1. Introduction

Beyond a large diversity of concepts, welfare refers principally to the subjective psychological state of the individual, as related to its internal and external environment (Fraser, 1999; Rushen, 2003; Broom, in this book). Since we are not yet at the stage of being able to read directly animals’ feelings and emotions, we try to infer those from measurable indices that we know or suppose to be related to them. Most of these measures – including behaviour, biology, production traits and pathology – derive from the study of emotions / stress / adaptation psychophysiology and physiopathology (Dantzer and Mormède, 1983). I do not intend to write an exhaustive review of the abundant literature available on this topic, but rather give a personal view of the limitations of these approaches and the challenges faced by researchers in future investigations. Comments and discussions will focus mostly on four major issues:

The autonomic nervous system (ANS), (Cannon, 1935) and the hypothalamic-pituitary-adrenocortical (HPA) axis (Selye, 1936) have the front of the stage in stress studies (Mormède, 1995). However, these neuroendocrine systems are primarily involved in metabolic homeostasis and particularly in the regulation of energy fluxes (Dallman et al., 1995). In a teleological perspective, the reason why these systems are activated by stressors is that they are able to produce energetic metabolites either from energy storage tissues (the ANS mobilizes fat from adipose tissues and glycogen from liver) or by transformation of proteins into energetic metabolites (neoglucogenesis enhanced by glucocorticoid hormones). This energy supply is used by the defense mechanisms to cope with the stressor. Consequently, any change in HPA axis or ANS functional parameter is not necessarily the response to a stressful stimulus, but can reflect their involvement in homeostatic metabolic processes. The best example is the increase of cortisol levels induced by meals that are not usually considered as stressors. Furthermore, metabolic adaptive changes do not necessarily require activation of these systems, but can sometimes shut them down, depending upon the specific demands of the situation. For instance the response of the ANS to early weaning in pigs is an inhibition that can bee seen by the reduction of catecholamine levels in urine (Hay et al., 2001). This is an energy saving mechanism adapted to the deficit resulting from weaning-induced starvation. These metabolic influences on neuroendocrine changes must be taken into consideration when interpreting experimental data.

The second point of discussion is the duration of the stimulus. This dimension of the response plays a pivotal conceptual role in the “general adaptation syndrome” as described by Selye, with the three successive phases, alarm, resistance and exhaustion (Selye, 1956). The biological responses to acute challenges (such as delivery, castration, weaning, mixing of animals from different social groups, restraint, transportation, slaughter) have been studied extensively and, like most stressors, activate biological stress systems in a more or less standardized manner (alarm phase). This common pattern of response is at the origin of the stress concept that was defined par Selye as the “non-specific response of the body to any demand made upon it” (Selye, 1973). Note however that this non-specificity is mostly the result of the uniqueness of the response of the HPA axis, i.e. an increase in circulating cortisol levels that is exquisitely sensitive to “any demand”, whatever its nature and intensity. However, if the stimulus is maintained for some time blood cortisol returns to control levels even if the sustained activation of the HPA axis can be detected by different approaches like dynamic testing. Since many factors challenging animal welfare are long lasting – this is the case for most influences from the physical and social environment of the animals – more attention should be given to the exploration of chronic readjustments of adaptation mechanisms (resistance phase or allostasis) (McEwen, 1998).

The third issue is the huge individual variability seen in the basic functioning of adaptation mechanisms and in their responses to environmental challenges. This variability has a multiple origin, genetic, developmental, and experiential. Although it may not be of primary importance in longitudinal protocols with the same animals studied in basal conditions and then submitted to control or experimental situation, it has to be taken into consideration in field studies in which the history of the animal is not readily available.

The fourth issue is the multivariate nature of the interaction of the individual with its environment that we have to keep in mind when assessing welfare.