Introducing biological realism into the study of developmental plasticity in behaviour

There is increasing attention for integrating mechanistic and functional approaches to the study of (behavioural) development. As environments are mostly unstable, it is now often assumed that genetic parental information is in many cases not sufficient for offspring to become optimally adapted to the environment and that early environmental cues, either indirectly via the parents or from direct experience, are necessary to prepare them for a specific environment later in life. To study whether these early developmental processes are adaptive and through which mechanism, not only the early environmental cues but also how they impinge on the later-life environmental context has therefore to be taken into account when measuring the animal's performance. We first discuss at the conceptual level six ways in which interactions between influences of different time windows during development may act (consolidation, cumulative information gathering and priming, compensation, buffering, matching and mismatching, context dependent trait expression). In addition we discuss how different environmental factors during the same time window may interact in shaping the phenotype during development. Next we discuss the pros and cons of several experimental designs for testing these interaction effects, highlighting the necessity for full, reciprocal designs and the importance of adjusting the nature and time of manipulation to the animal's adaptive capacity. We then review support for the interaction effects from both theoretical models and animal experiments in different taxa. This demonstrates indeed the existence of interactions at multiple levels, including different environmental factors, different time windows and between generations. As a consequence, development is a life-long, environment-dependent process and therefore manipulating only the early environment without taking interaction effects with other and later environmental influences into account may lead to wrong conclusions and may also explain inconsistent results in the literature.