A model for avian primary moult

The regression methods frequently used to estimate the parameters associated with primary moult in birds are unsatisfactory. Results obtained using least squares regression, and various ad hoc adaptations, are so obviously incorrect that many authors have fitted lines 'by eye' (Newton 1968, Thomas & Dartnall 1971, Elliott et al. 1976, Morrison 1976, Appleton & Minton 1978). In a comparison of seven regression methods, estimates of the average starting date varied between 29 June and 31 July, completion date between 2 and 24 October, and duration of moult between 72 and 109 days for the Redshank Tringo totonus, in spite of the very large sample of 1482 observations (Summers et al. 1983). In this paper we present a new approach to the analysis of primary moult and develop a mathematical model specifically designed for moult data.