HOW BIRDS DEVELOP SONG DIALECTS

ROBERT E. LEMON Department of Biology McGill University Montreal, Quebec, Canada

This paper considers how dialects develop in of their territorial proclamations throughout the songs of territorial oscines. The term the breeding season. Both sexes have a reper- “dialect” implies on the one hand some simi- toire of several distinctive songs, as many as larities in the repertoires of sound patterns, 8 to 12 being used frequently by each indi- here called “syllables,” and their sequences vidual (Lemon 1965, Lemon and Herzog in songs of birds on neighboring territories. 1969). They also have a repertoire of calls It also implies some dissimilarities among used for communication at close distances and birds of different localities. Much evidence in different contexts (Lemon 1968a). indicates that copying is the major basis of The songs consist of repetitions of sound the similarities, while the dissimilarities arise patterns called syllables, each type consisting from a variety of sources. These include of one to three individual sound units. For active generation of new syllables and new example, the first syllables of all three males in sequential combinations, and the interaction figure l-Al, Bl, and Cl-have one part only, of time-limited copying and dispersal. Hence, whereas in A5, and B5, the syllables have where dialects occur they result from opposing three parts. Any particular song consists of tendencies toward conformity and toward in- one to three types of syllables occurring in dividuality of repertoires. Not all species can definite orders or permutations as seen in be said to exhibit dialects (indeed the term A2, A3, A4, and others. Each of these permu- itself is used quite imprecisely); but species tations is called a “song type.” In the present obviously vary in the extent to which their examples only about half of the song types of songs conform within a population. In spite each bird are shown. of this variation, the bases of development Neighboring Cardinals have very similar seem the same throughout the species studied, songs which are often different from those of although there is varying emphasis one way Cardinals elsewhere. Of the examples from or another. An important source of variation three males shown in figure 1, two of the and confusion is the size of the repertoire males were neighbors on immediately ad- and the number of sequences in which the jacent territories at London, Ontario, while items are presented. the other resided near Melbourne, about 32 Most emphasis in this paper is placed on km direct flight from the former location. the family Fringillidae because this group The similarities of the songs of birds A and B shows a wide range in the expression of dia- are obvious although some individual differ- lects and is the best studied of any avian ences are evident in A6 and 136. family with regard to song development. I In some cases, little or no difference is de- shall first describe the situation as noted in the tectable among the songs used in different Cardinal ( Car&n& cardinalis), for this spe- localities, as is shown by the songs in Al, Bl, cies has been studied in both field and labora- and Cl. In fact, some syllables and song tory, and it sings in a manner that is easy to types occur widely throughout Ontario and appreciate in the visual displays of sound adjacent United States with little change. spectograms. Findings in this species are re- On the other hand, differences between lated to other fringillids and finally to other local populations that are the expressions of oscines. The consequence of these consider- dialects are also obvious and can be con- ations is a general model of song development sidered under three categories (Lemon 1966). for oscines. First, the syllables are very similar, yet there are slight and consistent differences occurring AN EXAMPLE OF DIALECTS locally. Such differences are seen in the syl- The Cardinal is a common bird of gardens lables of rows 2 and 3. On the other hand, the and woodlands, especially in the eastern syllables may be so different that they can United States, southern Ontario and Mexico. hardly be considered of the same class, as Both males and females sing well, although in rows 4 and 6. Of course, there are many the males sing much more frequently as part examples in which the distinction between

[3851 The Condor 77:385406, 1975 386 ROBERT E. LEMON

MELBOURNE

“\I \\

FIGURE 1. The similarities and dissimilarities of songs of two male Cardinals from contiguous territories at London, Ontario, and one from about 35 km away near Melbourne, Ontario. Songs toward the upper half of the figure show those syllables and song types most similar among the two places. Only about half the repertoires are shown. (In this and subsequent figures, the unnumbered base marks are 0 kHz. Exceptions are indicated otherwise, as in some examples in Figures 2, 7 and 8.)

different classes becomes quite arbitrary; 2A. This bird had been color-banded as a indeed this is to be expected since I shall nestling and had accompanied his banded argue that new syllable classes arise through father for the next three months after leaving the modification of existing forms. Finally, the nest; obviously this young male had been the syllables may occur in combinations and exposed to the local songs. permutations peculiar to a locality, as seen I took several Cardinals from the nest ap- in row 3. Further examples of these differ- proximately five days after hatching and let ences will be shown later. them hear recorded songs including some from localities other than their own. Eventually BASES OF COMMON PATTERNS they reproduced the foreign patterns, as in In considering the development of dialects, figure 2B. Conversely, individuals taken from we shall deal first with those aspects that the nest but isolated from song failed to re- contribute to the sharing of patterns by neigh- produce the syllables and song types of their boring territory holders. We shall examine parents, with a few exceptions to be con- first the developmental processes, leaving the sidered later. Instead, they developed reper- ecological features till later. The occurrence toires that were smaller than those of wild of patterns common to several individuals Cardinals; the syllables produced were usually could result from copying or from the in- one-parted only, lacking two or three temporal heritance of influences among birds of com- subdivisions characteristic of many sung in mon lineage. The evidence supports the the wild, and most were slurred in one direc- view that both aspects are involved, although tion in their frequency sweeps, as in figure that for copying is much clearer. 2Cl. When reared together, these isolated Cardinals developed syllables of which most Evidence of copying or imitation is seen were common to all birds in the group. When both in wild Cardinals and in birds reared reared separately, they had repertoires mainly under controlled conditions (Lemon and Scott unique to each bird. 1966, Dittus and Lemon 1969). Young male Evidence of copying was noted also in the Cardinals were captured in their first winter wild in two young males that settled in terri- and were isolated from further exposure to the tories immediately adjacent to an older male songs of other Cardinals. The following spring that had certain peculiarities of song, two they sang most of the syllables common to of which subsequently appeared in their the locality, an example being shown in figure songs. The young birds reproduced a syllable HOW BIRDS DEVELOP SONG DIALECTS 387

FIGURE 2. Examples of Cardinal songs illustrating copying and improvisation. AI is the song of a wild Cardinal reproduced as in A2 by his male offspring the spring after the latter left the nest. Bl is a model song played from a tape recorder and B2 is a copy of it reproduced by a tutored male. Cl and 2 are songs improvised by captive males reared in the absence of model songs from wild Cardinals. The first syllables in Cl are similar to syllables sung by some wild Cardinals. Dl is a song that was peculiar to a certain banded wild Cardinal and D2 and D3 are apparent copies of the same as reproduced by two young males which settled on territories adjacent to the first male. El is a sequence of syllables that was peculiar to the same adult male as in Dl and that was reproduced as in E2 by one of the young males which settled beside him. The more usual form of song is that in E3.

which up to that time had been peculiar to TiTihen acquiring a territory for the first the older bird (fig. 2D) and they omitted a time, if the young Cardinals settle early, say syllable which most birds in the area, except in February or March, then song development the older neighbors, sang with another (fig. will be completed on the territory. However, 2E). if they settle later, in April and May, develop- The copying process terminates with the ment may already be completed. Once devel- final development of the motor patterns. oped, the songs remain stable, as was noted Until this time, the Cardinals are open to in- in wild adults over as long as five or more fluences from other individuals both during years. their first summer and in the following spring. The foregoing evidence indicates the im- Hence, a young Cardinal settling on a terri- portance of sensory information in the de- tory for the first time may copy much from velopment of song. Further evidence of this his immediate neighbors if he settles before is seen from the effects of experimental deaf- motor development is complete. Four young ening. Cardinals deafened after exposure to males exposed to one set of recorded songs songs but before motor development failed to in summer and to another set the following reproduce what they had heard (fig. 3Al) spring copied from both sets. The subject (Dittus and Lemon 1970). Such birds de- which copied least from the first tape was veloped very small repertoires and the syl- least exposed to it. The period when Car- lables were sung quietly (in some instances dinals first become sensitive to songs is not they were sung with great vigor but in known precisely but a young male captured silence), they were less pure with strong three weeks after leaving the nest later repro- harmonics and were accompanied by many duced some of the peculiarities of the dialect short pulses or clicks somewhat reminiscent where he was captured. of certain calls (fig. 3A2). There was also 388 ROBERT E. LEMON

trols exerted by the central nervous system. Such features might include the tendency to repeat syllables in numbers inversely propor- tional to their duration, to repeat the song types in long series or bouts, and to organize the sequences of syllables and song types into predictable sequences (Lemon and Chatfield 1971). Even here, sensory information prob- ably has an effect. In other words, the sensory and motor systems probably function together rather than independently. For example, the songs of one Cardinal can influence the choice of song type and the number of repetitions of such by another (Lemon 1968b). Further comment is needed on the songs of Cardinals reared in isolation from an early age. Syllables of such birds generally differed in configuration from those of their parents, although they were typically pure whistles. Some syllables, however, had practically the same configurations as those of wild Cardinals. In certain cases such syllables were occasion- ally identical to those of the birds ’ parents but perhaps more often they were the same as those of birds located considerable distances away. Two examples can be seen in fig. 2C: The first syllables in Cl are similar to those shown in figure 1, A3, B3, C3; and the last I I 1.0 SEC syllables in C2 are similar to those in 2D. Sylla- bles like this were widespread in Ontario but FIGURE 3. Songs of deafened Cardinals. Al and I did not find them in the repertoires of all A2 are those sung by Cardinals reared in isolation isolates. These syllables were apparently im- without singing experience before deafening. Some portions of the syllables are reminiscent of calls, provised. They might have been innate, but especially those at the end of Al. A2 had exposure more probably they arose through improvisa- to song but had not developed song before deafening. tion in which limits to the possible outcome Bl is the song of a Cardinal before deafening while were set so that some syllables developed in B2 is a version of the same song months after deaf- this way turn out to be the same as those in ening. Although generally the same afterward, there is some hesitancy evident in the first syllable and the wild. This point is elaborated further un- the harmonic on the first part of the syllable is more der improvisation. emphatic. The picture that emerges from these results is that during development of song, a Cardinal some deterioration of songs in birds deafened reproduces the model of song provided nor- after motor development, but much of this mally by other individuals. A memory of the was in the loss of purity of tone through the model may be retained for some months after addition of stronger harmonics. In such birds exposure or may be established during the the original patterns remained recognizable period of motor development itself. In the (fig. 3B). absence of a model provided from another Features of the motor system are probably individual, Cardinals reared in isolation im- also important in determining various aspects provise their own syllables and song types, of song, but the forms of sounds produced by which usually differ from those of wild Car- different song birds are not correlated closely dinals. However, there are enough similarities with the structure of the syrinx, which is in syllable form and in the general features of essentially the same throughout oscines (Miski- whistling and repetition to indicate that in- men 1951). In Cardinals the sounds of herited influences determine these features to deafened birds had roughly the same fre- some extent. These influences are partly quencies emphasized as those of wild birds, sensory and partly motor or the interaction be- between 1.5 and 4 kHz and this may relate tween the two. Further elaboration on de- in part to the syrinx and its dimensions. velopmental processes is given in the next Other features, however, probably reflect con- section. HOW BIRDS DEVELOP SONG DIALECTS 389

DIFFERENTIATION OF DIALECTS : DEVELOPMENT OF NEW SYLLABLE TYPES Aside from the sharing of song patterns among individuals, it is characteristic of dialects that the vocal patterns are distributed locally. Hence we must also deal with the question of how the local repertoires become differenti- ated with respect to each other. Since copying is the major source of the similarities of neighbors ’ songs, local differ- ences develop first in individuals and then spread to others through copying. Hence I shall describe sources of individual variation and thereby show that each singer may con- FIGURE 4. Some individual peculiarities of song. tribute to the differentiation of song in the A variant of the second syllable in Al with strong locality where he is situated. There are several harmonics as sung by one Cardinal only. A2 and possible sources of individual peculiarities, be- A3 show the more common form with little or no evidence of harmonics. In Bl, from a bird reared in ginning with the least important. isolation and tutored from tape recordings, a syllable Because copying involves matching of the from a recorded song is sung in the same song with a motor output with a particular model, hy- syllable apparently improvised. The model song with pothetically one might anticipate errors in the the first syllable type is shown in B2. Cl and C2 are process, either through improper storage of syllables from wild male Cardinals cantured during their first winter. The syllables diffk from those the model because of faulty hearing or because heard in the locality where captured and adjacent of noisy conditions. Also, since Cardinals may areas and were apparently improvised as in the same develop their songs some months after actual manner as with the isolates. Dl and D2 show the exposure to the model, there may be problems inclusion of syllables of unusually high frequency found in the songs of two wild Cardinals. of recall. All of these points are speculative and it is very difficult to find examples that might be copied from each series, those with more ex- considered errors; in fact, only one example perience with the earlier recordings copied seems reasonable. Most Cardinals sing syl- more from them. lables of relatively pure tone without promi- Related to this problem is the time at which nent harmonics. One male Cardinal produced the birds actually achieve or become holders what was apparently a copy of a common of territories. Cardinals may acquire terri- syllable but with exceptionally strong har- tories as early as January and as late as May, monics (fig. 4A), even though the rest of his and those which are later may have com- repertoire was normal. pleted their development before settling. In spite of the possibility of errors, evidence Another source of differences among in- indicates that other sources of variation are dividuals and dialects is evident in captive more important. One of these is the effect of birds but is more difficult to assess in wild the duration of sensitivity and its interaction individuals. The repertoires of captive Car- with dispersal. As noted, song development dinals exposed to model songs included not ends when the motor patterns are stabilized. only copies or variants of these models but Cardinals have a long breeding season in also syllables that were more similar to those Ontario and young may leave the nest from developed by totally isolated birds. Again early May to September in the extreme. A such syllables were usually complete without few Cardinals exhibiting juvenal plumage of subdivisions in time, and their slurring was the year and unaccompanied by adults have most often in one direction only such as the been heard to sing fairly well late in the sea- second syllable in fig. 4Bl. Hence, in addition son in August. This may indicate that individ- to copying, these birds apparently improvised uals leaving the nest earlier might develop from models other than those provided by the their patterns earlier the following spring than recorded songs played to them. In four males other individuals and because of that might be exposed to two sets of recorded songs in both less susceptible to new patterns at that time. autumn and spring, the improvised syllables The few experimental Cardinals exposed to accounted for one third of the total repertoires, two series of recorded songs in autumn and while in birds exposed to one set only, they spring differed in the extent to which they comprised almost half. 390 ROBERT E. LEMON

\\

1.0 SEC I FIGURE 5. The emergence of developed song from subsong in a Cardinal isolated from five days after hatching and without further exposure to Cardinal songs. A shows a portion of subsong beginning with a series of calls, a) to d), and beginning song elements apparently occurring with the calls. B is a portion of subsong one week later than A showing syllables that appear to be the forerunners of those stabilized in C and D, the latter recorded seven months later. Evidence of calls decreases throughout. The structure of the song syl- lables seems related to the immediately preceding syllables.

Improvisation of this sort may happen under cesses, two of which are elaborated here. natural conditions also. Cardinals captured First is the use of calls, which are well de- in the wild during their first winter after ex- veloped by the time of song development, to posure to song the previous summer not only provide an initial basis for comparison and reproduced copies of the local dialect, but also to facilitate the motor processes through produced syllables not found in the locality auditory feedback. Second is the modifica- nor at several places in Ontario (fig. 4C). tion of the calls and early song elements by a Nor were the syllables considered copies of phenomenon called “drift,” named after the songs from other species. If such improvised process as seen in the domestic fowl (Andrew syllables are retained in wild Cardinals, they 1969). In drift, the elements are immediately would be difficult to distinguish from similar repeated but with some modification; the re- syllables copied from models unless they had sult is a variant of an earlier form that may some distinguishing peculiarity. One such eventually become stabilized as a new sylla- peculiarity would be a syllable of exceptionally ble. high frequency, for although during develop- In Cardinals four calls commonly used in ment sounds of relatively high frequency are subsong are similar in form and occur in simi- produced, say of 6 kHz or more, most emphasis lar contexts. The chuck is a short broad- in the final repertoire is at lower frequencies. frequency pulse with lower frequencies em- Two birds in Ontario sang peculiarly high phasized; the chip is similarly short but of syllables not shared by their neighbors and more restricted and higher frequency of 4 to which may have had the origin just described 8 kHz; the chitier is a series of short repeated (fig. 4D). pulses but is of high frequency, 6 to 10 kHz; and the fourth, the pee-too, differs somewhat IMPROVISATION OF SYLLABLES in being slightly longer with downward The foregoing has shown that development of slurring and in having two separate portions song involves either the replication of a model of high and low frequency. The chip is a based on memories of songs of other birds, or contact call used by pairs and flocks. All alternatively the improvisation of a repertoire four calls are used in agonistic encounters of syllables without the aid of such models. within the species, while the chip is also used Although these two possibilities seem unre- in mobbing predators and the chitter is also lated, they probably involve the same pro- a warning of aerial predators. HOW BIRDS DEVELOP SONG DIALECTS 391

I I 0 I 0 1.0 SEC

FIGURE 6. Drift of syllables in the songs of adult wild Cardinals. A to D show drift related to acceleration with associated decreases in frequency in A and B and increases in C and D. E and F show changes in over- all confinuration: G and H show the addition of Darts to the syllables as the songs progress, although in H the additions are often incomplete.

The interaction of the calls and drift can songs shown has two syllables of essentially be seen in portions of developing song or sub- the same slurring, as if one had developed song of an individual reared without exposure through drift from the other. to songs from other birds or recordings. The bout of subsong shown in fig. SA begins with DRIFT IN DEVELOPED SONGS three chucks, (a), followed by a chip, (b), Certain forms of drift can be seen clearly in the pee of pee-too, (c), and the same fol- developed songs of Cardinals where the ef- lowed immediately by a chip, (d). There fects may be gradual (fig. 6A) or rather follows a series of calls and varied elements abrupt (fig. 6B-E). The result of the drift which are more truly the beginning elements is often a form similar to the original as in of song. These elements are not random ex- these examples, or it may be a syllable more pressions of sound but rather each is usually different as in fig. 6F where the second part related to the immediately preceding element of the syllable is eventually slurred downward although they show some variation from it, whereas initially it was upwards. In fig. 6G, and although some of the variation probably H, are examples where parts are added to reflects lack of motor control. The structure syllables as the songs progress, although in H of the chuck calls in the first sample is such this addition was not always complete, as if that they are indistinguishable from other tran- reflecting incomplete facilitation of the sing- sient sounds with emphatic beginnings and ing process. endings, but their occurrence in the middle Where drift does not involve the addition of syllables suggests they are indeed calls. or omission of parts, conceivably it could re- In view of the two voices frequently heard in sult through a change in the rate of singing songs, a call and a song element could occur such as one can alter the frequency/time re- simultaneously from different sources. lationships of recorded sounds by varying the The occurrence of the described calls was speed of playback. When this is done changes very noticeable in bouts of developing song in speed have directly proportional effects on in wild Cardinals even when they sang clearly frequency and inversely proportional effects recognizable syllables (Lemon and Scott on time periods. 1966). In fig. 6C the drift in this two-part syllable A later portion of a bout (fig. 5B) shows involved both a shortening of the syllable and more immediate repetition and more gradual the interval between syllables as well as an modification or drift. At this time there ap- increase in frequency (pitch). The intervals pear the forerunners of syllables of the final between syllables before the shift were about repertoire as shown in fig. 5C and D. In these 0.06 s and 0.03 to 0.035 s after, while the dura- later samples the calls are less obvious or en- tions of the syllables were approximately 0.21 tirely absent. Note also that each of the two s before and 0.11 s after. As an index of change 392 ROBERT E. LEMON

I I I 010 LOSEC

FIGURE 7. The reordering of syllable sequences. Al is the combination of two syllable classes by a captive bird tutored with the two separate songs in A2 and A3. Bl is an unusual sequence of syllables in a young captive wild Cardinal which had some months in the wild before capture. The more usual sequence is in B2. Cl is an unusual combination in a captive bird as in Bl with the more usual combinations of the particular syllables in C2 and C3. Dl is an example similar to Cl but as recorded from a Cardinal in the wild. in frequency only the lower limit of the second in fig. 6A, the duration of successive intervals part of the syllable was measured, the value decreased, but the syllables became longer being approximately 1.9 kHz before the shift in time and lower in frequency. The maxi- and 2.3 kHz after. The changes in both in- mum frequencies of the four su,ccessive syl- tervals and durations were roughly equal after lables were 3.3, 2.7, 2.6, and 2.5 kHz, while the shift, being about one-half the original the durations of the syllables were 0.28, 0.32, value, but the change in frequency was much 0.35, and 0.37 s. The successive intervals less than expected by the simple acceleration were 0.17, 0.07, 0.06, and 0.07 s. Figure 6B model because the minimum frequency of the shows another example of a drop in maximum second part of the syllable should have been frequency (from 3.4 to 2.9 kHz) and an in- approximately 3.8 kHz instead of 2.3 kHz. crease in duration (from 2.1 to 2.4 s ) , although Note also that the maximum frequency of the in this case the intervals also shortened (from second part of the syllable is lower rather than 0.06 to 0.05 s). higher after the drift. Drift in developed songs may occur at dif- In fig. 6D a similar situation applies. The ferent positions in consecutive songs. For ex- durations of the syllables were reduced from ample, the sudden change noted in fig. 6C in four successive songs by the same bird oc- about 0.13 s before drift to about 0.05 s after; curred between the following syllables num- although it was difficult to obtain accurate bered from the beginning of each utterance: measures of minimum and maximum fre- 9-10, 12-13, 13-14, 12-13. In some examples, quencies, it is clear that the increase in fre- although not in this one particularly, the usual quency was hardly proportional to the change sudden change did not always occur even in in duration. Also, the successive intervals be- relatively long utterances. tween syllables shortened progressively from These examples show that changes in the 0.15 s to 0.08 s without appreciable change in duration and frequency of the syllable cor- the form of the syllable until the interval be- relate with changes in the rate of singing. came about 0.06 s. Sometimes, however, this correlation is posi- In the example of more gradual drift shown tive and sometimes it is negative so that it is HOW BIRDS DEVELOP SONG DIALECTS 393 not exactly as one would expect by the simple scripts representing the number of repetitions. acceleration model. Of course, this model The usual sequences in this area were WBW, takes no account either of the situation where C, UY, with Z following at the end of many parts are added or omitted to syllables or of permutations, and for the second WBW, A, further changes in configuration. In conclu- and T. Incidentally, both of these males ac- sion, it seems possible that what is described companied these songs with series of calls. as drift may result from several underlying in- Gradually more certainty in the sequences fluences. of syllables is established and less frequent Although the examples considered here permutations and combinations are eliminated. have been taken from the developed songs of In one example I heard a wild male sing 35 Cardinals, they reflect influences present permutations during his first season but only earlier in the development of song. There is 18 in the second season, only about half of reason to believe that these influences play which were used commonly. Figure 7Dl an important part in the development of new shows an unusual combination from another syllable types as is described in the next bird consisting of two syllables usually sung section. alone, as shown in D2 and D3. The uncom- mon permutations which remain usually re- SEPARATION AND REORDERING OF sult from two syllables which are more often SYLLABLE TYPES sung separately and less often in the same Once new variants of syllables have been song type. Lemon and Chatfield (1971) produced by drift, they may be either sepa- showed detailed examples, particularly from rated and sung alone in separate song types or one Cardinal called “Chambers,” which in- recombined with other syllables into new song cluded syllables designated A and T which types. This separation process can be ob- were most often sung alone, but were also served in the developing song or subsong. sung in combination as AT or TA and ATA. Initially the calls and developing syllables are Whether the permutations and combina- often presented in long series lasting several tions are determined by any rules relating to seconds overall but as development progresses the forms of the syllables is difficult to deter- the utterances shorten to an average duration mine. Most often a Cardinal achieves a com- of only two to four seconds in the Cardinal, bination used by the bird he copies, but often and fewer kinds of syllables are sung in the long syllables are associated with short syl- same utterance. Thus the different syllables lables as in fig. 1B and C. This may partly ex- come to be sung either alone or with only one plain why the permutations seen in these two or two other types in the same song. examples occur widely throughout southern Examples of the separation of syllable types Ontario and the adjacent United States. were seen in experimental birds provided with model songs containing two syllable types EMERGENCE OF NOVEL SYLLABLES which the birds sang independently (fig. 7A). If the foregoing model is true, then we should Evidence of this reordering of syllables was expect to find in the repertoires of individual noted too in birds captured in the wild during Cardinals syllables that are very similar and their first winter. In fig. 7Bl is a different yet are used independently. Two examples sequence of syllables sung by such a bird are presented here that apparently reflect whereas the more usual sequence is seen be- these processes. low. At the time of recording, this Cardinal Near Melbourne, Ontario, most Cardinals had not stabilized its singing completely and used two very similar syllables, examples from was still singing exceptionally long sequences. four individuals being shown in fig. SA. The In fig. 7Cl is yet another example of an un- syllables on the left half of the figure were usual sequence from a Cardinal captured in sung at rates of roughly three to four per second the wild, which was formed from syllables while those on the right were sung at nearly taken from two otherwise separate songs five per second. Often there were consistent heard from other birds in the area (C2, C3). differences in these syllables between individ- This final sorting of sequences was noted ual birds, but among the group as a whole the also under natural conditions. Using symbols only consistent difference was that the initial to represent the syllables (Lemon and Scott part of the syllables swept upwards more 1966) two young males were heard singing rapidly in those syllables sung faster. In some the following long utterances: the first sang syllables, however, as seen in the first three utterances of AWCZCY and AWXSYCAW, individuals especially, certain portions were the number of syllables not being recorded; either present or absent. the second sang W1A1W12ABT5Wk, the sub- Another example shows even more combina- 394 ROBERT E. LEMON

I I 1.0SEC

FIGURE 8. The emergence of new syllable classes in Cardinals. In A the songs are presented in pairs on the left and right, each pair being from one of four birds from Melbourne, Ontario. The two similar syllables were sung by each bird in separate songs. One consistent difference between them was that‘ the syllables on the right side of A were shorter and were sung at a faster rate than those on the left. In B, songs are again shown in pairs, but in this figure the‘ songs of each pair come from a different bird at London, Ontario. Each pair shows one of five arrangements of syllables differing by the presence or absence of common parts, Bl to 4, or in the fusion of parts as in 5. Birds in the locality usually had Bl and 2 and most also had one of the following three.

tions of variation, all being recorded from birds least of establishing their integrity. Otherwise, at London, Ontario, and living within a small constant mixing of populations would mediate area. Again a syllable similar to that in fig. against local peculiarities. Obviously geo- 8A is shown in 8B1, with examples from two graphic barriers may play an important role different birds. In this figure variation is seen in some species. For the Cardinal in eastern more definitely as the presence or absence North America, however, such barriers are of certain parts of the syllable. Two parts often only minor, if present at all. Hence we occur in the syllables in Bl; a third part is must look to other factors. added in B2; one is absent in B3; two parts Cardinals are essentially sedentary or non- are absent in B4; while in B5, it appears that migratory, although the adults may leave their two parts have been fused. All male Cardinals territories in winter to join feeding flocks. in the area had at least two of the syllables They are highly loyal to their breeding ter- and many had three. ritories from year to year; D. M. Scott (pers. It is possible that examples of a new syllable comm.) has found that of 100 to 200 adults type copied by young neighbors shown earlier banded, none was found to change his breed- in figure 3 and shown again here in figure ing territory any farther than the distance of 8B4, arose originally when a Cardinal dropped one territory from the original held. Further some portions of a syllable already in the there is a relatively high return of adults from area. Alternatively the simplicity of this syl- year to year, about 70%, obviously reflecting a lable in question suggests that it was im- relatively low mortality over winter. Of those provised without regard to any model from birds acquiring territories for the first time, a another individual. large number are the offspring of local parents and may settle between 2 and 10 or 12 ter- ritories distance from the territory where they LIFE HISTORY ASPECTS RELEVANT were reared (P. Smith, from D. M. Scott). TO DIALECTS Less detailed, but similar results were reported For dialects to occur, there must be some basis by Laskey ( 1944) in Tennessee closer to the of separation of the local populations or at center of distribution of the species. HOW BIRDS DEVELOP SONG DIALECTS 395

TABLE I. Certain fringillid species grouped according to the degree of conformity seen in the songs of males on immediately adjacent territories. Identical superscriptions (a to d) indicate species that are congeneric or nearly so. + or - indicates presence of dialects as determined by reports in the literature.

Group 1 Group 2 Group 3 High Conformity Moderate Conformity Low Conformity

“Cardinal Chaffinch + dOregon Junco - “Pyrrhuloxia “White-throated Sparrow - dMexican Junco - Rufous-sided‘ Towhee + Rufous-sided‘ Towhee bBrown Towhee - White-crowned‘ Sparrow + Rufous-collared‘ Sparrow Chipping Sparrow - Song Sparrow z Indigo Bunting - Rose-breasted Grosbeak ?

Cardinals are relatively recent in Ontario, dinals ( Lemon 1966)) a member of this species having been reported first in 1901 at Point would have to travel perhaps some hundreds Pelee at the western end of Lake Erie. From of miles to get to an area where most of the that area many Cardinals apparently moved repertoire is unfamiliar. Of course this need into Ontario by following the Thames River, not be so with other species. which offered suitable habitat for breeding In conclusion, the basis of dialect integrity and winter survival. As a result the first Car- in Cardinals is found principally in the habits dinals occurred at London in 1915 and by mid- of territorial fidelity and a low rate of annual 1930s’ they were fairly common in that area. mortality. Dispersal occurs, but most often Other Cardinals apparently entered Ontario over short distances. Over longer distances it via the Niagara Peninsula. probably accounts for widespread similarities From an analysis of banding records prin- of song as noted with figure 1. cipally from the United States, Dow and Scott (1971) found that 13% of 1500 recaptures DEVELOPMENT OF DIALECTS IN were outside the 10 minute block of latitude OTHER FRINGILLIDS and longitude in which they were banded. Of Having considered the development of dialects course these data do not indicate what propor- in one species in some detail, I wish next to tion of the population actually disperses. More examine other species, first dealing with those birds dispersed during their first year than of the same family, particularly since the during their second and they travelled farther. fringillids have been studied better than any Most of these dispersed less than 30 miles but other group. the upper limit was 550 miles. Hence, most We have seen that in Cardinals the situa- Cardinals disperse to areas where some of the tion with regard to dialects is complex be- songs will be very similar to those they had cause some syllables and song types are wide- heard from their parents. Yet the occurrence spread while others are local. This raises the of some local differentiation at even shorter question as to whether there is any precise distances (Lemon 1966)) indicates how effec- definition of dialects. Actually there is not. tive some sources of differentiation of song I shall not review in detail the literature on must be, particularly the modification through this matter because some authors take ac- drift. count of whole song types, or just certain Although many males acquire their terri- syllables at certain positions in the song types. tories and mate successfully during their first Also, the birds ’ own assessment of differences full season after leaving the nest, some young in songs is not likely to coincide with ours. males fail to acquire mates and in subsequent Indeed, in playback experiments we some- years they are found on a different territory. times find that species differ in their responses Other young males appear to wander through and in ways that are sometimes surprising. their first potential breeding season without Some examples will be cited in the next holding a territory, at least permanently. It sections. My primary concern here is to account for has been suggested that for other species at the conformities and non-conformities in the least, young birds might be influenced in their repertoires of birds on neighboring territories choice of nesting area by the similarity of the and in other localities. Therefore, we need to songs of that area relative to those of the consider to what extent different species show selecting individuals ’ home area. It is impor- conformities; we will also consider the inter- tant to note that because of the widespread pretations of the researchers on whose studies distribution of much of the repertoire of Car- the analysis is made as to whether the species 396 ROBERT E. LEMON show dialects. The assessment of the degree Copying as the basis of conformity has been of conformity in the songs of birds on neigh- clearly established in the White-crowned Spar- boring territories is here made according to row (Marler and Tamura 1964, Marler 1972). three levels, high, medium, or low. This ar- In this species, however, copying appears re- rangement is arbitrary and does not take ac- stricted to patterns heard only during the first count of the wide differences in the sizes of the one to seven weeks. Some of the songs de- repertoires. A summary of the species con- veloped by isolated White-crowned Sparrows sidered (table 1) also shows the occurrence in the absence of models from other birds are of dialects. reminiscent of the songs of White-throated Those species showing high conformity have Sparrows (Zonotrichia albicollis) in being com- both syllables and sequences of syllables in prised entirely of whistles of sustained rather song types essentially identical. Those in the than modulated frequency. Hence there may medium category have some syllables in com- be some inherited influence on this aspect. mon and possibly also some sequences of Some of the published audiospectrograms have same, while those in the low category ap- examples suggestive of drift in this species. parently have no syllables in common and The importance of auditory feedback for nor- hence no sequences either. The degree of con- mal song development in White-crowned Spar- formity and the occurrence of dialects need rows was demonstrated by Konishi (1965a). not agree, for there may be widespread con- Development has not been studied in the formity without local differences. Closely re- Pyrrhuloxia or the Rufous-sided Towhee, nor lated species may occur in different groups, is drift immediately evident from the pub- supporting the view that the bases of song lished audiospectrograms. However, many of development are the same although reflecting the syllables used in separate songs are differences in emphasis. strikingly similar, as if from a common origin. Information on population dynamics is GROUP 1. HIGH CONFORMITY available from White-crowned Sparrows. They This group includes species clearly con- may be either migratory as in populations in sidered to have dialects. These are the Car- the Sierra mountains of California (Morton, dinal and its close relative, the Pyrrhuloxia Horstmann and Osborn 1972), but in the San ( Pyrrhuloxia sinuata), the Rufous-sided Tow- Francisco area they are often permanent resi- hee ( Pipilo erythrophthalmus), and the White- dents (Blanchard 1941). In both areas there crowned Sparrow (Zonotrichia leucophrys) . is a high degree of return to the same territory The first three species have several song pat- in successive years. The annual mortality of terns per bird, each consisting of repetitions of adult birds in both areas is roughly 50%. one to three syllable types; both syllables and their permutations in the songs are strikingly GROUP 2. MODERATE CONFORMITY similar among different individuals. Cardinals This group contains species that have con- have 9 to 12 commonly used songs, most often siderably more individuality than was shown of two syllables types (Lemon 1965, Lemon in the previous group, but some of which still and Herzog 1969); Pyrrhuloxias have 10 to 15 show evidence of dialects. They show more songs of mostly one syllable type (Lemon and conformity than species in the third group to Herzog 1969); Towhees in western United be considered. States have nine songs or sometimes fewer, The Chaffinch (Fringilla coebbs) of Europe each consisting of two syllable classes has been considered to exhibit dialects but ( Kroodsma 1971). The White-crowned Spar- there is also apparently a considerable amount row differs with usually only one or sometimes of individual variation. One problem in as- two songs per bird, each beginning with sessing the situation is that published audio- whistles of sustained frequency followed by spectrograms are limited and the principal two trills of repeated and rapidly slurred syl- study on geographic variation (Marler 1952) lables (Marler and Tamura 1962). was done by ear. Each male has two to five With Cardinals and White-crowned Spar- song types each consisting of two or three rows, playing songs that are obviously differ- phrases of repeated syllables followed by a ent from those of the locality concerned evokes non-repeated terminal flourish. The songs of significantly less response than the local song neighboring Chaffinches may be sufficiently (Lemon 1967, Milligan and Verner 1971). In similar to exhibit the matching characteristic the first three species, individuals frequently of Cardinals and towhees (Marler 1956), but interact by matching each others’ choice of there is also much individual variation with song type (Lemon 1968b, Lemon and Herzog the result that the chief similarity is in the 1969, Kroodsma 1971). terminal flourish. HOW BIRDS DEVELOP SONG DIALECTS 397

In studies of song development, Thorpe (Lemon and Harris 1974). The syllable (1958a, 1958b, 1961) and Nottebohm (1968) classes are the same in this species over wide demonstrated that imitation was the chief geographic areas (Borror and Gunn 1965) and basis of similarity among Chaffinches and that are much the same in the two areas just men- isolated individuals developed song patterns tioned. Hence, in this species although neigh- which in complexity of syllable structure and bors may or may not share the same syllable sequence were clearly related to the exposure sequences, if they do, there is yet no clear that they had received. Those birds without evidence of dialects, although significant dif- exposure improvised songs with very simple ferences occur in the frequencies of syllables. rapidly slurred syllables and simple terminal Playback experiments in Algonquin Park flourishes. The period of sensitivity seems sim- revealed greater responses to songs from birds ilar to that of Cardinals, lasting normally un- on non-neighboring territories than to those til motor development in the first spring after of immediate neighbors which were familiar leaving the nest. Nottebohm (1969a) sub- to the experimental subjects ( Falls 1969). The sequently showed that the development could same applied to birds in New Brunswick, be delayed by castration but may be resumed which also failed to show any differential re- with injections of testosterone. There is evi- sponses to songs from other localities (Lemon dence that Chaffinches exhibit subsong each and Harris 1974). spring as adults, but it is probable that this The Song Sparrow (Melospiza melodia) is reflects low motivation rather than true de- important in our considerations for it has been velopment of song. There is evidence also fairly well studied, its organization indicates that in subsong, Chaffinches may incorporate an intermediate position in its degree of con- sounds employed by other species but that formity between neighbors and the results of these are excluded from the final repertoire these studies have led to some confusion. (Thorpe 1961). Drift is not explicitly de- Song Sparrows live in open areas with rela- scribed in the species but is indicated in the tively small trees, shrubs, and patches of grass; audiospectrograms. in such areas near open water they often occur Two members of the genus Zonotrichia in high densities with territories much less fall within this group, the White-throated than 1 acre in size (Tompa 1963, Harris and Sparrow of eastern North America and the Lemon 1972). Each male has several song Chingolo or Rufous-collared Sparrow (2. patterns, up to 15 in California (Mulligan cupensis) of South America. The songs of 1966) and up to 9 or 10 in Quebec. The songs the White-throat consist almost entirely of are complex, perhaps the most complex of all whistled syllables of sustained frequencies, fringillid songs. They consist of many varied although successive syllables may shift in fre- syllables, some whistled with many sub-parts, quency relative to each other. As noted, sus- some tremolos, and some very brief sounds. tained whistles are characteristic of White- Some syllables are repeated in trills. The same crowned Sparrows, and they are also a feature syllable may occur in different songs of the of other North American members of the same bird, but this applies only to a small genus, the Golden-crowned Sparrow (2. portion of the total of an individuals’ reper- atricnpikz) , and Harris ’ Sparrow (2. querula). toire. Male Song Sparrows on neighboring The Chingolo does not show this tendency, territories rarely have identical songs, but they though, for its songs begin with varied se- share some syllables (Borror 1965, Mulligan quences of non-repeated slurred whistles, fol- 1966, Harris and Lemon 1972). lowed by a trill of rapidly slurred syllables. Although dialects in Song Sparrows are not Neighboring Chingolos frequently differ in as evident to human observers as those in their initial parts of the songs, although there other species, Song Sparrows have shown sig- are instances where one form may predominate nificantly less response to songs of their own locally, but the claim of dialects in this species species from only a few miles distance as com- (Nottebohm 1969b) is based primarily on the pared to songs from the experimental subjects ’ terminal trill, which does vary locally (see own locality (Harris and Lemon 1974). This also King 1972). difference in response was more noticeable White-throated Sparrows offer a different than that to non-neighbors ’ and neighbors ’ situation. In the region of Algonquin Park, On- songs from the subjects ’ own localities, in spite tario, most individuals follow a few sequences of the obvious individual variation in the songs of amplitude-modulated whistles. However, in (Harris and Lemon, in preparation). coastal localities in New Brunswick there is Mulligan (1966) studied the development much greater individuality, with the result of song in this species. Some individuals were that no particular sequence predominates placed while in the egg under an incubating 398 ROBERT E. LEMON

Canary, (Se&us canaria) and were subse- Cardinals was published earlier (Lemon and quently reared by Canaries, while others were Scott 1966). This point aside, however, their hand-reared after having been taken from statement implies far more than has been the nest after hatching. These individuals demonstrated. were eventually reared either separately or My own interpretation is that Song Sparrows together in groups, with no major differences are no different from other fringillids, all of in the results, Mulligan concluded of the songs which show evidence of copying and im- developed by these birds that, “If one were provisation. Where they differ is in the degree to mix these songs with recordings from wild to which each of these influences prevails. Song Sparrows, it would be difficult to dis- Probably Song Sparrows in isolation actually tinguish them from the latter.” However, it is develop some syllables essentially identical to also clear that the repertoires were smaller those of their parents or others. But as noted, than those of wild birds, both in numbers of wild Song Sparrows share only a portion of syllables and song types, and there were addi- their repertoires. Also, the size of the reper- tional quantitative differences as well. Also, toire and the complexity of the patterns pro- three siblings reared by the Canaries sang duced help to obscure the similarities with quite different songs. There was no attempt other species. to assess whether the syllables were identical Another species considered in this group is in form with those of the parents or members the Indigo Bunting (Passerina cyarwa) . Mem- of the wild population (Mulligan, pers. bers of the genus vary in the number of syl- comm. ) , although some presumably could lable types per song and in the tendency to have been. Some individuals who were allowed repeat the syllables within the song (Thomp- to hear tape recordings of Song Sparrow songs son 1968). Indigo Buntings usually have one showed clear evidence of copying. This fact song type per bird, consisting on the average plus the occurrence of similar syllables in im- of four syllable types (called “figures” by mediate neighbors in the wild gives strong Thompson) which are repeated one to three evidence that imitation normally plays a role times each. In local populations, the number in the development of song in this species. of syllables in all individuals together seems One individual who was deafened before song relatively constant at approximately 60 to 80 development eventually sang, but his reper- (Thompson 1970, Shiovitz and Thompson toire was relatively small and the syllables 1970, Emlen 1971a). However, none of these differed in quality from those developed by syllables is common to all birds in each pop- other isolates and wild Song Sparrows. ulation; in fact, the mean number of individ- Song Sparrows in the San Francisco area uals in which a syllable occurred in a sample where Mulligan studied are highly sedentary of 38 to 46 buntings in Michigan was about 6. and distinct populations have been recognized. Not surprisingly therefore, there is relatively From year to year there is little change in the little sharing of entire sequences of syllables, territorial positions of particular males (John- which in the same samples was limited to less ston 1956). Dispersal of young is often over than 25% of all birds. In another sample from short distances with a median of 185 m, about New York from 36 birds, the figure was less 100 m less than in Ohio. In Ohio, where at 7%. Where the syllable sequences were the Song Sparrows are more migratory, there is same, more often than not, the buntings shar- also a high incidence of return to the same ter- ing them held contiguous territories. In spite ritory, but there appears to be more movement of the large number of syllables in each sample than is seen in Cardinals (Nice 1937). there was about 90% sharing among each of In the literature, the Song Sparrow has been three sites in Kentucky, New York, and Michi- treated as a special case with a development gan, In this respect this species is like the of song different from that of other fringillids White-throated Sparrow. (Marler 1967, Nottebohm 1972b). As an ex- Within a population of buntings, many of ample, Konishi and Nottebohm (1969) wrote, the syllables are similar, as if derived from the “These birds produced songs indistinguishable same origin. Shiovitz and Thompson (1970) from the average song of wild Song Spar- cited a case of a bird which had two very rows. . . . This is, perhaps, the first and only similar syllables yet apparently employed them reliable study of passerine birds in which the differently in the song. Evidence from bunt- audiospectrograms did not show any qualita- ings reared under controlled conditions (Rice tive difference between isolate and wild-type and Thompson 1968) indicates that the de- songs.” Actually, audio-spectrographic evi- tails of the syllables are normally copied from dence showing syllables of isolated Cardinals other birds, and isolates develop songs which remarkably similar to certain syllables of wild exhibit syllables different from those in the HOW BIRDS DEVELOP SONG DIALECTS 399

wild. The captive birds showed ability to syllable type repeated in each song while the copy throughout the spring after leaving the latter employed three to four syllable types nest for a longer period of time than observed per song. Although accurate samples were in captive Chaffinches and Cardinals. Whether not taken, some individuals have more than such is the case in the wild is not known. one song type. Syllables in populations show Thompson (1970) suggested that the simi- a high degree of individuality, no two birds larities of syllables are the result of copying being found to have the same. Therefore, no that may begin when the young birds leave evidence of dialects could be found. the nest but may also continue the following The Oregon Juncos raised in two groups in spring, when they settle on a territory. He two seasons each developed songs, the larger ruled out the possibility of effect of the ter- repertoires occurring in the larger group ( Mar- ritory while the buntings winter in Central ler et al. 1962). Some of the songs consisted America and during their return migration. of two syllables. In the smaller group, the It is possible that some young return to ter- syllables differed among all the birds, whereas ritories adjacent to those of their fathers, but in the larger group, at least three syllables he thought this unlikely as a general rule. were shared by two or three birds. These Perhaps more so than with Cardinals, young birds also showed evidence of copying songs Indigo Buntings do not necessarily retain the from sparrows and towhees. Some individuals territories first acquired, but instead change also developed new syllables during their in the second year. This could account for the second season, based on material copied from relatively low sharing of sequences in a pop- others, sometimes of another species, or modi- ulation. The individuality in the songs is ap- fied from a syllable sung previously. At least parently involved in the markedly greater re- one published audiospectrogram shows evi- sponse levels to songs of non-neighbors than dence of drift. to songs of immediate neighbors (Emlen In the Arizona Juncos, the situation was 1971b). somewhat different, for most juncos reared Rose-breasted Grosbeaks (Pheucticus ludo- together had songs of fewer syllable classes vicianus) sing songs with syllables of variable than those of the wild population (Marler order, which are predictable on first and 1967). Birds reared together developed songs second order Markov chain models (Lemon with a number of syllables comparable to those and Chatfield 1973). The repertoires may in the wild population, while an isolated in- have over 20 syllables per bird, roughly a third dividual had fewer syllables per song. There of which may be common to neighbors. was evidence of copying among the group Konishi (1965b) reared Black-headed Gros- reared together as well as in the bird exposed beaks (Pheucticus melanocephalus) in isola- to recorded songs. tion and also deafened one. The isolates were The importance of auditory feedback in the fairly normal in their syllables although they development of song in these species was did not sing any of the syllables found in shown by Konishi ( 1964b), and deafened birds the single wild individual studied. The lacked stability in the syllables developed. deafened bird also sang but his syllables were Differences between the species occurred but not as stable as those of the birds with hearing. these may reflect differences in the experi- Rufous-sided Towhees may also be con- mental procedure (Konishi and Nottebohm sidered in this group as well as in that of high 1969). conformity, for in the eastern United States Hence in these species we see not only a they show considerable individual diversity high degree of individual variation in natural in syllable sequences, although some syllables populations, but also evidence of copying in are shared ( Borror 1959b). captive birds. Therefore, the juncos do not differ radically from the species previously GROUP 3. LOW CONFORMITY considered. The significance of the individ- uality of song is not known. I have observed This group contains species in which individ- singing by flocks of migrating Slate-colored ual variation clearly predominates over group Juncos (.I. h. hyemalis) where there might be similarity. The main examples are from the an emphasis on individuality. The rapid genus Junco, which is considered closely re- modulations of the syllables also suggest that lated to Zonotrichia and Melospixa. Two the prime information used may not be fre- species studied, the Oregon Junco (J. hyemalis quency but rather amplitude modulations in oreganus) (Konishi 1964a) and the Arizona time. Junco (J. phaeonotus; Marler and Isaac 1961) Other species such as the Brown Towhee differed in that the former usually had one (Pipilo fuscus) (Marler and Isaac 1960) and 400 ROBERT E. LEMON the Chipping Sparrow (Spixella pusilla) of calls that are apparently the counterparts (Borror 1959a) have been described as show- of territorial songs. Neighboring males share ing much individual variation but little is much of their repertoires independently of known of the develonment of their sonas. the females. Males or females when interact- ing often select calls similar to those sung by COPYING, DRIFT AND CALLS IN SONGS their partner, but when mates interact they OF OTHER SPECIES usually select calls that are different. In de- We may ask how generally the results so de- velopment it appears that partners actively scribed apply to song birds other than avoid copying the calls of the opposite sex. fringillids. Copying has been shown in a How recognition of a Mynahs’ sexual role is variety of species (Lanyon 1960, Thorpe 1961, determined is not known and the species are Marler and Hamilton 1966). The older litera- monomorphic. ture also refers to innateness of song, often In certain African shrikes, males and fe- meaning the general characteristics of song males develop special repertoires by which and not necessarily the precise configurations they interact with great precision. Although of the syllables, although there may be some the repertoires are often unique, the unique- valid exceptions. A few relevant examples in- ness in some species at least is primarily in cluding aspects additional to copying are dis- the sequences of syllables in the songs; birds cussed here. of a locality may share a common repertoire Copying combined with improvisation of of syllables (Thorpe 1972, Wickler 1972). new syllables has been well documented in Thorpe attributed to shrikes an ability to the European Blackbird (Turdus me&a) improvise syllables based on musical prin- ( Thielcke-Poltz and Thielcke 1960, Hall- ciples. Craggs 1962). Calls are also important in de- African indigobirds (Viduinae) exhibit song velopment. The organization of the song is dialects copied partly from their brood probably similar to that of fringillids in that hosts as well as from their own species (Payne although the repertoire is large, the sequences 1973). Payne attributed differences between are often highly ordered. A highly ordered birds and dialects to errors in copying although sequence is seen also in the close relative the evidence seems limited. Drift as a basis of Mistle Thrush ( Turdus wiscivorus) (Isaac and syllable variation seems evident in Splendid Marler 1963), which also shows local varia- Sunbirds (Cinnyris coccinigaster; Grimes tions of repertoire. Blackbirds also show fre- 1974). quent alternations between syllable classes (Todt I970), somewhat in the same form as SOCIAL INFLUENCES ON SOURCES the alternations of song types in short bouts OF MODELS by Cardinals. One other important aspect in the develop- Dialect variation is claimed in Short- ment of song in birds is that of the social in- toed Treecreepers (Certhia brachydactyla) fluences on the selection of material to copy. (Thielcke 1969, 1972). Drift is evident in The selection of material can be important successive syllables with intergrades occurring in maintaining species isolation, especially for between the most different syllable types. close relatives, the songs of which are often Calls appear to be the forerunners of syllables rather similar, probably because of common in Certhia and the syllables of C. brachy- inherited influences as well as originating from dactyla and C. familiaris are very similar; a particular repertoire. Examples of simi- some calls and elements in songs appear to larities are seen in creepers (Thielcke 1964), be homologous ( Thielcke 1964). Cardinals, and Pyrrhuloxias (Lemon and In the Old World warblers of the genus Herzog 1969). Phylloscopus excellent examples of drift can There are many examples where birds fail be seen ( Thielcke and Linsenmair 1963). In- to observe their species limits but instead copy deed much of the variation seen occurs about other species, sometimes unrelated. Such ex- a basic plan somewhat similar to the additions amples occur in fringillids (Borror 1968, or subtractions to the syllables noted earlier Baptista 1972)) meadowlarks ( Sturnella) in Cardinals. (Lanyon 1957)) and wrens (Troglodytes and Indian Hill Mynahs (Gracula religiosa) Thryomanes) (Kroodsma 1973). In some re- (Bertram 1970) not only show high con- lated species, however, copying the other formities in the calls among neighbors rela- species ’ repertoire may be useful in repelling tive to birds of other localities, but also the them, as may be the case in tits (Parus) conformities are confined to members of the (Thielcke 1969) and creepers (Thielcke 1970, same sex. Males and females have repertoires 1972). HOW BIRDS DEVELOP SONG DIALECTS 401

Mimicry of other species or sources has TABLE 2. A schema relating variation in bird song occurred among captive birds reared with to developmental processes. others of a different species (Scott 1902, Lanyon 1960). Of course, it also occurs nat- A. NATURAL VARIATION IN REPERTOIRES urally among some species such as the Mock- Conformity t--) Individuality ingbird (Mimus polyglottos), which has a B. DEVELOPMENTAL INTERACTIONS large repertoire (Borror and Reese 1956). Imitation * Improvisation Mimicry by nest parasites of songs and calls ( Copying) (Time-Dependency of Copying, of their host species occurs regularly in African Dispersal of Young viduine finches (Nicolai 1964, 1973, Payne Reordering of Syllabics )Drift ’ 1973). C. CONSTRAINTS These results indicate that the association 1. Sensory-motor of auditory and visual stimuli is an important 2. Social basis of determining which sounds a species 3. Behavioral-Ecological copies, except in the mimics like the Mock- ingbird. Confirming experimental evidence of this conclusion comes from the studies of Im- agree with the birds ’ own perception of the melmann (1969) in the cross-rearing of songs they hear. Hence, to human observers estrildid finches. Zebra Finches (Taeniopygia the great diversity of song in Song Sparrows guttata), when reared by Bengalese Finches obscures differences between populations that (Lonchura striata), copied songs only from are apparently evident to the birds themselves the foster father even when they could see and as judged by their lessened response to foreign hear their true father. The only exception to songs (Harris and Lemon 1974). In spite of this was a case where a Zebra Finch began these criticisms, the concept has brought at- to assist the rearing of the young wherein they tention to the bases of variation of repertoires copied from the conspecific parent. and their behavioral and evolutionary sig- nificance (Thielcke 1970). A GENERAL INTERPRETATION Local differences of song may fail to ap- The song repertoires of birds may be con- pear in two cases: either the conformity is sidered a multi-dimensional form of behavior. widespread, or there is practically no local Some dimensions are the numbers of syllables conformity at all. White-throated Sparrows of the repertoire, and the sequential and tem- may show both features, tending more toward poral rules governing them. Another is the conformity in the repertoire of syllables and degree to which the repertoires conform their sequences in the center of their range in among birds of the same species, and yet an- Ontario, but toward individuality at the other is the geographical distribution of the periphery in New Brunswick; in both areas previous dimensions. Our main concern here they use essentially the same repertoire of is with the last two, especially with the be- syllables. Somewhat similarly, the Indigo havioral bases for the conformities and non- Bunting populations seem to have the same conformities among the songs of birds on repertoires in New York, Kentucky, and Michi- neighboring territories and in different lo- gan; however, the songs of individuals usually calities. differ except for some pairs and trios of bunt- Whether dialects occur depends in part on ings which are most often on immediately the degree of conformity in the songs of adjacent territories. In this case, the move- neighbors and also on the extent to which the ment of males to new territories in their local populations diverge. To some extent, the second breeding season may be the basis of concept of dialect is artificial, for many cita- the individuality established after copying has tions of dialects fail to consider the actual occurred in the first season. geographic distributions of the patterns; that In these two species, the limits to the varia- is, the differences between localities could be tions of the syllables may reflect certain limits disjunctive or clinal. A further complication on the particular frequency-time configura- is that derived from dimensions of the reper- tions. In White-throated Sparrows, the songs toires cited above. As a result, situations such are varied primarily by amplitude modulations as that described by King (1972) for Chin- only. The result is a repertoire of only five or golos may be so complex that the concept is six syllable classes in a population. Indigo difficult to apply. Instead, one sees clines, Buntings modulate their whistles in the fre- mosaics and other disjunctions, and even quency dimension also, so that the number of changes of repertoire with the season. The syllables is much larger. There is a caution dialects as defined by researchers may not about assuming limits to the amount of varia- 402 ROBERT E. LEMON tion, for the assessment of the number of that involves the replication by the motor sys- syllable classes is somewhat arbitrary as the tem of a model obtained normally through classifier recognizes distinct groups. If the audition. Auditory feedback plays a dominant evidence were considered more thoroughly, role in the development of the songs as shown variation might seem more continuous. In by experiments where lack of audition has Cardinals, classifying syllables within a local been forced upon a bird either through deaf- population is fairly easy, whereas the syllable ening or exposure to very loud noise (Marler classes often intergrade between localities et al. 1972). The importance of hearing is (Lemon 1966). In spite of this comment, syl- undoubted, but proprioceptive information lables in different geographic areas are often should also be available during development similar, apparently because of the interaction of the motor patterns, not only from the of imitation with dispersal. syrinx, but also from the rib cage muscula- The sharing of repertoires by neighbors on ture, and may be the main source of informa- adjacent territories is obvious in some cases tion for those birds which develop their songs and includes most or all of the repertoires; in after deafening. There is a suggestion, too, other cases it is less obvious. The extent to of proprioception from the results of sectioning which the conformity exists is seen as the out- of mixed nerves to the syrinx. Sectioning of come of two main developmental interactions, the left nerves in Chaffinches (Nottebohm imitation and improvisation (table 2). 1970, 1971) greatly distorts the sounds. In Imitation is the copying, usually by young, White-throated Sparrows (Lemon 1973) there of the songs of adults, often the copiers ’ own is evidence of not only qualitative distortion parents. The copying most often occurs in the but also of changes in the durations of the first few months of life and perhaps in most syllables, especially at the beginning of the species is confined to the period that ends song. with the motor development of the songs. It has been suggested that there is an innate Therefore, it coincides approximately with the template which is the basis of the songs of time when young males first acquire a terri- deafened birds (Marler 1964, 1972, Konishi tory. Imitation is also influenced by social 1965a). Konishi suggested that the song de- constraints which result more often in the ex- veloped by a deafened Song Sparrow was clusion of sounds from individuals and mem- evidence of such a template, but it could also bers of species of no social consequence to have developed through proprioception and the copying individual. There also appear to improvisation. Also one might ask what be- be sensory and motor constraints on the kinds comes of the innate template in birds which of material that may be copied by particular develop their songs normally. species. This might lead to the conclusion The concept of a template that is innate that there are “species-specific” differences or otherwise is also misleading because it does which determine the variation among species. not distinguish between sensory and motor No doubt this is true to some extent. For ex- information. During development, when a ample, the hearing of juncos and House Spar- bird reproduces the song it has heard from an- rows (Passer domesticus) differs considerably other individual, it is reproducing information (Konishi 1969, 1970). However, closely re- that is basically sensory in origin. If the model lated species, especially of the same genus, of development presented here is correct, how- are not likely to differ much in their sensory ever, replication may involve initially the ap- or motor systems, particularly when such sys- plication of this information to the calls. The tems must remain flexible enough to handle calls in some deafened birds appeared to de- variation resulting from experience. velop normally (Dittus and Lemon 1970) and Individuality in the repertoire occurs from on this basis they may have been innate, in several causes, chiefly generating new syllable the sense of the classical isolation experiment types through drift and reordering of syllable (Lorenz 1965). Other evidence suggests that sequences in the songs. The models used for experience is involved in development of calls the syllables may be those provided by other (Nottebohm 1972a, Mundinger 1970), and it individuals and copied from them, or they is possible that the calls which occur while the may arise through gradual development based birds are still in the egg have an influence on drift applied initially to calls. Other fac- which is essential to normal development of tors influence individuality, particularly the those calls which appear later. In a bird interaction of copying with dispersal, with the which is deafened, these calls could be the temporal limits on development, and with the basis of improvisation and might be described relative annual changeover of populations. as the innate template. This template, however, Development of song is a complex process is motor, not sensory, although it may be con- HOW BIRDS DEVELOP SONG DIALECTS 403 verted to sensory information via feedback. My point is that the apparent innateness in The template concept also fails to recognize the calls of fowl and doves may reflect a the importance of drift, reordering of sylla- much simpler situation than in oscines where bles, and repetition of syllables, all of which normally the song elements require a much are probably mainly motor in their control. higher level of neuromuscular control. Indeed, Much of the research on development dis- probably this is the reason why such innateness cussed here, including my own, was done is not readily apparent in higher oscines. TO under the influence of the historical context, make a crude analogy, the natural sound of a where one of the prime questions of ethologists violin string can be obtained when plucked by was whether behavior was innate or learned. either a musician or an untrained novice, but There has been much discussion of this prob- only the musician can make the strings vibrate lem, which need not be reviewed here. As in a harmonically controlled fashion over a far as the studies of development are con- length of time. cerned, however, the very nature of the ques- This discussion need not pose a threat to tion supposes answers in the same framework: the recognition that some aspects of vocal namely, that song is either innate or learned. behavior may be common to oscines and Thus observations made were applied to this others, such as drift in calls of chicks, al- framework, whether the application was valid though the underlying bases of drift may be or not. somewhat different. The occurrence of drift There have been claims that song or parts in fowl should also raise the question of how of song are innate (Lanyon 1960). It is not it is related to the calls of adults. clear in these examples though, how the in- Of related practical problems, one is that nateness is derived. Is it through the coding the standards for comparing the songs of ex- of information in the central nervous system perimental birds have not been chosen con- which programs muscle contradictions in the sistently. The best standards would surely be syrinx and associated respiratory structures, or the songs of the experimental birds ’ own is it related also to the structures of the syrinx? parents, and barring those, of other members For example, the crows of domestic fowl of the parental populations when conformity (Gallus gallus var. domesticus) (Konishi in such populations is obvious. In many cases 1963) and the calls of Ring Doves (StwptopeZia members of the population are not a valid risoria) (Nottebohm and Nottebohm 1971) standard because of the large degree of in- appear normal even in birds deafened at an dividuality in the songs of some species, so early age. These particular calls, however, the idea of an average song of a species may should not be equated to the song elements hardly apply. of oscines, for the former often have strong Also, experimental birds often vary greatly harmonics and show little or no rapid and in their responses under controlled conditions highly precise frequency modulation so char- and the unnaturalness of the tutoring from a acteristic of song elements in the latter. There tape recorder does not help. are exceptions to this. Certain alarm shrieks of Much of the confusion in the literature lies oscines when held in the hand have strong har- in the nature of song itself and in the fact monics, (e.g. Cardinals) (Dittus and Lemon that repertoires may be so complex that simple 1970), as do the sounds produced by oscines understanding is difficult, as in the Song after the hypoglossus has been sectioned, Sparrow. Also, even closely related species shown especially clearly in White-throated differ greatly in the size of the repertoire and Sparrows (Lemon 1973). These last two ex- the degree of conformity. There seems little amples show the lack of precise control of the doubt that the size of the repertoire is deter- syrinx so characteristic of song syllables. Other mined to a great extent by inherited influences, examples are the short contact calls or chips although proof is limited to obvious differences of Cardinals ( Dittus and Lemon 1970). Al- between species. However, as noticed in Car- though more sophisticated analysis may yet dinals, a strong tendency to conformity has a reveal some differences, these calls seem nor- restrictive influence on the inclusion of cer- mal in deafened birds. But this normality may tain patterns from unimportant individuals or simply reflect the relative simplicity with of improvised models. On the other hand, in which the sounds can be produced, that is, a experimental situations where the material quick passage of air over tightened mem- available to copy is relatively low, a bird may branes, followed immediately by damping, and include material that would otherwise be ex- the frequency structure is probably related to cluded. Further, where social stimulation is the dimensions of the membranes as well as missing, the repertoire may never develop to to the tension applied to them. its normal extent. 404 ROBERT E. LEMON

SUMMARY LITERATURE CITED 1. Individual songbirds frequently share, in ANDREW, R. J. 1969. The effects of testosterone varying degrees, their repertoires with others on avian vocalizations. In R. A. Hinde [ea.] on neighboring territories. Hence birds within Bird vocalizations, pp. 97-130. Cambridge Uni- versity Press. a locality show some degree of conformity in BAPTISTA, L. F. 1972. Wild House Finch sings their songs. At the same time, repertoires White-crowned Sparrow song. Z. Tierpsychol. differ between localities, giving expression to 30:266-270. the term “dialect.” This paper considers how BERTRAM, B. 1970. The vocal behaviour of the Indian Hill Mynah, GrucuZu religiosa. Anim. dialects form, illustrating first with the Car- Behav. Monogr. 3:79-192. dinal, and then considering other species, BLANCHARD, B. D. 1941. The White-crowned Spar- especially fringillids. rows (Zonotmchia’ Zeucophrys) of the Pacific seaboard: environment and annual cycle. Univ. 2. Conformity is seen principally as the result Calif. Publ. Zool. 46:1-178. of copying by young, a process which is limited BORROR, D. J. 1959a. Songs of the Chipping Spar- in most species to a period from the time row. Ohio T. Sci. 59:347-356. when the young leave the nest until the motor BORROR, D. J. 1959b. Variation in the songs of the Rufous-sided Towhee. Wilson Bull. 71:54-72. patterns develop the following spring. Reared BORROR, D. J. 1965. Song variation in Maine Song without experience of song, birds develop Sparrows. Auk 77:5-37. some features such as general structures of BORROR, D. J. 1968. Unusual songs of passerine syllables and repetitions of syllables and songs. birds. Ohio J. Sci. 68:129-138. BORROR. D. T. AND C. R. REESE. 1956. Mockinn- These facts suggest inherited motor and I ” bird imitations of Carolina Wren. Bull. Mass sensory influences. Audubon Sot. 40:245-256; 309-318. BORROR, D. J. AND W. W. H. GUNN. 1965. Varia- 3. Individuality of songs is determined by a tion in White-throated Sparrow songs. Auk 82: variety of influences, the major ones being: 26-47. the interactions of the time-limited copying DITTUS, W. P. J. AND R. E. LEMON. 1969. Effects process with dispersal of the young arising of song tutoring and acoustic isolation on the from the area of rearing to their new place, songs of Cardinals. Anim. Behav. 17:523-533. DITTUS, W. P. J. AND R. E. LEMON. 1970. Auditory the active modification of syllables through feedback in the singing of Cardinals. Ibis 112: drift, and the reordering of syllables within 544-548 songs. These individual characters are in turn Dow, D. D. AND D. M. SCOTT. 1971. Dispersal and copied by young males. In this way local range expansion by the Cardinal: An analysis peculiarities are built up within a population. of banding records Can. J. Zool. 49:185-198. EMLEN, S. T. 1971a. Geographic variation in 4. The integrity of a local dialect is main- Indigo Bunting song (Passer&u cyuneu). Anim. tained in Cardinals by a high return of adults Behav. 19:407408. EMLEN, S. T. 1971b. The role of song in individual to their territories from year to year and low recognition in the Indigo Bunting. Z. Tierpsychol. annual mortality. Also, young often disperse 28:241-246. to relatively short distances from home nests. FALLS, J. B. 1969. Functions of territorial song in However, many individuals disperse over con- the White-throated Sparrow. In R. A. Hinde [ed.] siderable distances and probably introduce Bird vocalizations, pp. 207-232. Cambridge Uni- versity Press. novel patterns to certain localities. Such GRIMES, L. G. 1974. Dialects and geographical dispersal also probably has the countering variation in the song of the Splendid Sunbird. effect of making some patterns widespread Ibis 116:314-329. over fairly large areas as in Ontario. HALL-CRACGS, J. 1962. The development of song in the Blackbird, Tz~rdus me&a. Ibis 104:277- 5. In other species, conformity varies from 300. high to low, although in all groups there is HARRIS, M. AND R. E. LEMON. 1972. Songs of Song evidence of copying and improvisation. The Sparrows ( Melospizu melodiu) : individual varia- tion and dialects. Can. J. Zool. 50:301-309. differences depend to a great extent on which HARRIS, M. AND R. E. LEMON. 1974. Songs of Song tendencies dominate. Sparrows: responses of males to songs of different localities. Condor 76:3344. 6. A model of development based on calls, HARRIS, M. AND R. E. LEMON. In prep. Songs of copying, improvisation and reordering of syl- Song Sparrows: responses of males to songs of lables is presented. This model is seen to neighbors and non-neighbors. apply to oscines in general. I?,4MELMANN, K. 1969. Song development in the Zebra Finch and other estrildid finches. In R. A. 7. Problems associated with the development Hinde [ed.] Bird vocalizations, pp. 61-74. Cam- of song are discussed, particularly those based bridge University Press. on the concept of innateness. ISAAC, D. AND P. MARLER. 1963. Ordering of sequences of singing behaviour of Mistle Thrushes I thank my colleagues in this research, D. M. in relationship to timing. Anim. Behav. 11:179- Scott, W. Dittus and M. Harris. 188. HOW BIRDS DEVELOP SONG DIALECTS 405

JOHNSTON, R. F. 1956. Population structure in salt LEMON, R. E. AND D. M. SCOTT. 1966. On the de- marsh Song Sparrows. Condor 58~2444; 254- velopment of song in young Cardinals. Can. J. 272. Zool. 44: 191-199. KING, J. R. 1972. Variation in the song of the LORENZ, K. 1965. Evolution and modification of Rufous-collared Sparrow Zonotrichia caper&s, in behavior. Univ. Chicago Press. Northwestern Argentina. Z. Tierpsychol. 30:344- MARLER, P. 1952. Variation in the song of the 373. Chaffinch. Ibis 94:458-472. KOXISHI, M. 1963. The role of auditory feedback MARLER, P. 1956. Behavior of the Chaffinch in the vocal behavior of the Domestic Fowl. Z. Fringi2Za coelebs. Behaviour Supp. 5 Brill, Leiden. Tierpsychol. 20: 349-369. MARLER, P. 1964. Inheritance and learning in the KOXISHI, M. 1964a. Song variation in a population development of animal communication signals. In of Oregon Juncos. Condor 66:423436. R. G. Busnel Led.1 Acoustic behaviour of animals. KONISHI, M. 1964b. Effects of deafening on song Elsevier, Amsterdam. development in two species of juncos. Condor MARLER, P. 1967. Comnarative studv of song de- 66:85-102. velopment in sparrows. Proc. XIV Inte&atl. KONISHI, M. 1965a. The role of auditory feedback Ornithol. Congr., 231-234. in the control of vocalizations in the White- MARLER, P. 1972. A comparative approach to vocal crowned Sparrow. Z. Tierpsychol. 22:770-783. learning: song development in White-crowned KONISHI, M. 1965b. Effects of deafening on song sparrows. J. Comp. Physiol. Psychol. 71: (2) l-25. development in American Robins and Black: MARLER, P. AND lli.‘ J. HAMILTON III. 1966. Mech- headed Grosbeaks. Z. Tierusvchol. 22:584-599. anisms of animal behavior. Wiley, New York. KONISHI, M 1969. Hearing, iingle unit analysis, MARLER, P. AND D. ISAAC. 1960. Song variation and vocalizations in songbirds. Science 166: 1178- in a population of Brown Towhees. Condor 62: 1181. 272-283. KONISHI, M. 1970. Comparative nemophysiological MARLER, P. AND D. ISAAC. 1961. Song variation in studies of hearing and vocalizations in song- a population of Mexican Juncos. Wilson Bull. birds. Z. vgl. Physiol. 66~257-272. 73: 193-206. KONISHI, M. ANU F. NOTTEBOHM. 1969. Experi- MARLER, P., M. KREITH, AND M. TAMURA. 1962. mental studies in the ontogeny of avian vocaliza- Song development in hand-raised Oregon Juncos. tions. In R. A. Hinde [ed.] Bird vocalizations, Auk 79: 12-30. pp. 2948. Cambridge University Press. MARLER, P. AND M. TAMURA. 1962. Song “dialects” KROODSMA, D. 1971. Song variations and singing in three populations of White-crowned Sparrows. behavior in the Rufous-sided Towhee, Pfpilo Condor 64:368-377. errrthrovhthalmus oreaanus. Condor 73:303-308. MAHLER, P. AND M. TAMURA. 1964. Culturally KROOD&IA, D. 1973. Co-existence of Bewicks’ transmitted patterns of vocal behavior in Spar- Wrens and House Wrens. Auk 90:341-352. rows. Science 146: 1483-1486. LAN~ON, W. E. 1957. The comparative biology MARI.ER, P., P. MUNDIN~ER, M. WASER AND A. of the Meadowlarks (Sturnella) in Wisconsin. LUTJEN. 1972. Effects of acoustical stimula- Publ. Nuttall Ornith. Club 1. Cambridge, Mass. tion and deprivation on song development in LANYON, W. E. 1960. The ontogeny of vocaliza- Red-winged blackbirds ( Agelaius vhoeniceus tions in birds. In W. E. Lanyon and W. N. ) . Anim. Bihav. 20:586-666.- ’ Tavolga [eds.] Sounds and communication. Am. Inst. Biol. Sci., Washington. MARLER, P. AND P. MUNDINGER. 1971. Vocal learn- LASKEY. A. 1944. A study of the Cardinal in ing in birds. In H. Moltz ted.] Ontogeny of Tennessee. Wilson Bull.* 56:2744. vertebrate behavior. Acad. Press, N.Y. LEMON, R. E. 1965. The song repertoires of Car- MILLIGAN, M. M. AND J. VERNER. 1971. Inter- dinals (Richmondena cardinalis) at London, populational song dialect discrimination in the Ontario. Can. J. Zool. 43:550-569. White-crowned Svarrow. Condor 73:208-213. LE~XON, R. E. 1966. Geographic variation in the MISKIMEN, M. 195i. Sound production in pas- song of Cardinals Can. J. Zool. 44:413-428. serine birds. Auk 68:493-504. LESION, R. E. 1967. The response of Cardinals to MORTON, M. L., J. L. HORSIXANN, AND J. M. OSBORN. songs of different dialects. Anim. Behav. 15: 1972. Reproductive cycle and nesting success 538545. of the mountain White-crowned Sparrow (Zono- LEMON, R. E. 1968a. The displays and call notes trichia leucophrys oriantha) in ’ the Central of Cardinals. Can. J. Zool. 46:141-151. Sierra Nevada. Condor 74: 152-163. LEMON, R. E. 1968b. The relation between the MULLIGAN, J. A. 1966. Singing behavior and its organization and function of song in Cardinals. development in the Song Sparrow Melospiza Behaviour 32: 158-178. melodia. Univ. Calif. Publ. Zool. 81:1-76. LE~~oN, R. E. 1973. Nervous control of the syrinx MUNDINGER, P. C. 1970. Vocal imitation and the in the White-throated Sparrow (Zonotrichia recognition of finch calls. Science 168:480-482. albicollis). J. Zool. Lond. 171: 131-140. NICE, M. M. 1937. Studies in the life history LEMON, R. E. AND C. CHATFIELD. 1971. Organiza- of the Song Sparrow. Trans. Linn. Sot. N.Y. IV. tion of song in Cardinals. Anim. Behav. 19:1-17. NICOLAI, J. 1964. Der Brutparasitismus der Vidu- LEAION, R. E. AND C. CHATFIELD. 1973. Organiza- inae als ethologisches Problem. Z. Tierpsychol. tion of song of Rose-breasted Grosbeaks. Anim. 21: 129-204. Behav. 21328-44. NICOLAI, J. 1973. Das Lernprogramm in der LEMON, R. E. AND M. HARRIS. 1974. The question Gesangs aushildung der Strohwitwe Tetmeruru of song dialects in the songs of White-throated fischeri Reichenow. Z. Tierpsychol. 32: 113-138. Sparrows. Can. J. Zool. 52:83-98. NOTTEBOHM, F. 1968. Auditory experience and LEMON, R. E. AND A. HERZOG. 1969. The vocal song development in the Chaffinch (FTingilla behavior of Cardinals and Pyrrhuloxias in Texas. coelebs): ontogeny of a complex motor pat,- Condor 7 I: 1-15. tern. Ibis 110:549-568. 406 ROBERT E. LEMON

NOTTEBOHM, F. 1969a. The “critical period” for THIELCKE, G. 1972. Waldbaun&ufer (Certhia fami- song learning. Ibis 111:386-387. liaris) ahmen artfreundes signal nach und re- NOTTEBOHM, F. 196913. The song of the Chingolo agieren darauf. J. Omithol. 113:287-296. Zonottichia cape&s in Argentina: description THIELCKE, G. AND K. E. LINSENMAIR. 1963. Zur and evaluation of a system of dialects. Condor geographisches Variation des Gesanges des Zilp- 71:299-315. zalps, Phylloscopus collybita, in Mittel-und Sud- NOTTEBOHM, F. 1970. Ontogeny of bird song. westeuropa mit einem Vergleich des Gesanges Science 167:950-956. des Fitis, Phylloscopus trochilus. J. Ornithol. NOTTEBOHM, F. 1971. Neural lateralization of vocal 104:372-402. control in a passerine bird. I. Song. J. Exp. THIELCKE-POLTZ, H. AND G. THIELCKE. 1960. Zool. 177: 229-262. Akustisches Lernen verschieden alten schollieo- NOTTEBOHM, F. 1972a. Neural lateralization of lienter Amseln ( Turdus merula L. ) und die Ent- vocal control in a passerine bird. II. Subsong, wicklung erlernter Motive ohne und mit kunst- calls, and a theory of !earning. J. Exp. Zool. lichem Einfluss von Testosteron. Z. Tierpsychol. 179 :35-50. 17:211-244. NO~EBOH;M, F. 1972b. The origins of vocal learn- THOMPSON, W. L. 1968. The songs of five species ing. Am. Nat. 106:116-140. of Passerina. Behaviour 31:261-287. NOTTEBOH~~, F. AND M. E. NOTTEBOHM. 1971. Vo- THO~~PSON, W. L. 1970. Song variation in a popu- calizations and breeding behaviour of surgically lation of Indigo Buntings. Auk 87:58-71. deafened ring doves ( Steptopelia risoria). Anim. Behav. 19:313-327. THORPE, W. H. 1958a. The learning of song pat- PAY-, R. B. 1973. Behavior, mimetic songs and terns by birds with especial reference to the song song dialects, and relationships of the parasitic of the Chaffinch, Fringilla coelebs. Ibis 100: indigobirds (Vi&a) of Africa. Ornitl~ol. 535-570. Monogr. 11. THORPE, W. H. 195813. Further studies on the pro- RICE J. AND W. L. THOMPSON. 1968. Song devel- cess of song learning in the Chaffinch (Fringilla opment in the Indigo Bunting. Anim. Behav. coelebs gengleri) . Nature (Lond. ). 182:554- 16:462-469. 557. SCOTT, W. E. D. 1902. Data on song in birds: the THORPE, W. H. 1961. Bird-song-the biology of acquisition of new songs. Science 15:178-181. vocal communication and expression in birds. SHIOVITZ, K. A. AND W. L. TIIOMPSON. 1970. Geo- Cambridge Universitv Press. graphic variation in the song composition of the THORPE, W. -H. 1972. ’ Duetting and antiphonal Indigo Bunting, Passerina cyanea. Anim. Behav. song in birds. Behaviour SUDDI.__ 18. 18: 151-158. TODT, P. 1970. Gesang und gesangliche korre- THIELCKE, G. 1964. Zur Phylogenese einiger Lau- spondez der Amsel. Naturwiss. 57:61-66. tausserungen der europaischen Baumlaufer ( Cer- TOMPA, F. S. 1963. Behavioral response of Song thia brachydactyla Brehm und Certhia familiaris Sparrows to different environment conditions. L.). Z. Zoo]. Syst. Evol. 2:383-413. Proc. XIII Internatl. Ornithol. Congr., 729-739. THIELCKE, G 1969. Die Reaktion von Tamien und WICKLER, W. 1972. Aufbau and Paarspezifitat des Kohlmeise (Parus ater, P. major) auf den Gesang Gesangsduettes von Laniarius funebris ( Aves, nahverwandter Formen. J. Omith. 110: 148-157. Passeriformes, Laniidae). Z. Tierpsychol. 30: THIELCKE, G. 1970. Lernen von Gesang als mog- 464476. lither Schrittmacher der Evolution. Z. Zool. Svst. Evol. 8:80%820. Accepted for publication 3 July 1974.