Correlates of Egg Rejection in Hosts of the Brown-Headed Cowbird

Abstract

We conducted a comparative analysis of eight potential correlates of egg rejection in hosts of the parasitic Brown-headed Cowbird (Molothrus ater) to test the evolutionary equilibrium and evolutionary lag hypotheses as explanations for the acceptance of cowbird parasitism. The analyses generally supported evolutionary lag. Historic contact with cowbirds may explain why hosts that have recently come into contact with cowbirds accept parasitism, but it does not account for acceptance by hosts with long histories of contact with cowbirds. Egg predation by hosts, nest sanitation, population size, and egg appearance were not correlated with rejection. Larger species that typically build larger nests were more likely to reject. Large hosts may have been parasitized more frequently in the past, possibly due to their more easily found nests or superiority as hosts, and as a result, may have had more opportunity to evolve rejection. Rejection was also correlated with taxonomic affiliation, suggesting that once rejection evolves it is maintained, which implies that rejection is not costly and thus argues against an evolutionary equilibrium. Not surprisingly, hosts with large bills were more likely to reject. This may be a corollary of the tendency for large hosts, which tend to have larger bills, to reject. An evolutionary equilibrium may exist for hosts with eggs that resemble cowbird eggs, depending on the costs to host reproductive success and the likelihood of committing recognition errors. Nevertheless, some hosts have been in contact with cowbirds for a long time, build large nests, have large bills, have a “favorable” phylogeny, and lay eggs that differ from cowbird eggs, yet accept cowbird parasitism. Chance may play a role in the accumulation of the necessary recombinants and mutations necessary for the evolution of rejection.

Factores Correlacionados con el Rechazo de Huevos por parte de Hospederos de Molothrus ater

Resumen. Realizamos un análisis comparativo de ocho factores que potencialmente podrían estar correlacionados con el rechazo de huevos por parte de hospederos del parásito de cría Molothrus ater para poner a prueba las hipótesis de equilibrio evolutivo y de demora en la respuesta evolutiva propuestas para explicar la aceptación del parasitismo. Los análisis generalmente apoyaron la hipótesis de la demora en la respuesta evolutiva. El contacto histórico con Molothrus podría explicar por qué hospederos que sólo recientemente han entrado en contacto con estas aves aceptan el parasitismo, pero no explica por qué hay hospederos que tienen largas historias de contacto con el parásito y aceptan ser parasitadas. La depredación de huevos por parte del hospedero, el comportamiento de aseo del nido, el tamaño poblacional y la apariencia de los huevos no estuvieron correlacionados con el rechazo de huevos parásitos. Las especies grandes, que típicamente construyen nidos más grandes, presentaron una probabilidad más alta de rechazar huevos. Los hospederos más grandes podrían haber sido parasitados más frecuentemente en el pasado, posiblemente debido a que sus nidos se encuentran con mayor facilidad o a que son mejores hospederos y, como resultado, podrían haber tenido mayores oportunidades evolutivas para desarrollar el comportamiento de rechazar huevos. El comportamiento de rechazo también estuvo correlacionado con la filiación taxonómica, lo que sugiere que una vez que éste evoluciona se mantiene, implicando que no es costoso, lo que a su vez es un argumento en contra de la hipótesis del equilibrio evolutivo. De forma poco sorprendente, los hospederos con picos grandes fueron más propensos a rechazar huevos. Esto podría ser un corolario de la tendencia de los hospederos grandes (los cuales tienden a tener picos más grandes) a rechazar huevos. Es posible que exista un equilibrio evolutivo para los casos de hospederos que tienen huevos que se asemejan a los de Molothrus, dependiendo de los costos sobre el éxito reproductivo del hospedero y la probabilidad de cometer errores de reconocimiento. Sin embargo, algunos hospederos han estado en contacto con Molothrus por períodos prolongados, construyen nidos grandes, tienen picos grandes, tienen afinidades filogenéticas “favorables” y ponen huevos que difieren de los del parásito, y aún así aceptan el parasitismo. El azar podría jugar un papel importante en la acumulación de los recombinantes y mutaciones necesarios para que el comportamiento de rechazo pueda evolucionar.

Introduction

Hosts of avian brood parasites often raise fewer of their own young when parasitized and, in some instances, only the parasite is reared (Rothstein 1990, Lorenzana and Sealy 1999). Despite these costs, many hosts accept parasitism. Two opposing hypotheses have been proposed to explain this enigma. The evolutionary equilibrium hypothesis suggests that acceptance of parasitism is more adaptive than rejection, because rejection entails costs that outweigh the benefits of this behavior. These putative costs include damage to the host's own eggs during egg rejection, mistakenly rejecting its own eggs, or retaliation by the brood parasite for rejecting its eggs (Zahavi 1979, Rohwer and Spaw 1988, Lotem et al. 1995). In contrast, the evolutionary lag hypothesis states that rejection is almost always more adaptive than acceptance, but there is some time lag after parasitism begins and before egg rejection evolves (Rothstein 1975, 1990, Davies and Brooke 1989, Sealy 1996). It is difficult to test these two alternatives, especially evolutionary lag, because support for it is usually by default (Sealy 1996, Rothstein and Robinson 1998).

Instead of directly testing the evolutionary lag and equilibrium hypotheses, we used a comparative approach to determine which characteristics distinguish accepters and rejecters of Brown-headed Cowbird (Molothrus ater) eggs. Rothstein (1975) proposed that six factors probably were most important in the evolution of egg rejection: (1) eggs that differ in appearance from those of the cowbird, (2) long history of contact with the cowbird, (3) large population size, (4) well-developed nest sanitation, (5) large bills, and (6) large, easily found nests. Rothstein suggested that bill size and easily found nests were the most important, but he did not statistically analyze these correlates. Two other potentially important traits he examined were egg predation and taxonomic affiliation, but concluded that these were probably not important. Since Rothstein's study (1975), numerous other host species have been tested for egg rejection; thus these characteristics merit reexamination. We compared these eight characteristics in rejecters and accepters to determine which are important in the evolution of cowbird-egg rejection.

Methods

Hypothesized Correlates of Egg Rejection

Correlate 1: Hosts with the longest history of contact with cowbirds should be more likely to reject cowbird eggs

Estimating historic contact between cowbirds and their hosts is difficult because the ranges for cowbirds and their hosts were different in the past. Cowbirds prefer open areas, and typically avoid unfragmented habitat (Robinson, Thompson et al. 1995) and so were largely confined to the Great Plains of North America prior to European settlement (Friedmann 1929, Mayfield 1965, Rothstein 1994). Therefore, hosts in the Great Plains and in open habitats have ostensibly been in contact with cowbirds the longest and have had the most time to evolve rejection (Rothstein 1975). We considered hosts to have had long histories of contact with cowbirds if they nest in relatively open habitats and their ranges overlap the historic range of the cowbird (see figure 10.1 in Davies 2001; Hosoi and Rothstein 2000).

Correlate 2: Large hosts were parasitized more frequently in the past, and as a consequence should be more likely to reject

Contrary to present patterns of cowbird parasitism, hosts approximately the same size as the cowbird (male: 49 g, female: 39 g) or larger may have been parasitized more frequently in the past (Rothstein 1975, Mason 1980, Peer and Bollinger 1997a, 2000). Both Bronzed (M. aeneus) and Shiny (M. bonariensis) Cowbirds parasitize larger hosts more frequently (Post and Wiley 1977, Carter 1986, Mason 1986a, 1986b). It is unclear whether this is a result of large species being superior hosts (Rothstein 1975, Fraga 1985, Wiley 1986, Kleven et al. 1999) or simply because these hosts build larger nests that are easier for cowbirds to locate (Rothstein 1975). We compared host masses using Dunning (1984, 1993) and present them as the mean mass of males and females, with the exception of Couch's Kingbird (see Table 1 for scientific names), for which only the mass of males is known. We also compared nest sizes using outside diameter and outside depth.

Correlate 3: Hosts that practice nest sanitation should be more likely to evolve rejection

Nest sanitation may be a preadaptation for the evolution of egg rejection because the same mechanics are used in both behaviors (Rothstein 1975). Birds grasp fecal sacs and eggshells between their mandibles, carry them away from their nests, and drop them, actions similar to egg rejection, although rejecters often gently place eggs on the substrate rather than dropping them (Peer and Sealy 2004).

Correlate 4: Egg predators should be more likely to evolve egg rejection

Egg predators may be more likely to evolve rejection because the mechanics involved in this behavior are also similar to those in egg rejection (Rothstein 1975, Ortega and Cruz 1988, Peer and Bollinger 1997a). Egg predators either grasp eggs between their mandibles or puncture eggs with open mandibles and then eat them. Similarly, rejecters puncture or grasp eggs, and this may enhance the likelihood of egg rejection.

Correlate 5: Hosts with large populations should have larger and, hence, more variable gene pools, which make them more likely to evolve rejection

Species with larger populations (i.e., more diverse gene pools) may be more likely to evolve rejection (Rothstein 1975). Population status was determined using Peterjohn et al. (1994) and Price et al. (1995); both sources are based on Breeding Bird Survey (BBS) data. Species that were observed on <14 BBS routes were classified as small populations. Peterjohn et al. (1994) determined this as the minimum sample required to analyze short-term population trends and this method largely agreed with our subjective assessment of large versus small populations based on the figures in Price et al. (1995). The exceptions were Couch's Kingbird, Crissal Thrasher, and Curve-billed Thrasher, which range into Mexico and Central America where there are no BBS routes. We classified these as having large populations based on their large geographical range. Peterjohn et al. (1994) did not distinguish between Florida, Western and Island Scrub-Jays. We classified the Florida and Island Scrub-Jay populations as small: the Florida Scrub-Jay population is estimated at 10 000 individuals and the Island Scrub-Jay at 12 500 (Woolfenden and Fitzpatrick 1996, Kelsey and Collins 2000).

Correlate 6: Rejection should be common to the members of a taxonomic unit

Rejection may evolve once within a lineage and be retained in the ancestors during speciation (Rothstein 1990, 2001, Bolen et al. 2000, Peer and Sealy 2004). We considered rejection to be common to a taxonomic unit if ≥2 species exhibited it.

Correlate 7: Hosts with eggs that differ in appearance from cowbird eggs should be more likely to recognize and reject cowbird eggs

Presumably it is more difficult for hosts with eggs similar to cowbird eggs to recognize that they have been parasitized (Rothstein 1975, Burhans and Freeman 1997, Peer et al. 2000). Host eggs were compared to cowbird eggs on the basis of size, background color, and spotting pattern. Each accepter and rejecter species received points when their egg traits matched those of a cowbird egg, and a zero when their traits did not match. Host eggs received one point if their mean length was within 4 mm of a cowbird egg (21 mm) and one point if their mean width was within 4 mm (16 mm). If length or width were >4 mm different they received a score of zero. Host eggs received one point if their background color matched the cowbird's white to creamy-white background color. Finally, host eggs received one point if they were spotted, and two points if the spots were the same color as the brown spots on cowbird eggs. Thus, host species could receive total scores ranging from 0 to 5, and the scores for each accepter and rejecter were then compared. Eggs were evaluated using Baicich and Harrison (1997).

Correlate 8: Hosts with large bills can better handle eggs than those with smaller bills

Species with large bills may be more able to manipulate and reject parasitic eggs (Rothstein 1975, Rohwer and Spaw 1988). Rothstein (1975) compared the ratio of bill lengths of hosts to the width of the cowbird egg to determine whether hosts could grasp cowbird eggs. Rohwer and Spaw (1988) attempted to improve this measure with a “grasp index” by multiplying bill length by bill width, measured at the base of the bill. A shortcoming of the latter index is that the width of the bill at its base is fairly inconsequential, considering that eggs are grasped at the bill tip (S. I. Rothstein, pers. comm.). This results in values that are misleading in terms of rejection abilities, which is evident in a comparison of the Cedar Waxwing and Gray Catbird (Rothstein, pers. comm.). Both species have essentially the same grasp indices as calculated by Rohwer and Spaw (1988), 230 and 232, respectively. Accordingly, both should have similar rejection abilities. However, the catbird is a grasp-ejecter (Lorenzana and Sealy 2001) and the waxwing is a puncture-ejecter (Rothstein 1976). Hosts with large bills grasp cowbird eggs between their mandibles and remove them, whereas smaller hosts are forced to puncture-eject cowbird eggs and remove them in their open beaks (e.g., Bullock's and Baltimore Orioles; Rothstein 1977, Sealy and Neudorf 1995) or remove them on their closed beaks or piecemeal (e.g., Eastern Warbling Vireos; Sealy 1996). In contrast to their similar grasp indices, catbird bills are 17% longer than waxwing bills, thus explaining why they grasp-eject instead of puncture-eject cowbird eggs. Bill length predicts rejection ability more reliably (S. I. Rothstein, pers. comm.) and we used this measure to compare the rejection abilities of accepters and rejecters.

We report bill lengths of the species measured by Rohwer and Spaw (1988). For other species, we measured tomial length from the commisure to the tip of the upper mandible (Rothstein 1975, Rohwer and Spaw 1988) of five adult females of each species. These measurements were taken at the Field Museum of Natural History, Chicago, the University of Manitoba Zoology Museum, and Museum of Systematics and Ecology at the University of California, Santa Barbara.

In the analyses of these eight correlates we used only host species that have been experimentally tested for egg rejection in at least four nests. In a similar analysis, Rohwer and Spaw (1988) classified species as accepters if reports indicated more than 20% of their nests were parasitized. This method has since been proven unreliable. For example, Davis and Sealy (2000) reported cowbird parasitism frequency in excess of 40% in a population of Western Meadowlarks in Manitoba, but Peer et al. (2000) found that Western Meadowlarks were rejecters in Illinois.

Ejection of eggs from nests, pecking eggs, egg burial, and desertion of parasitized nests have been considered methods of “rejection.” However, birds desert nests for a variety of reasons (Rothstein 1975) and without carefully controlled experiments (e.g., Hill and Sealy 1994) it is impossible to determine whether desertion is a response to parasitism or to some other disturbance. The significance of egg burial is also questionable as it is often just a continuation of the nest-building process (Rothstein 1975). An exception is the Yellow Warbler, which buries parasitized clutches (Sealy 1995). However, we do not consider the Yellow Warbler a rejecter species because the warblers do not recognize cowbird eggs and the cue used to bury cowbird eggs is unclear (Sealy and Lorenzana 1998). By contrast, ejection and pecking are direct responses to the egg and are “true” rejections (Rothstein 1975, Peer and Bollinger 1997a). Most “rejections” recorded in accepter species are via nest desertion, whereas those by “true” rejecters are through ejection (Rothstein 1975). Nest desertion as an antiparasite adaptation has been reviewed recently (Hosoi and Rothstein 2000); therefore, we focused our analyses on species that eject or puncture-eject parasitic eggs.

There are 26 potential hosts known to reject cowbird eggs at a frequency of ≥75% (“rejecters”), four that reject at 21–74% of nests (“intermediate rejecters”), and 32 at ≤20% of nests (“accepters,” Table 1). For these analyses, we classified two intermediate rejecters as rejecters. Northern Mockingbirds reject 60% of Brown-headed Cowbird eggs, but reject 69% of Bronzed Cowbird eggs and 78% of Shiny Cowbird eggs (Table 1). Eastern Meadowlarks reject 36% of Brown-headed Cowbird eggs and 40% of nonmimetic eggs (Table 1), and the frequency of rejection may be increasing in this host (Peer et al. 2000). These two hosts differ from accepter species in that they eject rather than desert or bury foreign eggs (Peer et al. 2000, 2002). Likewise, we classified a current accepter, the Common Grackle, as a rejecter species for the analyses. This species is unique because of its low level of true egg rejection and the fact that it is rarely parasitized (only 30 cases documented; Peer et al. 2001). It has been suggested that it once rejected close to 100% of cowbird eggs, but has subsequently lost most of this behavior in the absence of parasitism (Peer and Bollinger 1997a, Peer and Sealy 2004). Thus, we assumed that this grackle should possess correlates required for the evolution of egg rejection.

Great-tailed Grackles, Boat-tailed Grackles, Loggerhead Shrikes, and Yellow-billed and Black-billed Magpies reject Brown-headed Cowbird eggs or nonmimetic eggs (Table 1) and are sympatric with Brown-headed Cowbirds. However, the two grackle species are unsuitable hosts for Brown-headed Cowbirds because they are too large (rejection in these species may have evolved in response to Giant Cowbird [Molothrus oryzivora] parasitism; Peer and Sealy 2000a, 2004). Loggerhead Shrikes and magpies may have evolved rejection in response to Common Cuckoo (Cuculus canorus) parasitism (Bolen et al. 2000, Rothstein 2001). Only a single record of cowbird parasitism has been recorded for the shrike (DeGeus and Best 1991) and there are no records of cowbird parasitism for magpies, which also are too large to be suitable hosts of the cowbird. However, other members of the shrike clade are regular cuckoo hosts, and European Magpies (Pica pica) are parasitized by cuckoos in Eurasia, where they are rejecters; hence, rejection may have evolved in response to this form of parasitism and has been retained during speciation (Bolen et al. 2000, Rothstein 2001). As a consequence, we did not include these five species in the analyses. Moreover, there are no closely related accepters to compare with these five rejecters; thus they would not be included in our comparative analyses regardless of our classification. We did, however, include these species in the taxonomic analysis because we were interested in determining whether rejection is common to groups of related individuals.

Cowbirds rarely parasitize species that are unsuitable hosts because of diets, inappropriate feeding methods, or inaccessible nests, including Mourning Doves (<10 records of parasitism; Peer and Bollinger 1998), Black-billed Cuckoos (7 records; Friedmann and Kiff 1985), Barn Swallows (<20 records; Friedmann 1963, Friedmann et al. 1977), Mountain Bluebirds (4 records; Hébert 1999), House Wrens (9 records; Friedmann and Kiff 1985), and American Goldfinches (6% of 802 nests; Middleton 1991). Because these hosts are rarely parasitized, there is no apparent selection for rejection; thus we excluded them from the analyses. Although Mourning Doves and Black-billed Cuckoos are intermediate rejecters, they are also not included in the analyses (Table 1). Rejection in Mourning Doves may have evolved as a manifestation of nest usurpation (Peer and Bollinger 1998). Black-billed Cuckoos are parasitized by conspecifics and Yellow-billed Cuckoos (Coccyzus americanus; Nolan and Thompson 1975, Fleischer et al. 1985, Hughes 1997), and rejection in this species may have evolved in response to these forms of parasitism. Cedar Waxwing diets are also incompatible with cowbirds; however, they are frequently parasitized and are rejecters (Rothstein 1976). For this reason, we included Cedar Waxwings in the analyses, but again, there was no closely related accepter species for comparison.

We assumed that all birds can distinguish between cowbird eggs and their own. The visual ability of each cowbird host is unknown, but birds' color vision exceeds that of humans in both spectrum and acuity (Sillman 1973, Gill 1990). Therefore, all hosts probably can visually identify cowbird eggs as long as their own eggs differ sufficiently from cowbird eggs in appearance, even if they do not alter their behavior accordingly.

Statistical Analyses

The data were analyzed using the pairwise comparative method (Møller and Birkhead 1992). Closely related species frequently inherit traits from common ancestors, so treating species as independent data points incorrectly inflates the number of degrees of freedom and can lead to erroneous conclusions (Harvey and Pagel 1991). The pairwise comparative method controls for phylogeny by comparing pairs of closely related species that differ in the trait of interest. We used this method to compare up to 12 species pairs in which the data were discrete and the phylogenies for many of these species were unresolved. Other comparative analyses require continuous data or resolved phylogenies (Harvey and Pagel 1991). Rejecters were compared to the most closely related accepter species according to the classification of the American Ornithologists' Union (1998). We also compared the eastern and western subspecies of the Warbling Vireo, which may be distinct species (Sibley and Monroe 1990, Sealy 1996). If the relationships were not clear, pairs were chosen at random among species (within the same families) that were equally likely to be their closest relatives. Pairs were compared using a sign test and all tests were one-tailed because we predicted a priori that these correlates would affect rejection. The accepted level of significance was P < 0.05. One correlate that was not analyzed using the comparative method was phylogenetic effects, and we used a Fisher exact test for this correlate. We wanted to determine whether rejection was common to members of a taxonomic unit; controlling for phylogeny would defeat this purpose.

Results

We found that five species pairs differed in their length of contact with cowbirds, and in each case the species with the longer period of contact showed greater rejection behavior (Table 2; sign test, P = 0.03). Because historic contact with cowbirds affects rejection, we did not include the Western Warbling Vireo, Wood Thrush, Rusty Blackbird, and Le Conte's and California Thrashers in this, or the remaining analyses, as these five hosts have been in contact with cowbirds for a relatively short period. All seven rejecter species were larger than the accepter with which they were paired (Table 2; sign test, P = 0.02), and had larger nests than accepters (Table 2; sign test, P = 0.02).

A pairwise comparative analysis was not possible for nest sanitation because all accepters and rejecters for which there are data practice nest sanitation (Appendix). There was no apparent correlation between egg predation and egg rejection; only one of the seven pairs differed, Orchard and Baltimore Orioles (Appendix). Pairwise comparisons were not possible for population size because most hosts had large populations (Appendix). The exceptions were the Florida and Island Scrub-Jays, and the Le Conte's and California Thrashers. However, the thrashers were excluded from this analysis because of their short history of contact with cowbirds, and both scrub-jay species are rejecters.

Taxonomic affiliation was correlated with rejection; 26 of 29 rejecters (including Common Grackle, Northern Mockingbird, and Eastern Meadowlark) were related to other rejecters, whereas only 12 of 31 accepters were related to rejecters (Tyrannidae [3], Turdidae [2], Mimidae [2], Icteridae [5]; Fisher exact test, P < 0.001). The 29 rejecter species are represented by eight families: Tyrannidae (5), Laniidae (1), Vireonidae (1), Corvidae (6), Turdidae (1), Mimidae (6), Bombycillidae (2), and Icteridae (7).

Our analysis did not reveal a relationship between egg appearance and rejection, and the trend was opposite to our prediction (sign test, P > 0.05; Table 2). All seven rejecters had larger bills than their accepter cohorts (sign test, P = 0.02; Table 2, Appendix).

Discussion

Historic Contact with Cowbirds

Clearly hosts must be parasitized before they evolve rejection, unless they have inherited this behavior from an ancestor (Rothstein 1975, 2001, Peer 1998). Hosts that nest in open habitats and whose breeding ranges include the Great Plains and their immediate borders have been in contact with the Brown-headed Cowbird the longest (Friedmann 1929, Mayfield 1965, Hosoi and Rothstein 2000, Davies 2001), and presumably have had the longest time to evolve rejection (Mayfield 1965, Rothstein 1975, Peer et al. 2000). Our analysis demonstrated a correlation between historic contact with cowbirds and rejection.

Despite evidence that supports this correlate, there were seven accepter-rejecter pairs with similar lengths of historic contact with cowbirds and another six accepter species (Chipping Sparrow, Clay-colored Sparrow, Vesper Sparrow, Lark Sparrow, Lark Bunting, and Dickcissel) that probably have been in contact with cowbirds throughout much of their evolutionary histories and lay eggs that differ from cowbird eggs (Appendix). Thus, this correlate may explain why hosts that only recently have come into contact with cowbirds accept parasitism, but it does not explain why some hosts with long histories of contact are accepters.

Host Mass and Nest Size

Both body mass and nest size were correlated with rejection behavior, which suggests that large hosts were parasitized more frequently in the past and are now avoided because many are rejecters (Peer and Bollinger 1997a, 2000). This is similar to patterns of Shiny and Bronzed Cowbird parasitism; these species tend to parasitize hosts that are as large or larger than themselves, and consequently most rejecters of their eggs are larger (Carter 1986, Mason 1986b, Peer and Sealy 1999).

Brown-headed Cowbirds may have parasitized large hosts more frequently in the past because they are superior hosts that can better defend their nests and provide more food (Rothstein 1975). For example, Wood Thrushes are one of the few large accepters, presumably because they nest in forests and have only recently come into contact with cowbirds. Cowbirds multiply parasitize them (Trine 2000). Wood Thrushes are capable of rearing more than a single cowbird per nest, in contrast to smaller hosts (Trine 2000). Shiny Cowbirds also multiply parasitize larger hosts (Mason 1986a) and their growth rates are positively correlated with host size (Fraga 1985, Wiley 1986). Likewise, success of Bronzed Cowbirds is highest with large hosts (Carter 1986), and Common Cuckoo nestlings reared by larger hosts grow significantly faster and are larger at fledging than nestlings reared by smaller hosts (Kleven et al. 1999). One apparent exception is the Northern Cardinal, which fledges fewer cowbirds than some smaller hosts. However, much of this difference is due to egg removal by hosts or cowbirds (Scott and Lemon 1996).

Another possibility is that larger host nests are easier to locate (Rothstein 1975). There is some evidence linking nest visibility and the likelihood of parasitism. Forest-nesting species were more likely to be parasitized when surrounded by open canopies (Brittingham and Temple 1996), as were heavier and, hence, larger Song Sparrow nests (McLaren and Sealy 2003) as well as Song Sparrow nests with less lateral cover (Larison et al. 1998). In contrast, Smith (1981) found no relationship between cover and parasitism on Song Sparrows, and Burhans and Thompson (1998) obtained mixed evidence for cover and the likelihood of parasitism in Field Sparrows and Indigo Buntings (Passerina cyanea). Some hosts, such as the Western Wood Pewee, build relatively easily found nests and are rarely parasitized, whereas hosts like the Willow Flycatcher (Empidonax traillii) build better-concealed nests in the same area and are parasitized much more frequently (Curson et al. 1998; M. J. Whitfield, pers. comm.).

While the visibilities of nests of some larger hosts sometimes differ (Murphy et al. 1997), these hosts must build larger nests to accommodate their young and their nests may be more difficult to hide. Several large rejecter species (e.g., kingbirds, grackles, American Robins, Northern Mockingbirds, Scissor-tailed Flycatchers) also nest in dead trees, isolated trees with little vegetation surrounding them, telephone poles, and other artificial structures, where their nests are in plain view (Regosin 1998, BDP, unpubl. data). Rothstein (1975) noted that nests of rejecters were easier to find when conducting experimental parasitism studies, and 89% of 1256 nests found by Peer and Sealy (1999) were nests of rejecters.

The correlation between host body mass and rejection also supports the role of historic contact in the evolution of rejection. Species that were sympatric with cowbirds were not necessarily parasitized at the same frequencies. For example, parasitism frequencies vary significantly in different regions of North America for the Red-winged Blackbird, Wood Thrush, and numerous grassland hosts (Robinson, Rothstein, et al. 1995, Peer et al. 2000). Once the large hosts evolved rejection, cowbirds may have been forced to switch to new, smaller hosts (Rothstein 1975, Mason 1980, Peer and Bollinger 1997a, 2000). Accordingly, the small hosts would have had less time to evolve rejection; hence, most are accepters. Therefore, historic parasitism, in the context of host body mass and nest size, may account for the presence or absence of rejection and support evolutionary lag.

Nest Sanitation

Rothstein (1975) suggested that nest sanitation may be a preadaptation for the evolution of egg rejection (see also Ortega and Cruz 1988). Heightened nest sanitation may be manifested in the removal of foreign objects, such as cowbird eggs, and the mechanics involved in nest sanitation are similar to those used in egg rejection (Rothstein 1975). However, we, like Rothstein (1975), were unable to find a pattern based on this criterion and this correlate probably does not influence egg rejection behavior.

Egg Predation

It has been suggested that egg predators evolve rejection more readily because of the similarities between these behaviors (Rothstein 1975, Ortega and Cruz 1988, Peer and Bollinger 1997a). We found no correlation between egg predation and egg rejection. In addition, a number of accepters of cuckoo eggs are also egg predators (Yom-Tov 1976, Soler and Møller 1990). Despite the similarity in mechanics between the two behaviors, egg predators are not more likely to have evolved egg rejection.

Population Size

The only accepters with small populations were the Le Conte's and California Thrashers, but both were excluded from this analysis because of their short history of contact with cowbirds. Population size (i.e., gene pool) may constrain the two thrasher species from developing rejection. Florida and Island Scrub-Jays were the only species with small populations included in this analysis. The scrub-jays apparently inherited rejection from an ancestor as both are isolated from parasitism (Peer and Rothstein, unpubl. data). However, the remaining accepters all have large populations and therefore this correlate does not appear to speed the evolution of egg rejection.

Taxonomic Affiliation

Rejection is common to all jays and many of the kingbirds and mimids that have been tested (see also Peer and Sealy 2000b). The Loggerhead Shrike may have inherited rejection from an ancestor that was parasitized in the Old World by Common Cuckoos (Rothstein 2001). Similarly, the Boat-tailed Grackle may never have been parasitized by cowbirds, but possibly inherited rejection from its common ancestor with the Great-tailed Grackle that evolved rejection in response to Giant Cowbird parasitism (Peer and Sealy 2000a, 2004). The fact that rejection tended to be common within lineages was a problem in itself for our comparative analyses because in some cases there were no accepter species with which to compare the rejecter species (such as the corvids). This result suggests that once rejection evolves in a lineage it may be retained during speciation (Rothstein 1990, 2001, Peer and Bollinger 1997a, Peer and Sealy 2004), or alternatively that rejection is easily evolved in these groups (Peer and Bollinger 1997a). This correlate also supports evolutionary lag because it indicates that rejection is not costly in the absence of parasitism in some birds (Rothstein 1990, 2001, Peer and Sealy 2004). Nevertheless, it does not explain how rejection evolved in the first place.

Egg Appearance

We did not find a relationship between egg appearance and rejection. This was in part due to the fact that none of the species with eggs that closely match cowbird eggs (i.e., received scores of 5) were included in this analysis. Thus, all of the eggs of hosts used were at least somewhat dissimilar from cowbird eggs. There is little doubt that egg appearance is an important constraint on the evolution of rejection. Only eight host species received scores of 5 and all are accepters: Prothonotary Warbler, Yellow Warbler (but see Sealy 1995), Yellow-breasted Chat, Rose-breasted Grosbeak, Northern Cardinal, Song Sparrow, Grasshopper Sparrow, and Field Sparrow (Appendix). This is particularly evident in the Yellow-breasted Chat, which ejects immaculate eggs (54%) more frequently than spotted cowbird eggs that resemble its own (9%; Burhans and Freeman 1997). Several other hosts also reject eggs that differ strongly from their own eggs: Chalk-browed Mockingbirds (Mimus saturninus; Fraga 1985), Brown-and-Yellow Marshbirds (Pseudoleistes virescens; Mermoz and Reboreda 1994), Western Meadowlarks (Peer et al. 2000), and Yellow-winged Blackbirds (Agelaius thilius; Fraga 1985). Grasshopper Sparrows and Lark Sparrows, whose eggs also resemble cowbird eggs, tend to reject undersized eggs that differ from their own (Peer et al. 2000). There may be an evolutionary equilibrium in these hosts depending on the costs of rearing cowbirds. For example, parasitized Northern Cardinals suffer relatively small losses (Scott and Lemon 1996). It may be more beneficial for cardinals to accept cowbird eggs than risk rejecting their own eggs. Indeed, Northern Cardinals rejected 18 of 55 immaculate eggs and only 1 of 18 spotted eggs in a recent study (D. E. Burhans et al., unpubl. data).

Bill Size

Rothstein (1975) and Rohwer and Spaw (1988) found that rejecters had larger bills than accepters, but these authors did not corroborate their conclusions statistically. Pairwise comparisons confirm their conclusions as all seven rejecters had larger bills than their accepter cohorts. Rohwer and Spaw (1988) argued that small bills are a significant constraint in the evolution of rejection, and suggested that it is less costly for smaller hosts to accept parasitism and raise cowbirds than to attempt to puncture-eject the cowbird eggs. The host's bill may deflect off the thick-shelled cowbird egg (Spaw and Rohwer 1987, Picman 1989), and damage some of the host's eggs. According to the “puncture-resistance” hypothesis (Rohwer and Spaw 1988), bill-size constraints have resulted in an evolutionary equilibrium that accounts for the acceptance of cowbird eggs. However, Sealy (1996) found that the 15-g Eastern Warbling Vireo rejects cowbirds eggs via puncture-ejection and they lost only 0.3 eggs for every cowbird egg ejected. The cost of acceptance clearly exceeds that of rejection in Warbling Vireos, which typically raise only cowbirds and none of their own young when they accept parasitism (Rothstein et al. 1980).

Sealy (1996) noted that the Western Warbling Vireo, which is slightly smaller than the Eastern Warbling Vireo in both bill size (16.2 ± 0.93 mm vs. 17.6 ± 0.94 mm, respectively) and mass (12 g vs. 15 g), may be below the minimum size required to evolve ejection. Dickcissels have even smaller bills (Appendix), but eject 100% of undersized eggs without damaging their own eggs; however, Dickcissels reject only a small percentage of normal-sized cowbird eggs, sometimes damaging their own eggs in the process (Peer et al. 2000). This implies that the smaller bill of the Dickcissel may be a constraint when attempting to reject cowbird eggs. Dickcissels are also able to raise some of their own young along with cowbirds (Zimmerman 1983); therefore the costs of rejection might outweigh the benefits for some Dickcissels, suggesting an evolutionary equilibrium (Peer et al. 2000), but this needs further study.

Considering that the Eastern Warbling Vireo is the smallest known rejecter of cowbird eggs, it is reasonable to expect that species with bills as large or larger than the vireo should also be able to reject cowbird eggs. However, 10 such species are accepters (Appendix). This suggests that the puncture resistance hypothesis (an example of evolutionary equilibrium), does not account for acceptance by these hosts, although the costs of rejection have not been determined in these species.

Another apparent example of evolutionary lag pertaining to bill size is the case of Western and Eastern Meadowlarks. Western Meadowlarks reject more frequently where both meadowlarks are sympatric and consequently may have had the same length of contact with cowbirds (Peer et al. 2000). The Eastern Meadowlark's bill is slightly smaller than the Western Meadowlark's bill, but is clearly large enough for it to reject without damaging its own eggs (Appendix, Peer et al. 2000). The necessary mutations and recombinants for the evolution of rejection may have arisen first in Western Meadowlarks.

The tendency for rejecters to have larger bills may also be explained by the fact that larger birds typically have larger bills (Appendix). Our analyses suggest that larger hosts were parasitized more frequently in the past and have had more time to evolve rejection, thereby supporting evolutionary lag. This is further supported by studies of the Common Cuckoo. In contrast to hosts of the Brown-headed Cowbird, most cuckoo hosts in Europe and Africa are rejecters, but there is no relationship between bill size and acceptance (Rothstein 1992). Large hosts are typically grasp-ejecters, medium-sized hosts puncture-eject, and small hosts desert parasitized nests (Davies and Brooke 1989, Moksnes et al. 1991). Similar to the shells of cowbird eggs, cuckoo (Cuculus, Clamator) eggshells also are unusually strong (Brooker and Brooker 1991, Picman and Pribil 1997). The difference between cuckoo-host systems and cowbird-host systems is likely due to the longer history of association between cuckoos and their hosts than cowbirds and their hosts (Rothstein et al. 2002). Given enough time (i.e., evolutionary lag), perhaps most cowbird hosts will evolve rejection similar to cuckoo hosts in Europe and Africa.

Conclusions

In his original analysis, Rothstein (1975) identified seven rejecter species and 23 accepters. Currently, we are aware of 30 hosts that demonstrate intermediate levels of rejection or higher, three more including the Yellow-breasted Chat, Northern Cardinal, and Chestnut-collared Longspur, which reject nonmimetic eggs at intermediate frequencies, and one species, the Common Grackle, that may have lost rejection behavior. Of the remaining 28 accepter species, the Yellow Warbler regularly responds to natural parasitism by burying cowbird eggs, and Field Sparrows desert nests, especially after observing cowbirds at their nests (Sealy 1995, Strausberger and Burhans 2001). Two other hosts that have not been tested for rejection, the Blue-gray Gnatcatcher and Bell's Vireo, also frequently desert parasitized nests (Goguen and Mathews 1996, Budnik et al. 2001, Kershner et al. 2001, Kus 2002). Thus, relatively few of the cowbird's 227 known hosts (DeGeus and Best 1991, Ortega 1998) are known to reject parasitism.

Our analyses generally support evolutionary lag for the widespread acceptance of cowbird eggs. Egg predation, nest sanitation, population size, and egg appearance were not correlated with rejection. Historic contact may explain why hosts that have only recently come into contact with cowbirds accept parasitism, and why large hosts that may have been parasitized more frequently are also more likely to reject than smaller hosts. Rejection was also correlated with taxonomic affiliation, indicating that once rejection evolves it is often maintained. This also supports the evolutionary lag hypothesis. However, taxonomic affiliation does not explain how rejection evolved in the first place. Large bills appear to facilitate the evolution of rejection by making the removal of cowbird eggs easier, but small bills are not necessarily a constraint because all hosts have the option of deserting parasitized nests. That hosts with larger bills are more likely to reject may simply be a result of large hosts tending to have larger bills and these hosts were ostensibly parasitized more frequently in the past. Despite the lack of a relationship between egg appearance and rejection, other studies have demonstrated that hosts with eggs that match cowbird eggs are less likely to reject. Thus, an evolutionary equilibrium may exist between these hosts and cowbirds depending on the costs to host reproductive success and the likelihood of committing recognition errors. Nevertheless, some hosts that have been in contact with cowbirds for a long time, build large nests, have large bills, have a “favorable” phylogeny, and lay eggs that differ from cowbirds', still accept cowbird parasitism (e.g., Red-winged Blackbirds). Chance may play a role in the accumulation of the necessary recombinants and mutations necessary for the evolution of rejection (Rothstein 1975). More species should be tested to resolve these issues as fewer than half of the hosts known to have successfully reared cowbirds have been tested experimentally.

Acknowledgments

We thank Diane Neudorf for advice on pairwise comparative analyses. Comments from Dirk Burhans, James Briskie, Catherine Ortega, and Mary Whitfield improved the manuscript. David Willard of the Field Museum of Natural History, Chicago, allowed us to take bill measurements, and Peter Lowther, Stephen Rothstein, and David Willard took additional measurements for which we are grateful. This study was supported by a NSERC grant to SGS and by a G. A. Lubinsky Memorial Scholarship from the Department of Zoology, University of Manitoba, awarded to BDP.

Literature Cited

Author notes