Anim. Behav., 1995,49,244247 Individual and sex differences in the provisioning calls of European bee-eaters C. M. LESSELLS*, C. L. ROWE? & P. K. McGREGORt *Netherlands Institute of Ecology, Boterhoeksestraat 22, P. 0. Box 40, 6666 ZG Heteren, The Netherlands TDepartment of Life Science, University of Nottingham, Nottingham NG7 2RD, UK (Received 8 April 1994; initial acceptance 10 June 1994; final acceptance 14 July 1994; MS. number: ~~-1005) In social species, selection will often favour the ability to discriminate individuals or different classes of individuals. In colonially breeding European bee-eaters, Merops apiaster, helpers at the nest preferentially aid close relatives (Lessells 1990), and cross-fostering experiments show that chicks subsequently help foster rather than genetic parents (M. I. Avery, personal communication). Playback experiments show that parents can rec- ognize their own chicks acoustically a few days before they fledge (Lessells et al. 1991). Casual observations suggest that chicks also might learn to identify their parents individually before they leave the nest: fledged chicks do not beg indis- criminately from adult bee-eaters, and sometimes follow their parents back to the nest (C. M. Lessells, personal observation). If chicks can rec- ognize individual parents they are likely to do so largely or exclusively by acoustic means, because calls offer the only feasible means of communi- cation between birds inside and outside their metre-long burrows. An ability to recognize pro- visioning adults could be important to chicks in making future decisions as to who to help. The sex of the adult might be as important as its identity because helpers are less related on average to the chicks than are either of the breeding pair. A chick that is fed by more than two adults could there- fore choose the adult of the minority sex as a recipient of future help as a way of maximizing its relatedness to the chicks that it helps. As a first step in evaluating the possibility of recognition of adults by chicks, we investigated the calls made by parent bee-eaterswhen visiting the nest to feed the chicks (‘provisioning calls’). Our aim was to deter- mine what information such calls contained about the identity or sexof the calling bird, by analysing spectrograms of the calls of known individuals. Parent birds make provisioning calls at nearly every visit to the nest with food, either just before they land or when they are perched at the nest burrow entrance. Often they call only once, but they may make a repeated series of calls before entering the burrow. We recorded 193 calls made by 28 individual birds on 112 visits on 38 bird- dates using an Electret EM1 16 microphone, sited by or just inside the nest entrance, and a Sony WM-D6C cassette recorder. Recordings were made between 11 July and 5 August 1992 at the bee-eater colony at Mas des Sarcelles in the Camargue, southern France (Lessells 1990). Methods of marking birds for field identification and determining their sex are described by Lessells (1990). Chicks were between 4 and 23 days old when their parents were recorded and recordings at the same nest on different dates were made between 5 and 9 days apart. Recordings were filtered (Kemo VBF/3 600 Hz high pass, series enabled), digitized with an 8 bit, 22 kHz A/D converter and displayed with Canary 1.1 software as waveforms and spectrograms (bandwidth 706 Hz, grid resolution 144ms by 22 Hz, 75% overlap, smooth setting). We quantified the similarity between calls in two ways. First, we used 11 time or frequency variables (see Fig. 1 for details). Second, we used the cross-correlation routine of the Canary pack- age to generate cross-correlation coefficients (rC) representing the similarity between two spectro- grams (discussed more fully in Charif et al. 1993, pp. 51-64; Fig. 2 gives illustrative r, values). We did not carry out multivariate analyseson the time and frequency variables because the Canary cross- correlation coefficients provide a measure of the overall similarity of calls. We analysed the univariate measures using nested ANOVA. Because of the extremely un- balanced design (many birds were recorded on only one date), we carried out two separate analy- ses. In the first, we included all calls, and nested individual within sex. This provided maximum degrees of freedom for testing for a sex effect, but