The Chick-A-Dee Call System of the Mexican Chickadee’

The Condor 96:70-X2 0 The Cooper Ornithological Society 1994

THE CHICK-A-DEE CALL SYSTEM OF THE MEXICAN CHICKADEE’

MILLICENT SIGLER FICKEN Department of BiologicalSciences, University of Wisconsin-Milwaukee,Milwaukee, WI 53211

ELIZABETH D. HAILMAN Department of Dairy Science,University of Wisconsin-Madison,Madison, WI 53706

JACK P. HAILMAN Department of Zoology, Universityof Wisconsin-Madison,Madison, WI 53706

Abstract. Chick-a-dee calls of the Mexican Chickadee (Parus sclateri)are composed of combinations of three common note types (A, C and D) and one very rare type (B). Calls have the invariant sequenceof notes A-B-C-D, where any note type may be omitted, given once or repeated a variable number of times before transiting to the next type. The B and C notes are phonologically similar to the B and C notes of chick-a-dee calls of the Black- capped Chickadee (P. atricapillus), but the A note is markedly different and the D note somewhat different from equivalent notes of the congener.A total of 2,07 1 calls recorded yielded 60 different call types,and Zipf-Mandelbrot plots show that the call systemis “open”; as the sample size is increased new call types will be found without demonstrable bound. In relatively undisturbed contexts (with mate on territory, in fall flocks, alone in fall) birds gave mainly [A][D] calls with lesser numbers of [A] and [C] calls, where brackets indicate variable repetition of note types. In disturbed contexts(mobbing plastic Great Homed Owl, mobbing speakerplaying calls of the Northern Pygmy-Owl, observer sitting under the nest cavity) the birds gave more [C] calls with [A][C] as well. In the longestmobbing sessionto owl calls, birds gave mainly [A] calls when approaching,switched to [C] calls while flying about the speaker,and then resumed[A] callsand moved offwhen the playbackwas stopped. Outside of human language,this is the second truly combinatorial system of vocal communication found in animals, the first being chick-a-deecalls of the Black-cappedChickadee. This studyprovides the first data substantiatingquantitative differencesin caikfrom different contexts,an important step toward understandingwhat kinds of information combinatorial chick-a-dee calls encode. Kev words: Parus sclateri: Mexican Chickadee;vocalizations; calls; syntax; mobbing; flocks.

INTRODUCTION ture of the Mexican Chickadee’s calls, (2) com- All tit speciesof the subgenusPoecile (genusPar- paresthese results with those from Black-capped us) appear to give combinatorial “chick-a-dee” Chickadees, and (3) documents differences in calls (Hailman 1989) but this complicated call calling in different contexts. system has heretofore been analyzed quantita- Chick-a-dee calls are of special interest be- tively only in the Black-capped Chickadee, P. cause these compose the only presently docu- atricupillus (Hailman et al. 1985, 1987; Hailman mented combinatorial system of animal com- and Ficken 1986). Chick-a-dee calls of the Mex- munication outside of human language. The ican Chickadee (P. sclateri) were first recorded Black-capped Chickadee uses four note types by Dixon and Martin (1979) whose spectro- (Ficken et al. 1978) in combination to createhun- grams (their figures la and b, p. 422) show two dredsofdifferent call types (Hailman et al. 1985). kinds of notes. Ficken (1990a) found that in a The note types (designatedA, B, C and D) occur sample of over 1,000 recorded calls, four types in the fixed sequenceA-B-C-D, where any note of notes were given but one of them only rarely type may be omitted, given once, or repeated a (seealso Ficken and Nocedal 1992). The present variable number of times before transiting to the study, based on further field work including next note type within the call. This straightfor- mobbing experiments, (1) describes the acous- ward phonological syntax was shown to be log- tical features and analyzes the syntactical struc- ically explicit by writing a “characteristic function” (correctly working algorithm) for a Turing machine (Hailman and Ficken 1986). However, ’ Received 12 March 1993. Accepted 9 August 1993. systematic departures from first-order Markov

[701 MEXICAN CHICKADEE CALLS 7 1 chains of transitions between note types show (150 Hz filter band width). Note types (described that far more complicated rules underlie the pre- in the Results) were classifiedby eye and written cise structure of calls (Hailman et al. 1987). De- on data sheets.In all, 2,045 calls were analyzed spite extensive analyses of syntactical structure, (a few occurredin other contextswith insufficient the communicative significance of chick-a-dee sample sizesfor analysis). Data on the note com- calls remains elusive. The present study of the position and context of each call were read into Mexican Chickadee provides not only the first a database and made accessibleto mainframes comparative data on syntax in another species and microcomputers for analyses.Analyses were but also an important first step toward under- basically a subsetof those performed on calls of standing differences in calls in different behav- Black-capped Chickadees by Hailman et al. ioral contexts. (1985) designed by J.P.H. and programmed by E.D.H. Most of the special analytical programs METHODS were written in PASCAL and run on microcom- All recordingswere made at or in the vicinity of puters. Rustler Park in the Chiricahua Mountains (Co- Contingency tables were analyzed statistically chise County) in southeasternArizona. Record- using “computer-intensive” methods of the pub- ings were made by M.S.F. in October 1985, and licly available software MONTE CARLO RXC May and October 1986 (Ficken 1990a); further written for the Apple Macintosh by W. R. Engels field work by the three authors togetherwas con- of the University of Wisconsin-Madison, De- ducted in April 1990 and March 199 1, by M.S.F. partment of Genetics. Such analyses provide a in May 199 1, and by the three authors together more sensitiveand accurateassessment than Chi- in April 1992. square methods often applied to such data (En- Calls were divided into four principal contexts gels 1988). At least 1,000 trials were run for a of recording: (1) with mate on the breeding ter- given test to createthe distribution against which ritory, (2) in mixed-species flocks in the fall, (3) the data were compared. Sequential analysis by when disturbed by the presence of an observer the method of Markov chains was done with near the nest, and (4) when mobbing a speaker UNCERT, publicly available software for DOS playing the call of a Northern Pygmy-Owl, Gluu- computers copyrighted by E.D.H. and J.P.H. cidium gnoma. In addition, small samplesof vocalizations were taped in two other contexts: (5) RESULTS lone birds in the fall and (6) mobbing a plastic model of a Great Horned Owl, Bubo virginianus. PHONOLOGY The mobbing experiments using owl tapes were Chick-a-dee call systemsas a whole are doubtless conducted at five different sites. The partially homologousamong speciespossessing them. Two diurnal Northern Pygmy-Owl takes small birds of the Mexican Chickadee’s note types differ and is common in the study site; during one ex- somewhat in acoustical structure from those of periment, an owl answered the playback from the Black-capped Chickadee, so they might not acrossa canyon. In two experiments tapes were be homologous. Unlike the basically chevron- played back from a Marantz PMD 430 Profes- shaped A note of the Black-capped Chickadee, sional cassetterecorder through a Realistic Min- the Mexican Chickadee’s A note (Fig. 1) is fre- imus-0.6 amplified speaker near the ground on quency modulated at a high rate, giving it a buzz- a log or stump. In the other three experiments, ing aspectto the human ear. Buzzing notes char- tapes were played back from a Sony Walkman acterize other vocalizations of the Mexican Professional cassette recorder through a Sony Chickadeeas well (Dixon and Martin 1979; Fick- SRS-27 amplified speaker.All playbacks attract- en 1990a, 1990b). The A note usually sweeps ed a number of small forest speciesbesides Mex- downward in frequency during its duration (Fig. ican Chickadees. 1) but the magnitude of the decline is variable. Most of the recordings were made on a Sony Dixon and Martin (1979, Fig. 1a) showed a call Walkman Professional cassette tape recorder with three A notes that lacked this frequency (details in Ficken 1990a, 1990b), with further downsweepcompletely, and there seemsto be a recordings of mobbing experiments made with continuum in the slope of the downsweep. The a Sony 8 mm camcorder. Tapes were analyzed A note was common (1965 notes recorded),com- by M.S.F. with a Kay 7800 Digital Sona-Graph prising 28.4% of all notes.

\ 1 72 M. S. FICKEN, E. D. HAILMAN AND J. P. HAILMAN a

0 0 0.5 1 1.5 Time (s)

C d 8

0 0 0.5 1 Time (s)

FIGURE 1. Sound spectrogramsillustrating the A and D notes of chick-a-dee calls of the Mexican Chickadee, selectedto illustrate some of the phonologicalvariation found in notes. The frequency and time scalesare the same for all spectra.(a) Commonest of all calls is the call type AD, with typical down-slurred A note. Notice the banded structure of the D note. (b) A call showing the typical A/D contraction of the final A and first D notes of calls containing both types. (c) The contracted A/D may begin a call; notice the noisy (unbanded) structureof the D notes. (d) Rare phonologicalstructures occur, as in this call with an A-like introductory note followed by an A/D-like structureof long A component and short D component. MEXICAN CHICKADEE CALLS 73

TABLE 1. Durations, highest frequencies, and lowest frequencies of note types in chick-a-dee calls of the Mexican Chickadee. Entries are mean I standard deviation (sample size).

Call type: duration Fig. No. of Duration Lowest Highest Note(s) call type (msec) frequency (kHz) frequency (kHz)

Single A - 151 i 35 (13) 5.7 i 0.7 (13) 8.0 * 0.5 (13) “Low frequency” A 2a 208 i 28 (14) 3.4 i 0.2 (14) 6.3 * 0.6 (14) AA: 331 & 92 (16) - First A 127 -c 53 (16) 5.5 i 0.6 (16) 8.1 + 0.6 (16) Second A 120 * 35 (16) 5.7 i 0.8 (15) 7.3 ? 1.8 (16) AA/D: 521 i 118 (20) lb First A 170 f 32 (20) 5.8 ? 0.4 (20) 8.4 f 0.2 (20) Second A (of A/D) 50 t 31 (20) 4.2 YZ0.7 (18) 8.2 + 0.3 (20) D (of A/D) 212 -c 67 (20) 3.2 * 0.3 (19) 6.2 f 0.5 (18) A/D: 425 2 107 (17) - A (of A/D) 100 f 76 (17) 4.4 -+-0.6 (16) 8.3 f 0.2 (16) D (of A/D) 319 f 95 (17) 3.2 + 0.2 (17) 6.4 f 0.5 (12) A/D D: 690 i 181 (18) lc A (of A/D) 87 f 59 (18) 4.3 f 0.7 (15) 8.2 + 0.3 (16) First D (of A/D) 245 + 117 (18) 3.1 + 0.3 (14) 5.6 + 0.3 (15) Second D 270 f 85 (18) 2.9 + 0.4 (14) 5.5 i 0.3 (15) [C] (First of a series) 2e 23 f 2 (14) 2.8 f 0.3 (14) 7.6 * 0.1 (13)

The Mexican Chickadee’s D note more closely brief A notes (Fig. 2b) but are much shorter and resembles that of the Black-capped Chickadee, never combined with other notes into multi-note and is typically the longest note-type in duration calls. These notes appear to be related to or var- (Fig. l)-even longer than that of the Black- iants of “tseets,” of which Ficken (1990a) re- capped Chickadee. The D note varies from hav- corded only six examples. These A-like notes are ing a distinctly banded frequency structure (Fig. not included in the analyses below. la) to being nearly uniformly noisy (Fig. 1c). The The B note, which was very rare (3 occurrences Mexican Chickadee’s D note often has an onset in 6,9 18 notes), is chevron-shaped (Fig. 2d) and “spike” of higher frequency (also evident in spec- similar to the B note of the Black-capped Chick- trograms of Dixon and Martin 1979, and Ficken adee. The C note, which was the commonest type 1990a). The 1,080 D notes recorded comprise (3,870 recorded, comprising 55.9% of all notes), 15.6% of the sample. When A and D notes occur is a “noisy chevron” (Figs. 2c, d), again similar in the same call (always in the order A-D), they to the C note of the Black-capped Chickadee. may be independently uttered (Fig. 1a), or more The C notes were usually uttered in distinct cou- typically the last A in a sequence merges with plets (Fig. 2e), another variation of phonology the first D note to form a sort of contracted note worthy of further study. we designateby “A/D” (Figs. 1b, c). The A part The duration of a given type of note varies, in of this contraction is variable in duration, but some casesmarkedly (Table l), depending upon usually much shorter than a typical A. Rarely, the type of call and where the note falls within peculiar phonations occur in recordings, as in the call. Intemote intervals vary from about 50 Figure Id where an A-like note occursbefore an to 90 msec for various types, but the only sys- A/D-like contraction. tematic variation found was a bimodal distri- It might be that some variations in A notes bution of C-C intervals (shorter intervals within are communicatively significant. Single A notes, than between couplets).The interval between the not combined with other As or other note types, last note of one call and the first note of the next are particularly variable. Among the 6,9 18 notes is one to two orders of magnitude longer than recorded, 34 single A notes of unusually low intemote intervals within calls, thus defining the acoustical frequency were found (Fig. 2a). These call as a natural unit of phonation. may represent a rare but distinct subtype of A Notes of the Mexican Chickadee (especially notes, treated here as A notes but deserving of the A and D notes) are noticeably longer than further study. The Mexican Chickadee also com- those of the Black-capped Chickadee (Table 1) monly utters high frequency notes that resemble whereas the number of notes in a call is notice-

\ 1 14 M. S. FICKEN, E. D. HAILMAN AND J. P. HAILMAN a b

0.5 1 Time (s)

ACC

Time (s) FIGURE 2. Spectrogramsillustrating other notes of chick-a-dee calls (cf. Fig. 1). The frequencyand time scales are the same for all spectra.(a) A low-frequency A note that may representa specialsubtype. (b) High-frequency notes resemblingabbreviated A notes, which appear to be “tseet” notes related to flight intention. (c) An A note with two C notes. (d) The rare B note followed by a seriesof C notes. (e) The couplet structureof C notes, found only in calls composed of this note type alone. MEXICAN CHICKADEE CALLS 75

BEGIN END CALL CALL (silence) (silence)

FIGURE 3. Kinematic graph of first-order transitions between notes within chick-a-dee calls. The sizes of the letters A, B, C and D are approximately proportional to the probabilities of occurrenceof those note types (except that the rare B would not be visible if shown in correct proportion). The widths of the arrows are proportional to the probabilities of transition, such that outgoing arrows sum to unity. (Rare transitions that would have invisibly thin lines are omitted: B either begins a call or follows A.) Recurrent arrows represent repetitions of the same note type. ably fewer than in the other species.The chick- transits to D, or ends the call; if transiting to D, a-dee calls of the Mexican Chickadee are from that note repeatsor ends the call. If the call begins one to at least 14 notes in length, which is a with C, that note most commonly repeats, or smaller maximum note-length than in calls of ends the call. the Black-capped Chickadee. The longest in From Figure 3 one may infer that calls will Hailman et al. (1985) was 24 notes, and that commonly have the structure [A], [C] or [A][D], sample did not include predator contexts, where where the enclosure of the note type in square calls are apparently often longer (Ape1 1985). brackets denotes possible repetition. Other call Table 1 contains several examples of how the structuresshould be much rarer. Categoriessuch placement of a note affectsits duration. For ex- as [A][D] are termed sequence types, and each ample, the A note of single-A calls is longer than may include many distinct call types such as AD, either note of an AA call, but not as long as the AAD, AAAAD, ADD, ADDDD, and so on. first A that is followed by an A/D contraction to There are 15 such sequencetypes possible, but compose a call. Similarly the D part of an A/D becauseof the rarity of B notes five of these did contraction is shorter if the contraction is pre- not occur in the record: [B], [B][D], [A][B][C], ceded by an A note than if the contraction is [B][C][D], and [A][B][C][D]. The commonest se- itself the entire call. The frequency characteris- quence types were: 735 [A][D] calls, 664 [C] and tics of notes may also vary somewhat according 5 17 [A], as expected from Figure 3. Two se- to their placement within calls (Table 1). quence types occurred with sufficient frequency to be of interest: 88 [A][C] and 30 [C][D]. The SYNTAX remaining typeswere very rare: 7 [D], 2 [A][C][D], Markov chain analysis showedthat the sequence and 1 each [A][B], [B][C], and [A][B][D]. of notes within a call follows the same rule as in The departureof Mexican Chickadeecalls from the Black-capped Chickadee: note types always a first-order Markov process was small. The occur in the order A-B-C-D (no exceptions in maximum uncertainty possible with four note- 2,07 1 calls), where any note type may be omitted types is log, 4 = 2 bits/note, but the unequal entirely, given once, or repeated a variable num- frequencies of note types yielded a zero-order ber of times. The first-order Markov chain (Fig. Markov chain of only 1.2 1 bits/note. First-order 3) reveals the principal syntactic structure. Calls analysis reveals a large drop to only 0.36 bit/ begin with A or C (less than 1% begin with B or note, which would be a true first-order Markov D). If beginning with A, that note-type repeats, process only if all uncertainty were removed.

\ 1 16 M. S. FICKEN, E. D. HAILMAN AND J. P. HAILMAN

Second-order analysis showed a small drop to technique (see Methods). All pairs of contexts 0.21 bit/note, meaning that long-range con- differed significantly (P’s < 0.001) except two straints are relatively unimportant in the calls. involving the small sample size of lone birds: the Put differently, the next note to occur can be difference from “with mate” is marginal (P = predicted well from the previous one but know- 0.06) and from “fall flocks” non-significant (P = ing longer ordered-stringsof previousnotes hardly 0.19). improves predictability. When undisturbed on territory with the mate, almost three-quarters of calls were of the [A][D] OPENNESS sequencetype (Fig. 5a). When in mixed-species A major finding in the Black-capped Chickadee fall flocks (Fig. 5b) or alone (Fig. 5c), only about was that the call system is “open” (Hailman et half of calls were [A][D], with [A] and [C] calls al. 1985): as the sample size of calls is increased, making up most of the other half. In the former the number of different call types increaseswith- case,there were also a few calls of the [A][C] type out demonstrable bound. A call type is defined and other types. When there was an external dis- as any different combination of notes; thus A, turbance (right column of diagrams in Fig. 5), CC, AC, AAACC and ACCCCCC are all differ- the distribution of sequencetypes was markedly ent call types. In this study we found 60 different different, with [C] and [A][C] together always call types in the 2,071 calls recorded, and as in making up at least half of all calls. When dis- the Black-cappedChickadee thesecall types var- turbed by a Great Homed Owl model, birds gave ied widely in frequency of occurrence. For ex- many [A][C] calls (Fig. 5d). When mobbing a ample, there were 783 calls of the type AD but Northern Pygmy-Owl tape, [A] and [C] calls were only one of the type ADDDD. commoner (Fig. 5e), and when disturbed by the The test for opennessis made by a Zipf-Man- observer sitting about 6 m from the nest, [C] calls delbrot plot of the probability of occurrence (P) predominated (Fig. 5f). as a function of occurrence rank (r) on log axes. OWL-TAPE PLAYBACK EXPERIMENTS Mandelbrot’s (1953) formula P = i(r + k) ( for a fitted function has three constants:the intercept Mobbing experiments provide a means of study- (i), curvature factor(k) and asymptotic slope (s), ing calls in detail becausethere is a specificlocus which is always negative. To show openness it in spaceto which the mobbing birds attend and is not necessaryto provide a fitted function but all individuals present behave similarly. Exper- merely to show that the plotted curve reachesa iments were conducted at five sites from March non-zero asymptotic slope. Figure 4 shows this to May of various years. The experiments, using property for the Mexican Chickadee, both with playback tapes of Northern Pygmy-Owl calls, all data combined (top curve) and plotted sep- generally attracted one pair of Mexican Chick- arately by the six contexts.Even the curvesthem- adees, often with small songbirds of other spe- selves for the different contexts are remarkably cies. The chickadeesgave 20 to 50 calls and then similar, reaching about the same asymptotic departed with the playback still running. One slopesat their right-hand sides.The curves dem- experiment brought in several chickadeesfor a onstrate that the call system of the Mexican long period, and the results are analyzed sepa- Chickadee, like that of the Black-capped Chick- rately below. The results of the other four ex- adee, is open and hence more new call types will periments are shown in Figure 6, which plots the be found if the sample is increased. cumulative frequency of each call sequencetype through the experiment. DIFFERENCES AMONG CONTEXTS Every experiment shown in Figure 6 yielded The distribution of callsamong the sequencetypes a similar but unique set of results. The experi- differed according to context of recording (Fig. ment of 22 March was made about half-way on 5). For purposesof statistical comparison among the road to Rustler Park. This playback probably contexts, calls were consideredin five categories: evoked calls from a single bird, which did not [A], [Cl, [A][D], [A][C] and “other” (all the re- closely approach the speaker. In all four exper- maining sequencetypes combined). As each re- iments one sequencetype predominated, and that cording context was independent of the others, type of call was given at an approximately equal every possible pair of contexts was compared in rate (constant slopes in Fig. 6) throughout the a 2 x 5 contingency table using the Monte Carlo experiment. The predominant type was [A][D] MEXICAN CHICKADEE CALLS 77

10 100 r (occurrence rank) FIGURE 4. Zipf-Mandelbrot plots for all calls combined (top) and separatedby contexts.Note that probability valuesare labeled only for the top curve; each successivecurve is displaceddownward by one log unit to separate the curves for clarity. The occurrenceprobability of the commonest call is about 0.2-0.3 for all curves. The fact that curves reach an asymptotic non-zero slopeshows that the call systemis “open”: new call typeswill be foundas the samplesize increases. in the March and April experiments and [C] in One experiment (Fig. 7) was unusual in several the experiments conducted in May. Only [A] and respects.First, the playback attracted at least four [C] calls were common to all experiments, [A][C] Mexican Chickadees(it was difficult to be certain and [A][D] occurred in three experiments each, of the precisecount becauseof densevegetation), and three sequence types were unique to one instead of a single bird or a pair. Second, the experiment: [C][D] in March and [A][B][D] and birds continued mobbing for a sufficient period [A][C][D] in April. Our qualitative observations that we could stop the playback while they were suggestedthat the [C] calls were given when birds still present, and so record post-playback calls. were most agitated by the playbacks and closest And third, much (but not all) of the experiment to the speaker. The differences among experi- was recorded on videotape. ments may therefore indicate greater agitation Birds began moving toward the site giving during the nesting season (May) than earlier mainly [A] calls. While still approaching,the birds (March and April). began giving [C] calls in addition to [A] calls, and 78 M. S. FICKEN. E. D. HAILMAN AND J. P. HAILMAN

UNDISTURBED DISTURBED

a. d.

plastic great with mate horned owl (N=36)

b. e.

mixed- mob species Pygmy fall flocks owl tape (N=1013) (N=287)

alone observer in fall at the nes (N=35) (N=443)

FIGURE 5. Pie diagrams of the distributions of call sequencetypes in six contexts of recording. (The sum of samplesizes over all diagramsis lessthan the total numberof recordingsbecause a few recordingsdo not belong in any of the six contextsshown.) at approximately the same rate (same slopes in (changing to a noticeably shallower slope). Tap- Fig. 7). As they moved still closer, birds began ing ceasedas the birds were moving off. giving [A][D] calls as well, until assembledat the These results were consistent with those re- speaker site. Here they stopped giving [A] calls corded when birds were intensely disturbed by altogether (zero slope in figure) and soon after an observer sitting under the nest hole (Fig. Sf, stoppedgiving [A][D], while abruptly increasing above), where they also gave mainly [C] calls. the rate of [C] calls (steep slope in Fig. 7). When Videotapes of the owl-playback experiment of playback was terminated, calling immediately Figure 7 show that [C] calls were given primarily changed again, with [A] calls immediately re- by perchedbirds that were pivoting on a branch, suming (steepslope) and [C] calls becoming rarer although they were sometimes given in flight. (It MEXICAN CHICKADEE CALLS 79

20 22 March

0 0 10 20 30 40

15 15 May 15 May [Cl site 1 site 2 10 /

[Al, IN’2 & [ND1 5

-- 0 0 10 20 30 40 0 10 20 SUCCESSIVE CALLS SUCCESSIVE CALLS FIGURE 6. Cumulative frequencyplots of call sequencetypes given by Mexican Chickadee pairs (a lone bird on 22 March) while mobbing a speaker playing back sounds of a Northern Pygmy-Owl. March and April experiments were dominated by [A][D] calls whereas May trials yielded mainly [C] calls. is often difficult to tell from the videotape which capped Chickadee (Hailman et al. 1985). There call is being given by which bird.) Calls contain- are four note types, they are combined into unit ing the A note were given during approach (and calls with very short internote intervals com- retreat) from the site, but there is a difference pared with intervals between calls, and they oc- between [A] and [A][D] calls. In mobbing trials, cur in the invariable sequenceA-B-C-D within [A] calls were most often strings such as AA, calls, where any note type may be omitted en- AAA, and AAAA, unlike single A notes given tirely, given once, or repeated a variable number in other contexts; furthermore, A-strings given of times. The note types are structurally similar in mobbing trials seemed sometimes to be given in the two chickadees,with B and C notes being in flight. By contrast, [A][D] calls were given only nearly identical. The calls are given by both sex- by perched birds. These birds were moving to- es, throughout the year, and in a wide variety of ward the site in short flights, but seemed never contexts.In both speciesthe [A][D] sequencetype to utter any call containing a D note while ac- is the commonest overall. Both call systemsare tually in flight. open to new call types as the sample size increases, and are the only communication systemsout- side of human language to be proven open. DISCUSSION The chick-a-dee calls of the two species do COMPARISONS WITH THE BLACK-CAPPED differ, however, in many ways. The A notes of CHICKADEE the Mexican Chickadee are markedly different The overall structure of chick-a-dee calling in the from those of the Black-capped Chickadee, and Mexican Chickadee is similar to that of the Black- even the D notes are dissimilar (as noted by Dix- 80 M. S. FICKEN, E. D. HAILMAN AND J. P. HAILMAN

playback birds birds moving mobbing approaching i closer i speaker I 50 1 14 April site 1

MD1

20 40 60 80 100 SUCCESSIVECALLS FIGURE 7. Cumulative frequencyplots (as in Fig. 6) of a particularly long mobbing sessionwith at least four Mexican Chickadeescalling. Birds first gave [A] calls at a distance, then added [C] calls when drawing closer, and finally abandoned the former entirely when near the speaker.When the owl tape was turned off, [A] calls abruptly resumed.

on and Martin 1979). In both cases, though, the The rarity of B notes renders the Mexican note types may reflect origins from a common Chickadee’s calls less diverse than those of its ancestral source: the Black-capped Chickadee’s congener, and the total note length of calls is A note is a single chevron, the Mexican Chick- much shorter. The more variable duration of the adee’s a continuous chain of short chevrons run Mexican Chickadee’s notes may in some sense together (Fig. 1, above) so as to sound distinctly replace the more variable number of note repe- buzzing to the human ear. Similarly, the D notes titions given by the Black-cappedChickadee. The of both species are the longest note type, cover rarity of B notes and the shorter note length of a wide frequency spectrum, and may be distinctly calls means that the Mexican Chickadee’s utter- banded into emphasized frequency components ances tend to be syntactically simpler, although or more uniformly distributed to composea noisy not necessarily semantically simpler (especially note. if duration of notes encodesuseful information). Major differences are in duration, with the A Good contextual analyses of Black-capped notes of the Mexican Chickadee being highly Chickadee calls have not been published, but variable in duration according to the other note some comparison of mobbing situations between types with which they occur (means from 120 to the two speciesmay be made preliminarily. The 170 msec in Table 1, for A notes not contracted Mexican Chickadee utters primarily [C] or [A][C] with D’s). By comparison, the A notes of the calls when greatly disturbed or mobbing (Figs. Black-capped Chickadee are uniformly shorter 5-7, above). An unpublished thesisof Ape1(1985) at about 45 msec (Hailman et al. 1985: 195, Ta- showed that Black-capped Chickadees give pri- ble 1). Similarly, the Mexican Chickadee’s un- marily [D] calls or other sequencetypes contain- contracted D note averages 270 msec (Table 1, ing D notes. Similarly, Lambrechts and J. P. above), whereas the Black-capped Chickadee’s Hailman (in prep.) found that the Black-capped D notes average about 130 msec (Hailman et al. Chickadeesmobbing a stuffed owl near the nest 1985: 195, Table 1). gave elevated proportions of [D] and [A][B][D] MEXICAN CHICKADEE CALLS 8 1 calls, with hardly any calls ofany kind containing marily to the position and movements of the C notes. There are important differences in the caller in space,especially relative to external ob- detailed contexts among these three studies, but jects such as conspecifics,the nest, a human ob- the differencein resultsis so striking as to suggest server, or a potential predator. The present re- a fundamentally different semantic structure in sultson the Mexican Chickadeesupport that view the two chickadee species. By Marler’s (1955) and give it further substance.In contexts where classic criterion that mobbing calls should con- birds are moving through the environment with sistof short,broad frequencynotes repeated,both little obvious disturbance except that caused by chickadee speciesqualify. The Mexican Chick- conspecifics,Mexican Chickadees utter primar- adee’s D notes may be too long for effectively ily [A] and [A][D] calls (Fig. 5a-c, above). These rapid repetition, whereas its C note more nearly same two sequencetypes characterize birds mov- meets the required characteristics.However, the ing toward the site of an owl playback (Figs. 6 C note of the Black-capped Chickadee is very and 7, above), and sometimes when leaving the similar to that of the Mexican Chickadee, so it site (Fig. 7). The D note is always given while remains unclear why the former speciesdo not perched whereas the A note seems to be given seem to employ C notes in its mobbing calls. In sometimes in flight as well. any case, the former impression that D and A preliminary interpretation may therefore be D-like notes were the “mobbing calls” of Parus formulated as follows. A notes indicate a rest- (e.g., Thielcke 1968) is too simple a view of the lessness(a bird ready to fly or already in flight) complex vocalizations of tits. whereas a D note indicates that the caller is perched and hence indexes its location in space. SEMANTICS The [A] sequencetype in undisturbed birds con- What chick-a-dee calls “mean” (what kind of sists mainly of single A notes, but in birds ap- information they encode) is a formidable prob- proaching a mobbing site AA and AAA call types lem becauseof the complexity of the calling sys- are especially common. Thus, the repetition of tem and the fact that chick-a-dee calls occur in A might indicate speed of movement or high virtually every definable context. At one extreme “intensity” of restlessness.The [C] calls of the is the possibility that chick-a-dee syntax is sim- Mexican Chickadee are given most commonly ply some by-product of phonation, such that all in disturbed situations (Fig. 5d-f, above), where calls are semantically equivalent, carrying little [A][C] calls are also common. Furthermore, when more information than the speciesidentity and intensely mobbing an owl playback, the birds position in space of the caller. At the other ex- gave nearly pure [C] calls (Fig. 7) as they did to treme is the possibility that each of the hundreds the human observer at the nest (Fig. 5f). The C of call types has its own distinct meaning, like note is most often given by a perched bird that words in a dictionary. Hailman et al. (1985, 1987) is pivoting while it calls, but may sometimes be favored an intermediate view for the Black- given in flight. In the Black-capped Chickadee a cappedChickadee, suggestingthat each note type C note given in flight has been correlated prelim- means somethingdifferent and repetitions of note inarily with a sudden swerve or change in direc- types within calls serve as modifiers denoting tion (Hailman et al. 1985) but our videotapes “intensity.” The present results on the Mexican cannot document the same phenomenon for cer- Chickadee support this hypothesis by the find- tain in the Mexican Chickadee. That [C] calls, ings of distinctly different suitesof sequencetypes and even a few [A][C] calls, occur in relatively in different contexts (Fig. 5, above) and details undisturbed situations (Figs. 5a-c) may indicate of calls given during mobbing situations (e.g., mild disturbance by conspecifics.Therefore, C Fig. 7, above). Comparable contextual differ- notes seem to indicate a disturbing stimulus and ences have yet to be reported for the Black- the tendency to change direction, so that when capped Chickadee. combined with A notes in [A][C] calls further Even though different suites of sequencetypes indicate a tendency to move some distance, as characterize different contexts, the fact that the when flying past a playback speakeror owl mod- same note types, sequencetypes and call types el. can occur in different contexts suggeststhat note In summary, the relatively simple structure of types encode information common to a variety Mexican Chickadee calls provides a clearer pic- of situations. Hailman et al. (1985, 1987) sug- ture of the information encoded by note types gestedthat this encoded information related pri- than is currently available from data on Black- 82 M. S. FICKEN, E. D. HAILMAN AND J. P. HAILMAN capped Chickadees. Basically, A notes indicate FICKEN, M. S. 1990a. Vocal repertoire of the Mexican a tendency to move some distance in space, C Chickadee.I. Calls. J. Field Ornithol. 61:380-387. FICKEN,M. S. 1990b. Vocal repertoire of the Mexican notes denote a disturbing stimulus and a ten- Chickadee. II. Sona and song-like vocalizations. dency to alter direction, and D notes connote a J. Field Omithol. 61:388-395. perchedbird. (The B note of the Mexican Chick- FICKEN,M. S., R. W. FICKEN,AND S. R. WITKIN. 1978. adee cannot be analyzed because of its rarity). The vocal repertoire of the Black-capped Chick- Repetitions of notes within calls indicate the “in- adee. Auk 9534-48. FICKEN, M. S., AND J. NOCEDAL. 1992. Mexican tensity” of the behavior that correlates with the Chickadee, p. l-l 1. In A. Poole, P. Stettenheim, note types. Whether or not combinations of these and F. Gills [eds.], The birds of North America, notes into sequencetypes such as [A][C], [A][D] no. 8. American Ornithologists’ Union and Acad- or [C][D] encode something more than the con- emv of Natural Sciences. Washinaton. DC and Philadelphia. catenation of the separatemeanings of the notes HAILMAN,J. P. 1989. The organization of major vo- is a derivative question of considerable difficulty calizations in the Paridae. Wilson Bull. 101:305- that cannot be addressedwith present data. 343. HAILMAN, J. P., AND M. S. FICKEN. 1986. Combi- natorial animal communication with computable ACKNOWLEDGMENTS syntax: chick-a-dee calling qualifies as ‘language’ by structural linguistics. Anim. Behav. 34: 1899- We thank Sally Spofford for suggestingstudy sites.We 1901. used the facilities of the Southwest Research Station HAILMAN, J. P., M. S. FICKEN,AND R. W. FICKEN. ofthe American Museum ofNatural History and thank 1985. The ‘chick-a-dee’ callsofParus atricapillus: the staff of that institution. a recombinant system of animal communication comparedwith written English.Semiotica 56: 19l- LITERATURE CITED 224. HAILMAN, J. P., M. S. FICKEN.AND R. W. FICKEN. APEL, K. M. 1985. Antipredatorbehavior in the Black- 1987. Constraints on the structure of combina- cappedChickadee (Parus atricapillus).Ph.D.diss., torial “chick-a-dee” calls. Ethology 75:62-80. Universitv of Wisconsin. Milwaukee. WI. MANDELBROT.B. 1953. Contribution a la theorie DIXON, K. L., AND D. J. MA&IN. 1979. Notes on the mathematique des jeux de communication. Publ. vocalizations of the Mexican Chickadee. Condor L’Inst. Stat. L’Univ. Paris 2:5-50. 81:421-423. MARLER,P. 1955. Characteristics of some animal ENGELS,W. R. 1988. Monte Carlo RxC contingency calls. Nature 176:6-g. table test. Computer softwarefor Apple Macintosh THIELCKE,G. 1968. Gemeinsamesder Gattung Par- with on-line documentation. Dept. of Genetics, us. Ein bioakustischerBeitrag zur Systematik.Vo- Univ. of Wisconsin, Madison, WI. gelwelt (Beihefte) 1:147-l 64.