation comprising polygyny and cooperative polygy-
nandry, involving a coalition of usually genetically unre-
lated males, who usually shared paternity. However, in
both systems, extra-group paternity also occurred, and
with 63% of multi-offspring broods exhibiting multiple
paternity, sperm competition was clearly intense. This is
consistent with our initial hypothesis that the phalloid or-
gan may have evolved in response to sperm competition.
The phalloid organ was not intromittent during copula-
tion, but appears to be a stimulatory organ facilitating
male orgasm and ejaculation. The males of no other bird
species are known to experience orgasm. However, why
buffalo weavers require a unique stimulatory phalloid or-
gan, why it takes males so long to ejaculate and how
these features are linked with intense sperm competition
remain unclear. We consider three possibilities.
Female choice for a large cloacal protuberance
Protracted copulation may allow females to discriminate
between males on the basis of either their copulatory
performance or the stimulation through their cloacal re-
480
481
gion that males provide during mounting. In passerine
birds, the male’s seminal glomera swell with spermato-
zoa during the breeding season to form a cloacal protu-
berance which can be 10–20 mm high in medium-sized
passerines (Salt 1954; Wolfson 1954; Birkhead et al.
1993b). The size of the cloacal protuberance both within
and between species reflects the number of sperm stored
in the seminal glomera and the number of sperm avail-
able for ejaculation (Birkhead et al. 1993b; 1995a).
Moreover, all else being equal, the more sperm a male
transfers at ejaculation the greater his likelihood of fertil-
isation (Parker 1990; Birkhead et al. 1995b; Colegrave et
al. 1995). Females that preferentially allow themselves
to be inseminated by males with a relatively large clo-
acal protuberance may gain both direct benefits in terms
of sperm numbers and fertility, and indirect benefits
through sons inheriting a large cloacal protuberance.
Cloacal protuberance size may therefore be under selec-
tion by female choice. The hypertrophy of the male buf-
falo weaver’s cloacal wall may exploit any female pref-
erence for a large cloacal protuberance more cheaply
than by the production of additional spermatozoa (see
Ryan et al. 1990; Enquist and Arak 1993). Females may
therefore prefer the hypertrophied cloaca of the buffalo
weaver. Such a feature may then evolve through ‘run-
away selection’ (Fisher 1958; Eberhard 1985; Andersson
1994) and may result in the eventual evolution of a phal-
loid organ. Although this explanation can potentially ex-
plain the evolution of the phalloid organ, it does not ex-
plain why the phalloid organ is stimulatory.
Prolonged copulation as a paternity guard
In a number of insects and other invertebrates, protracted
mounting and copulation are thought to have evolved as
paternity guards (Thornhill and Alcock 1983). In most
birds, copulations are rather brief, and in most passerines
typically last only a few seconds (Birkhead et al. 1987).
However, in some species, mounting can be protracted.
In vasa parrots, Caracopsis vasa and C. niger, partners
may remain in a copulatory tie for up to 1 h (Wilkinson
and Birkhead 1995). Mounting is also prolonged in the
aquatic warbler, Acrocephalus paludicola: the male re-
mains mounted on the female’s back for 30 min, during
which time he inseminates several times. In this species,
sperm competition is intense (Schulze-Hagen et al.
1993), and protracted mounting and copulation may
serve as a paternity guard by occupying females at criti-
cal times of their ovulatory cycle (Schulze-Hagen et al.
1995). Male buffalo weavers, by requiring stimulation of
the phalloid organ for ejaculation, may occupy the time
and energy of the female and thus reduce her likelihood
of copulating with other males during her fertile period.
However, the long process of copulation may be costly,
making pairs conspicuous to both other males and preda-
tors, although such an effect may be ameliorated by cop-
ulating at some distance from the colony and in dense
vegetation.
Female choice for stimulating males
The copulatory behaviour of some rodents is similar to
that of the buffalo weaver in that males typically mount
females several times prior to ejaculation (Dewsbury
1972, 1984). In rats (Rattus norvegicus), a male’s likeli-
hood of fertilisation covaries with the number of mounts
he performs during copulation (Adler 1969). This is for
two reasons: (1) the likelihood that spermatozoa are trans-
ported up the female’s reproductive tract depends upon the
number of mounts (Matthews and Adler 1978) and (2) im-
plantation of the embryo occurs only if the female is stim-
ulated appropriately during copulation (Dewsbury 1984).
Embryo implantation is irrelevant in birds. However, stim-
ulation by the male buffalo weaver may influence the
transport of spermatozoa within the female reproductive
tract, for example, by facilitating sperm transport to the
sperm storage tubules or the infundibulum, where fertil-
isation takes place (see also Edvardsson and Arnquist
2000). Male buffalo weavers which can maintain a pro-
longed series of mounts until ejaculation may therefore be
more likely to fertilise a female’s ova than males which
cannot do so. By delaying ejaculation, the phalloid organ
may facilitate stimulation of the female and increase a
male’s likelihood of fertilisation.
In summary, the function of the phalloid organ is to
facilitate male orgasm and ejaculation. It may have
evolved in response to the high level of sperm competi-
tion that exists in the red-billed buffalo weaver. The
presence of the phalloid organ may increase a male’s
likelihood of achieving successful copulation, insemina-
tion and/or fertilisation, although the mechanism remains
obscure, as does any role of the female.
Acknowledgements This study was supported by a grant from
the National Geographic Society (to T.R.B.), the Lindeth Charita-
ble Trust, the Louise Hiom Fund, the University of Sheffield
Foreign Travel and General Research Funds, an NERC grant (to
T.B.) and a University of Sheffield Postgraduate Scholarship (to
M.W.). Permission for the study was granted by the Ministry of
Environment and Tourism (Namibia), and the Ministry of Youth,
Family and Health (Hessen, Germany) and the experiments con-
ducted in this study were completed within the current laws of the
country in which they were performed. We are extremely grateful
to Peter Berthold and Bernd Leisler for very generous hospitality
at the Max-Planck Institute at Vogelwarte Radolfzell and to Karl-
Heinz Siebenrock for expert technical assistance there. We are es-
pecially indebted to John and Celia Mendelsohn for unlimited hos-
pitality and logistical help in Namibia. We are also very grateful to
Rob Simmons and Phoebe Barnard for their hospitality and assis-
tance, to the many farmers upon whose land the majority of this
work took place, particularly Helmut Finke, Eki Freyer, Wolfgang
Teubner, Harry Schneider-Waterberg and Jan Visser, and to An-
drew Krupa for technical assistance. Finally we thank two anony-
mous reviewers for their helpful comments on the manuscript.
References
Adler NT (1969) Effects of the male’s copulatory behaviour on
successful pregnancy of the female rat. J Comp Physiol Psy-
chol 69:613–622
Andersson MB (1994) Sexual selection. Princeton University
Press, Princeton, NJ
Jeffreys AJ, Wilson V, Thein SL (1985) Hypervariable ‘minisatel-
lite’ regions in human DNA. Nature 314:67–73
Jensen RAC, Clinning CF (1983) Birds of the Etosha National
Park. Directorate of Nature Conservation and Recreation Re-
sorts, Windhoek
Kemp A, Kemp M (1974) Observations on the buffalo weaver.
Bokmakierie 26: 55–58
King AS (1981) Phallus. In: King AS, McLelland J (eds) Form
and function in birds, vol 2. Academic Press, London, pp
107–147
Lesson RP (1831) Traité d’ornithologie. Bertrand, Paris
Lifjeld JT, Robertson RJ (1992) Female control of extra-pair fertil-
isation in tree swallows. Behav Ecol Sociobiol 31:89–96
Maclean GL (1973) The sociable weaver, part 3: breeding biology
and moult. Ostrich 44:219–240
Maclean GL (1993) Roberts birds of southern Africa, 6th edn.
John Voelcker Bird Book Fund, Cape Town
Matthews MK Jr, Adler NT (1978) Systematic interrelationship of
mating, vaginal plug position, and sperm transport in the rat.
Physiol Behav 20:303–309
Parker GA (1990) Sperm competition games: sneaks and extra-
pair matings. Proc R Soc Lond B 242:127–133
Pellatt, EJ, Birkhead TR (1994) Ejaculate size in zebra finches
Taeniopygia guttata and a method for obtaining ejaculates
from passerine birds. Ibis 136:97–106
Ryan M, Fox JH, Wilczynski W, Rand AS (1990) Sexual selection
for sensory exploitation in the frog Physalaemus pustulosus.
Nature 343:66–67
Salt WR (1954) The structure of the cloacal protuberance of the
vesper sparrow (Pooecetes gramineus) and certain other pas-
serine birds. Auk 71:64–73
Schulze-Hagen K, Swatschek I, Dyrcz A, Wink M (1993) Multiple
Vaterschaften in Bruten des Seggenrohrsängers Acrocephalus
paludicola: erste Ergebnisse des DNA-fingerprintings. J Orni-
thol 134:145–154
Schulze-Hagen K, Leisler B, Birkhead TR, Dyrcz A (1995) Pro-
longed copulation, sperm reserves and sperm competition in
the aquatic warbler Acrocephalus paludicola. Ibis 137:85–91
Seutin G, White B, Boag PT (1991) Preservation of avian blood
and tissue samples for DNA analyses. Can J Zool 69:82–90
Siegel S, Castellan NJ (1988) Nonparametric statistics for the be-
havioural sciences. McGraw-Hill, New York
Smith RL (ed) (1984) Sperm competition and the evolution of ani-
mal mating systems. Academic Press, Orlando, Fla
Sokal RR, Rohlf FJ (1995) Biometry, 3rd edn. Freeman, New
York
Sushkin PP (1927) On the anatomy and classification of the weav-
er birds. Bull Am Mus Nat Hist 57:1–32
Thornhill R, Alcock J (1983). The evolution of insect mating sys-
tems. Harvard University Press, Cambridge, Mass
Wetton JH, Carter RE, Parkin DT, Walters D (1987) Demographic
study of a wild house sparrow population by DNA fingerprint-
ing. Nature 327:147–149
Wilkinson R, Birkhead TR (1995) Copulation behaviour in the
vasa parrots Coracopsis vasa and C. nigra. Ibis 137:117–119
Winterbottom M, Burke T, Birkhead TR (1999) A stimulatory
phalloid organ in a weaver bird. Nature 399:28
Wolfson AW (1954) Notes on the cloacal protuberance, seminal
vesicles and a possible copulatory organ in male passerine
birds. Bull Chicago Acad Sci 10:1–22
482
Bennun L (1994) The contribution of helpers to feeding nestlings
in grey-capped social weavers, Pseudonigrita arnaudi. Anim
Behav 47:1047–1056
Bentz GD (1983) Myology and histology of the phalloid organ of
the buffalo weaver (Bubalornis albirostris). Auk 100:501–504
Birkhead TR, Møller AP (1992) Sperm competition in birds: evo-
lutionary causes and consequences. Academic Press, London
Birkhead TR, Møller AP (eds) (1998) Sperm competition and sex-
ual selection. Academic Press, London
Birkhead, TR, Atkin L, Møller AP (1987) Copulation behaviour of
birds. Behaviour 101:101–138
Birkhead TR, Burke T, Zann R, Hunter FM, Krupa AP (1990)
Extra-pair paternity and intraspecific brood parasitism in wild
zebra finches Taeniopygia guttata, revealed by DNA finger-
printing. Behav Ecol Sociobiol 27:315–324
Birkhead TR, Stanback MT, Simmons RE (1993a) The phalloid
organ of buffalo weavers Bubalornis. Ibis 135:326–331
Birkhead TR, Briskie JV Møller AP (1993b). Male sperm reserves
and copulation frequency in birds. Behav Ecol Sociobiol 32:
85–93
Birkhead TR, Fletcher F, Pellatt EJ, Staples A (1995a) Ejaculate
quality and the success of extra-pair copulations in the zebra
finch. Nature 377:422–423
Birkhead TR, Wishart GJ, Biggins JD (1995b) Sperm precedence
in the domestic fowl. Proc R Soc Lond B 261:285–292
Bruford M, Hanotte O, Brookfield JFY, Burke T (1998) Multilo-
cus and single-locus DNA fingerprinting. In: Hoelzel AR (ed)
Molecular genetic analysis of populations: a practical ap-
proach, 2nd edn. IRL, Oxford, pp 287–336
Burke T, Bruford MW (1987) DNA fingerprinting in birds. Nature
327:149–152
Colegrave N, Birkhead TR, Lessells CM (1995) Sperm prece-
dence in zebra finches does not require special mechanisms of
sperm competition. Proc R Soc Lond B 259:223–228
Collias NE, Collias EC (1978) Cooperative breeding behaviour in
the white-browed sparrow weaver. Auk 95:472–484
Crook JH (1958) Etudes sur le comportement social de Bubalornis
a. albirostris. Alauda 26:161–195
Darwin C (1871) The descent of man, and selection in relation to
sex. Murray, London
Davies NB (1992) Dunnock behaviour and social evolution. Ox-
ford University Press, Oxford
Dewsbury DA (1972) Patterns of copulatory behaviour in male
mammals. Q Rev Biol 47:1–33
Dewsbury DA (1984) Sperm competition in muroid rodents. In:
Smith RL (ed) Sperm competition and the evolution of animal
mating systems. Academic Press, London, pp 547–571
Eberhard WG (1985) Sexual selection and animal genitalia. Har-
vard University Press, Cambridge, Mass
Edvardsson M, Arnqvist G (2000) Copulatory courtship and cryp-
tic female choice in red flour beetles Tribolium castneum. Proc
R Soc Lond B 267:1–5
Enquist M, Arak A (1993) Selection of exaggerated male traits by
female aesthetic sense. Nature 361:446–448
Fisher RA (1958) The genetical theory of natural selection, 2nd
edn. Dover, New York
Hartert E (1927) Specimens of Textor niger. Ibis 37:51–53
Hoesch W (1952) Über die Funktion des phalloiden Organs.
J Ornithol 93:362–363
Jamieson IG, Quinn JS, Rose PA, White BN (1994) Shared pater-
nity among non-relatives is a result of an egalitarian mating
system in a communally breeding bird, the pukeko. Proc R
Soc Lond B 257:271–277
