Behavioral Ecology 2007 MacFarlane 21 33
Behavioral Ecology 2007 MacFarlane 21 33
Behavioral Ecology 2007 MacFarlane 21 33
doi:10.1093/beheco/arl065
Advance Access publication 20 October 2006
lthough rarely addressed in the literature, sexual inter- in other species both sexes participate in same-sex mounting
A actions between individuals of the same sex occur in birds,
with over 130 avian species worldwide being documented as
(e.g., the Tasmanian native hen, Gallinula mortierii; Ridpath
1972). Such mounting may involve cloacal contact or attempted
engaging in same-sex sexual, same-sex, or homosexual behav- cloacal contact.
iors (Bagemihl 1999). Courtship behaviors employed for adver- It seems that species may also vary in the type of same-sex
tising to prospective mates in opposite-sex sexual interactions sexual behavior expressed: in some species, only same-sex
also occur among individuals of the same sex. Same-sex court- courtship has been documented (e.g., male regent bowerbirds,
ship activities may involve elaborate displays, synchronized Sericulus chrysocephalus; Lenz 1994) or only mounting (e.g., the
dances, gift-giving ceremonies, or behaviors at specific display great crested grebe Podiceps cristatus; Bagemihl 1999), whereas
areas including bowers, arenas, or leks. Courtship behavior in other species same-sex courtship and mounting co-occur
often results in same-sex mounting and even copulatory behav- (e.g., the purple swamphen, Porphyrio porphyrio; Jamieson and
ior, and for a subset of species, same-sex pair-bonding and Craig 1987). Moreover, the frequency of occurrence varies
long-term same-sex associations have been reported (e.g., the greatly interspecifically.
black swan, Cygnus atratus; Braithwaite 1981). Although the majority of individuals that engage in
In some species, only one sex has been observed to court same-sex sexual activity usually also engage in heterosexual
the same sex (e.g., female black-faced sheathbills, Chionis interactions (Mills 1991), this pattern of behavior appears to
minor; Bried et al. 1999), whereas in other species both sexes be inconsistent with traditional evolutionary theory. What
participate in same-sex courtship (e.g., the galah, Eolophus adaptive value, if any, does interacting sexually with an indi-
roseicapillus; Rogers and McCulloch 1981). The same variation vidual of the same sex provide if the behavior does not directly
among species occurs for mounting: in some species, only one contribute to an individuals fitness?
sex has been observed to mount the same sex (e.g., male Descriptive cross-species accounts of same-sex sexual behav-
cattle egrets, Ardea ibis; Fujioka and Yamagishi 1981), whereas ior exist for mammals (Dagg 1984; Tyler 1984; Vasey 1995)
and birds (Bagemihl 1999) yet little, if any attention, has been
directed at quantitatively examining factors that may influ-
Address correspondence to G.R. MacFarlane. E-mail: geoff. ence its expression and maintenance across taxonomic
[email protected]. groups. It is not known why some species exhibit only male
Received 2 March 2006; revised 25 July 2006; accepted 5 August male (MM) or only femalefemale (FF) sexual behavior,
2006. whereas other species exhibit both. Nor is it known why
same-sex sexual behavior is common in some species and rare
The Author 2006. Published by Oxford University Press on behalf of
the International Society for Behavioral Ecology. All rights reserved.
For permissions, please e-mail: [email protected]
22 Behavioral Ecology
in others. We were interested in investigating whether 1) the interaction with adult males, or observe interactions between
reproductive social organization and 2) the developmental malefemale pairs during early development. A significant
state at hatching has any bearing on the occurrence of court- proportion of these male and female individuals exhib-
ship and/or mounting behavior between members of the ited same-sex preferences in adulthood (Adkins-Regan and
same sex, and we chose to investigate this in avian species. Krakauer 2000). The sex ratio of conspecifics and later social
The dominant social mating system of avian species may be experience has also been observed to affect sexual imprinting
broadly classified according to the degree of polygamy or, preferences in altricial species (Oetting et al. 1995). For these
more specifically, whether a species is predominantly socially various reasons, the expression of same-sex sexual behavior
monogamous, facultatively polygamous, or obligately polyga- may also be related to the state of development at hatching
mous based on the number of potential mates acquired by across species.
one or both of the sexes and the longevity/exclusivity of the We tested our hypotheses in a phylogenetic context, using
association between individuals (Owens and Hartley 1998). a data set of avian species of the world for which same-sex
Social mating system also has implications for the social sexual behavior has been reported.
structure of the species and the potential opportunity for
social and/or sexual interactions among individuals. The
temporal and spatial distribution/availability of the sexes, METHODS
the patterns of breeding and nonbreeding dispersal, and
Data collection
the degree of parental care provided by each sex are all
related to social mating system (Temrin and Tullberg 1995; A survey of the literature published in scientific journals was
Because same-sex sexual behavior has rarely been the focus ship behavior may serve other related functions such as the
of study in the ornithological literature, the occurrence of development and/or maintenance of pair bonds and some-
same-sex sexual behavior may be underreported due to sci- times serve dual or multiple nonsexual functions, though
entific or theoretical difficulties in ascribing to it reproduc- these instances were not included in our study (Armstrong
tive, social function or adaptive significance (Tyler 1984; 1965; Immelmann 1977). Mounting (pseudocopulatory) be-
Bagemihl 1999). Thus, the species list collated may represent havior describes attempted and/or achieved cloacal contact
a subset of potential species in which individuals interact or manipulation (or insertion of the intromittent organ) and
sexually with the same-sex. Furthermore, we acknowledge is similar in form to that reported for opposite-sex encounters
that expression and frequency of same-sex behavior may in for the same species (Birkhead et al. 1986; Marchant and
some instances reflect the anecdotal nature of some refer- Higgins 1990; Lombardo et al. 1994; Kaplan and Rogers
ences to same-sex sexual behavior, the contexts in which species 2001).
are studied behaviorally, and the biased way in which species Based on these definitions, 4 dichotomous variables were
are chosen to study and/or observed in the field for a variety constructed for species in which same-sex sexual behavior was
of reasons (Clarke 1997). Nonetheless, for species in which evident; MM courtship, MM mounting, FF courtship, and FF
same-sex sexual behavior has been recorded, it is instructive mounting. For each variable, 0 equates to the absence of any
to attempt to compare patterns of sexual behavior with other reports of same-sex behaviors in this species and 1 equates to
behavioral traits. Only species in which accounts of same-sex the presence of same-sex behaviors in this species.
sexual behavior could be verified with original data were
included in the data set. Wherever possible, species were Relative frequency of same-sex sexual behavior
Table 1
Survey of the expression and frequency of same-sex sexual behavior in avifaunal taxa under noncaptive conditions (N 80)
Species MMca MMmb FFcc FFmd MMFRe FFFRf POLYg A/Ph Referencesi
Struthio camelus 1 0 0 0 1 0 3 4 1
Centrocercus urophasianus 0 1 1 1 1 1 3 4 2
Biziura lobata 1 0 0 0 1 0 3 4 3
Cygnus atratus 1 1 0 0 2 0 2 4 4
Anser anser 1 1 0 0 2 0 2 4 5
Anser caerulescens 0 1 1 0 1 1 2 4 6, 7
Branta canadensis 1 0 1 1 2 2 2 4 8
Tadorna tadornoides 0 0 1 0 0 2 1 4 9
Anas platyrhynchos 1 0 1 1 1 2 1 4 7, 10, 11
Anas discors 1 0 0 0 1 0 2 4 7
Aythya affinis 0 1 0 0 1 0 2 4 7
Melanerpes formicivorus 1 1 1 1 1 1 2 1 12
Dinopium benghalense 0 1 0 0 1 0 2 1 13
Coracias cyanogaster 0 1 0 1 1 1 2 1 14
Dacelo novaeguineae 0 1 0 0 1 0 2 1 15, 16
Ceryle rudis 0 1 0 0 1 0 2 1 7
Table 1, continued
Species MMca MMmb FFcc FFmd MMFRe FFFRf POLYg A/Ph Referencesi
Corvus monedula 0 0 1 0 0 1 2 1 69
Ptiloris victoriae 1 0 0 0 3 0 3 1 70
Ptiloris magnificus 1 0 0 0 1 0 3 1 63
Sialia sialis 1 0 0 0 1 0 2 1 71
Tachycineta bicolor 0 1 0 0 1 0 2 1 72
Riparia riparia 0 1 0 0 2 0 2 1 7
Hirundo pyrrhonota 0 1 0 0 1 0 2 1 73
Pseudonigrita arnaudi 0 1 0 0 1 0 2 1 74
Euplectes orix 1 0 0 0 2 0 3 1 75
Fringilla coelebs 1 0 1 0 1 1 1 1 7
Xanthocephalus xanthocephalus 0 1 0 0 1 0 3 1 76
Statistical analyses rent popular approaches to the analysis of such data have
mainly used Felsensteins (1985) phylogenetically indepen-
Ordinary linear modeling techniques assume that the resid-
dent contrasts (or equivalently, generalized least squares linear
uals are independent and normally distributed with zero
modeling methods; Garland and Ives 2000; Rohlf 2001). These
mean. However, in situations where the errors are not nor-
mally distributed, linear models can be generalized to cases techniques are appropriate if the residuals are normally distrib-
where the residuals come from some other distribution in the uted. When the residuals are not independent and/or nor-
exponential family (such as Binomial, Poisson, etc.), produc- mally distributed and GLMs are the method of choice, GLMs
ing generalized linear models (GLMs; McCullagh and Nelder are most often fitted using generalized estimating equation
1989). We used GLMs to examine the relationship between (GEE) methods (Liang and Zeger 1986; Hardin and Hilbe
the occurrence and frequency of same-sex sexual behaviors 2003). GEE methods employ a GLM approach, incorporating
(assuming statistical independence) and our independent the phylogenetic relatedness among species as a correlation
variables (degree of polygamy and state of development of matrix, to specify dependence among observations in the
young) and to perform comparisons among differing degrees model. Paradis and Claude (2002) first introduced GEE meth-
of polygamy and developmental state of young. We assumed ods for phylogenetic comparative analyses. Traits analyzed via
the Poisson distribution for categorical dependent variables this procedure may be continuous, categorical, or dichoto-
(frequency of same-sex sexual behavior) and the binomial mous, and two or more traits may be analyzed simultaneously.
distribution for dichotomous dependent variables (same-sex The derivation of the correct df for statistical hypothesis
courtship, same-sex mounting). Significant differences among tests is an issue in phylogenetic GEE models. The problem
levels of our independent variables were assessed via the Wald arises because there is only one single cluster of observations
statistic (compared with z or t distribution dependent on sam- in a phylogenetic GEE model (i.e., one phylogeny), whereas
ple size) (Harrell 2001). GEE models are usually fitted to data from multiple clusters.
Usual statistical methods assume that each data point is an Paradis and Claude (2002) proposed a heuristic method to
independent sample. However, closely related species may not approximate the appropriate df using the phylogeny, based
be statistically independent as they share a common evolu- on the ratio of the distance of nodes to tips to the distance
tionary history (Harvey and Pagel 1991). Recent studies sug- from the root to tips. Their method performed well in simu-
gest that this lack of independence is an important property lations. We used the method of Paradis and Claude (2002) in
of virtually all cross-species data sets (Freckleton et al. 2002; order to examine how the phylogenetic correlations among
Blomberg et al. 2003), although correlations due to phylogeny species affected our conclusions from the analyses that treated
may be fewer for behavioral data (Blomberg et al. 2003). Cur- species as independent.
26 Behavioral Ecology
A matrix of fixed correlations specifying dependence was significantly higher in species exhibiting obligate polygamy
among species observations was derived from shared branch compared with monogamy or facultative polygamy (Table 2,
lengths (fixed branch lengths 1) based on the phylogenetic species data; Figure 1a). MM mounting was also significantly
topology of Sibley and Ahlquist (1990) above the family level. higher in species exhibiting facultative polygamy compared
Relationships among genera and among species within genera with those exhibiting monogamy or obligate polygamy (Table
were based on Sibley and Ahlquist (1990) where possible, sup- 2, species data; Figure 1b). The frequency of MM sexual be-
plemented with a composite of several phylogenies (Strauch havior increased with degree of polygamy. Frequency of oc-
1985; Livezey 1986; Sheldon 1987; Christidis and Schodde currence of MM sexual behavior (predominantly courtship)
1992; Sheldon and Winkler 1993; Bjorklund 1994; Crochet was significantly higher in obligate polygamy compared with
et al. 2000) and the taxonomy of Sibley and Monroe (1990) monogamy (Table 2, species data; Figure 1c).
(N 80, phylogenetic df 30.49) (see Appendix 1, electronic When the data were corrected for phylogenetic relatedness,
Supplementary data). The resultant matrices were again fit- a number of relationships became nonsignificant. The signif-
ted, specifying Poisson or binomial distributions. Significant icant differences remaining were a significantly greater occur-
differences among parameter estimates for each category of rence of MM mounting in facultative polygamy compared
the degree of polygamy and developmental state of young at with monogamy and significantly greater frequency of
hatching were again assessed via the Wald statistic. Analyses occurrence of MM sexual behavior in obligate polygamy com-
were performed using R (R Development Core Team 2005) pared with monogamy and facultative polygamy (Table 2,
with the Analyses of Phylogenetics and Evolution (APE 1.3-1) contrast data).
package (Paradis et al. 2004).
Table 2
Wald statistics for same-sex sexual behavior parameter estimates employing species level (raw) data (z)
and contrast data (t), assessing differences among levels of the independent variable the degree of
polygamy
MM sexual behavior and state of development of in semialtricial species compared with altricial species (Table
young at hatching 3, species data; Figure 2b).
When the data were corrected for phylogenetic relatedness,
Expression of MM sexual behavior was equally distributed
the significantly greater occurrence of MM mounting in semi-
across categories of developmental state at hatching. No sig-
nificant differences among developmental state categories altricial compared with altricial species was maintained. In
were observed in neither the expression of MM courtship addition, there were significantly greater occurrences of MM
nor the frequency of MM sexual behavior (Figure 2a,c). How- mounting in semiprecocial and precocial species when com-
ever, the occurrence of MM mounting was significantly higher pared with altricial species (Table 3, contrast data).
28 Behavioral Ecology
FF sexual behavior and state of development of compared with altricial species (Table 3, species data;
young at hatching Figure 2f).
When the data for females were corrected for phylogenetic
The occurrence of FF courtship was significantly higher in
relatedness, a subset of relationships were maintained. The
precocial than altricial species (Table 3, species data; Figure
2d). The expression of FF mounting was significantly higher expression of FF courtship was significantly greater in semi-
in semiprecocial and precocial species compared with altricial precocial and semialtricial than altricial species. FF mounting
species (Table 3, species data; Figure 2e). was significantly higher in precocial, semiprecocial, and semi-
The frequency of FF sexual behavior was significantly altricial species compared with altricial species. The frequency
higher in precocial, semiprecocial, and semialtricial of FF sexual behavior was significantly higher in semiprecocial
MacFarlane et al. Same-sex sexual behavior in birds 29
Table 3
Wald statistics for same-sex sexual behavior parameter estimates employing species level (raw) data (z)
and contrast data (t), assessing differences among levels of the independent variable the state of
development of young
species compared with altricial species (Table 3, contrast females also court and mount individuals of the same sex, al-
data). though the expression and frequency of occurrence is greater
for females. For a number of monogamous species, longer term
FF bonding and FF coparenting of offspring from malefemale
DISCUSSION
(MF) extrapair copulations occur (e.g., Larus novaehollandiae;
Our analyses suggest that consistent patterns exist between Mills 1991).
the expression/frequency of same-sex sexual behavior and Why was FF sexual behavior observed more frequently in
social mating systems in avian taxa and that the interplay monogamous species when traditional selection theory would
among these variables differs in males and females. Expres- predict otherwise? Adaptive hypotheses offered argue that
sion of FF sexual behavior, that is, mounting and especially FF FF pairings may arise in species in which biparental care is
courtship, was highest in monogamous species and signifi- required for survival of offspring and when opportunities
cantly lower in species exhibiting facultative and obligate po- for malefemale pairing are limited demographically and/or
lygamy. As the degree of polygamy increased, the frequency of behaviorally (Conover 1984). A pattern has been observed in
FF sexual behavior decreased, being especially low in species many representatives of Laridae in which supernormal clutches
exhibiting obligate polygamy. The opposite trend was found (associated with FF pairing, courtship, mounting, and copar-
for males. Although the occurrence of MM courtship and the enting) are more frequent when the population exhibits a
frequency of MM sexual behavior (courtship and/or mount- female-biased sex ratio (Hatch 1993; Bagemihl 1999; Nisbet
ing) were distributed across all mating systems, both were and Hatch 1999). Perhaps, the most compelling evidence in
lower in monogamous systems. The frequency of MM sexual support of this sex ratio bias hypothesis for FF pairs is avail-
behavior increased significantly with the degree of polygamy. able from formal hypothesis testing conducted on colonies
Most relationships for FF sexual behavior with social mating of the ring-billed gull, Larus delawarensis, and the Californian
system were maintained after phylogenetic correction. Con- gull, Larus californicus. These studies experimentally manipu-
versely, some relationships for males were maintained, lated the sex ratio to produce a female bias by removing
whereas others became nonsignificant suggesting some pat- males. The frequency of supernormal clutches of 46 eggs
terns were an artifact of phylogenetic relatedness. MM mount- (a measure or index of FF pairings) was greater in colonies
ing remained significantly higher in species displaying from which males had been removed than it was in control
facultative polygamy, and the frequency of MM sexual behav- colonies (Conover and Hunt 1984). It may be advantageous
ior remained significantly greater in obligate polygamy com- for females to pair with other females and coparent while
pared with monogamy and facultative polygamy after seeking extrapair reproductive opportunity with paired males
controlling for the effects of phylogeny. Most species within (to obtain fertilized eggs). Mills (1991) found that although
the obligate polygamy category exhibited MM courtship, yet FF pairs in silver gulls, L. novaehollandiae, have a lower overall
few exhibited MM mounting. Many of the species within this lifetime reproductive success than MF pairs (0.85 FF vs. 1.91 MF,
category are the closely related bowerbirds and birds of para- mean number of offspring fledged in lifetime), over 50% of
dise, which may explain variation in significant relationships females in populations do not breed at all. The effects of de-
observed after phylogenetic consideration. mographic biases may hold promise in terms of adaptive explan-
In socially monogamous systems, both sexes benefit from ations for the formation and maintenance of FF associations
socioreproductive exclusivity (Reichard 2003) and the fre- within these monogamous species, especially when opportuni-
quency of pair-specific display, coordinated displays, courtship, ties for MF extrapair copulation exist outside the pair bond
sexual solicitation, and mounting are more often bidirectional (Birkhead and Mller 1995).
and occur before, during, and after pairing (Bastock 1967; The frequency of MM sexual behavior was lower in monog-
Wachtmeister 2001). Under these conditions, both males and amous species, and increased as the degree of polygamy
30 Behavioral Ecology
increased, as would be predicted within a traditional adaptive vide care of offspring. Alternatively, there is always the possi-
framework. This pattern is opposite to the pattern for females. bility that dispersal makes it less likely that observers will
Unlike FF sexual behavior though, which is predominantly detect same-sex interactions among females. It is also interest-
restricted to monogamous species and almost absent in polyg- ing to note that obligately polygamous species that display
amous species, MM sexual behavior does occur across all spe- variations of polyandry as a reproductive strategy are largely
cies. Some species that are primarily socially monogamous absent from the data set, with the exception of Struthio camelus
(within the context of other polygamous associations) do (predominantly polygynandrous; Kimwele and Graves 2003),
display MM sexual behavior, form long-term MM pairs, and and that in this species FF sexual behavior has not been ob-
may even coparent (e.g., A. anser, Huber and Martys 1993; served although we would perhaps predict otherwise.
C. atratus, Braithwaite 1981). In the black swan, C. atratus, An alternative interpretation of patterns consistent with se-
males may temporarily associate with a female, mate, and then lection theory may be that the sex that actively solicits in het-
break the bond once the eggs have been laid or they may erosexual interactions, that is, the competitive sex, is the one
usurp nests or steal eggs from MF clutches. These male pairs participating in same-sex sexual behavior. For polygynous spe-
perform parenting duties, including incubation and caring cies, this role is played by males and for polyandrous species
for cygnets (Braithwaite 1981). females. For monogamous species, both sexes may solicit.
Most species displaying facultative polygamy are also so- Thus, MM interactions would be predicted to occur more
cially monogamous with varying degrees of polygamy and frequently in polygynous species (as observed in our data
cooperative breeding arrangements, although MF extrapair set) and FF interactions to occur more frequently in polyan-
copulations are predominantly male solicited (e.g., Alca torda; drous species. In the absence of polyandrous species in the
in birds, enough variation is present to permit examination Bolhuis JJ. 1991. Mechanisms of avian imprinting: a review. Biol Rev.
of same-sex sexual behavior in relation to mating system 66:303345.
and developmental state at hatching and to document social Bossema I, Roemers E. 1985. Mating strategy including mate choice,
conditions under which it is more likely to occur for both in mallards. Ardea. 73:147157.
males and females. The frequency of MM sexual behavior in- Braithwaite LW. 1981. Ecological studies of the black swan. III. Behav-
iour and social organisation. Aust Wildl Res. 8:135146.
creased with degree of polygamy, although occurred across all Bried J, Duriez O, Juin G. 1999. A first case of female-female pairing in
mating systems. FF sexual behavior occurred predominantly the black-faced sheathbill, Chionis minor. Emu. 99:292294.
in socially monogamous species. FF sexual behavior was also Brown LH, Urban EK, Newman K. 1982. The birds of Africa. London:
more likely to occur in species in which monogamy occurs Academic Press.
together with the production of precocial offspring; that Buchanan MO. 1966. Homosexual behaviour in wild orange-fronted
is, in monogamous species that are exceptions to the more parakeets. Condor. 68(4):399400.
common altricial mode of development. The social context Campbell K. 1993. Hammerkops. East Afr Nat Hist Soc Bull. (January
conferred by a particular mating system and mode of devel- April): 11.
opment are likely contributing factors to the expression and Castro I. 2001. The stitchbird, Notiomystis cincta. In: Higgins PJ, Peter
frequency of same-sex sexual behavior. In some cases, same- JM, Steele WK, editors. Handbook of Australian, New Zealand and
sex sexual behavior may serve social and/or sexual functions Antarctic birds. Vol. 5. Melbourne (Australia): Oxford University
Press. p. 954966.
not necessarily related to reproduction, confer indirect fitness
Christidis L, Schodde R. 1992. Relationships among the birds-of-
benefits, and/or be adaptively neutral, co-occurring within paradise (Paradisaeidae) and bowerbirds (Ptilonorhynchidae), pro-
the context of reproductive opportunity. Attempting to as- tein evidence. Aust J Zool. 40:343353.
Fujioka M. 1988. Extrapair copulations in little egrets (Egretta garzetta). Livezey BC. 1986. A phylogenetic analysis of recent anseriform using
Paper presented at annual meeting of the Animal Behaviour Soci- morphological characters. Auk. 103:737754.
ety; Aug; Missoula (MT): University of Montana. 83 p. Lombardo MP, Bosman RM, Faro CA, Houtteman SG, Kluisza TS.
Fujioka M, Yamagishi S. 1981. Extramarital and pair copulations in the 1994. Homosexual copulations by male tree swallows. Wilson Bull.
cattle egret. Auk. 98:134144. 106:555557.
Garland TJ, Ives AR. 2000. Using the past to predict the present: Longcore JR. 1967. Unusual behaviour of the yellow-headed black-
confidence intervals for regression equations in phylogenetic com- bird. Wilson Bull. 79(1):117118.
parative methods. Am Nat. 155:346364. Lowe VT. 1966. Notes on the musk duck, Biziura lobata. Emu. 65:
Garnett ST. 1978. The behaviour patterns of the dusky moorhen, Gal- 279289.
linula tenebrosa Gould (Aves: Rallidae). Aust Wildl Res. 5:363384. MacRoberts MH, MacRoberts BR. 1976. Social organization and
Gowaty P. 1996. Battle of the sexes and the origin of monogamy. In: behaviour of the acorn woodpecker in central coastal California.
Black J, editor. Partnerships in birds: the study of monogamy. Ornithol Monogr. 21:1115.
Oxford: Oxford University Press. p. 2152. Makkink GF. 1942. Contribution to the knowledge of the behaviour of
Grafen A. 1989. The phylogenetic regression. Philos Trans R Soc Lond oyster-catcher (Haematopus ostralegus). Ardea. 31:2374.
B Biol Sci. 326(1233):119157. Marchant S, Higgins PJ. 1990. Handbook of Australian, New Zealand
Hale WG, Ashcroft RP. 1982. Pair formation and pair maintenance in and Antarctic birds. Vol. 1. Ratites to ducks. Melbourne (Australia):
the redshank, Tringa tetanus. Ibis. 125:323. Oxford University Press.
Hardin JW, Hilbe JM. 2003. Generalized estimating equations. New Marshall AJ. 1954. Bower birds: their displays and breeding cycles.
York: Chapman & Hall/CRC. Oxford: Oxford University Press.
Harrell FE Jr. 2001. Regression modelling strategies: with applications McCullagh P, Nelder JA. 1989. Generalized linear models. New York:
to linear models, logistic regression, and survival analysis. New York: Chapman & Hall/CRC.
Poole A, Stettenheim P, Gill F. 1994. The birds of North America: life Snow BK. 1972. A field study of the calfbird Perissocephalus tricolor. Ibis.
histories for the 21st century. Philadelphia (PA): Academy of 114:139162.
Natural Sciences and American Ornithologists Union. Snow DW. 1963. The display of the blue-backed manakin, Chiroxiphia
R Development Core Team. 2005. R: a language and environment for pareola. Zoologica. 48:167176.
statistical computing [Internet]. Vienna (Austria): R Foundation Stiles FG. 1982. Aggressive and courtship displays of the male annas
for Statistical Computing. [cited 2006 Feb 24]. ISBN 3-900051-07-0. hummingbird. Condor. 84:208225.
Available from: http://www.R-project.org. Stiles FG, Wolf LL. 1979. Ecology and evolution of lek mating behav-
Ramo C. 1993. Extra-pair copulations of gray herons nesting at high iour in the long-tailed hermit hummingbird. Ornithol Monogr.
densities. Ardea. 81:115120. 27:177.
Reed CEM. 1993. Black-winged stilt and black stilt. In: Marchant S, Strahan R. 1996. Finches, bowerbirds & other passerines of Australia.
Higgins PJ, editors. Handbook of Australian, New Zealand and Pymble (Australia): Angus & Robertson.
Antarctic birds. Vol. 2. Melbourne (Australia): Oxford University Strauch JG. 1985. The phylogeny of the Alcidae. Auk. 102:520539.
Press. p. 758780. Temrin H, Tullberg B. 1995. A phylogenetic analysis of the evolution
Reichard U. 2003. Monogamy: past and present. In: Reichard U, of avian mating systems in relation to altricial and precocial young.
Boesch C, editors. Monogamy: mating strategies and partner- Behav Ecol. 6:296307.
ships in birds, humans and other mammals. Cambridge (UK): Trail PW, Koutnik DL. 1986. Courtship disruption at the lek in the
Cambridge University Press. p. 325. Guianan cock-of-the-rock. Ethology. 73:197218.
Reynolds JD. 1996. Animal breeding systems. TREE. 11(2):6872. Tyler PA. 1984. Homosexual behaviour in animals. In: Howells K,
Ridpath MG. 1972. The Tasmanian native hen, Tribonyx mortierii. I. editor. The psychology of sexual diversity. Oxford: Blackwell Scien-
Patterns of behaviour. CSIRO Wildl Res. 17:151. tific. p. 4262.
Riggert TL. 1977. The biology of the mountain duck on Rottnest Van Rhijn J, Groothuis T. 1985. Bi-parental care and the basis for
Island, western Australia. Wildl Monogr. 52:167. alternative bond types among gulls, with special reference to