A number of theoretical studies have predicted that avian predation risk is mass dependent. Models of bird flight predict that increased mass will reduce flight velocity, making birds more vulnerable to predation. Empirical studies supporting this have demonstrated a significant relationship between mass and routine flight velocity (when birds are not alarmed). Studies investigating the effects of mass upon flight velocity when birds are alarmed, however, tend to show that the mass/velocity relationship is less marked. We found a real difference in the relationship between mass and flight velocity in zebra finches, Taeniopyga guttata, when comparing flights of alarmed and unalarmed birds. Despite the strong mass/routine velocity relationship already shown for zebra finches, mass tended to be a poor predictor of the flight velocity of alarmed birds within the natural weight range of the species. This difference appeared to be partly due to alarmed birds increasing their velocity more when heavy than when light, in comparison to their respective predicted routine flight velocities for their weights. As a result, mass/velocity regression slopes tended to be more shallow for alarmed birds. Consequently, increases in body mass within the natural weight range of a bird, may have less effect on the flight velocities of alarmed birds than they do on routine flight velocities. We therefore recommend caution in the use of weight as a predictor of predation risk in birds without examining its effects upon the flight velocity of alarmed birds. We suggest some explanations for the differences in the mass/velocity relationship between the flights of alarmed and unalarmed birds. (C) 1998 The Association for the Study of Animal Behaviour
