768 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol 123. No. 4. December 2011 
Tail piuiips/sec 
FIG. 1. The probability that a Black Phoebe is successful at foraging decreased 
tail prior to a foraging bout (n = 14 observations of eight individuals). 
as a function of how much it pumps its 
defined as an individual moving within a 2-m 
radius of the speakers. 
Sire Impact .—Data from the pre-playback and 
control (e.g., House Finch vocalization) trials 
were analyzed with repeated measures ANOVA. 
Individuals, sites, and individuals within sites 
were included in the model as random factors. 
Statistical Analyses .—All hypotheses consisted 
of observations of 10 individual birds. Data were 
analyzed by a GLM with substrate type as the 
fixed factor and individual as the random effect 
for the balance hypothesis. A /-test was used to 
examine whether foraging birds used higher tail¬ 
pumping rates than non-foraging individuals, and 
logistic regression was used to test whether 
successful foraging could be predicted by tail¬ 
pumping rate. Tail pumping rates were analyzed 
with a GLM tor the intruder hypothesis with 
vocalization type as the fixed factor and individ¬ 
ual as the random effect. Interactions of vocali¬ 
zation type and individual not statistically signif¬ 
icant iP > 0.25). were dropped from the analyst} 
and only the main effect of vocalization type wa> 
analyzed by ANOVA with a Tukey post-hoc test 
Approach proportions were analyzed by a good¬ 
ness of fit (G-) test. Data from the signal to 
predators hypothesis were analyzed by a GLM 
tTdn rvn PU T ir ? ra,CS as thc response, vocaliza- 
type ttS the f,xed factor, and individual as the 
random factor. Interactions between the fixed and 
random factor, if not statistically significant (P > 
0.25), were dropped from the analysis and the 
main elfect of vocalization type was analyzed by 
ANOVA with a Tukey post-hoc test. A G 2 test 
was used for approach and call proportions, and 
Pearson’s r was used to examine correlations 
between the number of approaches and calls. Data 
were transformed as appropriate to meet statistical 
assumptions and analyses were performed with 
Systat 11 (Systat 2004). Tail pumping rates are 
reported as means ± SE. 
RESULTS 
Balance Hypothesis .—There were no signifi¬ 
cant differences in tail pumping rates, controlling 
for the effects of individuals, between birds 
perched on ‘stable' substrates and those perched 
on ‘unstable’ substrates (F l 60 = 0.023. P = 0.88). 
Foraging Enhancement Hypothesis.— Tail¬ 
pumping rales were similar between foraging 
and non-foraging individuals (r 45 = -0.85. P ~ 
0.40). However, the probability of success of 
those eight individuals that did forage could be 
predicted by tail pumping rate (logistic regression 
deviance G 2 = 4.44. df = 1. P = 0.035; Fig. If- 
Increased foraging success was shown by those 
individuals that tail pumped at a lower rate. 
