General and Comparative Endocrinology

Transcription

1 General and Comparative Endocrinology 161 (2009) Contents lists available at ScienceDirect General and Comparative Endocrinology journal homepage: Review Identifying hormonal habituation in field studies of stress Nicole E. Cyr a,b, *, L. Michael Romero a a Department of Biology, Tufts University, Medford, MA 02155, USA b Department of Chemistry, Wellesley College, Wellesley, MA 02481, USA article info abstract Article history: Received 7 January 2008 Revised 9 November 2008 Accepted 3 February 2009 Available online 11 February 2009 Keywords: Catecholamines Chronic stress Field studies Glucocorticoids Habituation Habituation is a term commonly used to explain a decrement in response intensity to a repeated stimulus or set of stimuli. In the stress literature, hormonal habituation is often used to describe a situation where an individual has learned to perceive a repeated stressor as innocuous, and thus the intensity of the release of hormonal stress mediators reduces over time. Consequently, a habituated individual is not considered stressed. There are, however, situations where an individual may be chronically stressed despite a reduction in the response intensity of hormonal stress mediators to a repeated stimulus. These alternative explanations are rarely considered in field studies even though a false conclusion that an individual has habituated (i.e., is not stressed) may lead to false conclusions regarding the animal s overall physiology and health. The present paper provides four alternative explanations for an observed attenuation in the response of hormonal stress mediators to a repeated stimulus or set of stimuli which lead to six criteria that define habituation in a field context. Furthermore, we propose four diagnostic tests to help distinguish hormonal habituation from these alternative explanations in field studies. These tests will help identify hormonal habituation in free-living animals and prevent potential problems of falsely describing an individual or population of individuals as habituated. Ó 2009 Elsevier Inc. All rights reserved. 1. Introduction The intensity of hormonal, physiological, and behavioral responses to a repeated noxious stimulus often wanes over time. This process is not only frequently interpreted as habituation; it is often used to define habituation. However, the term habituation usually implies that the individual became familiar with the stimulus and thus no longer perceived the stimulus to be stressful. Most researchers include the concept of familiarization in their definition of habituation. Indeed, familiarization (i.e., learning that the stimulus is not harmful) is a central component of how habituation is defined in the stress literature (e.g., Gray, 1987; Willner, 1993). However, there are other reasons why hormonal and physiological responses to stressors attenuate over time and using a broad definition of habituation obfuscates the identification of other causal factors that may have separate mechanisms and consequences. The impetus for the present paper comes from what we consider the over-use of habituation as the interpretation for decreased hormone response intensities to a repeated stimulus or set of stimuli in nature. Habituation is only one of four possible explanations for an observed decrease in hormone response intensity, but is often the only explanation that is considered. This could * Corresponding author. Address: Department of Chemistry, Wellesley College, Wellesley, MA 02481, USA. Fax: address: ncyr@wellesley.edu (N.E. Cyr). have important consequences for field studies. Individuals or populations of individuals thought to have habituated to a stressor may falsely be considered to no longer be chronically stressed when, in fact, they remain chronically stressed despite a reduced stress response. For example, a major current issue in marine mammal conservation is how whales do, or do not, cope with anthropogenic noise (NRC, 2005). It would obviously be beneficial if whales habituate to anthropogenic noise, but falsely concluding that whales habituate could lead to ignoring noise impacts and causing potential population declines. The goals of this paper are to present a brief history of the concept of habituation, discuss the potential problems with interpreting an attenuated hormonal stress response as habituation, to list other explanations for a decreased hormonal stress response, and to suggest tests for hormonal habituation that will be useful in field studies The stress response Responses to stress are often divided into two categories: acute and chronic. Acute responses are those that take place in response to short-term stressors and have a definitive onset and last for only a few hours. Chronic stress is defined as either multiple, frequent exposure to stressors and/or long term constant exposure to stressors, and is usually measured by changes in hormones of the hypothalamic pituitary adrenal (HPA) axis such as glucocorticoids (GCs) and corticotrophin releasing factor (CRF), the catecholamines /$ - see front matter Ó 2009 Elsevier Inc. All rights reserved. doi: /j.ygcen

2 296 N.E. Cyr, L.M. Romero / General and Comparative Endocrinology 161 (2009) (CAs) epinephrine and norepinephrine, and immune factors such as cytokines and lymphocytes. These hormones and cytokines are often referred to as primary stress mediators (reviewed in McEwen, 1998; McEwen and Wingfield, 2003) and we will use this nomenclature here. For the purpose of this review we focus on the hormones of the HPA axis and CAs because the vast majority of stress research has focused on these primary stress mediators. During chronic stress, primary stress mediators typically increase followed by general dysregulation (Dallman and Bhatnagar, 2001). Chronic stress has received much attention in a variety of fields such as biomedicine and comparative biology primarily because chronically stressed individuals frequently experience disease (metabolic, physiological, and infectious), the proximate cause of which is often an excess of primary stress mediators. In the short-term, or in response to an acute challenge, the stress response is believed to be adaptive (Sapolsky et al., 2000; Wingfield and Romero, 2001). However, long term release of primary stress mediators such as CAs and GCs can result in disruption of normal physiological function (McEwen, 1998). For example, cardiovascular consequences of excess CAs include hypertension, myocardial infarction, increased cardiac output and arrhythmias (Rupp, 1999). Chronically elevated GCs have been associated with hyperglycemia, neuronal cell death, and suppression of the immune and reproductive systems (reviewed in Wingfield and Romero, 2001). There are also behavioral responses to chronic stress, but they are diverse and extraordinarily context dependent and beyond the scope of this review. Although most biomedical studies that induce chronic stress in laboratory animals demonstrate a significant increase in primary stress mediators (typically GCs and CAs), a substantial minority of studies report either no change or a decrease in these mediators (reviewed in Dallman and Bhatnagar, 2001). However, even when these primary stress mediators do not change or decrease, chronically stressed animals can still suffer from disease (Blanchard et al., 1995; Costoli et al., 2004). For example, in a laboratory-based social stress model involving dominant and subordinate rodents there are two classes of subordinates, those that show elevated GC concentrations and those that do not, yet both classes show evidence of disease (Blanchard et al., 1995). Notwithstanding these data, habituation is the typical interpretation when animals reduce the intensity of their stress response to a particular stressor and, once habituation has been deemed to have occurred, the animals are no longer considered to be chronically stressed (Willner, 1993). Consequently, increased CA and GC concentrations are thought to indicate chronically stressed individuals, and a lack of increased CA and GC concentrations are thought to indicate nonchronically stressed, or habituated, individuals. As we will present below, this is not always a reasonable assumption. Conservation biologists have recently begun using primary stress mediators, predominantly GCs, to determine whether individuals or populations of individuals are chronically stressed, especially animals experiencing human disturbances and translocation (e.g., Cabezas et al., 2007; Creel et al., 2002; Fowler, 1999; Wasser et al., 1997). In general, hormonal habituation and chronic stress are considered opposite states where a chronically stressed animal (or population of animals) is at risk of disease and a habituated animal (or population of animals) is not at risk. For example, in a recent review which evaluated methods of quantifying human disturbance on wild animals Tarlow and Blumstein (2007) state that chronic stress can be caused by the inability of an animal to habituate to human disturbance. The authors go on to review the current literature in which some studies have found a positive correlation between GC concentrations and human disturbance and others have found the opposite. In each case, elevated GCs were assumed to indicate stress; whereas low GCs were assumed to indicate hormonal habituation. However, there are circumstances where an individual will reduce the intensity of their response to a particular stressor or set of stressors, yet still experience chronic stress (discussed in detail below). Clearly, a precise definition of hormonal habituation, as well as knowing when it occurs and when it does not occur, is critical for determining whether an animal is chronically stressed Historical definitions of habituation Habituation has been studied extensively and here we present a very brief history of the definitions of habituation, and which components we feel are useful for identifying hormonal habituation in free-living animals. The phenomenon of habituation has been most comprehensively studied in terms of reflex systems. One of the first definitions of habituation, which is still used in the literature today, comes from a review of neural responses in the intact organism written by Harris (1943) in which he defines habituation as: response decrement as a result of repeated stimulation. Harris (1943) also noted that habituation can be distinguished from other causal factors such as trauma, injury, and aging because it is reversible. This became a controversial point because later research showed that not all habituated responses recovered spontaneously after the stimulus was removed as Harris described, but instead some recovered on a longer time scale or not at all (e.g., Sharpless and Jasper, 1956; Thorpe, 1963). As a result, the phenomenon of habituation was split into two categories (1) short-term habituation in which response intensity recovers spontaneously after stimulus withdrawal and (2) long-term habituation in which the response intensity recovers long after the stimulus is ended or the response remains attenuated (Sharpless and Jasper, 1956). Later, Thompson and Spencer (1966) listed nine parametric characteristics of habituation for what they termed the operational definition of habituation. These characteristics of habituation were: 1. Given that a particular stimulus elicits a response, repeated applications of the stimulus result in decreased response (habituation). The decrease is usually a negative exponential function of the number of stimulus presentations. 2. If the stimulus is withheld, the response tends to recover over time (spontaneous recovery). 3. If repeated series of habituation training and spontaneous recovery are given, habituation becomes successively more rapid (this may be called potentiation of habituation). 4. Other things being equal, the more rapid the frequency of stimulation, the more rapid and/or more pronounced is the habituation. 5. The weaker the stimulus, the more rapid and/or more pronounced is the habituation. Strong stimuli may yield no significant habituation. 6. The effects of the habituation training may proceed beyond the zero or asymptotic response level. 7. Habituation of response to a given stimulus exhibits stimulus generalization to other stimuli. 8. Presentation of another (usually strong) stimulus results in recovery of the habituated response (dishabituation). 9. Upon repeated application of the dishabituatory stimulus, the amount of dishabituation produced habituates (this might be called habituation of dishabituation). Although not universally accepted, many authors in the fields of behavior, psychology, and physiology extend the definition of habituation to include an aspect of learning (e.g., Domjan, 1996; Dubovicky and Jezova, 2004; McCarty et al., 1992; Thorpe, 1963). In this case, habituation is used to describe a situation where an

3 N.E. Cyr, L.M. Romero / General and Comparative Endocrinology 161 (2009) individual learns to ignore innocuous stimuli. We feel that this is an important addition to the concept of habituation, and is an implicit assumption in almost all field studies of chronic stress. Consequently, the nine parametric characteristics of habituation described by Thompson and Spencer (1966) together with the concept that habituated animals have learned to perceive a repeated stressor as innocuous, provide a strong basis for determining whether an animal has habituated to a stimulus in nature. We use this as a foundation for determining whether hormonal habituation has occurred to free-living animals exposed to a stressor or set of stressors. 2. Explanations for attenuated stress response to repeated stimuli There are four possible explanations for a decrease in the response to a repeated stressor one of which is hormonal habituation (Table 1). Below we list each of the four explanations and provide an example of the three explanations that are not habituation to illustrate the distinction between that explanation and hormonal habituation Physiological responses vary due to seasonal/life history changes The vertebrate stress response is activated by stressors that are real and/or perceived, and an animal may perceive an event as stressful only at certain times of the year (Romero, 2002). Furthermore, vertebrates vary both their basal and stress-induced GC concentrations during the year (Romero, 2002). Many of these changes in GC concentrations can occur on a short time scale. For example molting birds rapidly reduce their basal and stress-induced GC concentrations (Romero, 2002), a change that is partially mediated by decreases in the capacity to secrete GCs (Romero, 2001). Attenuated CA and GC responses may be explained by a regulated change due to different life-history demands. Furthermore, as Landys et al. (2006) persuasively argue, changes in baseline GC secretion that are part of the normal progression of life-history stages should not be misconstrued as indicating stress. Because Table 1 Explanations for a decrease in the intensity of the physiological response to a stressor or set of stressors. Explanation Characteristics 1. Seasonal/life history changes A natural shift in perception of a stressor such that the animal perceives a given stimulus as noxious at certain times of the year, and thus activates the stress response. However, the animal does not perceive that same stimulus as stressful at other times of the year, therefore does not activate the stress response A natural shift in concentrations of a physiological mediator (e.g., CAs and GCs). CAs and GCs vary in their concentration under basal and stress-induced conditions seasonally 2. Habituation At first, the animal perceives a given stimulus as noxious With repetition the animal learns to perceive that stimulus as innocuous, and thus reduces the intensity of their stress response to that particular stimulus 3. Physiological desensitization without habituation With repetition, the physiological response to the stimulus decreases, but the animal does not learn to adapt to the stimulus. The animal, thus, perceives the stimulus as noxious even with repetition 4. Exhaustion There is a breakdown of the physiological system such that the animal is too fatigued to maintain a stress response with repetition of a stimulus responses to stressors are specific to each life-history stage, comparisons of responses in order to diagnose habituation must be made within, not between, life-history stages. Although this may seem a trivial point, life-history stages can change extremely rapidly in many species. For example, several arctic breeding species can change reproduction stages in a day (Wingfield and Hunt, 2002), which is accompanied by changes in GC responses to capture stress (Holberton and Wingfield, 2003). It would clearly be inappropriate to compare responses from two different reproductive stages and conclude that the lower response reflected habituation Hormonal habituation Most authors define habituation as a decrease in response intensity as a novel stressor becomes familiar with repetition (Dubovicky and Jezova, 2004; e.g., Gray, 1987; Grissom et al., 2007; McCarty et al., 1992). Once a stressor becomes familiar the animal may no longer perceive that stimulus as harmful, and thus the stress response is no longer necessary and the term chronic stress is not appropriate. In other words, the stressor becomes innocuous and the individual ignores it Desensitization of the physiological response without habituation to the stressor Severe stressors often do not conform to certain characteristics of habituation and instead cause changes in the entire physiology of stress hormone release (Marti and Armario, 1998; Tache et al., 1976). Few of Thompson and Spencer s (1966) characteristics of habituation apply, and the 5th characteristic specifically indicates that habituation may not occur for strong stimuli. For example, CAs and GCs may be attenuated because their respective axes (sympathetic adrenal medulla and hypothalamic pituitary adrenal) have been altered to a continued or repeated challenge. Mechanisms include downregulation of receptors, dysregulation of negative feedback, downregulation of hormone production (e.g., changes in gene transcription rates), changes in concentrations of synthetic enzymes, and changes in rates of hormone clearance. At the same time, the animals may not have concluded the stimulus is innocuous. The important criterion that the animal no longer perceives the stressor as stressful is not met. Thus, the animal may experience stress despite reduced release of primary stress mediators. Note that distinguishing between habituation and desensitization is a central aspect of our hypothesis of what constitutes habituation and verifying this distinction should be confirmed experimentally. Importantly, desensitization by itself may be costly as it may compromise the animal s ability to respond to other stimuli Exhaustion Hans Selye (1946) proposed a concept of physiological exhaustion wherein CA and GC concentrations decrease because they can no longer be maintained. In other words, the attenuated response resulted not from habituation, but from the breakdown of the physiological system. This was part of his three-stage model of the stress response where in the final stage the individual is so fatigued that it is impossible to maintain a stress response and all primary stress mediators decrease in their concentrations. There is precedent for this idea in the medical literature. For example, reduced GC output occurs in patients with fibromyalgia and chronic fatigue syndrome, and similarly, individuals with posttraumatic stress disorder present lower basal and sometimes lower stress-induced GC concentrations (McEwen, 1998) at least during certain times of the day (Miller et al., 2007). Although it is unclear

4 298 N.E. Cyr, L.M. Romero / General and Comparative Endocrinology 161 (2009) how relevant these human diseases are to animals in the wild, certain human disturbances such as pollution may create functionally equivalent physiological changes (e.g., Hontela, 1998). 3. Defining and testing hormonal habituation in free-living animals The preceding four potential explanations for an attenuated response of primary stress mediators suggests that historical definitions of habituation are insufficient when applied to free-living animals. Integrating historical definitions with conditions facing free-living animals, however, indicates that six criteria are sufficient for defining habituation in this context. Furthermore, four diagnostic tests, based upon these six criteria, can help determine whether a free-living animal has or has not habituated to a repeated stimulus Criteria for habituation in field studies Thompson and Spencer s (1966) list of parametric characteristics of habituation helped to define habituation to a neural stimulus. Similarly, we have developed criteria for determining hormonal habituation in field studies (Table 2). These six criteria can be considered an operational definition of hormonal habituation for use in the field. Critical to this definition is that habituation can only be interpreted in reference to a specific and explicitly defined stimulus (i.e., habituation is stimulus-specific). For habituation to be the interpretation in field studies, the response must show the following criteria Criterion 1: the response to the same stimulus must decrease over time The concept of an attenuated response to a repeated stimulus is common to all definitions of habituation including neural, behavioral, or physiological habituation (Domjan, 1996; Gray, 1987; McCarty et al., 1992; Thompson and Spencer, 1966); and thus, is our first criterion for hormonal habituation Criterion 2: the decreased response should not generalize to other stimuli (dishabituation) Dishabituation (also called facilitation) of GC and CA responses is a phenomenon that has been consistently demonstrated in the biomedical literature where an animal that had habituated to a repeated stimulus shows a normal or exaggerated response to a different (novel) stimulus (Dallman and Bhatnagar, 2001; McCarty et al., 1992). In other words, hormonal habituation is not a generalized response; it is a reduced response to one specific stimulus. The concept of dishabituation was the 8th characteristic of Thompson Table 2 A list of the six criteria that compose our operational definition of hormonal habituation for use in field studies of stress. Criterion number Criterion description 1 The response to the same stimulus must decrease over time 2 The decreased response should not generalize to other stimuli (dishabituation) 3 The decreased response should result from learned, not physiological, changes (the capacity to respond should not be reduced) 4 The decreased response should not lead to decreases in health 5 The decrease should be in response to the stimulus, not in the prestimulus state (learning that a stimulus is innocuous should return basal primary stress mediator levels to the original pre-stimulus state) 6 The decreased response should be compared to responses within a single life-history stage and Spenser s (1966) definition of neural habituation, and is critical to our definition of hormonal habituation because it emphasizes that the reduction in response intensity is due to the specific repeated or constant stimulus, and that habituation does not hinder the animal s ability to mount a stress response to other stimuli Criterion 3:the decreased response should result from learned, not physiological, changes (the capacity to respond should not be reduced) Criterion 3 explicitly adds a learning component to the concept of habituation wherein the process of habituation occurs because the animal learns to ignore an innocuous stimulus. Learning is a critical component because it directly links habituation to the brain s decision making process whereby a stimulus is determined to be or not to be a stressor. If the actual capacity to respond is reduced, thereby resulting in a decreased response, then whether or not the animal perceives the stimulus to be a stressor is no longer relevant it could not respond even if it wanted to. This would be counter to what most researchers would consider habituation. In terms of our definition of hormonal habituation, habituation occurs because the animal learned that the repeated stimulus is not harmful and no longer responds to it, but perceives the novel stimulus as harmful and responds with an appropriate increase in primary stress mediators. Note that criterion 3 (learning) is thus linked to criterion 2 (dishabituation). The stress axis, therefore, should retain the capacity to function normally when presented with a novel acute stressor Criterion 4: the decreased response should not lead to decreases in health Animals that exhibit physiological desensitization or exhaustion are likely to show changes in weight as well as other indicators of failing health because they are chronically stressed. During the process of hormonal habituation the initial increases in primary stress mediators may or may not lead to health costs, but over time, as the hormonal response wanes, any health consequences should improve. Once the animal has fully habituated and no longer perceives the stressor as stressful, the hormonal response should cease and the animal should no longer experience any health cost, barring any lasting effects from the earlier condition. Thus, deteriorating health could indicate that the animal has failed to habituate to the specific stimulus in question and is chronically stressed, irrespective of the attenuated hormonal response. Here we focus on long-term exposure to repeated or constant stressors. There are certainly cases where animals exposed to short-term stressors do not incur any health consequences. However, animals that experience long term (chronic) stress are expected to suffer health costs Criterion 5: the decrease should be in response to the stimulus, not in the pre-stimulus state (learning that a stimulus is innocuous should return basal primary stress mediator levels to the original pre-stimulus state) Habituated animals decrease the intensity (either the magnitude or duration) of their stress response to a repeated or constant stressor because they learned the stressor was not harmful. Consequently, baseline concentrations of primary stress mediators should not be affected. Primary stress mediators such as GCs and CAs circulate at baseline concentrations and are elevated during acute stress to stress-induced concentrations. It is the stress-induced concentrations of these primary mediators that enable the animal to cope with the stressor (Sapolsky et al., 2000). A habituated animal responds to a stimulus by first increasing these primary mediators because the stimulus is stressful, then decreasing the response each time the same stressor is presented because the animal perceives the stimulus as less harmful each time. Hence, habituation is a physiological change in the magnitude or duration of the stress-induced, not baseline, concentrations

5 N.E. Cyr, L.M. Romero / General and Comparative Endocrinology 161 (2009) of primary stress mediators. Therefore, it is expected that only the stress-induced concentrations of primary stress mediators are affected. If baseline concentrations of primary stress mediators are affected then the animal has not habituated Criterion 6: the decreased response should be compared to responses within a single life-history stage Given that many primary stress mediators change with life-history stage (Landys et al., 2006; Romero, 2002), it is important to compare responses of these mediators within the same life-history stage. Otherwise an observed decrease in the response intensity of a primary stress mediator may be due to a shift in life-history stage rather than habituation Diagnostic tests for evaluating habituation in field studies We propose four diagnostic tests for determining whether habituation has occurred to a stressor or set of stressors in field studies (Table 3) Diagnostic test 1: dishabituation Dishabituation is the common phenomenon that occurs when an animal that has habituated to a specific stimulus shows a normal or exaggerated response to a novel stimulus (Thorpe, 1963). Several studies have demonstrated dishabituation of the primary stress mediators GCs and CAs (Dallman and Bhatnagar, 2001; McCarty et al., 1992). For our definition of habituation, we also include a learning component where habituation occurs because the animal learned that the repeated stimulus is not harmful and no longer responds to it, but perceives the novel stimulus as harmful and responds with an appropriate increase in primary stress mediators (criterion 3). The stress axis, consequently, has the capacity to function normally when presented with a novel acute stressor. Therefore, dishabituation can be used as a diagnostic test for habituation. For example, the adrenal gland should be functioning normally and an ACTH (the primary GC secretagog) injection would test for alterations in adrenal responsiveness. Similarly, an injection of corticotrophin releasing factor (CRF) and arginine vasopressin (AVP or the homolog AVT depending upon the species) would test for changes in pituitary responsiveness (Romero and Sapolsky, 1996). If, however, the animal has not habituated to the repeated stressor, as in the case of physiological desensitization or exhaustion, the introduction of a novel stressor or a challenge with ACTH, CRF, or AVT (AVP) is not expected to cause dishabituation. In the case of exhaustion, the system is too fatigued to respond and in the case of physiological desensitization adding a new stressor should not alter the animal s state (i.e., chronic stress) and the stress response should remain attenuated. The release of GCs and CAs is likely to be altered during physiological desensitization or exhaustion. For example, studies that do not show a normal increase in GC concentrations following ACTH (Walker et al., 2006) or AVT injection (Rich and Romero, 2005) are not examples of habituation. Consequently, exposure to a novel stressor, or injections of ACTH, CRF, or AVP (AVT) could test for habituation of the endocrine response. Although we recognize the difficulty of conducting all of these tests in the field, ideally all tests would be useful to diagnose dishabituation. If all tests are impossible, ACTH injection is the test most likely to be useful. However, it is best to conduct multiple endocrine challenges to identify the level at which the HPA axis has been affected. By only conducting one test an important physiological alteration may be missed. For example, only presenting the ACTH challenge would test the state of the adrenals, but would not test whether the anterior pituitary is releasing ACTH in response to CRF Diagnostic test 2: health consequences In the short-term animals in the process of habituation may incur health problems due to elevated primary stress mediators. However, over the long term, habituated animals should not experience health consequences (see criterion 4). In contrast, animals that experience long term (chronic) stress are expected to suffer health costs. These include, but are not limited to, effects such as decreased immune function, reproductive failure, lethargy, decreased body condition, increased parasite load, behavioral Table 3 Diagnostic tests for detecting habituation of primary stress mediators in field studies. Test Description Which alternative explanations can be distinguished from habituation Dishabituation Health and fitness Changes in baseline versus stress-induced concentrations of stress mediators Control for life-history stage 1. Expose the animal to a novel stressor and look for a normal or exaggerated response in stress mediators Habituation is a specific response to a repeated stimulus, however when an animal is exposed to a new stimulus stress mediators increase (i.e., the stress response functions normally). If the response was generalized and stress mediators continued to be blunted during a novel stressor, that would not be considered habituation, but would be consistent with either physiological desensitization or exhaustion 2. Inject ACTH to measure adrenal responsiveness A habituated animal should mount an appropriate response to ACTH, AVP(T), and CRF injection, whereas animals experiencing physiological desensitization or exhaustion may not. Although performing all three tests is ideal ACTH injection would be the preferred first choice 3. Inject AVP (AVT) and/or CRF to measure pituitary responsiveness 1. Monitor the health and fitness of individuals. For example, compromised immune responses, survival, reproductive function and/or success. Over time, measure the baseline concentrations of stress mediators as well as the response of these mediators to an acute stimulus Animals that exhibit Physiological desensitization or Exhaustion are likely to experience failing health or fitness costs because they are chronically stressed. Habituated animals have learned to ignore the stimulus and are not stressed, thus habituated animals are not expected to incur health and fitness costs associated with chronic stress Habituated animals will reduce their stress response to a repeated stressor, but baseline concentrations of stress mediators should not be altered. However, both baseline and stress-induced concentrations may decrease during physiological desensitization (e.g., Rich and Romero, 2005) 1. Compare hormone levels over different times of the year/ Taking measurements over the course of the year during different lifehistory stages will control for seasonal/life history changes day/life history-stage 2. Compare hormone levels within life-history stages Comparisons of separate populations (or the same population under different environmental conditions at different times) should be made within the same life-history stage. This will also control for seasonal/life history changes

6 300 N.E. Cyr, L.M. Romero / General and Comparative Endocrinology 161 (2009) alterations (e.g., anxiety), energy dysregulation, and hypertension. Thus, long term health problems can be used as diagnostic for hormonal habituation. Note that, under laboratory conditions, the health measures listed above can be related to chronic stress even in the absence of detectable changes in primary stress mediators (Blanchard et al., 1995). It will be important to experimentally verify this under field conditions as well. In addition to health costs, reproductive fitness costs may also indicate that an animal has or has not habituated. For example, if an animal learns to ignore a stimulus and primary stress mediators such as GCs return to normal, the animal should not suffer long term fitness costs related to the stimulus (e.g., the animal should return to normal reproductive success). In contrast, a chronically stressed animal is expected to incur fitness costs. Fitness costs may indicate chronic stress, but this should be interpreted with caution because other factors not related to chronic stress can also decrease fitness. Therefore, while changes in fitness may present a clue that an animal has or has not habituated, a change in fitness is not a diagnostic Diagnostic test 3: changes in baseline versus stress-induced concentrations of primary stress mediators When baseline primary stress mediator concentrations in animals subject to a potential stressor are compared to those in non-stressed animals, there are three possible outcomes (1) either baseline primary stress mediators are elevated, (2) baseline primary stress mediators are equivalent, or (3) baseline primary stress mediators are reduced compared to a non-stressed animal. Elevated baseline levels indicate that an animal is currently mounting a stress response, and consequently has not habituated. If the baseline levels are equal to a non-stressed animal, the animal may have habituated, but a final determination depends upon the response of the primary stress mediator to the chronic stressor (i.e., the definition of habituation still requires that the stress-induced levels of the mediator be attenuated compared to a naïve (nonhabituated) animal, see criterion 1). If baseline primary mediators are lower than a non-stressed animal, this indicates physiological desensitization or exhaustion because a truly habituated animal has learned that the stimulus is innocuous so that baseline physiology should be identical to a non-stressed animal Diagnostic test 4: control for life-history stage (multiple measurements over time) For those stressors that are of relatively short duration (e.g., inclement weather, predator attacks, ecotourism, etc.), physiological responses to repeated stressors should only be compared within a life-history stage. Otherwise, physiological changes resulting from different life-history demands would be a potential confounding variable. For long-term stressors (e.g., famine, social instability, anthropogenic changes, etc.), or where life-history stage cannot be controlled, the response to a repeated stressor should be measured multiple times over the course of the year or during different life-history stages to control for natural variation in primary stress mediators caused by life-history demands (see our 6th criterion). It is not always possible to observe the entire process of habituation in the field. For example, if one is interested in learning whether a population of animals has habituated to tourism the animals may already show an attenuated stress response to human presence, such that the researcher has missed the process of gradual decline in that response. In these cases, we suggest testing for dishabituation (diagnostic test 1) and testing for changes in health and fitness (diagnostic test 2) relative to other conspecific populations. Together, these four diagnostic tests will help to distinguish hormonal habituation from other explanations for an observed decrease in response intensity to a repeated stressor or set of repeated stressors in field studies. While we understand that conducting all four diagnostic tests will not always be feasible, each test is important because each represents a different aspect of habituation and together they can accurately identify hormonal habituation. 4. Examples of interpretation The following four examples presented below illustrate why some decreases in the response of primary stress mediators such as GCs and CAs do not fit our criteria of hormonal habituation. They are all examples from nontraditional model species and field studies and illustrate why an attenuation of stress responses in wild animals may not be easy to interpret in the context of habituation. Furthermore, they show why an uncritical interpretation of habituation can lead to incorrect conclusions about the potential implications of the work Weather In nature, animals must learn to cope with adverse weather conditions. GC concentrations have been shown to increase during storms (Astheimer et al., 1995; Rogers et al., 1993; Smith et al., 1994; e.g., Wingfield et al., 1983) theoretically to facilitate survival during such inclement conditions. Animals frequently exposed to adverse weather may elevate GC concentrations each time or may habituate, especially if the conditions are only moderately severe. One might assume that a reduced GC response to a repeated challenge such as low temperatures signifies habituation, but that may not be the case. For example, Romero et al. (2000) showed that lower temperatures stimulated GC release during molt, but not during the breeding season in arctic snow buntings (Plectrophenax nivalis) and red polls (Carduelis flammea). Were the birds habituating to lower temperatures during breeding? The comparison between breeding and molt is not appropriate to answering this question. There are other explanations, based on life history theory, which can explain the lower response during breeding. Wingfield (1994) proposed that the reduction in sensitivity to stressors was a regulated process to avoid the negative impact of chronic GC concentrations in disrupting breeding, an important life-history stage. To conclude habituation to explain the attenuation would violate our 6th criterion for habituation that comparisons should be made within a single life-history stage. Therefore, seasonal/life history changes (Table 1, explanation 1) is the best explanation for the attenuation of GC release to a repeated stressor in this example Rotating unpredictable stressors Kant et al. (1983) designed a protocol that rotated varying intermittent stressors so that the animals could not habituate, and thus would remain chronically stressed throughout the duration of the protocol. Most studies using this protocol on laboratory rats report an increase in GCs, which has been a measure of its success in inducing chronic stress. However, studies that employed this protocol using another species (European starlings, Sturnus vulgaris) found a decrease in both baseline and stress-induced plasma GCs (Rich and Romero, 2005) and heart rate (Cyr and Romero, 2008). Several lines of evidence indicate that the attenuation of GCs cannot be explained by habituation; primarily because there is no evidence that the birds learned that the stimuli were not harmful. First, birds lost weight throughout the protocol (Rich and Romero, 2005) and weight loss can be an indicator of stress (Gray et al., 1990; Wingfield et al., 1997). This violates criterion 4. Second, Rich and Romero (2005) showed that the primary regulation of GC secretion shifts from adrenocorticotropin hormone (ACTH) under

7 N.E. Cyr, L.M. Romero / General and Comparative Endocrinology 161 (2009) normal conditions to arginine vasotocin (AVT) under chronic psychological stress (i.e., the rate limiting step in the pathway leading to GC secretion shifts from ACTH to AVT), indicating that the attenuation did not result from learning that the stimulus was not stressful. This violates criterion 3. Third, whereas a decrease in stress-induced responses would clearly indicate habituation, Rich and Romero (2005) and Cyr and Romero (2007) show decreases in the baseline levels as well. This violates criterion 5. If stressors were no longer being interpreted as being stressful, then there should be no decrease in baseline levels (see Sections and for further discussion). Importantly, similar results occurred when these laboratory conditions were repeated in the field. Nesting free-living European starling females subjected to a similar rotation of chronic rotating stressors also showed lower plasma baseline GCs (Cyr and Romero, 2007). The similarity of the response in the captive and free-living birds suggests that the captive studies adequately modeled the physiological changes in the free-living birds. Furthermore, these females fledged fewer young than starlings not exposed to the rotating stressors. In addition, surviving young in broods from starlings exposed to stressors were not in better condition than young of unstressed broods. These results suggest a fitness cost to the chronic stress protocol used in this study (Cyr and Romero, 2007). Although fitness is not a diagnostic test for habituation, fitness costs may provide a clue that the animals in this study were experiencing chronic stress as stress is one reason that fitness would decrease. Despite the diminished plasma GC response over time, the birds responses to the protocol did not fit two of our criteria for habituation (criteria 4 and 5 were not met) and based upon the parallel laboratory data, were unlikely to have satisfied two others (criteria 2 and 3). Rather, reduced GC concentrations observed in these studies are better explained by physiological desensitization (Table 1, explanation 3). Furthermore, a recent study using the same chronic stress protocol showed that metabolites of GCs measured in fecal samples were higher in chronically stressed than unstressed free-living starlings (Cyr and Romero, 2008), which violates criterion 1 for habituation. Fecal GC metabolites provide a longer term measure of GC concentrations than plasma samples. Therefore, these changes likely show that even though the GC response to each of the four acute stressors decreased during chronic stress, the frequency of GC release to acute stressors was high so that over the course of the day birds exposed to the chronic stress protocol actually had higher CORT then unstressed birds. This result also highlights the issue that changes in primary stress mediators may depend on the measurement technique Ecotourism Many researchers are interested in understanding whether animals habituate to anthropogenic stressors. The effect of tourism as an anthropogenic stressor on wild species has been investigated extensively (e.g., Fowler, 1999; Mullner et al., 2004; Romero and Wikelski, 2002; Walker et al., 2005, 2006; Yorio et al., 2001). For example, Magellanic penguins (Spheniscus magellanicus) nesting in highly visited tourist areas showed behavioral and physiological desensitization to humans (Fowler, 1999; Walker et al., 2006; Yorio and Boersma, 1992). Penguins in areas of high tourism responded to a tourist-like stressor (human presence for 15 min near their nest) with fewer head turns and lower GC concentrations compared to penguins nesting away from tourist areas (Walker et al., 2006). Does this attenuation fit the criteria of habituation? Two lines of evidence suggest not. First, these penguins also exhibited a lower GC response to capture and restraint. Walker et al. (2006) used a typical 30 min restraint protocol for which the animal was captured, removed from the nest, and restrained with an opaque bag placed over the head. Given that this is presumably a novel stressor to the animal, one would expect to observe dishabituation in the habituated, tourist-exposed penguins, but this was not the case (violates criterion 2). Second, tourist-exposed penguins had a lower GC response to ACTH injection than penguins not exposed to tourists (violates criterion 3). Consequently, the data suggest that the stress response of tourist-exposed penguins has physiologically acclimated, but that the animals have not fully habituated to humans (i.e., still perceive human visitation as a stressor) and as such are chronically stressed (Table 1, 3rd explanation) Chronic contaminant exposure Several researchers have become interested in how chronic exposure to contaminants affects GC responses (e.g., Franceschini et al., 2008; Hontela, 1998; Norris et al., 1999). A particularly illuminating example is shown by Norris et al. s (1999) work on trout (Salmo trutta) living in metal-contaminated water. These animals did not appear unhealthy in their environment and had been living in these waters for many years. However, when subjected to capture and handling, contaminated trout initiated an identical GC response yet failed to sustain that GC response compared to uncontaminated controls. Furthermore, all the contaminated fish died during the procedure, whereas none of the controls died. These contaminated animals were surviving in their environment, but the introduction of an acute stressor disrupted their hypothalamic pituitary axis to such an extent that it caused death. It is unlikely that anyone would make the mistake of concluding that the attenuated GC response in contaminated fish represented habituation. However, the concern is for field studies where there is hidden contamination (contamination unbeknownst to the researchers). If this had been the case, Norris et al. (1999) could potentially have concluded that some of the fish (the contaminated ones) had habituated. However, the contaminated fish clearly suffered health consequences, which violates our 4th criterion. The reduced GC response to capture and handling in Norris et al. s (1999) example resulted from exhaustion of the HPA axis (Table 1, 4th explanation), most likely as a side-effect of toxic changes in the interrenal gland resulting in an inability to sustain a GC response, not habituation. Hidden toxic effects on the HPA axis may be an important alternative to habituation. 5. Conclusions Habituation is an extremely important concept in the field of stress, but there is no standard definition of hormonal habituation for wild animals. We favor an expanded definition of hormonal habituation beyond simply a response decrement due to repeated stimulation because, using this definition alone, habituation cannot be distinguished from other factors that might cause a decrease in response intensity (Table 1). Moreover, this is a potentially dangerous definition of habituation in conservation contexts because animals demonstrating a decrease in their stress response may falsely be considered not chronically stressed. There are many negative consequences associated with a decrease in CAs and GCs such as hypoglycemia, hypotension, and fatigue. Furthermore, the stress response is thought to be adaptive in that it helps an animal survive stressors. Consequently, an inability to mount a proper stress response would presumably decrease fitness. These problems may be ignored if populations are naively assumed to have habituated. Furthermore, conservation decisions may be made based on false identification of habituated populations. Clearly, a simple attenuation of GC and/or CA responses is insufficient to conclude that habituation has occurred and studies of habituation would greatly