Frequency of bird hybrids: does detectability make all the difference?

Abstract

The number of hybrids recorded in the field may be biased because these data may simply reflect the detectability of a hybrid pairing. A questionnaire was constructed for assessing the detectability of natural avian hybrids. Ratings from 17 independent experienced observers were subjected to factor analysis and all observers loaded onto the same single factor. The reliability was extremely high (Cronbach’s alpha=0.96). Initially, I found a weak positive and significant correlation between the number of hybrids and their detectability index (r _s=0.268; P=0.015; n=81). However, the respective correlations within three orders and one family containing enough hybrids for analysis revealed non-significant results. Significant differences existed among orders with respect to both detectability and hybrid frequency. At the family level, differences existed in detectability, but not in the number of hybrids recorded. Furthermore, I found no publication bias. These results suggest a general influence of detectability and on higher taxonomic levels, and that either detectability or the number of hybrids vary between orders possibly due to phylogenetic effects. Data about hybrids should be used in a cautious manner in analyses across a wide range of species, but seem unbiased when restricted to analyses within orders and families. As a rule of thumb, distinctness of hybrids seems to be a function of the plumage difference between the hybridising species.