Prandial correlations and the structure of feeding behaviour

Prandial correlations measure the strength of the effects of between-meal interval length on the size of subsequent meals (pre-prandial) and vice versa (post-prandial) and provide insight into the control of food intake. In this paper we develop hypotheses about prandial correlations by using feeding behaviour data consisting of more than 300,000 meals that were recorded from at least 290 animals from each of six genotypes (four broiler lines, ducks and turkeys). The objectives of the study were to (i) obtain estimates of pre- and post-prandial correlations in the three species, (ii) quantify in probabilistic terms exactly how interval length affects meal size and vice versa and (iii) test whether prandial correlations were affected by selection intensity for growth in broilers. Pre-prandial correlations were statistically significant but interval length had almost no effect on meal size in ducks and turkeys and explained only 7–18% of the variation in meal size in broilers. Post-prandial correlations were higher than pre-prandial correlations and meal size explained between 8% and 42% of the variation in subsequent interval length. Selection intensity for growth did not systematically affect pre- or post-prandial correlations in broilers. The probability of animals starting a meal within the next 10 min ( P start ) remained close to zero for longer the larger the preceding meal was but meal size did not affect the rate of the rise in P start with time since the last meal. Similarly, the probability of animals ending a meal before consuming another g of feed ( P end ) was higher at the start of a meal the shorter the preceding interval length but the rate of the rise in P end with amount of food consumed was not affected by previous interval length. Analyses of effects of meal size on changes in P start and of interval length on changes in P end provide more information that is relevant for the structure of feeding behaviour than prandial correlations alone. Such analyses require an appropriate method to group feeding behaviour into meals and data sets of uniform feeding behaviour. A combination of the behavioural probabilities highlighted here with analyses of physiological changes occurring at the same time appears promising for the elucidation of the control of food intake.