Table 1.
Factors | Total amount of food consumed within 5 min | Fish activity (average distance moved within 2s) | Average position where the first 2 food items before and after the simulated heron attack were consumed | ||||||
---|---|---|---|---|---|---|---|---|---|
df | χ2 | P | df | χ2 | P | df | χ2 | P | |
+Treatment | 8,3 | 17.968 | 0.0005 | 9,3 | 2.279 | 0.5165 | 23,3 | 1.891 | 0.5953 |
+Treatment:time point | 11,3 | 15.221 | 0.0016 | 12,3 | 2.763 | 0.4296 | 32,9 | 12.196 | 0.2025 |
+Treatment:time interval | 35,3 | 1.7218 | 0.6321 | ||||||
+Treatment:time point: time interval | 47,12 | 13.258 | 0.3505 | ||||||
422 observations on 110 fish | 405 observations on 110 fish | 518 observations 89 on fish |
df, degrees of freedom. For the number of food items consumed and the feeding position, we used generalized linear mixed models with Poisson error family. For fish activity, we used linear mixed models after log-transforming the data. We included time point as fixed factor and used fish identity as random effects including the repeat to account for the presence of intraindividual variation between repeats. For the feeding position, we additionally included the time interval in the recording (i.e., before vs. after the simulated heron attack) both in the random effects and as a fixed effect. Subsequently, we added the treatment and its interactions with repeat. Test statistics and Markov chain Monte Carlo-estimated P values are for the comparison with the preceding model.