Anlrn. Behav., 1981, 29, 918-923 SONG MATCHING IN THE GREAT TIT PARUS MAJOR L. BY JOHN R. KREBS, RUTH ASHCROFT & KARL VAN ORSDOL* Edward Grey Institute, Zoology Department, South Parks Road, Oxford OX1 3PS *Department of Applied Biology, Pembroke Street, Cambridge CB2 3D X Abstract. Great tits respond to playback of song from their own repertoire by singing the same song type as that on the tape. This is referred to as song matching. There is a good correlation between the probability of matching and response strength as indicated by latency, approach to the speaker and total number of songs. We propose that matching functions as a graded signal in territorial disputes. Song matching occurs when a bird responds to playback or a rival by singing phrases from its repertoire which resemble the stimulus song. Matching in playback experiments or song duels between rivals has been observed in many species (e.g. Hinde 1958; Lemon 1968a, 1968b, 1974; Bertram 1970; Todt 1970, 1971; Kroodsma 1971; Br6mond cited in Armstrong 1973) but not in all (Fails & Krebs 1975; Martin 1980), and there have been several suggestions about the significance of matching in communi- cation, but little circumstantial or direct evidence (Falls & Krebs 1975). One likely function of matching, suggested by Br6mond (cited in Armstrong 1973), is that it enables a territorial bird to direct its 'keep out' signal at a particular rival or intruder. In this paper we present circum- stantial evidence for a related idea, that matching acts as a graded signal. Our hypothesis is that a territory holder signals the likelihood of attack- ing an intruder by the probability of matching the intruders' song. This idea, as with Br6mond's, is appropriate for species in which song is used as a method of territorial defence. We studied matching of playback songs in the great tit, a species in which the territorial function of song has been established (Krebs 1977a; Krebs et al. 1978). Methods General Methods The playbacks were done with a Uher 4000IC or Nagra IIIB tape recorder linked to a Nagra DH speaker-amplifier by 20 m of cable. We plotted the territories of experimental birds be- fore starting the tests, and recorded their song repertoires using a Beyer dynamic microphone in a 24-inch (60-cm) Grampian parabolic re- flector. The songs used for playback were trans- scribed through a high-pass filter (2 kHz) onto 10-s continuous loops (Cousino Audiovendor). The 10-s loop contained one or two song bursts. (The great tit song usually consists of a 2 or 3 note phrase uttered in short bursts of about 6-10 repetitions, lasting 2-3 s.) The playback tests consisted of 2 min of play- back (12 cycles of the loop) followed by 8 min of silence. During the 10 min of a test, we re- corded the following aspects of the bird's re- sponse: (a) song type(s) used, (b) amount of song (recorded as number of song bursts); (c) closest approach to speaker; and (d) latency to first response (song, call, or approach within 20 m). Birds were almost always silent and some distance from the speaker at the start of a test, so that the first response was easily recognizable. All observations were made standing 20 m from the speaker, which was hung in a tree at a height of 2 m near the boundary of the territory and pointing towards the territory centre. Experimental Procedure We tested whether or not great tits match in response to playback by recording the reper- toires of test birds and playing to each individual different songs out of its own repertoire. If matching occurs, a bird will be more likely than expected by chance to sing the same song type as that on the tape in its first response. We tested each of seven individuals 60 times. The 60 trials each lasted 10 rain (see above) and were done in two groups of 30 on two mornings separated by 8-12 days. On each morning the 30 trials were run consecutively over a 5-h period starting at dawn. Six of the birds were tested with three songs chosen at random from their reper- toires. Each song was played 10 times on both mornings, and the order of presentation was determined by a pseudo-random sequence. The seventh bird had only two songs in its repertoire (Table I) and was tested with both of these songs, each being played 15 times on both mornings. The experiment was done in March and April 1974 (before the start of nest building) at Plas Newydd Estate, Anglesey, and Penrhyn Estate, Gwynedd, both in North Wales. 918