Introduction

Brood parasitism is a prevalent breeding strategy in birds, with about 1% of all species being obligate brood parasites (Davies, 2000). These species, being able to reproduce only by laying eggs in the nests of other species (the hosts), rely on these unrelated foster parents to care for the parasitic young (Johnsgard, 1997; Davies, 2000). This type of parental-care parasitism (Roldán and Soler, 2011) reduces the fitness of the hosts to a variable extent, providing selective pressures on hosts to evolve defences against brood parasites, which in turn would select for the evolution of counter-defences in brood parasites, leading to selection for improved host defences, further parasitic adaptations, and so on (Davies, 2000). this escalation in defences and counter-defences both by hosts and by parasites, gives rise to a coevolutionary arms race (Dawkins and Krebs, 1979).

The most common and effective host defence against brood parasitism is recognition and subsequent rejection of parasitic eggs (Rohwer and Spaw, 1988; Rothstein, 1990; Davies, 2000). Egg rejection may be costly because hosts may sometimes eject one of their own eggs (recognition errors; Davies and Brooke, 1988; Marchetti, 1992; Davies et al., 1996) and/or may also damage their own eggs while trying to eject the parasitic egg (ejection costs; Davies and Brooke, 1988; Rohwer and Spaw, 1988; Marchetti, 1992; Røskaft et al., 1993). Rejection costs are especially critical when small-sized host species in which ejection behaviour has not evolved desert their nests as a response to parasitism (Hoover, 2003; Servedio and Hauber, 2006). It is widely accepted that recognition errors as well as ejection costs would select against the maintenance of egg discrimination in the absence of brood parasitism (Rothstein, 1990; Brooke et al., 1998; Davies, 2000, 2011). However, it has recently been emphasised that rejection behaviour persists in many host species despite rejection costs and recognition errors (Bolen et al., 2000; Rothstein, 2001; Underwood et al., 2004; Peer et al., 2007; Martín-Vivaldi et al., 2012; Medina and Langmore, 2015). In fact, it has been shown that successful resistance, i.e. a rejection rate of about 100% in species that are not currently used as hosts by brood parasites, is a more frequent long-term coevolutionary outcome in the relationships between brood parasites and their hosts than previously suspected (Soler, 2014), meaning that rejection behaviour has insignificant costs in these host species.

Aims and methodology

This article synthesises data on egg rejection by potential host species of brood parasites on different continents. This comparison was first done by Rothstein, 1990, 1992) but an updated review is needed since much more experimental data is now available. Furthermore, the main aim of the present paper is to examine differences in rejection rates among hosts of the three geographical areas taking into account the sizes of the species involved, which has significant evolutionary implications.

The data used in this review (see  table S1 in Supplementary Electronic material (Art2_Soler M_Supplementary Material.pdf), and table 1) are based on the Appendix in Soler (2014) to which new information and species have been added. The body mass of all species has been taken from the Handbook of the Birds of the World (Del Hoyo et al., 2003-2011). The mass of the cattle tyrant Machetornis rixosa, which is not provided in the above-mentioned handbook, was taken from Dunning (1993). Host status was determined from the cited references for each species or from Friedmann et al. (1977) and/or Davies (2000). For cowbird host status, if not mentioned in the original paper and if that species is not included in Friedmann et al. (1977), the species is considered a non-host. The reported rejection rate both in the Supplementary Electronic Material and fig. 1 is that corresponding to non-mimetic eggs (or models), according to the criterion of authors in each paper. Sporadic nest desertions (1–3 nests) in eggrecognition experiments with low sample sizes were not considered as rejections because, without an appropriate control, these desertions could be due to other factors unrelated to experimental parasitism (Soler et al., 2011). Most rejection rates of nonmimetic eggs of European hosts of the common cuckoo were taken from Martín-Vivaldi et al. (2012). When considering the effect of host size, I have considered ‘small’ to indicate a mass of less than 30 g, ‘medium’ to be a mass of 30–75 g, and ‘large’ to be more than 75 g. For coevolutionary outcomes, following Soler (2014), a lack of rejection was considered to apply when the rejection rate was 10% or less, successful resistance when egg rejection was 90% or more, and intermediate rejection rate for percentages in between.

Brood parasites in Europe and America

Interspecific avian brood parasitism has evolved independently seven times in birds (Sorenson and Payne, 2002). Brood parasites are distributed throughout all continents, being especially frequent in Africa but also in South-east Asia and Australia (Davies, 2000). Brood parasites in Europe and America, although scarcer, have been extensively studied and have provided the basis for the theoretical background in brood-parasitism studies. These well-known species include the common cuckoo Cuculus canorus and great spotted cuckoo Clamator glandarius in Europe, and the brown-headed cowbird Molothrus ater and shiny cowbird M. bonariensis in America.

Fig. 1.

—Number of potential host species presenting various rejection rates of experimentally introduced non-mimetic eggs in three different geographical regions.

[Número de especies hospedadoras potenciales que presentan diferentes tasas de expulsión de huevos no miméticos en tres áreas geográficas diferentes.]

There are major differences between these four brood parasite species with respect to their parasitic strategies and the responses of their hosts. For instance, both European cuckoos are specialist brood parasites (the common cuckoo at the individual female level (Gibbs et al., 2000)), parasitising only one or just a few host species, while brownheaded as well as shiny cowbirds are generalists. Nevertheless, being a generalist is not characteristic of all parasitic cowbirds; one species in South America, the screaming cowbird M. rufoaxillaris, is an extreme specialist, using one species as the host almost exclusively (Fraga, 1998). On the other hand, there is another difference with respect to the evolution of the coloration of parasitic eggs and the egg-rejection strategies used by their hosts: host rejection of non-mimetic eggs has selected for the evolution of host-egg mimicry by the common cuckoo (Davies and Brooke, 1988) resulting in sympatric host-specific races -or gentes- that lay eggs matching those of each particular host species (Brooke and Davies, 1988; Moksnes and Røskaft, 1995; Gibbs et al., 1996; Stoddard and Stevens, 2010). It has frequently been emphasised that cowbirds have not evolved egg mimicry (Rothstein, 1990; Gibbs et al., 1997; Davies, 2000; Stokke et al., 2002) but egg mimicry has not evolved in the great spotted cuckoo either (Soler et al., 2003). thus, absence of egg mimicry is not a special characteristic of parasitic cowbirds.

Egg rejection rates by European and American hosts

Brood parasitism in general, and rejection rates in particular, have been reviewed recently (Davies, 2011; Kilner and Langmore, 2011; Feeney et al., 2014; Soler, 2014; Medina and Langmore, 2015) but differences in rejection rates between Europe and America have not been discussed since Rothstein 1990, 1992). The most important and clear difference between brood parasite-host interactions in Europe and America is the rejection rates. Rothstein (1990, 1992) emphasised that there were sharp differences in rejection rates between host species from Europe and America. In particular, most potential cowbird hosts in America show either close to 100% or 0% rejection of experimental non-mimetic eggs and most common cuckoo hosts in Europe show intermediate rejection rates. Since Rothstein's review, many new potential host species have been tested in egg-recognition experiments and we now have information on nearly three times as many hosts as in Rothstein's study (see  table S1 in Supplementary Electronic Material (Art2_Soler M_Supplementary Material.pdf)). As evident in fig. 1, which presents the number of potential host species relative to rejection rates in three different geographical areas, the scenario described by rothstein remains valid: intermediate rejection rates (i.e. between 20%–80%) are almost absent in Nearctic and Neotropical hosts (7.6% and 15%, respectively) while they are more frequent in Europe (i.e. common cuckoo hosts; 36%; χ² = 14.5, P = 0.001 and χ² = 5.01, P = 0.02, respectively; fig. 1). two apparent differences can be mentioned: firstly, the number of Neotropical hosts showing a rejection rate close to 0% has increased to some extent (from 13 out of 22 species (59.1%) in Rothstein's study to 27 out of 40 species (67.5%) now; fig. 1); and, secondly, the number of common cuckoo hosts showing a rejection rate close to 100% has also increased (from 6 out of 18 species (33.3%) in Rothstein's study to 23 out of 50 species (46%) now; fig. 1). However, differences are not statistically significant in either case (χ² = 0.44, P = 0.51 and χ² = 0.87, P = 0.35, respectively).

The more complete information available presented in fig. 1 suggests three main conclusions: an initial one that was suggested by Rothstein (1990) and two others derived from the new data set, which were not indicated by Rothstein's data:

Firstly, a rejection rate of close to 0% is much more frequent in Nearctic and Neotropical hosts (48.5% and 67.5%, respectively) than in European common cuckoo hosts (18.0% χ² = 11.57, P = 0.007 and χ² = 22.69, P < 0.0001, respectively; fig. 1). The main explanation for this difference is that cuckoos are an ancient lineage that evolved 65–144 million years ago (Davies, 2000), whereas cowbirds appeared only 2.8–3.8 million years ago (Rothstein et al., 2002).

Secondly, in European common cuckoo hosts, a rejection rate of close to 100% is more frequent than previously assumed by Rothstein (1990) (see above; fig. 1). Many recent studies have reported further cases of potential cuckoo host species that are not currently parasitised but exhibit very high rejection rates of non-mimetic eggs (Braa et al., 1992; Honza et al., 2004; Procházka and Honza, 2003; Stokke et al., 2004; Rutila et al., 2006; Vikan et al., 2009; Martín-Vivaldi et al., 2012) suggesting that they underwent high parasitism pressure in the past and that rejection behaviour is a cost-free defence in these species (reviewed in Soler, 2014).

Thirdly, a rejection rate close to 100% is very similar in Nearctic (43.9%) and in European (46%) hosts, and higher than that reported in Neotropical ones (17.5%) (χ2 = 7.76, P = 0.005 and χ2 = 8.12, P = 0.004, respectively; fig. 1). This last point has never previously been noted or discussed.

Host size and the evolution of egg-rejection behaviour

Three major potential long-term evolutionary outcomes of brood parasite-host coevolution can be distinguished (Soler, 2014): lack of rejection (when the rejection rate is 10% or less), successful resistance (when egg rejection is 90% or more) and coevolutionary cycles, which always occur in species showing an intermediate rejection rate (10%–90%; Soler, 2014). The results of simple exploratory comparisons (χ²), considering the 159 species included in  table S1 (Supplementary Electronic Material) (Art2_Soler M_Supplementary Material.pdf), show that host size has an important effect on the evolutionary outcomes reached with respect to egg-rejection rates (fig. 2; see also  table S1 in Supplementary Electronic Material (Art2_Soler M_Supplementary Material.pdf)). more importantly, firstly, lack of rejection is more frequent in small Nearctic and Neotropical host species than in small European hosts (χ² = 5.28, P = 0.02 and χ² = 8.67, P = 0.03, respectively; fig. 2a). Secondly, intermediate rejection rates are more abundant among small European host species compared to small Nearctic and Neotropical hosts (χ² = 11.54, P = 0.0007 and χ² = 24.53, P < 0.0001, respectively; fig. 2B). Thirdly, successful resistance is considerably more frequent (a) in small European host species compared to small Nearctic and Neotropical hosts (χ² = 27.02, P < 0.0001 in both cases; fig. 2C); (b) in medium-sized Nearctic host species compared to medium-sized European or Neotropical hosts (χ² = 30.22, P < 0.0001 and χ² = 26.47, P < 0.0001, respectively; fig. 2C); and (c) in large Nearctic host species compared to large European or Neotropical hosts (χ² = 15.48, P = 0.0001 in both cases; fig. 2C). Interestingly, an evolutionary outcome of successful resistance is rare among Neotropical hosts regardless of their size (fig. 2C). The first two results have been discussed elsewhere (Rothstein, 1990, 1992; Davies, 2000) but the third point has never before been mentioned, despite spectacular differences: of hosts presenting successful resistance, 17 out of 19 (89.5%) small hosts are from Europe while 16 out of 17 (94.1%) medium-sized and 11 out of 13 (84.6%) large hosts are from north America (fig. 2C). Undoubtedly, the 17 European hosts have evolved successful resistance as a response to parasitism by the common cuckoo, given that all of them are considered to be (or have been in the past) hosts of that species (Davies, 2000; see  table S1 in Supplementary Electronic Material (Art2_Soler M_Supplementary Material.pdf)). Which brood-parasite is responsible for the evolution of successful resistance in the medium-sized and large hosts of North America? With respect to medium-sized hosts, egg rejection pre sumably evolved as a response to parasitism by bronzed cowbirds Molothrus aenus, which usually parasitise hosts that are as large or larger than themselves (Carter, 1986), or by brown-headed cowbirds, which in the past could have preferentially parasitised large hosts (Peer and Sealy, 2004a) because these are able to provide more food and rear a larger number of parasitic nestlings (Rothstein, 1975a; Peer and Sealy, 2004a). Later, when most medium-sized hosts evolved egg-rejection ability, brown-headed cowbirds would have been forced to switch to smaller hosts (Rothstein, 1975a; Mason, 1980; Peer and Sealy, 2004a). In fact, currently, only two of the 16 medium-sized hosts are considered common hosts of the brown-headed cowbird, seven are only rarely parasitised, and seven have never been found to be parasitised (i.e. are considered non-host species; see  table S1 in Supplementary Electronic Material (Art2_Soler M_Supplementary Material.pdf)). However, the case of large hosts is different and deserves more detailed analysis.

Fig. 2.

—Number of potential host species presenting each coevolutionary outcome according to host body mass and continent. In terms of mass, hosts were considered to be small, medium and large when weighing < 30 g, 30–75g and > 75 g, respectively. For the coevolutionary outcomes, lack of rejection (fig. 2a) was considered to apply when the rejection rate was 10% or less, successful resistance (fig. 2C) when egg rejection was 90% or more, and intermediate rejection rate (fig. 2B) when in between.

[Número de especies hospedadoras potenciales que presentan cada uno de los resultados coevolutivos según peso y continente. En términos de peso, las especies hospedadoras fueron consideradas como pequeñas, intermedias y grandes cuando pesaban < 30, entre 30 y 75, y > 75 g, respectivamente. Para los resultados coevolutivos, se consideró ausencia de expulsión (fig. 2A) cuando la tasa de expulsión era 10% o menos, resistencia exitosa (fig. 2C) cuando la tasa de expulsión era 90% o más, y tasa de expulsión intermedia (fig. 2B) cuando dicha tasa de expulsión estaba entre 10% y 90%.]

Egg rejection by large-sized host species in North America vs. Europe

Several of the large hosts in North America are sympatric with larger cowbird species in Mexico and Central America (the bronzed cowbird and giant cowbird Molothrus oryzivora). Thus, rejection ability in some large hosts could have been selected under pressure from these cowbirds, especially in the case of grackles Quiscalus spp. as suggested by Peer and Sealy (2004a). However, this possibility is less likely in the case of corvids because giant cowbirds usually parasitise only large colonial icterids (Ortega, 1998).

Most potential large hosts in North America (11 out of 15, 73.3%; see  table S1 in Supplementary Electronic Material (Art2_Soler M_Supplementary Material.pdf)) not only exhibit successful resistance but in addition their high rejection ability has been maintained for a long period of time in the absence of conspecific brood parasitism (Bolen et al., 2000; Underwood et al., 2004), even after speciation processes (Bolen et al., 2000; Rothstein, 2001; Peer and Sealy, 2004b), signifying that they suffered from very costly brood parasitism and also that they won the arms race against their brood parasites (Rothstein, 2001; Soler, 2014).

It has been suggested that egg rejection in potential large hosts in North America has also been mediated by parasitism by cowbirds during the Pleistocene, about 10,000 years ago, when cowbirds were presumably more abundant (Peer et al., 2011a; Peer et al., 2011b). This possibility seems unlikely because most of these host species (see  table S1 in Supplementary Electronic Material (Art2_Soler M_Supplementary Material.pdf), and table 1) are too large to be considered appropriate cowbird hosts, given that when cowbirds parasitize large hosts, cowbird nestlings are not able to overcome the size advantage of host nestlings (Dearborn and Lichtenstein, 2002; Rivers et al., 2010) and most of these species are much larger than any current cowbird hosts. Thus, brownheaded cowbirds cannot have been responsible for provoking the strong selection pressures needed to favour the evolution and fixation of successful resistance in these large hosts.

The egg-rejection ability of current Nearctic species could also have evolved in an ancestral species that was parasitised by the common cuckoo in the Palaearctic (Bolen et al., 2000; Rothstein, 2001; Underwood et al., 2004). This second possibility could be correct in the case of some medium-sized species such as the loggerhead shrike Lanius ludovicianus and the American robin Turdus migratorius, which belong to Old World genera in which all species are rejectors (Lovászi and Moskát, 2004; Grim et al., 2011; Martín-Vivaldi et al., 2012).

Until now, the same explanation has been provided for most of the eight north American corvid species listed in table 1: rejection ability could have evolved as a response to common cuckoo parasitism in a common ancestor of the three species of scrub-jays Aphelocoma spp. and the blue jay Cyanocitta cristata (Peer et al., 2007) and in a common ancestor of the two magpies Pica spp. (Rothstein, 2001; Bolen et al., 2000; Underwood et al., 2004). However, this seems very unlikely in the first four species because rejection is not fixed in the corvid family, i.e. only some species are able to eject experimental non-mimetic eggs (Soler, 1990). Moreover, the common cuckoo does not parasitise corvids, other than the azure-winged magpie Cyanopica cyanus in Asia (Nakamura, 1990; Nakamura et al., 1998; but not the Iberian magpie C. cooki in Europe). The great spotted cuckoo does so but it only regularly parasitises magpies and carrion crows Corvus corone and only magpies show rejection capacity (Soler, 1990; see table 1).

Table 1

Relevant information on mass, host status, and rejection rate of non-mimetic eggs in all corvid species in which egg-recognition experiments have been performed on any continent.

[Información relevante sobre peso, estatus y tasa de expulsión de huevos no miméticos en todas las especies de córvidos en las que se han hecho experimentos de reconocimiento de huevos en cualquiera de los continentes.]

This possibility of rejection ability evolving as a response to cuckoo parasitism in a common ancestor is also very unlikely in the two North American magpie species, because rejection rates of non-mimetic eggs reported in both of them (100%; Bolen et al., 2000; Underwood et al., 2004) are higher than those reported in the Eurasian magpie not only in allopatric populations but even in populations that are sympatric with great spotted cuckoos, in which the rejection rate is significantly higher than in allopatric populations (Soler et al., 1999). In some allopatric European Eurasian magpie populations rejection rates have been reported to reach about 50% (52.9–57.9% in four of them; Soler et al. 1999) and even much lower; for example 25% in Eljovo, Bulgaria (Soler et al., 1999); and 0% in Uppsala, Sweden (Soler and Møller, 1990). Furthermore, in Seoul, South Korea, the rejection rate by magpies is also 0% (Sang-im Lee; personal communication). Even more, considering mimetic eggs, the rejection rate by North American black-billed magpies Pica hudsonia (50%) was much higher than those reported for magpies in European allopatric (with the great spotted cuckoo) populations (7.1%–26.3% ejection; Soler et al., 1999), and even higher than those reported for magpies in the majority of European sympatric populations: only in two of them was the ejection rate of mimetic eggs higher than 50%, while in six of them it was lower (18.8%–43.8%; Soler et al., 1999).

It has also been suggested by some authors as an unlikely possibility that egg recognition in North America corvids evolved as a response against an extinct brood parasite, either a cowbird (Peer and Rothstein, 2010) or a cuckoo (Rothstein, 2001; Peer et al., 2011b). Here I suggest that this is the most probable hypothesis for two main reasons:

Firstly, although recognition and rejection capacity could be maintained in absence of recognition and rejection costs (Davies et al., 1996; Rothstein, 2001; Peer et al., 2011a), what cannot be accepted is an increase and establishment of that capacity in North America magpie populations in the absence of selection pressures exerted by inter- or intraspecific brood parasitism.

Secondly, the American crow Corvus brachyrhynchos also presents a considerable rejection rate (21% of non-mimetic eggs and 8% of mimetic eggs Underwood et al., 2004; table 1) whereas in the other continents no species of the genus Corvus has been found to discriminate against non-mimetic eggs (Yom-Tov, 1976; Soler, 1990), even where it is the favourite host of a brood parasite, as in the case of the house crow Corvus splendens in India (Dewar, 1907; in Davies, 2000). Begum et al. (2012) reported a low ejection rate among house crows in Delhi (India; 1 out of 22), but this result is strange considering that Dewar (in Davies, 2000) found that house crows accepted chicken eggs and even golf balls experimentally introduced in their nests. Perhaps the experimental egg that was considered ejected disappeared for some other reason. Another strange result in Begum et al., (2012) concerns large-billed crows Corvus macrorhynchos since they found that seven out of 16 non-mimetic eggs were deserted (none was ejected); as the authors themselves stated, it is very difficult to understand why birds that are perfectly able to grasp-eject, desert nests instead of just ejecting the model egg. Clearly, a control group in which nests were not experimentally parasitised would be needed (Soler et al., 2011) to assess the significance of this result, given that nest desertion may occur in response to predation risk and other factors unrelated to brood parasitism, including disturbance by the researcher (Székely et al., 1996; Hosoi and Rothstein, 2000; Servedio and Hauber, 2006).

Thus, I suggest that the higher discrimination capacity shown by large potential North America host species, especially corvids, despite absence of selective pressures by brood parasites in recent history, strongly implies that rejection in these species evolved in response to selection pressures provoked by at least one extinct brood parasite in the New World, either an extinct large cowbird species or an extinct cuckoo species. The possibility of a large cowbird would be supported by the fact that high parasitism rates exert strong selection for host defences and parasitism rates on most cowbird hosts are usually much higher than those on cuckoo hosts. Thus, the high occurrence of large species with fixed rejection could have been provoked by an abundant and largesized extinct cowbird species. However, considering the high discrimination capacity retained by large potential host species in North America, the extinct brood parasite more likely was a Cuculidae species, because although virulence varies enormously, and many small host species of cowbirds are unable to raise any of their own chicks (Rothstein, 1975b), cowbirds are in general less virulent brood parasites than cuckoos, as shown by the low rejection rate (if any) evolved in contemporary cowbird hosts, regardless of the duration of sympatry (Davies, 2000). Numerous fossils of cowbirds (Lowther, 1993) and cuckoos (Davies, 2000) have been found. Thus, future research looking for fossils of brood parasites could solve the question of which of the two alternative hypotheses on the existence of an extinct brood parasite (cowbird or cuckoo) is correct. If it was a cuckoo species, it was probably larger than the great spotted cuckoo because parasitism by the great spotted cuckoo of its large carrion crow host (a similar size to C. brachyrhynchos) is not costly enough to favour the evolution of egg rejection in this host species: the cuckoo nestlings are unable to outcompete their larger host nestmates; Soler et al., 2002).

Conventional brood-parasitism theory frequently assumes that when a host overcomes its brood parasite, the parasite will be forced to switch to another host species with less-evolved defences (Davies and Brooke, 1989; i.e. the evolutionary alternation model; Nuismer and Thompson, 2006), while the possibility that the brood parasite could go extinct is hardly ever considered (Rothstein, 2001; Peer et al., 2011a, 2011b). This is probably the consequence of the traditional focus on the study of a pair of traits (i.e. defence-counterdefence). However, in consideration of the currently widely accepted idea that hosts have evolved several different lines of defence that can operate at different stages of the breeding cycle, and that brood parasites need to evolve counter-adaptations against all of these (Davies, 2011; Soler, 2014), the possibility of parasite extinction should to be considered a more probable event (Soler et al., 2014). For instance, individual-based simulations by Gilman et al. (2012) have suggested that when coevolution involves multiple traits, victims have an evolutionary advantage in any victim-exploiter system. this means that hosts need to overcome their brood parasite with only one effective defence in order to escape from brood parasitism, whereas brood parasites must overcome all effective host defences (Gilman et al., 2012).

Conclusion

Here I have argued that the high rejection capacity shown by large-sized potential hosts, especially corvids, from North America, in comparison to other European and Asian corvid species, evolved as a response to a highly virulent extinct brood parasite, either an extinct large cowbird or, more likely, an extinct cuckoo species, which lost the arms race against their large hosts. Recent studies in the field of brood parasitism have provided information on rejection rates in many potential host species that has allowed the identification of the crucial effect of host size on the evolution of egg-rejection behaviour, discussed in this paper. The challenge for experimental researchers in the near future will be to provide information about eggrecognition behaviour in more species, and more populations of the same species, in order to fully understand the coevolutionary relationships between brood parasites and their hosts with respect to egg-rejection defence.