MECHANISMS OF AVIAN EGG RECOGNITION: POSSIBLE LEARNED AND INNATE FACTORS STEPHEN I. ROTI-ISTEIN THE belief that at least some passerinebirds reject nonmimetic eggs placedin their nestswas confirmedlong ago by the experimentsof Swyn- nerto.n (1916, 1918) and Rensch (1924). The latter worker (Rensch 1925) went on to questionwhether such birds actually recognizetheir own eggs(rejection by true egg recognition)or whetherthey simply reject any eggthat differs from the majority (rejectionby discordancy).Rensch was interestedin egg recognitionbecause he believedit to be vital to the evolutionaryinteractions between brood parasitesand their hosts. Whether birds rejected eggson the basis of true recognitionor discordancy,the behaviorwould still function as an efficient antiparasiteadaptation because broodparasites generally deposit only one eggin eachhost nest. Although Rensch's(1925) experimentshave been widely interpretedas demonstrat- ing rejection by discordancy(e.g. Welty 1963), the bulk of his results actually indicate true egg recognition(Rothstein 1970). Numerous re- cent experimentsand a literature reviewsuggest that most or all passefines that reject foreign eggspractice true egg recognition (Rothstein 1974). Nearly all these recent experiments,though, were conductedon birds that had completedtheir clutch. The fact that birds that reject foreign eggspractice true egg recognitionafter completingtheir clutch prompts two significant questions: (1) Is this recognitionas highly developedat earlier stagesof the breedingcycle? (2) Which componentsof rejection by true egg recognitionare primarily learned and which are primarily innate? This paper reports on experimentsthat were designedto deal with these two questions. The new experimentsreported here were done on naturally breeding Gray Catbirds (Dumetella carolinensis). Previous experimentsshowed that catbirds are extremely intolerant of foreign eggs placed in their nests. Singleartificial or real cowbirdeggs experimentally added to 30 catbird nests in eastern North America (Connecticut, Maryland, and Michigan) were all removedby the catbirds. Catbirdsin westernNorth America (Manitoba and Nebraska) may be slightly more tolerant, with 22 of 25 nestsyielding ejectionsof singlecowbird eggs. In 17 additional catbird nests (from Connecticutand Maryland), where experimentsre- sulted in clutchesconsisting of only cowbirdeggs or of only one catbird egg and two or more cowbird or other type of foreign eggs, the catbirds ejectedthe foreigneggs in everyinstance (see Rothstein 1974 for deta/ls). 796 The Auk 91: 796-807. October 1974 October 1974] .4vian Egg Recognition 797 Friedmann (1963) has stated that while suchintolerance of foreign eggs servesas an efficient host defenseit is probably not evolvedin response to broodparasitism. I suggestthis intolerancecan be explainedonly in terms of an adaptation that evolved becauseit protects birds from brood parasites. This is the most parsimoniousexplanation, as rejectingforeign eggsdoes not seemto have any adaptive value in most birds, other than in the contextof broodparasitism (Rothstein 1970, MS). Althoughcatbird and cowbirdeggs differ little in size (averaging23.3 x 17.5 mm and 21.45 x 16.42 mm, respectively,according to Bent 1948, 1958), they are strikingly different in colorationand thus providea valu- able pair for experimentation.Catbird eggsare immaculatewith a blue- green ground color whereascowbird eggsusually have a whitish ground colorand are heavily mottled with brown and gray spots. Previouswork by Tschanz (1959) has shownthat in a nonpasserine, the CommonMurre (Uria aalge), egg recognitionis realized througha learning process. In the case of a passerine,the Village Weaverbird (Ploceus cucullatus), it has been suggestedthat egg recognitionis also learned (Victoria 1972). Given the variable nature of murre and weaver- bird eggsthese findings might not be unexpected.The eggsof most song- birds, the catbird included, show little variation and egg recognition through largely innate means might not be difficult to evolve. Never- thelessthe hypothesisof egg recognitionby learning will be supported by a demonstrationof greater tolerancetoward a foreign egg type in the early stagesof the breedingcycle than in later stages. If egg recognition has a learning component,greater tolerancein the early stagesof the nestingcycle would be expectedwhether the birds learn the appearance of their eggsanew with each breedingattempt, or whetherthe learning occursonly in completelynaive individualsduring the first nesting at- tempt of their lives. In the latter casethough, increased acceptance early in the cycle would occur only at those nests tended by naive birds. Un- fortunately it is impossibleto determine whether the catbirds tested in the experimentsreported here were naive or experienced;but given the high mortality of passerinesit is reasonableto assumethat within any seriesof nestsexperimented upon in the field somewill be tendedby naive and others by experiencedindividuals. Much of what follows is relevant only to the catbird and to other speciesthat I have termed "rejecters." Experimentswith cowbird eggs have shown little intraspecificvariation in responseto artificial cowbird parasitism. A minority of North American speciesreject cowbird eggs at nearly 100% rates. These are the rejecters. Most speciesaccept the eggs at nearly 100% rates and are termed "accepters." Rejecter and 798 STEPH• I. ROTHST•n• [Auk, Vol. 91 accepter specieshave presumably respondedand not responded,respec- tively, in a major way to the evolutionarypressures of brood parasites (Rothstein 1970). As accepter speciesdemonstrate little or no egg rec- ognition (except under special circumstances(Rothstein 1970, MS)), questionson their acquisitionof this behavior have reduced relevancy. METItODS AND CONTROLS FOR ARTIFICIAL EGGS The artificial eggs used in most experiments were cast in plaster of paris and painted with artist's paints of the acrylic polymer type. Artificial cowbird and catbird eggs were also shellacked. I have elsewhere (Rothstein 1970, 1974) described the production and properties of these eggs in greater detail. Suitable controls for artificiality of the eggs have been performed. For example, at four nests all four of the catbirds' own eggs were replaced with one artificial catbird egg plus three artificial cowbird eggs. In all four cases the cowbird eggs were ejected while the artificial catbird egg was left in the nest. At another five nests a single artificial catbird egg added to a nest with two or more real catbird eggswas accepted,where- as a single real or artificial cowbird egg was rejected under similar conditions at 52 of 55 nests from five localities in North America. These different acceptance rates (5-o vs. 3-52) are significant at P < 0.005 (Fisher Exact Probability Test, Siegel 1956). Throughout this paper data on which statistical tests have been per- formed will be reportedas in the precedingsentence. The entire two-by-two con- tingency table can be constructedfrom the information given (e.g. 5-0 vs. 3-52). To determinewhether catbirds possesstrue egg recognitionat an early stage in their breeding cycle and whether any recognitionthat occursis learned, the fol- lowing manipulation (experimentI) was performed: The first egg laid in 11 catbird nests was replaced with an artificial cowbird egg. The time interval between egg laying and initiation of the experimentis unknown, but it could have ranged from a few minutes to 4• h. The nestswere checkedlater the same day and then the next morning. In some casesadditional egg manipulations were conducted. Of the 11 nestsstudied 10 were in New Haven County, Connecticutduring May and June 1969 and 1970; the 11th nest, number 72-215, was in Anne Arundel County, Mary- land in May 1972. R•su•.Ts FROM PRt•A•¾ EX2•R•TS Resultsof experimentI are summarizedin Table 1. At three nests, the cowbirdegg was missingby the sameafternoon of the day it had beeninserted. At anotherthree, it wasmissing and a secondcatbird egg was in its place the next morning. The catbirdsat thesesix nestsclearly recognizedtheir own egg type even at the earlieststages of clutch initia- tion. At the threenests where the cowbirdegg was ejected by the same afternoon,it was ejectedin the absenceof any cuesor informationthat one or more catbird eggsmight have supplied. It cannot be determined whetherthe cowbirdeggs in the other three nestswere ejectedbefore or after the secondcatbird egg was laid. But it is of coursecertain that they werenot the minoritytype eggat the time of their ejection. It could be arguedthat the missingcowbird eggs in thesesix nestswere removed October 1974] Avian ,Egg Recognition 799 TABLE 1 RESPONSES OF CATBIRDS TO ARTItlCIAL COWBIRD EGGS PLACED IN TIllER NESTS' Status of cowbird egg Status of cowbird egg after 1 day after 2 days Experimental procedure and location Accepted Ejected Accepted Ejected I First catbird egg laid replaced with an artificial cowbird egg (experiment done within 4• h 5 6 4 7 after the first egg was laid); Connecticut, Maryland II One artificial cowbird egg added to clutch when there were at least two catbird eggs present; 4 25 2 27 Connecticut, Maryland, Michigan x A catbird egg was removed at the time the cowbird egg was added in 23 of the 29 experiment II nests. Whether or not a catbird egg is removed at the time of experimentation has no effect on response (Rothstein 1970). At one experiment II nest a real cowbird egg was used rather than an artificial one. It was ejected within a day.