CORE Metadata, citation and similar papers at core.ac.uk

Provided by Elsevier - Publisher Connector Current Biology 20, 1830–1833, October 26, 2010 ª2010 Elsevier Ltd All rights reserved DOI 10.1016/j.cub.2010.09.005 Report A Simple Rule Reduces Costs of Extragroup Parasitism in a Communally Breeding

Christina Riehl1,* In this study, I used a recently developed technique to isolate 1Department of Ecology and Evolutionary Biology, Princeton and genotype maternal DNA from cells on the external surface University, 106A Guyot Hall, Princeton, NJ 08544, USA of the eggshell [14], enabling noninvasive genetic identification of parasitic . I collected data on group membership and parasitism from a color-banded, genetically sampled popula- Summary tion of wild and conducted field experiments on the same population. This study is therefore the first to quantify How do cooperatively breeding groups resist invasion by conspecific brood parasitism in a cooperative nester, to parasitic ‘‘cheaters,’’ which dump their eggs in the com- show that parasites lay eggs out of synchrony with the hosts munal nest but provide no parental care [1,2]? Here I show they parasitize, and to experimentally determine the mecha- that Greater Anis (Crotophaga major), Neotropical nism by which hosts recognize asynchronous eggs. that nest in social groups containing several breeding The Greater is a Neotropical that nests on females [3], use a simple rule based on the timing of laying forested lake and river edges from central Panama to to recognize and reject eggs laid by extragroup parasites. I Argentina. Breeding groups typically consist of either two or experimentally confirmed that Greater Anis cannot recog- three unrelated, socially monogamous pairs that build a single nize parasitic eggs based on the appearance of host nest in which all of the females lay their eggs; lone pairs are phenotypes or on the number of eggs in the clutch. However, extremely rare. Prior to laying her first egg in the communal they can discriminate between freshly laid eggs and those nest, each female ejects any eggs that have already been that have already been incubated, and they accordingly eject laid by her fellow group members. After she lays her first asynchronous eggs. This mechanism is reliable in naturally egg, she typically stops removing eggs. Ejection, therefore, parasitized nests, because group members typically lay ceases only when all of the females in the group have begun their eggs in tight synchrony, whereas the majority of para- to lay, enforcing a tight reproductive synchrony among group sitic eggs are laid several days later. Rejection of asynchro- members [3,15]. Incubation begins with the penultimate or nous eggs therefore provides a rare empirical example of third-to-last egg, and late-hatching nestlings are unlikely to a complex, group-level behavior that arises through rela- survive [16]. All of the group members participate in parental tively simple ‘‘rules of thumb’’ without requiring advanced care, including incubation, nest defense, and food delivery to cognitive mechanisms such as learning, counting, or indi- nestlings. Group members also participate in complex, highly vidual recognition. ritualized group choruses, which may play a role in synchro- nizing reproduction and strengthening social bonds among Results and Discussion group members [3]. Nest parasitism was common in the study population near Conspecific brood parasitism, in which a parasitic female lays Barro Colorado Island, Panama. Over 4 years (2006–2009), her eggs in the nest of a conspecific host, is a common repro- 40% of Greater Ani nests were parasitized by females that ductive strategy in many species of birds [2]. Counter-adapta- were not members of the social group, accounting for 7% of tions to parasitism have evolved in several lineages, primarily all eggs laid in the population (n = 48 broods and 411 eggs involving recognition and rejection of parasitic eggs [4–6]. overall). However, parasites were rarely successful: 70% of Avian brood parasites and their hosts have therefore served parasitic eggs were ejected from the host nest (21 of 30), as model systems for understanding the evolutionary pres- whereas fewer than 2% of host eggs were ejected after all of sures that give rise to learning, counting, and recognition the host females had begun to lay (6 of 381). This dispropor- 2 [7,8]. However, egg recognition is notably rare in communally tionate rejection of parasitic eggs was significant (X 1 = 196, breeding species, in which more than one female will routinely p < 0.0001) and was positively correlated with differences in lay eggs in a single nest and contribute parental care to the joint the timing of egg laying between parasite and host females. clutch. In these systems, the benefits of accepting offspring The majority of parasitic eggs were laid several days after that are not one’s own—and the costs of mistakenly rejecting the onset of incubation at the host nest (Figure 1), and the the wrong offspring—have presumably constrained the evol- degree of asynchrony between the parasitic egg and the ution of individual recognition and discrimination [9–11]. This host clutch positively covaried with its probability of ejection presents an evolutionary paradox that has generated consider- (generalized linear mixed model with host nest as a random able theoretical interest [1,12,13] but little empirical data. If effect, controlling for host clutch size and parasitic egg 2 cooperatively breeding birds are unable to recognize their mass: log likelihood ratio X 1 = 6.9, df = 1, p < 0.01). Data own eggs, how do they prevent noncooperating parasites from naturally parasitized nests, therefore, show that Greater from laying eggs in the communal nest? Anis consistently accept parasitic eggs if they are laid in Little is known about the frequency of conspecific brood synchrony with the host clutch but reject them if they are laid parasitism in cooperatively breeding birds, primarily because more than a few days after the hosts have completed laying. it is difficult for researchers to discriminate between eggs Next, I used a series of field experiments to test the mecha- laid by group members and those laid by extragroup females. nism by which Greater Anis recognize asynchronous eggs. The pattern observed at naturally parasitized nests could be explained by either of the two mechanisms that have been *Correspondence: [email protected] posited to explain avian egg recognition: (1) recognition by Extragroup Parasitism in a Communal Breeder 1831

To distinguish between these two hypotheses, I experimen- 10 tally parasitized host nests with foreign eggs and determined 9 whether these ‘‘parasitic’’ eggs were accepted or rejected. 8 To ensure that all foreign eggs were within the range of varia- 7 tion that would be observed at naturally parasitized nests, I 6 used real Greater Ani eggs that were collected from unmanip- 5 ulated nests in the study population. I used a fully factorial, 4 randomized design in which host nests were parasitized % R 3 during either early or late incubation ( 2 days or 7 days after incubation began, respectively) with foreign eggs that appe- 2 ared to have been laid either synchronously or asynchronously 1 relative to the host clutch (Figure 3A). This experimental design 0 ensured that foreign eggs were always in the minority (that is,

Number of parasitic eggs of Number <-2 -1 to 0 1 to 2 3 to 4 5 to 6 >7 host eggs outnumbered foreign eggs in all four treatments) and that discordancy was tested solely on the basis of cha- Days after onset of incubation nges in egg phenotype over time. If recognition cues are learned from the host clutch, I Figure 1. Frequency Distribution of Parasitic Eggs at Naturally Parasitized predicted that hosts would accept foreign eggs in early incu- Nests, in Relation to the Onset of Incubation at the Host Nest bation, but not in late incubation. By contrast, if discrimination Day 0 was defined as the day on which incubation began. Parasitic eggs is based solely on the degree of mismatch between host and were either accepted (white) or rejected (black) by hosts. n = 30 parasitic parasitic eggs, I predicted that hosts would accept foreign eggs laid at 21 host nests. eggs that appeared to be synchronous with the host clutch, but not those that appeared to be asynchronous. In support learning or (2) recognition by discordancy [17]. By the first of the latter hypothesis, eggs that appeared to be asynchro- mechanism, hosts require a critical time period in which to nous relative to the host clutch were often ejected from the learn the appearance of the eggs in their clutch. This mecha- communal nest in both early and late incubation, whereas nism could explain the correlation between asynchrony and those that appeared to be synchronous with the host clutch rejection of parasitic Greater Ani eggs, because rejection of were always accepted (three-dimensional contingency test, 2 parasitic eggs is expected only if they are laid after the hosts X 4 = 15.1, p < 0.005; Figure 3B). Therefore, the results of this have imprinted on their own eggs. Individual female Greater experiment indicated that female Greater Anis do not need Anis do not lay eggs that are uniform in size or appearance, to learn the appearance of their own eggs in order to recognize and membership in nesting groups changes from year to and reject discordant, asynchronous eggs. year [16]. Therefore, group members would have to relearn Next, I hypothesized that if anis rely solely on laying the appearance of all the eggs in the communal clutch during synchrony to guide rejection decisions, then they should also each new nesting attempt, regardless of the age or experience discriminate against nonparasitic eggs that are laid asynchro- of each group member. By the second mechanism, hosts do nously relative to the rest of the clutch. However, such situa- not learn the appearance of their eggs. Rather, they simply tions are rare under natural circumstances: of 381 host eggs reject those that appear dissimilar to the majority of the eggs laid at 48 nests in the study population, only 12 (3.1%) were in the clutch. Hosts treat the more common egg type as their laid more than 2 days after incubation began. Therefore, I con- own, discriminating against discordant eggs that are more ducted a second experiment in which I created artificially likely to be parasitic. This mechanism could also explain the asynchronous clutches composed entirely of host eggs. I correlation between asynchrony and rejection in parasitic randomly selected one egg from each of 10 host clutches on Greater Ani eggs, because newly laid eggs differ in appear- the first day of incubation and retained it for 6 days, and then ance from those that have already been incubated [3]. Freshly I placed it back in its nest on the 7th day of incubation. This laid eggs are coated with vaterite, a white, chalky polymorph of egg therefore appeared to be freshly laid, whereas the other calcium carbonate [18]. This coating is rubbed and abraded eggs in the host clutch had been incubated for all 7 days. during incubation, gradually revealing the deep blue shell The asynchronous egg was ejected at 5 of the 10 nests, underneath (Figure 2). Therefore, natural wear on the external whereas none of the unmanipulated eggs were ejected (Fis- surface of the shell might enable anis to distinguish freshly laid her’s exact test, p < 0.001). eggs from those that have already been incubated and to Because the majority of parasitic females deposited their subsequently eject those that have been laid out of turn. eggs in the host nest so late in incubation that they were

Figure 2. Greater Ani Eggs Change Color over the Course of Incubation (A) Freshly laid eggs (day 0) are coated in white vaterite, which wears off during incubation to reveal the blue shell underneath. All three eggs in this series were laid by one female and sequentially removed from the nest (days 0, 4, and 7) in order to illustrate the natural progression of abrasion on the shell. (B) A communal nest that was naturally parasitized during late incubation. The host eggs all show substan- tial abrasion of the vaterite coating, whereas the freshly laid parasitic egg (center foreground) is still completely white. Current Biology Vol 20 No 20 1832

12 10 Figure 3. Hosts Reject Parasitic Eggs that Are 100 A ‘Synchronous’ ‘Asynchronous’ B Not Laid in Synchrony with the Host Clutch 90 11 80 (A) Schematic illustration of the experimental 10 design. Host nests were artificially parasitized in Early 70 incubation 60 either early or late incubation with a foreign egg 50 (indicated by an arrow) that appeared to have 40 been laid either synchronously or asynchro- 30 nously relative to the host clutch. Late 20 (B) Results of the experiment showing the incubation foreign Percentage 10 proportion of foreign eggs (+90% confidence eggs accepted (+90% CL) (+90% eggs accepted 0 limits) that were accepted by the hosts in each treatment. Sample sizes (number of host nests Sync Async Sync Async per treatment) are given above each bar. Early incubation Late incubation unlikely to hatch anyway, host rejection of asynchronous eggs breeders, for example, might be less sensitive than experi- might seem a redundant defense. However, previous work on enced breeders to the appearance of mismatched, asynchro- this species has shown that egg hatching rates are critically nous eggs in the communal clutch. limited by the number of eggs that can be incubated simulta- In summary, this study provides a rare empirical example of neously in the communal clutch [3]. Hatching rates are parasitic cheaters in an avian cooperative breeder, a strategy typically low (x = 84%; [3]) and decline sharply with increasing long predicted by theory. I found that female group-nesting clutch size, particularly for large clutches. Therefore, Greater Greater Anis do not distinguish between egg phenotypes laid Ani hosts should benefit from removing asynchronous eggs by group members and those laid by extragroup females; from the nest even if these eggs never hatch, because their rather, they accept all eggs that appear to have been laid in presence reduces the probability that viable eggs will hatch. synchrony with their own. These findings support previous Female Greater Anis frequently ejected eggs that were laid theoretical predictions that, in order for communal nesting asynchronously relative to the rest of the clutch, regardless systems to be evolutionarily stable, selection should favor of whether the egg was laid by a parasite or a host. Mistaken the acceptance of foreign eggs rather than the evolution of rejection of host eggs would seem to entail a substantial loss individual egg recognition. of fitness; indeed, the evolution of egg and chick recognition is thought to be fundamentally constrained by the risks and Experimental Procedures costs of such errors [19–21]. In this study, I found that the risk of mistakenly ejecting a host egg was relatively low, occur- Naturally Parasitized Nests ring at only 2% of unmanipulated nests in the study popula- Greater Anis were studied during four breeding seasons (June–October, tion. The cost of ejecting late-laid host eggs should be similarly 2006–2009) at Gatu´ n Lake, Panama, where anis nest in emergent vegetation along the shoreline. Details of the study area and monitoring methods are low: because the incubation period is brief (11–12 days) and given in previous studies [3,16]. In 2006, six nests were monitored, and 45 incubation ceases soon after the first eggs hatch, eggs that adults were trapped in mist nets, color banded, and genetically sampled. are laid more than 3 days after the onset of incubation are During 2007–2009, 27–58 nests were monitored and 21–30 adults were trap- unlikely to hatch, and those that do are nearly always outcom- ped yearly. In addition, 335 nestlings were color banded and genetically peted by older nestmates [16]. Ejection of asynchronous eggs, sampled over the 4 year period. A total of 132 nests were monitored during then, could potentially enforce synchrony among group the study. However, because the majority was depredated before hatching, this analysis was restricted to the 48 broods lacking helpers for which I had members in addition to reducing the costs of parasitism by complete information on genetic egg maternity, laying sequence, and egg nongroup members. fate. Of these groups, 35 contained two breeding females (four individuals A second type of recognition error is also possible in this total) and 13 contained three breeding females (six individuals total). Group 2 species: the ejection of an egg that appears to be discordant size did not influence the probability of being parasitized (X 1 = 0.02, p > 0.5) 2 based on the appearance of its shell but that was, in fact, or of rejecting parasitic eggs (X 1 = 1.4, p < 0.25). Nests were checked daily; I laid synchronously with the rest of the clutch. The data from considered incubation to have begun if either an incubating adult was present on the nest or the eggs were warm to the touch. nonparasitized nests suggests that this type of recognition error does occur, but rarely. Of the six host eggs that were ejected at unmanipulated nests, four were laid after the onset Genetic Analyses Maternal DNA was isolated from shed cells and blood stains on the surface of incubation (and probably would not have hatched). This of each egg on the day that it was laid [14], and it was genotyped with 12 provides additional evidence that rejection of asynchronous polymorphic microsatellite markers developed for the Greater Ani [24]. eggs enforces synchrony within the group and suggests that Details of DNA extraction and genotyping are given in previous studies the costs of recognition errors are negligible. The risk of [16,24]. I assigned genetic egg maternity to females in communal clutches such errors, albeit small, may explain why a substantial with the ‘‘identity check’’ function in CERVUS 3.0 [25,26], a maximum-likeli- number of parasitic and asynchronous eggs were accepted hood-based program for parentage analysis that can also be used to iden- tify repeat samples from the same individual. Nest parasitism was detected by hosts both at naturally parasitized nests and in the experi- when the number of unique maternal genotypes in the communal clutch ments. exceeded the number of breeding females in the social group. Parasitic Alternatively, variation in rejection rates across groups might eggs were usually easily identified because members of the social group be due to differences in the age and breeding experience of laid 3–7 eggs per clutch, whereas parasites rarely laid more than one. For group members [22,23]. Although the experiments presented clutches in which parasitism was suspected, samples from all eggs in the here show that female Greater Anis do not learn the specific clutch were genotyped again to check for typing error. I then used a poly- merase chain reaction-based method of molecular sexing to confirm that phenotypes of eggs in the host clutch, they do not rule out the foreign genotype was that of an extragroup female, and not contami- the possibility that longer-term experience may influence an nation from one of the males in the social group [27]. Finally, I cross-checked individual’s ability to detect discordant eggs. First-time the genotype of the parasitic egg against the genotypes of all color-banded Extragroup Parasitism in a Communal Breeder 1833

birds in the study population. In all 30 confirmed instances of parasitism, the 10. Koenig, W.D., Mumme, R.L., Stanback, M.T., and Pitelka, F.A. (1995). parasite’s genotype differed from that of any of the breeding females in Patterns and consequences of egg destruction among joint-nesting the social group at a minimum of two loci (typically at multiple loci). acorn woodpeckers. Anim. Behav. 50, 607–621. 11. Cariello, M., Lima, M.R., Schwabl, H., and Macedo, R.H. (2004). Egg Experimental Design characteristics are unreliable in determining maternity in communal For all experiments, nests to be manipulated were chosen at random from clutches of guira cuckoos (Guira guira). J. Avian Biol. 35, 117–124. the study population; no nest was manipulated more than once. I collected 12. Hamilton, W.D. (1964). The genetical evolution of social behaviour. II. naturally ejected eggs from underneath nests in 2006 to use in subsequent J. Theor. Biol. 7, 17–52. experimental manipulations (‘‘donor’’ eggs). Ejected eggs were weighed, 13. Bergmu¨ller, R., Russell, A.F., Johnstone, R.A., and Bshary, R. (2007). On and the contents were drained through a small hole drilled in one end of the further integration of cooperative breeding and cooperation theory. the shell. Prior to being used in experiments, the empty shells were refilled Behav. Processes 76, 170–181. to their original mass with water, and the hole was closed by gluing on 14. Schmaltz, G., Somers, C.M., Sharma, P., and Quinn, J.S. (2006). Non- a small piece of shell. Nests were checked daily following the addition of destructive sampling of maternal DNA from the external shell of bird the donor egg to determine whether it was accepted or ejected by the hosts. eggs. Conserv. Genet. 7, 543–549. Eggs that were relocated under the nest were considered to have been 15. Vehrencamp, S.L. (1977). Relative fecundity and parental effort in ejected; those that vanished were considered to have been depredated. communally nesting anis, Crotophaga sulcirostris. Science 197, This provided a potentially conservative measure of the number of eggs 403–405. ejected by hosts, because it is possible that some eggs could have been 16. Riehl, C. (2010). Egg ejection risk and hatching asynchrony predict egg depredated after being ejected (and incorrectly excluded). If the donor mass in a communally breeding cuckoo, the Greater Ani (Crotophaga egg remained in the host nest for more than 2 days, it was counted as major). Behav. Ecol. 21, 676–683. accepted. Two days was chosen as the cutoff time because, at naturally 17. Rothstein, S.I. (1975). Mechanisms of avian egg-recognition: Do birds parasitized nests, parasitic eggs that remained in the host nest for longer know their own eggs? Anim. Behav. 23, 268–278. than 2 days were never ejected. Statistical analyses and additional details 18. Board, R.G., and Perrott, H.R. (1979). Vaterite, a constituent of the of the experimental manipulations are given in the Supplemental Experi- eggshells of the nonparasitic cuckoos, Guira guira and Crotophagi mental Procedures. ani. Calcif. Tissue Int. 29, 63–69. 19. Lotem, A., Nakamura, H., and Zahavi, A. (1995). Constraints on egg Supplemental Information discrimination and cuckoo-host coevolution. Anim. Behav. 49, 1185–1209. Supplemental Information includes Supplemental Experimental Procedures 20. Davies, N.B., Brooke, M.de L., and Kacelnik, A. (1996). Recognition and can be found with this article online at doi:10.1016/j.cub.2010.09.005. errors and probability of parasitism determine whether reed warblers should accept or reject mimetic cuckoo eggs. Proc. R. Soc. Lond. B. Acknowledgments Biol. Sci. 263, 925–931. 21. Shizuka, D., and Lyon, B.E. (2010). Coots use hatch order to learn to This study was supported by graduate research fellowships from the Smith- recognize and reject conspecific brood parasitic chicks. Nature 463, sonian Tropical Research Institute and the National Science Foundation, as 223–226. well as by additional funding from the Max Planck Institute of Ornithology 22. Lotem, A., Nakamura, H., and Zahavi, A. (1992). Rejection of cuckoo (Department of Migration and Immuno-Ecology), the American Ornitholo- eggs in relation to host age: A possible evolutionary equilibrium. Behav. gists’ Union, the American Museum of Natural History, and the Program of Ecol. 3, 128–132. Latin American Studies at Princeton University. Research was approved 23. Langmore, N.E., Cockburn, A., Russell, A.F., and Kilner, R.M. (2009). by the Care and Use Committees of Princeton University and the Flexible cuckoo chick-rejection rules in the superb fairy-wren. Behav. Smithsonian Tropical Research Institute. Microsatellite genotyping was Ecol. 20, 978–984. carried out with the assistance of the Cornell Laboratory of Ornithology 24. Riehl, C., and Bogdanowicz, S.M. (2009). Isolation and characterization and the Naos Island Laboratory of the Smithsonian Tropical Research Insti- of microsatellite markers from the greater ani Crotophaga major (Aves: tute. L. Jara and W. Webber provided valuable field assistance, and M.C. Cuculidae). Mol. Ecol. Res. http://tomato.biol.trinity.edu/manuscripts/ Wikelski contributed advice and assistance on all aspects of the project. 9-6/mer-09-0270.pdf. 25. Marshall, T.C., Slate, J., Kruuk, L.E., and Pemberton, J.M. (1998). Statis- tical confidence for likelihood-based paternity inference in natural pop- Received: August 5, 2010 ulations. Mol. Ecol. 7, 639–655. Revised: August 26, 2010 26. Kalinowski, S.T., Taper, M.L., and Marshall, T.C. (2007). Revising how Accepted: August 31, 2010 the computer program CERVUS accommodates genotyping error Published online: September 30, 2010 increases success in paternity assignment. Mol. Ecol. 16, 1099–1106. 27. Griffiths, R., Double, M.C., Orr, K., and Dawson, R.J.G. (1998). A DNA References test to sex most birds. Mol. Ecol. 7, 1071–1075. 1. Sachs, J.L., and Rubenstein, D.R. (2007). The evolution of cooperative breeding; is there cheating? Behav. Processes 76, 131–137. 2. Lyon, B.E., and Eadie, J.M.A. (2008). Conspecific brood parasitism in birds: A life-history perspective. Annu. Rev. Ecol. Evol. Syst. 39, 343–363. 3. Riehl, C., and Jara, L. (2009). Natural history and reproductive biology of the communally breeding Greater Ani (Crotophaga major) at Gatu´ n Lake, Panama. Wils. J. Ornithol. 121, 679–687. 4. Jamieson, I.G., McRae, S.B., Simmons, R.E., and Trewby, M. (2000). High rates of conspecific brood parasitism and egg rejection in coots and moorhens in ephemeral wetlands in Namibia. Auk 117, 250–255. 5. Lahti, D.C., and Lahti, A.R. (2002). How precise is egg discrimination in weaverbirds? Anim. Behav. 63, 1135–1142. 6. Stokke, B.G., Rudolfsen, G., Moksnes, A., and Røskaft, E. (2004). Rejec- tion of conspecific eggs in chaffinches: The effect of age and clutch characteristics. Ethology 110, 459–470. 7. Rothstein, S.I. (1990). A model system for coevolution: Avian brood parasitism. Annu. Rev. Ecol. Syst. 21, 481–508. 8. Lyon, B.E. (2003). Egg recognition and counting reduce costs of avian conspecific brood parasitism. Nature 422, 495–499. 9. Frith, H.J., and Davies, S.J.J.F. (1961). Ecology of the magpie goose, An- seranas semipalmata Latham (Anatidae). CSIRO Wildl. Res. 6, 91–141.