The Auk 113(2):370-380, 1996 A COMPARATIVE ANALYSIS OF PASSERINE MOBBING CALLS MILLICENT SIGLERFICKEN TM AND JAMESPOPP 2 •Departmentof BiologicalSciences; and 2FieldStation, University of Wisconsin-Milwaukee, Milwaukee, Wisconsin53201, USA ABSTRACT.--Weexamined the acousticstructure of mobbingcalls of 52 speciesof passerines representingan array of taxa.Marler suggestedthat callsgiven during mobbinghave char- acteristicsenhancing locatability that include abrupt onsetsand a wide frequencyrange. Only aboutone-half of the specieswe studiedproduced calls during mobbingthat had these characteristics.However, clicks(probably produced by bill snapping)sometimes were given during divesat predators.Vocalizations given during mobbingwere quite diversein acoustic structure.Two trendswere noted:phylogenetic patterns; and possibleconvergence in certain speciesengaged in interspecificflocking. Received4 May 1995,accepted 7 September1995. ATTENTION CONCERNING the evolution of harassmentin causingthe predatorto leave the avian vocalizations has focused on song (e.g. vicinity, hasa considerableamount of empirical Kroodsmaand Miller 1982).However, nonsong support (Curio 1978, 1994). vocalizationsmay offer even better possibilities In a seminal work, Marler (1955) demonstrat- for understanding the evolution of acoustic ed the antitheticalacoustic structure of two types structure. Marlet (1957) pointed out that some of vocalizationselicited by predators.Calls giv- avian vocalizationssuch as song showed marked en when a hawk is flying overheadoften are specific distinctivehess, while others lacked high pitched, cover a narrow frequency range, these features and often were similar among and lack abrupt onsetsor terminations, while unrelated species,indicating the influence of those given during mobbing cover a wide fre- different selection pressureson diverse vocal quency range and show abrupt onsets.Calls in categories.Marlet (1955, 1957) and Marlet and the first category may have features making Hamilton (1966) pioneered studies of evolu- them difficult to localize, while mobbing calls tionary forcesshaping acoustic structure of avi- have features enhancing locatability (Marlet an calls, and it is surprisingthat little research 1955, 1959,Marlet and Hamilton 1966),and may has followed this lead. facilitate the recruitment of other individuals Our comparativestudy focusedon the acous- (Marlet 1959). While few experimental studies tic structureof soundsgiven during mobbing. demonstratedetectability of callsby the species A variety of behavior patterns with different emitting them, the "churr" mobbing call of the structuresand functions may be elicited by GreatTit (Parusmajor) has frequency character- predators,including distractiondisplays ("bro- isticswithin the bestdetectability range for the ken wing") near the nest or young, "distress" speciesand transmitsover longer distancesthan calls by captured individuals, and "alerting" the species'seeet calls given to a flying hawk calls when a predator is sighted. The term (Klump and Shalter 1984). When birds mob a "alarm" is sometimes used in the literature, but predator, they call and their actionsoften are its meaning is ambiguous and could include very conspicuous,including approachingand severalcategories of vocalizations.Klump and retreating. If the main function of mobbing is Shalter (1984), recognizing the need for distin- to induce the predator to move on, the partic- guishing among categoriesof antipredator be- ipation of numerousindividuals engagedin ha- havior, provided a comprehensive classifica- rassing the predator would be advantageous. tion. The distinguishing feature of mobbing vo- Sohogramsof the mobbing calls of seven spe- calizations is that they are given while ap- cies of British passetinesrepresenting several proaching the predator, with the callers different families are remarkably similar (Mar- frequently changing positions (Klump and let 1959:fig.16), suggestingconvergent evolu- Shalter 1984). Proposed functions of mobbing tion. are numerous(reviewed by Curio 1978),but the We examine both the structure of mobbing "move-on hypothesis,"stressing the role of the callsand a number of different hypothesescon- 37O April 1996] MobbingCalls 371 cerning their evolution: (1) Do mobbing calls corded.The expectationis, however, that with of passerinesbelonging to a wide variety of taxa a large and diverse assemblageof speciessam- confirm Marler's predictions of convergent pied, trends will allow tentative conclusions. acousticstructure in diverse taxa?Convergence We hope that the findings will encouragefur- would be indicatedby abrupt onsets(clicklike ther study of a complexphenomenon. patterns) and a wide frequency range (Marler 1959, Marler and Hamilton 1966). It is not clear METHODS whether theseattributes must always co-occur We recordedbirds at severalsites, both during the for easeof localization.By definition, a click breeding and nonbreeding seasons,and in response coversa wide frequencyrange and hasa sharp to a variety of stimuli (Table 1). In all cases,the birds' onset,but a call could cover a wide frequency responsesmet our criterion of mobbing (i.e. an ap- range and not be a click. In other words, some proach toward the stimulus, usually within about 10 calls might have one feature that would sup- m and often closer).In a few cases,especially with posedlyenhance locatability, but not necessar- owl playback,several species participated in the same ily both. While mobbing calls are often clicks mobbing event. Individuals were not color-banded, (Marler 1959),their most important feature is a but in many cases recordings were obtained from wide frequencyrange, and the soundsof dif- three or more individuals. ferent speciesmay show considerablevariation We useda SonyProfessional Walkman cassettetape recorderand a Nakamichi directional microphone ex- in other attributeswhile exhibiting this salient cept for calls obtained at nestswhere a Realisticultra- feature (Marler and Hamilton 1966). We ex- thin microphone was placed within I m of the nest. amineMarler's hypothesis not only by analysis Sonogramswere producedusing a Kay Digital 7800 of the characteristicsof mobbingcalls, but also Sona-graphat the 150 Hz band width setting (to pro- by comparisonof this categoryof calls with vide a compromisefor temporal and frequencymea- passerinenestling calls, where different selec- surements). Measurements were made from SOhO- tion pressuresare presumablyoperating (Popp gramsusing a ruler. We then obtainedmeasurements and Ficken 1991). (2) Are there also phyloge- of the following variableson five callsfor eachspecies netic trends in the acoustic structure of the calls? and calculatedmean values: (1) maximumfrequency Are thesephylogenetic patterns superimposed (kHz); (2) minimum frequency;(3) frequency range (maximum-minimumfrequency); and (4) call dura- on the predictedstructural convergence due to tions (measurementsgiven for individual units, al- selectionfor certaintypes of acousticstructure? though some were given in seriesof repetitions). (3) Doesthe acousticstructure of mobbingcalls In addition to measurements,calls were categorized provide insights concerning the functions of by their general configuration on SOhograms,a re- mobbing?Features promoting locatability might flectionof both temporaland frequencyaspects (Ta- supportthe move-onhypothesis, indicating the ble 2). This analysis was done blindly, as the cate- main function is recruitment of other individ- gorizer did not have information on the identity of uals in harassingthe predator. However, are the species.The categoriesare somewhatarbitrary, additional selection pressuresoperating? (4) but supplement the data obtained from measure- While callsimilarity among unrelated taxa could mentsin giving an overall view of what speciesshare certaincall similarities.The followingcategories were be due to natural selectionfavoring a narrow used:(I) Thin verticalline. These are very short-dura- range of "best" acousticdesign for mobbing tion soundscovering a wide frequency range. Some effectiveness,could convergent evolution of sound like a click, others like a sharp chip. (II) Thick calls of unrelated taxa in some cases be due to verticalbar. These calls are similar to the previous direct selectionto facilitate interspecificcom- category,but are longer in duration and the quality munication? differs. They often have a harsh sound and do not We recorded calls in a variety of situations sound like clicks. (III) Horizontalband. These calls are including natural mobbingof predators(and a longer in duration than those in the previous two few nest-holecompetitors), presentation of owl categories.The frequencyrange varies. In somecases, mounts,and playbacks,as well ascalls induced only a single band is given, but in others, stacksof bandsoccur (i.e. Black-cappedChickadee). (IV) Chev- by our presencenear nests.Recordings were ron. Calls are usually of short duration and have a made during both the breeding and nonbreed- chevron shape,with an ascendingfrequency, a peak, ing seasons.We probably have not sampledthe and a descendingfrequency. (V) Diversegroup. All of entire repertoire of mobbing calls from certain these callsare acousticallymore complexthan those species,and differentpredators might elicit calls in other categories,although they do not necessarily in some speciesthat differ from those we re- share common patterns. 372 F•CKENAND POPP [Auk, Vol. 113 TABLE1. Characteristicsof mobbingcalls. For specieswith two kinds of calls,shorter