Bill colour outperforms iris colour for sexing in Starlings Sturnus vulgaris in breeding condition

Reliable morphological criteria for sexing birds without the need for invasive procedures are scientifically and ethically valuable. Smith et al (2005) reported that iris colour assigns the genetic sex of Starlings Sturnus vulgaris with 98% accuracy, and bill base colour with 100% accuracy when birds are in breeding condition, but these figures have not been tested independently. We applied their criteria to 31 captive Starlings in breeding condition, using three observers blind to genetic sex, two of them new to Starling sexing. Bill base colour correctly predicted genetic sex for all 29 birds scored, with complete observer agreement. Mean iris score was correct for 27 of 31 (87.1%), fewer than the 94 of 96 reported by Smith et al (2005) for birds in breeding condition (Fisher exact test, P = 0.031), although with so few errors in either study this comparison is imprecise. Observer agreement was high, so these errors reflect iris colouration conflicting with genetic sex rather than inconsistent scoring; bill base colour was correct for all four misassigned birds. Body mass and tarsus length classified 77% and 58% of birds correctly, their published thresholds missing in opposite directions. We conclude that bill base colour should be the primary criterion for sexing Starlings in breeding condition; the accuracy of iris colour should not be assumed to be 98%.