The Condor 104:240±247 ᭧ The Cooper Ornithological Society 2002

CAVITY ADOPTION AND THE EVOLUTION OF COLONIALITY IN CAVITY-NESTING BIRDS

JESSICA R. EBERHARD1 Smithsonian Tropical Research Institute, Apdo. 2072, Balboa, Republic of Panama

Abstract. Among cavity-nesting birds, a distinction can be made between excavators, which dig their own cavities, and cavity-adopters, which nest in pre-existing cavities. His- torically, these two types of species have been grouped together as ``cavity-nesters,'' but it has become clear that the two nesting habits are associated with very different suites of life- history characters. This paper tests the hypothesis that cavity-adopters differ from excavators and other nest-building species in their propensity to evolve colonial breeding. Because of their dependence on pre-existing cavities, cavity-adopters presumably have less control than excavators over the location of their nests, and this could limit their ability to nest near conspeci®cs. A literature survey of the nesting behavior of 842 species in 17 bird families shows that coloniality almost never occurs in species that are obligate cavity-adopters. A phylogeny-based comparative analysis of nesting behavior in the Anseriformes indicates that in this group, colonial breeding has evolved less frequently in lineages of cavity-adopt- ers than would be expected by chance. Together, this evidence supports the hypothesis that colonial breeding systems are unlikely to evolve in lineages of cavity-adopters. Key words: Anseriformes, birds, cavity nesting, colonial nesting, comparative method.

EvolucioÂn de la Colonialidad en Aves que Nidi®can en Cavidades Resumen. Las aves que nidi®can en cavidades pueden dividirse en dos grupos: las es- pecies excavadoras, que construyen sus propias cavidades, y las no-excavadoras, que nidi- ®can en cavidades pre-existentes. HistoÂricamente, estos dos tipos de especies han sido agru- padas colectivamente como aves que anidan en cavidades, pero estudios recientes demues- tran que los dos tipos de nidi®cacioÂn estaÂn asociados con diferentes caracteres de sus his- torias de vida. En este estudio se pone a prueba la hipoÂtesis que especies no-excavadoras di®eren de las excavadoras y de otras especies que construyen nidos en cuanto a su pro- pensioÂn a evolucionar haÂbitos de reproduccioÂn colonial. Debido a que dependen de cavi- dades pre-existentes, las especies no-excavadoras presumiblemente tienen menos control que las excavadoras sobre la ubicacioÂn de sus nidos, y esto podrõÂa limitar su habilidad para nidi®car cerca de otros miembros de su especie. Una revisioÂn bibliogra®ca sobre el com- portamiento de nidi®cacioÂn de 842 especies pertenecientes a 17 familias demuestra que casi nunca se observa colonialidad en especies que obligatoriamente utilizan cavidades pre- existentes. Un anaÂlisis comparativo utilizando una ®logenia y datos del comportamiento de nidi®cacioÂn de miembros del orden Anseriformes indica que la colonialidad ha evolucionado con menor frecuencia de lo esperado al azar en linajes de especies no-excavadoras que utilizan cavidades. En conjunto, estos datos apoyan la hipoÂtesis que los sistemas de nidi®- cacioÂn colonial tienen baja probabilidad de evolucionar en linajes de no-excavadoras que anidan en cavidades.

INTRODUCTION banks, or in termitaria; cup and domed nests built with materials such as twigs, grass, mud, The great diversity of nests used by birds in- and spider silk; and elaborately woven nests like cludes the huge mounds of soil and decaying those built by weaverbirds, orioles, and oropen- vegetation used by megapodes; cavities in trees dolas (Collias and Collias 1984). Among cavity- or among rocks; burrows in the ground, in nesters, a distinction can be made between ex- cavators (which dig their own cavities) and cav- ity-adopters (which nest in pre-existing cavities). Manuscript received 11 June 2001; accepted 11 Jan- These two types of cavity-nesters have histori- uary 2002. cally been grouped together as having a single 1 Present address: Macaulay Library of Natural Sounds, Cornell Laboratory of Ornithology, 159 Sap- nest type, but it has become clear that the dis- sucker Woods Rd., Ithaca, NY 14850. E-mail: jre24@ tinction is an important one to make when study- cornell.edu ing the evolution of cavity-nesting lineages.

[240] EVOLUTION OF COLONIALITY IN CAVITY-NESTERS 241

Excavators differ from cavity-adopters in a of feeding territories, with aquatic habitat, and number of life-history traits, including nesting with nest exposure to predators (Rolland et al. success, clutch size, duration of incubation and 1998). There are several possible advantages to nestling periods, annual productivity, and adult nesting in colonies, including access to and se- survival (Martin and Li 1992). In a study of life- lection of favorable breeding sites, enhanced history traits of North American Passeriformes foraging ef®ciency, improved detection of and and Piciformes, Martin (1995) found that vari- defense against predators, and access to breed- ation in fecundity and adult survival is better ing partners (Siegel-Causey and Kharitonov explained by nest sites, and their associated nest 1990). Breeding pairs probably use the presence predation, than by food limitation. In fact, when or reproductive success of conspeci®cs as cues adult survivorship and annual fecundity are to indirectly assess the commodities necessary compared among species representing an array to breed (e.g., food, safety, health, mates, breed- of nest types, excavators rank among the highest ing sites), and then choose breeding sites that in survivorship and lowest in fecundity, while are near those of successful conspeci®cs (Dan- cavity-adopters generally have low survivorship chin and Wagner 1997). However, even if the and high fecundity (Martin 1995). In the case of above factors favor gregarious nesting, colonia- cavity-adopters, larger clutch sizes may have re- lity can only result if nesting substrates are sulted from the historically limited availability available in large enough patches to accommo- of nest sites rather than from nest failure rates date large groups. Because pre-existing cavities or food availability. Therefore, a consequence of suitable for breeding are not likely to be dense, the dependence on existing cavities may be se- obligate cavity adoption might be expected to lection for increased reproductive effort when a preclude the evolution of group nesting even if nest site is obtained (Martin and Li 1992, Martin selection favored group nesting. 1995). This paper examines the hypotheses that A recent paper by Collias (1997) argued that among cavity-nesting birds, excavators and cav- the diversity of nests built by passerine birds is ity-adopters differ in their propensity to evolve a key to explaining the adaptive radiation of this colonial breeding, and that obligate cavity adop- avian group during the late Tertiary. Collias em- tion tends to block the evolution of coloniality. phasized the importance of the ¯exibility in nest- I review data on the nest types and nest disper- site choice that accompanies the ability to build sion of several avian groups that include large a variety of nest types, and linked this ¯exibility numbers of cavity-nesting species, to examine with the radiation of passerines into new niches. the evidence that coloniality is unusual in cavi- Another context in which ¯exibility in nest-site ty-adopting species. Recognizing the importance choice might be important is in the evolution of of taking phylogenetic relationships into account colonial nesting (e.g., Eberhard 1998), particu- when testing this kind of evolutionary hypoth- larly among cavity-nesters, in which the evolu- esis, I also present a phylogeny-based compar- tion of social behavior may be affected by ative analysis of nesting behavior in the Anse- whether a species is an excavator or a cavity- riformes that tests the hypothesis that coloniality adopter. For coloniality to evolve, one funda- is unlikely to evolve among non-excavating cav- mental condition must be met: breeding pairs ity-adopters. must be able to nest in proximity to other breed- ing pairs. Because they depend on pre-existing METHODS cavities, cavity-adopters have much less control I obtained information on nesting behavior from over the location of their nests than do excava- the literature, and followed Sibley and Monroe's tors, and the breeding densities of obligate cav- (1990) taxonomic classi®cation of birds in tal- ity-adopters can be limited by the availability of lying the data. Rather than attempting to com- nest sites (Brawn and Balda 1988). Flexibility plete an exhaustive survey, I chose avian groups and control over nest placement could allow gre- on the basis of two criteria: that they contain a garious nesting by making it easier for breeding substantial number of cavity-nesting species, pairs to build nests near conspeci®c pairs. and that published information on their nesting A comparative analysis of coloniality in birds behavior be readily available. For each species has shown that coloniality is a relatively labile in the groups examined, I collected the follow- trait, and that it is correlated with the absence ing data: nest type (open nest or cavity nest), 242 JESSICA R. EBERHARD type of cavity-nester (excavator or cavity-adopt- MINSTATE and MAXSTATE functions; see er), and breeding dispersion (solitary or colo- Maddison and Maddison 1992). The observed nial). In species that were variable in a given numbers of gains and losses in coloniality were character, the predominant character state was obtained by mapping the dependent variable used (e.g., solitary nesters that occasionally form (nest dispersion) on the resolved tree; when nec- loose groups were coded as solitary). However, essary, ambiguity was resolved using the DEL- in cases where two character states appeared to TRAN option. For each set of calculations, I re- be equally common in a species, the character corded the probability of having as many or was coded as polymorphic. Species for which more than the observed number of gains in co- data on both nest type and nesting dispersion loniality, and as many or more than zero losses. were not available were omitted from the anal- Because the CCT only tests for a correlation, yses. In addition, one duck species (Heteronetta and not a cause-effect relationship between char- atricapilla, Anatidae) was omitted, since it a acters, I also tested the alternative hypothesis brood parasite. that obligate cavity-adoption is unlikely to I chose to study Anseriformes because a com- evolve in colonial lineages. I performed the test plete genus-level phylogeny is available for the using actual changes, repeating it 25 times using group, data on nesting are readily available, and different randomly resolved trees (the test was the relatively large amount of variation in both not repeated more than 25 times, as it became nest type and nesting dispersion makes the clear that further repetitions would not yield a group an ideal one for comparative analysis. Un- P-value that approached signi®cance). like most other avian orders, in Anseriformes The second comparative test that I used to test both coloniality and cavity-nesting have arisen the hypothesis that coloniality is more likely to multiple times, providing a large enough sample evolve in lineages that are not cavity-adopters is size and power to conduct a statistical test. Us- Pagel's (1994) method for the comparative anal- ing MacClade (Maddison and Maddison 1992), ysis of discrete characters. This method permits I redrew Livezey's (1986) phylogeny and ex- tests of hypotheses of correlated character evo- panded it by adding the corresponding species lution by comparing the ®t of two maximum- to the end of each genus branch. The node from likelihood models to the observed data, one that which each genus-group of species emerged was speci®es independent evolution of the two char- coded as a soft polytomy. I then mapped two acters, and the other that allows correlated evo- binary nesting characters onto the tree: nest type lution (comparison of the two models is called (0, nonexcavated cavity; 1, excavated cavity or the ``omnibus test''). When a model is ®tted to open nest), and nest dispersion (0, solitary; 1, the data, a set of transition rate probabilities is colonial). estimated; the parameters correspond to the state I used two different comparative tests to eval- changes among the four states that are possible uate the hypothesis that cavity adoption tends to with two binary characters. block the evolution of coloniality. The ®rst, With this method, it is possible to test whether MacClade's concentrated changes test (CCT; change in one character is contingent on the Maddison 1990), tests whether changes in a bi- state of the other. This is done by forcing the nary character (in this case the appearance of relevant transition parameters to be equal, and coloniality) occur more often than expected by then using a likelihood-ratio test (distributed as chance in particular parts of the phylogeny (in a ␹2 with 1 df) to compare the likelihood of this this case, in lineages that are not composed of model to the free-parameter correlated evolution obligate cavity adopters). Since the CCT can model. It is also possible to determine whether only be applied to dichotomous trees, the anse- a transition parameter is signi®cant by setting it riform tree was resolved arbitrarily using equal to zero and comparing the resulting like- MacClade's random polytomy resolution func- lihood with the likelihood of an unrestricted tion, and the test was repeated 100 times using model. 100 different randomly resolved trees. Probabil- Unlike Maddison's CCT, Pagel's method takes ities were calculated using 1000 simulations into account branch lengths in the phylogenetic with either ancestral state allowed. For each tree, tree under consideration, and represents the re- I ran simulations examining actual changes, and constructed character states on the tree's internal also reconstructed changes (with both the nodes as probability distributions rather than a EVOLUTION OF COLONIALITY IN CAVITY-NESTERS 243

TABLE 1. Taxa included in the survey of nesting behavior, and the numbers of cavity nesting and colonial species in each group. Sibley and Monroe's (1990) classi®cation was used in tallying the data. The number of species indicates the number included in the analysis. References: (1) Cramp and Simmons (1997), (2) del Hoyo et al. (1992), (3) Frith (1967), (4) Forshaw (1989), (5) Fry et al. (1992), (6) Gaston and Jones (1998), (7) Marchant and Higgins (1990), (8) Gomez-Dallmeier and Cringan (1990), (9) Johnsgard (1978), (10) Kemp (1995), (11) Madge and Burn (1988), (12) Turner and Rose (1989), (13) Warham (1990), and (14) Winkler et al. (1995).

Number of Colonial Number species Cavity cavity of analyzed nesters Colonial nesters Order Family genera (excludeda) (adopters) nesters (adopters) Reference Struthioniformes Apterygidae 1 3 (0) 3 (0) 0 0 (0) 7 Anseriformes Anhimidae 2 3 (0) 0 (0) 0 0 (0) 2 Anseranatidae 1 1 (0) 0 (0) 1 0 (0) 9 Dendrocygnidae 2 9 (0) 3 (3) 0 0 (0) 2, 3, 8, 9 Anatidae 39 129 (0) 22 (22) 12 0 (0) 1, 2, 11 Piciformes Picidae 26 131 (83) 131 (2) 3 3 (0) 14 Bucerotiformes Bucerotidae 8 54 (0) 54 (54) 0 0 (0) 10 Bucorvidae 1 2 (0) 2 (2) 0 0 (0) 10 Coraciiformes Coraciidae 2 10 (2) 10 (9) 0 0 (0) 5 Alcedinidae 3 17 (7) 17 (0) 0 0 (0) 5 Dacelonidae 9 33 (28) 33 (1) 0 0 (0) 5 Cerylidae 3 7 (2) 7 (0) 0 0 (0) 5 Meropidae 3 22 (4) 22 (0) 9 9 (0) 5 Psittaciformes Psittacidae 66 231 (118) 230 (207) 3 2 (1) 4 Ciconiiformes Laridae [Alcinae] 12 22 (1) 17 (13) 20 17 (12) 6 Procellariidae 22 90 (24) 75 (5) 85 71 (0) 13 Passeriformes Hirundinidae 14 78 (1) 44 (23) 16 6 (0) 12 Totals 214 842 (270) 670 (341) 149 108 (13) a Number of species for which data on nest type or nesting dispersion were missing. single set of inferred (parsimony or outgroup- available from the author. Data from the survey based) values. For this analysis I used an anse- are consistent with the hypothesis that colonial riform tree for which polytomies had been ran- nesting does not tend to occur in lineages of cav- domly resolved by MacClade. Branch lengths ity-adopters. In the above sample, 670 of 842 among genera were coded following Livezey species (80%) are cavity-nesters and 149 of 842 (1986) with one length unit per character (18%) breed colonially. One-hundred-eight of change; internodes that had been formed by 842 species (13%) are colonial cavity-nesters, polytomy resolution were given a length of 0.1, and except for one parrot and 12 auk species, and species branches were all given unit length. these colonial cavity-nesters excavate their own Since this method cannot handle polymorphic nests. The only exceptions are the Maroon-front- character coding, the few (4 of 149 taxa) cases ed Parrot (Rhynchopsitta terrisi), which nests in of polymorphism were coded to re¯ect the an- cliff crevices at high densities (Forshaw 1989), cestral character state (the state shared with the and several auks, which use crevices in rock taxon's closest relative). A signi®cance level of piles, cliffs, or talus slopes. All other colonial P Ͻ 0.05 was used for comparative tests using breeders either excavate their own nesting cav- both the CCT and Pagel's method. ities or build their own nests. There were only a few cases (not included in RESULTS the previous 108) in which a primarily cavity- Information on nesting behavior from 842 spe- adopting species nests colonially under certain cies in 214 genera, spanning 17 bird families conditions. Bulwer's Petrel (Bulweria bulwerii, and 8 orders, was obtained from the literature Procellariidae) has sometimes been found to nest (Table 1). Species for which data on both nest at high densities using closely spaced crevices type and nesting dispersion were not available among rocks on islands (Warham 1990). Among are not included in the above tally. A complete the Anseriformes, the Fulvous Whistling-Duck species-by-species listing of the survey results is (Dendrocygnidae: Dendrocygna bicolor) occa- 244 JESSICA R. EBERHARD sionally nests in colonies, but when it does it adoption tends to block the evolution of colonial builds a nest on the ground rather than occupy- breeding. Depending on the method used to cal- ing a cavity (Madge and Burn 1988); on small culate probabilities of obtaining the observed islands in Tamaulipas, Mexico, the Black-bellied number of gains of coloniality, the occurrence Whistling-Duck (Dendrocygna autumnalis) of colonial breeding was signi®cantly, or nearly nests at high densities on the ground (Markum signi®cantly, different from the null expectation and Baldassarre 1989); the Wood Duck (Anati- that the evolution of coloniality occurs at ran- dae: Aix sponsa) is sometimes loosely colonial dom with respect to nest type. When P-values if holes are available close together (del Hoyo were averaged over the 100 randomly resolved et al. 1992); and the Red-breasted Merganser trees, the following means were obtained: P ϭ (Anatidae: Mergus serrator) is often a gregari- 0.06 Ϯ 0.06 (SD) using actual changes; P ϭ ous nester, but is not an obligate cavity-nester 0.06 Ϯ 0.07 using MINSTATE ambiguity reduc- and often uses shallow depressions in the ground tion; and P ϭ 0.05 Ϯ 0.02 using MAXSTATE as well as hollows or crevices (Cramp and Sim- ambiguity reduction. These values are probably mons 1997), which allow it ¯exibility in nest conservative, since the concentrated changes test placement. Five sister species of African love- is susceptible to Type II error when the propor- bird (Psittacidae: Agapornis personata, A. lilia- tion of ``white'' branches (branches in which the nae, A. ®scheri, A. nigrigenis, and A. roseicollis) independent character has a value of 0) is Ͻ20% build cup or domed nests within cavities (Fors- (Lorch and Eadie 1999). In the case of the phy- haw 1989), and it is likely that this nest-building logeny tested here, 19% of the branches were permits the modi®cation of otherwise unsuitable white (cavity-adopter lineages). cavities, thereby giving breeding pairs increased An alternative hypothesis, that obligate cavi- ¯exibility in nest-site choice (Eberhard 1998). ty-adoption tends to be blocked by coloniality, Three of these lovebird species (A. personata, was not supported. In this case the CCT did not A. lilianae, and A. roseicollis) are known to be indicate a signi®cant association between cavity- colonial (Forshaw 1989, Juniper and Parr 1998). adoption and solitary nesting (averaged over 25 Other parrots have been reported to nest gregar- randomly resolved trees, P ϭ 0.38 Ϯ 0.07). iously when closely spaced cavities are avail- The likelihood-based comparative analysis of able: Glossopsitta porphyrocephala, Polytelis the full Anseriform dataset indicated that the alexandrae, Lathamus discolor, Aratinga wagle- evolution of nest type and nesting dispersion are ri, Ognorhynchus icterotis (Forshaw 1989), and correlated (P Ͻ 0.05). A contingent changes test Ara militaris (Rowley 1984). Furthermore, in a suggested that the transition from solitary to co- number of species descriptions, Turner and Rose lonial nesting is not contingent on nest type, and (1989) indicate that aggregation of breeding that the most likely transition is from colonial swallows depends on the availability of clumped cavity-nesting to colonial open-nesting. How- nest sites (e.g., Stelgidopteryx ru®collis and S. ever, this implies that the ancestors of colonial serripennis, Tachycineta thalassina, Notiocheli- open-nesters are most likely to be colonial cav- don murina and N. cyanoleuca, Progne chaly- ity-nesters. Given that colonial cavity-nesting is bea, Psalidoprocne fuliginosa, and Hirundo an- only observed under the unusual circumstances decola). Similarly, in his book on petrels War- noted above (and was not included as a character ham (1990:187) notes that in Oceanites, Garro- state on any terminal nodes), the results of this dia, and Pelagodroma ``burrow densities may be initial analysis are dif®cult to interpret biologi- high but those species using natural cavities un- cally. der stones are restricted by availability of suit- An examination of the distribution of nest able niches.'' In sum, these exceptions support types and nesting dispersion on the phylogeny the hypothesis that among cavity-nesters, colo- found that within the genus Mergus, two of the niality is associated with the ability to excavate three species are solitary cavity-nesters (as are cavities or construct a nest, and that the aggre- the nearest relatives) and the third species is a gation of cavity-adopters is dependent on cavity colonial open-nester, implying two character density. transitions over a very short branch. Since this Using Maddison's (1990) CCT, a comparative could cause the corresponding transition param- analysis of nesting behavior in the Anseriformes eter to be very large (M. Pagel, pers. comm.), supported the hypothesis that obligate cavity the analysis was repeated with a phylogeny in EVOLUTION OF COLONIALITY IN CAVITY-NESTERS 245

unlikely to evolve among obligate cavity-adopt- ers. The importance of nest-site choice in per- mitting coloniality is evident from the fact that for a number of taxa, gregarious nesting is con- tingent on the availability of closely spaced nest- ing sites. The evolution of coloniality is not a necessary consequence of the ability to excavate or build a nest (there are many excavators and open nesters that breed solitarily) but obligate cavity adoption does appear to limit the possi- FIGURE 1. Flow diagram showing the transitions in- bility of nesting close to conspeci®cs. volved in the evolution of coloniality relative to nest types in Anseriformes. ``Cavity-adopter'' refers to non- The facilitation of social nesting due to novel excavating cavity-nesters, while ``open/excavator'' nest-building behavior has been previously doc- nesters are species that use open nests or excavate their umented for the swallows (Winkler and Sheldon own cavities. Transition rate parameters are denoted by 1993). However, instead of focusing on the nest- q , in which the subscripts indicate the combination of ij building or nest-site-choice hypothesis presented character states at the beginning (i) and end (j)ofa given transition; possible character state combinations here, the authors argued that the evolution of are 1 (0,0), 2 (0,1), 3 (1,0) and 4 (1,1). Solid arrows dense coloniality in mud-nesting swallows oc- indicate pathways whose transition probabilities are curred because the closed nest cup counteracted signi®cantly greater than zero according to alternative the increased chances of forced extra-pair cop- model tests (Pagel 1994); dotted arrows represent path- ways that are not statistically signi®cant. ulations (EPCs) expected at high breeding den- sities. (An alternative interpretation, suggested by Wagner's [1993] hidden lek hypothesis, is which the above transition was removed (the that the pursuit of extra-pair copulations by fe- three species of Mergus were coded as solitary males promotes aggregated nesting.) cavity-nesters, resulting in one origin of colo- As Collias and Collias (1984) have pointed niality fewer than in the previous analysis). out, coloniality in swallows is not only associ- In this second analysis, the omnibus test again ated with the ability to build mud nests, but also indicated that the two characters are correlated with the ability to excavate burrows. Like nest (P Ͻ 0.03; Fig. 1). The transition parameters building, nest excavation permits ¯exibility in were biologically more reasonable, with very choosing a nest site, evidently facilitating the low transition probabilities estimated for transi- evolution of colonial nesting. In addition to the tions involving the colonial cavity-nesting state. colonial mud-nesters, the Bank Swallow (Ripa- None of the contingent changes tests was sig- ria riparia), which is a burrow-nester, is strongly ni®cant, so the data did not show that the change colonial (Turner and Rose 1989), and several of in either of the characters is dependent on the the cavity-adopting species included in the state of the other. However, the appearance of Winkler and Sheldon phylogeny occasionally coloniality in open-nesters was statistically sig- nest in small groups, depending on the avail- ni®cant, while the transition probability from ability of cavities (Turner and Rose 1989). This solitary to colonial breeding in cavity-nesters does not invalidate either the EPC-avoidance hy- was not signi®cantly greater than zero. Like the pothesis or the hidden lek hypothesis, but sup- CCT analysis, this is consistent with the hypoth- ports the idea that the evolution of a colonial esis that coloniality is more likely to evolve in breeding system is initially facilitated by the open-nesting (or excavator) species, and the lack ¯exibility of nest placement that goes with of signi®cance in the contingent changes test building or excavating a nest. may be due to the relatively low number of In the context discussed in this paper, cavity changes in the two binary characters. nesting, in particular for cavity-adopters, ap- pears to limit the choices available to breeding DISCUSSION pairs in selecting a nest site. Nevertheless, some Taken together, the survey of nesting behavior avian lineages show little variation in nest type, across a wide taxonomic range of birds and the and there are large groups of birds composed comparative analysis of nesting behavior in the almost entirely of cavity-nesters (e.g., Picifor- Anseriformes indicate that colonial breeding is mes and Coraciiformes), and the parrots (Psit- 246 JESSICA R. EBERHARD taciformes) are, with very few exceptions, all CRAMP, S., AND K. E. L. SIMMONS [EDS.]. 1997. Hand- obligate cavity-adopters. The striking sociability book of the birds of Europe, the Middle East and North Africa: the birds of the Western Palearctic. of most parrots, juxtaposed with the fact that Oxford University Press, New York. most species nest solitarily, led Ward and Zahavi DANCHIN, E., AND R. H. WAGNER. 1997. The evolution (1973) to call them ``suppressed colonial breed- of coloniality: the emergence of new perspectives. ers.'' The persistence of cavity-nesting and Trends in Ecology and Evolution 12:342±347. DEL HOYO, J., A. ELLIOTT, AND J. SARGATAL [EDS.]. roosting in otherwise ecologically and morpho- 1992. Handbook of the birds of the world. Vol. 1. logically diverse groups has been discussed by Lynx Editions, Barcelona. Ligon (1993), and he suggests that several fac- EBERHARD, J. R. 1998. Evolution of nest-building be- tors (inability to tolerate low temperatures, slow havior in Agapornis parrots. Auk 115:455±464. development times, high predation in tropical ar- FORSHAW, J. M. 1989. Parrots of the world. 3rd ed. Landsdowne Editions, Chipping-Norton, Sydney. eas, and lack of fecal sacs) might account for FRITH, H. J. 1967. Waterfowl of Australia. Halstead the pattern. Press, Sydney. The causes and consequences of nest-building FRY, C. H., K. FRY, AND A. HARRIS. 1992. King®shers, behavior are complex, and although in many bee-eaters and rollers. Princeton University Press, Princeton, NJ. cases it is dif®cult to establish the directionality GASTON,A.J.,AND I. L. JONES. 1998. The auks. Oxford of cause-effect relationships, some general pat- University Press, New York. terns are evident. There are strong associations GOMEZ-DALLMEIER,F.,AND A. T. CRINGAN. 1990. Bi- between the type of nest used by a bird and its ology, conservation and management of water- life-history characters (Martin 1995). The ability fowl in Venezuela. Editorial Ex Libris, Caracas. JOHNSGARD, P. A. 1978. Ducks, geese and swans of the to choose nest sites, discussed by Collias (1997) world. University of Nebraska Press, Lincoln, NE. in the context of moving into new ecological JUNIPER,T.,AND M. PARR 1998. Parrots: a guide to niches, could also affect the evolution of social parrots of the world. Yale University Press, New behavior by facilitating gregarious nesting. This Haven, CT. KEMP, A. 1995. The hornbills: Bucerotiformes. Oxford hypothesis is supported by a taxonomically University Press, New York. broad survey of cavity-nesting birds as well as LIGON, J. D. 1993. The role of phylogenetic history in a comparative analysis of nesting behavior in the the evolution of contemporary avian mating and Anseriformes, both of which indicate that ex- parental care systems. Current Ornithology 10:1± 46. cavators are more likely than cavity-adopters to LIVEZEY, B. C. 1986. A phylogenetic analysis of recent evolve colonial breeding systems, presumably anseriform genera using morphological characters. because excavators have more ¯exibility in Auk 103:737±754. choosing the location of their nests. LORCH, P. D., AND J. MCA. EADIE. 1999. Power of the concentrated changes test for correlated evolution. Systematic Biology 48:170±191. ACKNOWLEDGMENTS MADDISON, W. P. 1990. A method for testing the cor- I thank Princeton University's Department of Ecology related evolution of two binary characters: are and Evolution and the Smithsonian Tropical Research gains or losses concentrated on certain branches of a phylogenetic tree? Evolution 44:539±557. Institute for support; and W. Eberhard, P. Grant, K. MADDISON,W.P.,AND D. R. MADDISON. 1992. Mac- Harms, M. J. West-Eberhard, and anonymous review- Clade: analysis of phylogeny and character evo- ers for helpful comments and encouragement at vari- lution, v. 3.07. Sinauer Associates, Sunderland, ous stages of the manuscript's preparation. Thanks also MA. to M. Pagel who kindly made his analysis software MADGE, S., AND H. BURN. 1988. Waterfowl: an iden- available and answered questions about its implemen- ti®cation guide to the ducks, geese and swans of tation. the world. Houghton Mif¯in, Boston. MARCHANT, S., AND P. J. HIGGINS. 1990. Handbook of LITERATURE CITED Australian, New Zealand and Antarctic birds. Vol 1. Oxford University Press, Melbourne. BRAWN,J.D.,AND R. P. BALDA. 1988. Population bi- MARKUM,D.E.,AND G. A. BALDASSARRE. 1989. ology of cavity nesters in northern Arizona: do Ground nesting by Black-bellied Whistling-Ducks nest sites limit breeding densities? Condor 90:61± on islands in Mexico. Journal of Wildlife Man- 71. agement 53:707±713. COLLIAS, N. E. 1997. On the origin and evolution of MARTIN, T. E. 1995. Avian life history evolution in nest building by passerine birds. Condor 99:252± relation to nest sites, nest predation, and food. 270. Ecological Monographs 65:101±127. COLLIAS,N.E.,AND E. C. COLLIAS. 1984. Nest build- MARTIN, T. E., AND P. L I. 1992. Life history traits of ing and bird behavior. Princeton University Press, open- vs. cavity-nesting birds. Ecology 73:579± Princeton, NJ. 592. EVOLUTION OF COLONIALITY IN CAVITY-NESTERS 247

PAGEL, M. 1994. Detecting correlated evolution on WAGNER, R. H. 1993. The pursuit of extra-pair copu- phylogenies: a general method for the compara- lations by female birds: a new hypothesis of col- tive analysis of discrete characters. Proceedings of ony formation. Journal of Theoretical Biology the Royal Society of London Series B 255:37±45. 163:333±346. ROLLAND, C., E. DANCHIN, AND M. DE FRAIPONT. 1998. WARD,P.,AND A. ZAHAVI. 1973. The importance of The evolution of coloniality in birds in relation to food, habitat, predation, and life-history traits: a certain assemblages of birds as ``information cen- comparative analysis. American Naturalist 151: tres'' for food-®nding. Ibis 115:517±534. 514±529. WARHAM, H. 1990. The petrels: their ecology and ROWLEY, J. S. 1984. Breeding records of land birds in breeding systems. Academic Press, San Diego, Oaxaca, Mexico. Proceedings of the Western CA. Foundation of Vertebrate Zoology 2:73±224. WINKLER, H., D. A. CHRISTIE, AND D. NURNEY. 1995. SIBLEY, C. G., AND B. L. MONROE JR. 1990. Distribu- Woodpeckers: a guide to the woodpeckers, picu- tion and taxonomy of birds of the world. Yale lets and wrynecks of the world. Pica Press, East University Press, New Haven, CT. Sussex, UK. SIEGEL-CAUSEY, D., AND S. P. KHARITONOV. 1990. The evolution of coloniality. Current Ornithology 7: WINKLER,D.W.,AND F. H. SHELDON. 1993. The evo- 285±330. lution of nest construction in swallows (Hirundi- TURNER,A.K.,AND C. ROSE. 1989. A handbook to the nidae): a molecular phylogenetic perspective. Pro- swallows and martins of the world. Christopher ceedings of the National Academy of Sciences 90: Helm, London. 5705±5707.