Seeing the woodhoopoe for the trees: a response to Simmons et al. (2005)

Abstract

We conducted a preliminary mitochondrial DNA sequence study of woodhoopoes with the aim of acquiring diagnostic molecular markers to study interbreeding between two poorly differentiated forms (Cooper et al. 2001). This study revealed a surprisingly close relationship between 12 violetmantled birds from two localities in northwestern Namibia and green-mantled birds several thousand kilometres away in South Africa and Kenya; we also found negligible differentiation between these violet-mantled birds and a single green-mantled bird from a mixed flock in Namibia. These results led us to question the phenotypic basis on which these species are recognized, which we found to be inadequate for species diagnosis and inconsistent in application. We are not the first to do so: the Violet Woodhoopoe Phoeniculus damarensis was first described by Ogilvie-Grant (1901), but was subsequently synonomized with the Green Woodhoopoe P. purpureus by Peters (1945). Fry (1978) treats them as separate taxa, but Clancey (1985, p. 245) states that ‘the current policy of according the Violet Woodhoopoe specific status needs in-depth investigation’, a call reiterated by du Plessis (1997a). Observations of phenotypic variation are easily canalized by systematic interpretation; what one sees depends on what one expects to see. The central issue in Namibian woodhoopoes concerns the significance of the sole diagnostic trait ‐ adult mantle coloration (Clancey 1985). Simmons et al. (2005) apparently believe that variation in mantle coloration reflects a discrete difference between independently evolved species lineages, with distinct habitat preferences, distribution and behavioural ecology. They suggest that individuals with intermediate mantle coloration arise through hybridization, and that this process represents a threat to the locally endemic violet-mantled form. We propose an alternative interpretation in which the violet-mantled and intermediate forms result from clinal variation across an aridity gradient, with associated changes in habitat use and group size probably due to the varying availability of tree holes for this communally roosting species. We contend that the ecological gradient hypothesis is more parsimonious, and more consistent with the mitochondrial DNA sequence similarity and the extensive distributional overlap between these intergrading forms. We also contend that in the absence of a consistent phenotypic diagnosis these forms should be considered conspecific, pending further systematic research on geographical variation in green, violet and mixed groups of woodhoopoes in Namibia.