Table 1.

Summary of the results of the 16 first-order autoregressive models used to determine whether each of the eight summary wing kinematic parameters for the downstroke and the upstroke are associated with the state of the alula on each wingbeat (both flipped, both flat, one flipped and one flat). For each wing kinematic parameter, the mean value for both wings was used as the response variable in a first-order autoregressive model controlling for video sequence number. The F-statistics and associated p-values are calculated using sums of squares adjusted for the other terms in the model (i.e. type III sums of squares). The p-values give the probability of observing an F-statistic as large or larger than that which was observed, under the null hypothesis of no association [25]. Because we have quoted 16 separate p-values, using the frequentist approach of treating those that are less than 0.05 as significant would result in an inflated overall risk of type I error. We therefore used a false discovery rate method to control the expected proportion of false positives at the 5% level [27]. Of the 16 parameters, 14 were significantly associated with alula state, after controlling the overall false discovery rate (see text). Significant associations are shown in bold.

wing kinematic parameters	downstroke		upstroke
	p	F2,25 358	p	F2,25 358
stroke amplitude	<0.001	766	<0.001	530
stroke period	<0.001	52.5	<0.001	31.0
stroke plane	<0.001	248	<0.001	244
mid-stroke angle of incidence	<0.001	365	0.667	0.40
timing of rotation at end of stroke	<0.001	318	0.050	3.01
mid-stroke stroke angle	<0.001	519	<0.001	544
mid-stroke deviation angle	<0.001	426	<0.001	227
mid-stroke twist distribution	<0.001	16.5	<0.001	4.62