Abstract
Understanding the relationships among variables can be difficult if the variables covary with time. Using a simple, commonly used statistical procedure to control for the passage of time, we demonstrate that some hormonal changes traditionally assumed to be causally related to the transition from sexual to parental behaviour can be explained by seasonal changes in the hormones and the fact that parental behaviour follows sexual behaviour in the course of breeding. We quantified immunoreactive prolactin (ir-prolactin) and testosterone concentrations in laboratory-housed and free-living male American kestrels, Falco sparverius, over the course of the breeding season and during sexual and parental phases of reproduction. We found that ir-prolactin increased while testosterone decreased with the transition from sexual to parental behaviour. ir-Prolactin increased with date (i.e. seasonally), whereas testosterone peaked when the frequency of clutch initiation was greatest and then decreased with date thereafter. Controlling for the effects of date eliminated the change in ir-prolactin associated with the transition from sexual to parental behaviour, revealing that prolactin variation associated with this behavioural transition in male kestrels is due, in large part, to seasonal changes in hormone concentrations and the serial nature of these behavioural states. Controlling for the effects of date further revealed that the testosterone difference between sexual and parental males decreases seasonally. It is possible that other phenomena traditionally attributed to environmental, behavioural, social, or other variables also may be explained by the association between such variables and the passage of time. (C) 2004 Published by Elsevier Ltd on behalf or The Association for the Study of Animal Behaviour.