“Fit and Happy”: How Do We Measure Wild-Animal Suffering?

Last updated: 23 May 2018


In order to understand wild-animal welfare, we must be able to measure it. To target the most important causes of wild-animal suffering, it is important to understand which animals suffer the most and what causes their suffering. In this paper, I begin by reviewing theoretical arguments about wild-animal suffering, then move to discussing various empirical strategies for assessing the welfare of wild animals. I conclude with a brief discussion of how to reduce the time and expense of assessing wild-animal welfare.

Defining Animal Welfare

Defining what animal welfare is is philosophically fraught. Fraser (1995) distinguishes between three types of concepts. A type 1 concept is a single attribute which can be measured, such as the animal’s weight or height (ibid: 3). A type 2 concept cannot directly be measured but can be estimated based on various contributing measurable variables, such as the animal’s reproductive fitness (ibid: 3). A type 3 concept is value-laden and, to an extent, arbitrary, such as the aesthetic appeal of an animal: there is no single quantitative expression that expresses the aesthetic appeal of an animal, for all viewers and in all respects (ibid: 3).

Some have proposed type 1 operationalizations of animal welfare, such as the animal’s longevity, level of stress, productivity, immune competence, or reproductive success (Fraser, 1995: 5-7). However, these operationalizations generally fail to capture things we intuitively believe about animal welfare: for example, sex is stressful yet pleasurable, and an animal in chronic pain may be long-lived (ibid: 5-7). Others have proposed type 2 operationalizations, most commonly through various ways of ascertaining the animal’s preferences (ibid: 7-9). However, animals sometimes prefer things which are bad for them, perhaps due to ignorance or inability to consider long-term consequences (ibid: 7-9).

Fraser (1995) argues that welfare is instead a type 3 concept. In particular, one’s assessment of an animal’s welfare may depend on one’s values (ibid: 4-5). A veterinarian may prioritize the physical health of an animal; an ethologist, natural behavior; a theologian, reverence for the animal’s life and dignity; a Romantic poet, the animal living in what the poet perceives as harmony with nature (ibid: 4-5). For this reason, empirical estimates of an animal’s welfare will always be uncertain. Researchers may wish to use multiple methods which encompass many divergent sets of values, perhaps including values they do not share.

Two common philosophical definitions of animal quality of life are objective list theory and hedonism (Taylor and Mills, 2007: 56). Objective list theory says that an animal with a high quality of life is one for whom certain basic and species-specific needs are predictably satisfied, such as food, water, shelter, social interaction, environmental control, physical pleasure, and the absence of fear, pain, or distress (ibid: 56). Hedonism says that an animal with a high quality of life is one who has many positive experiences and few negative experiences (ibid: 56). Others define an animal’s quality of life as its own cognitive and affective assessment of how well its life is going, insofar as the animal is capable of making such an assessment (ibid: 57).

People who agree with objective list theory will tend to care more about objective standards for assessing wild-animal quality of life; those who agree with hedonism, stress and other potential measures of distress; and those who think an animal’s quality of life is its own self-assessment, self-report measures. However, in general these different definitions result in similar recommendations, and all methods of empirically assessing wild-animal welfare can inform the opinions of people with any stance on the definition of “quality of life.” For that reason, I will not discuss this distinction further.

Theoretical Arguments

Argument from Life History

Animals may invest a unit of reproductive effort either in producing more offspring or in increasing their offspring’s ability to survive and reproduce (Pianka, 1970). Most species tend to favor one strategy over the other. Although this is a very simplified view of the complex subject of life history strategies (Reznick, Bryant, and Bashey, 2002), it is sound enough for this argument to hold.

Species that have more offspring than what is required to replace previous generations can be expected to experience more suffering than happiness (Tomasik, 2016). Wild-animal deaths are often extraordinarily unpleasant, including such fates as starving to death, being eaten alive, or succumbing to parasite infection. Members of many species may have thousands or even tens of thousands of offspring at a time, and in order for the population to remain stable around the carrying capacity and avoid Malthusian crises, most of these offspring must die. Since they die before mating, it seems likely that their lives are quite short—and hence unlikely to contain enough pleasure to outweigh the pain of death.

Even if the overall population size at any point is the same, species with high birth rates and high death rates have more discrete individuals than species with low birth rates and low death rates, because the turnover rate is faster. For this reason, under many systems of population ethics, the suffering caused by high-turnover species outweighs the pleasure experienced by low-turnover species.

Invertebrates tend to have many times more offspring than than terrestrial vertebrates do (Pianka, 1970). However, many terrestrial vertebrates still have offspring at a rate well above replacement. For instance, a rat has ten to twelve offspring per litter (Ducommon, 2016). While that number is much smaller than the thousands of offspring an insect may have, it still means that, on average, eighty to ninety percent of rats will die before they get a chance to reproduce (and more if some rats have more than one litter). Moreover, aquatic vertebrates such as fish display a wide range of reproductive strategies, from investment in a single highly fit offspring to investment in many offspring (Pianka, 1970). People who care about invertebrates and aquatic vertebrates are likely to find the argument from life history more important, compared to people who care primarily or solely about terrestrial vertebrates. Therefore, even those who only care about certain animal groups should give the argument from life history some weight.


Theoretical arguments are valuable because they establish whether wild-animal quality of life is positive or negative. In general, most people who make theoretical arguments argue that wild-animal lives, as a whole, are not worth living. However, one limitation of these arguments is that they offer little insight into strategies for improving the lives of animals. In addition, while the argument explores one very important factor that affects the quality of wild-animal lives, it leaves other potentially important aspects unexplored.

It may be intractable to improve the lives of high-birth-rate species, particularly those that are quite small, such as insects. Many people who accept the theoretical arguments about wild-animal suffering support the destruction of wild-animal habitats for this reason, although many people also reject habitat destruction because they value ecosystem preservation for its own sake or are concerned about unexpected negative consequences.

The theoretical argument also points towards certain other lines of research that may be worth pursuing. For instance, it may be possible to encourage species to produce fewer offspring in which they invest more heavily, perhaps through making their environments more stable. Given how short the lives of most high-birth-rate species are, such strategies could, if tractable, yield results in a relatively short time.

Empirical Arguments


Among humans, the gold standard for measuring quality of life is considered to be self-report of how well an individual’s life maximizes their own values (McMillan, 2008: 185). Animals—like infants and nonverbal humans—cannot talk and so cannot report their quality of life (ibid: 193); thus, all assessments of animal quality of life are provisional and often unreliable.

The most obvious approach to solving this problem is to attempt to communicate with animals about their quality of life. In one landmark study, pigs were trained to press one lever following an injection of pentylenetetrazole, an anxiety-inducing drug, and to alternate levers following an injection of saline solution (Carey and Fry, 1995). The pigs were then exposed to a variety of anxiety-inducing and non-anxiety-inducing stimuli; they pressed one lever around anxiety-inducing stimuli and alternated levers around non-anxiety-inducing stimuli (ibid). This methodology may allow pigs to self-report their quality of life.

Pigs are exceptionally intelligent animals with complex behavior (Mendl, Held, and Byrne, 2010), which could mean that this self-report method—while useful for other intelligent animals such as corvids, cetaceans, elephants, and primates—is less useful for most animal species. Training animals to self-report may be inappropriate for normal wild-animal assessments, as the method is both labor-intensive and expensive. It may also habituate animals to humans, which can be dangerous for animals: habituated animals are more likely to expose themselves to dangers such as cars and hunters (Newsome & Rodger, 2013: 413; Orams, 2002/2006: 285). It is undesirable to cause harm to animals in the process of assessing their welfare.

Marian Dawkins (2008) has argued that improvements in animal welfare may be measured through the answers to two questions:

  1. Will it improve animal health?
  2. Will it give animals something they want?

A similar system operationalized animal welfare as “being fit and feeling good”—that is, the ability to sustain “health and vigor” across the lifespan, plus an absence of suffering and the presence of positive experiences like “comfort, companionship, and security” (Webster, Main, and Whay, 2004: S93-S94).

Dawkins argues that the common thread in the concept of suffering is that “[suffering states] are all states that are unpleasant enough that, if we could, we would endeavor to get out of them” (2008: 3). Positive reinforcers, like food or water, cause the animal to repeat the action that led to them, while negative reinforcers, like pain or a frightening stimulus, cause the animal to avoid the action that led to them (ibid: 3). Therefore, a behavioral definition of suffering is any emotional state caused by a negative reinforcer (ibid: 3). This definition allows one to objectively assess whether an animal is experiencing enjoyment, suffering or a neutral state by seeing whether the animal will work to obtain or avoid that state (ibid: 3). Furthermore, it may be possible to come up with a physiological and behavioral ‘profile’ of pleasure or pain in the animal (Mendl and Paul, 2004: S21). If the animal has a consistent physiological and behavioral reaction to being put in situations that, based on analogy to humans, probably elicit emotions of fear or happiness, then we can assume the animal is happy or afraid in all situations where they exhibit this physiological or behavioral reaction (ibid: S21). Developing such a profile would make assessing animal welfare easier, particularly in situations where a behaviorist analysis is expensive or otherwise intractable (ibid: S21).

However, given that animals may want things that are bad for their long-term health (such as to avoid a vaccine or to eat too many calories), it is not enough to simply assess an animal’s short-term pleasurable or painful experience. One must also assess the animal’s growth, health, and longevity, and incorporate those observations into the welfare assessment (Dawkins, 2008: 5-6; Webster, Main, and Whay, 2004: S94).

The advantage of this strategy is that it is based on the animal’s own self-assessment about its needs and wants, rather than on human assessments, which may be biased, self-serving, or anthropomorphic. Under this definition of suffering, in order to figure out whether animals in the wild are usually suffering, one must figure out how healthy they are and how much of their time they spend avoiding negative reinforcers (such as fear or distress) instead of seeking positive reinforcers (such as food or sex). While it may be difficult and undesirable to assess what wild animals want, for the reasons given above, it may be possible to draw conclusions from the numerous captive specimens of common wild animal species. Knowledge of zoo animals’ preferences may also apply to wild animals.

There are several ways used to figure out what an animal wants. A “choice test” offers the animal a choice between two or more options (Kirkden and Pajor, 2006: 34). For example, an animal may be given the choice between one room which has a bare floor and one room in which the floor is covered with straw. An “operant test” requires the animal to perform a certain activity, such as pressing a lever, to obtain a resource (ibid: 34-35).

Using a variant of the operant test, it may be possible to experimentally determine how important some activities are to animals by figuring out how high a ‘price’ they are willing to pay for them (Dawkins, 1990; Kirkden and Pajor, 2006: 35-37). For example, a researcher might place food behind a door. By attaching weights to the door, the researcher can determine how heavy a door the animal will push open to obtain a certain quantity of food, and compare this to how heavy a door the animal will push open to obtain a certain quantity of, say, painkillers or sedatives. To simplify the numerous experiments involved in determining a demand curve, one can instead determine the highest price an animal will pay for a particular resource (Fraser, 2008: 202-203). However, these economic models of animal motivational strength have never been empirically validated (Kirkden and Pajor, 2006: 40).

However, there are many limitations on this research. For example:

  • Animals may only be able to learn to work for a reward if the work led to the reward in the environment of evolutionary adaptedness (Fraser, 2008: 197).
  • They may fail to conceptualize the existence of rewards in their absence (Webster, 2008: 54-55).
  • They may have nonlinear demand curves, or may unduly prioritize short-term wants over long-term needs (Dawkins 1990).
  • They may be frightened by the unfamiliar test environment, producing non-representative results (Webster, 2008:54).
  • They may prefer situations they are familiar with, even if that leads to worse welfare (Fraser, 2008: 195-196).
  • They may demonstrate a preference for something the experimenter had not foreseen, such as the hallway between two experimental chambers (Webster, 2008: 53-54).
  • The times the experimenter offers the resource may not coincide with when the animal most wants the resource (Kirkden and Pajor, 2006: 34).
  • The animal’s preferences may shift depending on external variables such as the temperature (Kirkden and Pajor, 2006: 34).
  • Different resources may fulfill different needs, such as the animal preferring a large cage for exercising and a small cage for nesting (Kirkden and Pajor, 2006: 34).
  • The animal may be more easily satiated with one resource than another, even if the first resource is more strongly desired (Kirkden and Pajor, 2006: 35).
  • Animals may value resources differently depending on how deprived they have been of a particular resource (Kirkden and Pajor, 2006: 40-41).
  • Animals’ preferences may differ depending on whether they are socially isolated or in a group of other animals (Kirkden and Pajor, 2006: 42).
  • Animals may respond to aversive stimuli in certain ways (e.g. freezing) that make it difficult for them to perform operant behavior (Kirkden and Pajor, 2006: 39).

However, more careful experimental work can compensate for these flaws (Dawkins, 1990).For instance, one might only use animals that have been acclimatized to the test environment, or make sure to use work that relates obviously to the reward.

Criteria for Animal Quality of Life

Many people have invented criteria for measuring quality of life in animals, usually within the context of farmed animals. Many of these criteria may also be applied to wild animals.

One of the earliest attempts was the five freedoms created by the Farm Animal Welfare Council (FAWC) in Great Britain in the 1970s (FAWC, 2009). The five freedoms are a cornerstone of British governmental and industry policy and have been internationally influential (ibid). The freedoms are as follows (ibid):

  1. Freedom from hunger and thirst by ready access to water and a diet to maintain health and vigor.
  2. Freedom from discomfort by providing an appropriate environment.
  3. Freedom from pain, injury, or disease by prevention or rapid diagnosis and treatment.
  4. Freedom to express normal behavior by providing sufficient space, appropriate facilities, and appropriate company of the animal’s own kind.
  5. Freedom from fear and distress by ensuring conditions and treatment which avoid mental suffering.

It is important to note that #4 states “normal” rather than “natural”. Stress behaviors like stereotypy are natural in that animals evolved to do them, but they are considered abnormal behaviors which have an adverse effect on the animal’s welfare (FAWC, 2009). Similarly, in a wild-animal context, one may consider stressful situations like injury, famine or storms to be “natural” but not “normal”.

The Five Freedoms have been critiqued for their emphasis on avoiding suffering instead of actively ensuring a good life for the animal (FAWC, 2009). FAWC has responded by creating three categories: a good life, a life worth living, and a life not worth living (ibid). An animal with a life worth living has necessary, proportionate, and minimal suffering; its needs and certain wants (e.g. grooming, play) are provided for; and they may experience mutilation only if it is the lesser of two evils (ibid). Examples given of lives not worth living include having a debilitating and untreatable disease, a severe physical state like starvation or dehydration, or intense and chronic pain, fear, or distress (ibid). A good life is characterized as “disease controlled by the strictest measures and with minimal prevalence, normal behaviour, availability of environmental choices and harmless wants, a ban on most, if not all, mutilations, certain husbandry practices (including the manner of death) prescribed or forbidden, opportunities provided for an animal’s comfort, pleasure, interest, and confidence, and the highest standards of veterinary care” (FAWC, 2009: 16).


The Five Domains model, used to assess the effects of experiments on research animals, is based on the Five Freedoms (Mellor and Reid, 1994). However, the Five Domains frames the criteria not as freedoms animals ought to possess but as potential areas of welfare compromise (ibid). Interventions are graded based on their effects on each domain, as follows (ibid: 15-19):


  • Grade O: Fluids available in quantities that satisfy thirst. Food meets body maintenance requirements and the demands of pregnancy, lactation, exercise, growth, thermal challenge, recovery from illness/injury, etc.
  • Grade A: Water or food excesses or restrictions cause minor, readily reversed effects on physiological state, performance, or body condition.
  • Grade B: Water or food excesses or restrictions cause serious short-term or moderate long-term effects which are easily reversed once the animal resumes having a good-quality diet.
  • Grade C: Water or food excesses or restrictions cause levels of debility where euthanasia is appropriate.
  • Grade X: Water or food excesses or restrictions cause death.

Environmental challenge

  • Grade O: Environmental conditions do not require adaptive physical changes.
  • Grade A: Environmental conditions require adaptive physical changes but are within the animal’s homeostatic capacity.
  • Grade B: Environmental conditions elicit body responses beyond the animal’s physiological adaptive capacity, but untoward effects are reversed by restoration of benign conditions and, if necessary, therapeutic intervention.
  • Grade C: Environmental conditions cause levels of debility where euthanasia is appropriate.
  • Grade X: Environmental conditions cause death.
  1. Disease/injury/functional impairment
  • Grade O: No disease, injury, or functional impairment.
  • Grade A: Disease, injury, or functional impairment is within the animal’s homeostatic capacity with no or minor debility or incapacity.
  • Grade B: Moderate debility or incapacity from which complete recovery occurs spontaneously or can be easily effected therapeutically.
  • Grade C: Marked debility or incapacity where euthanasia is appropriate. Unpredictable death caused in a small proportion of animals.
  • Grade X: Extreme debility or incapacity. Unpleasant death.

Behavioral/interactive restriction

  • Grade O: Behavioral needs fully met.
  • Grade A: Behavioral needs not met in minor ways, such as a barren environment or mild and short-term restraint.
  • Grade B: Behavioral needs not met in marked short-term or moderate long-term ways, such as medium-term restrictions of instinctive behavior, unnaturally changing social groups, or treatment that results in reversible stereotypies.
  • Grade C: Behavioral needs not met in marked ways, such as marked and repeated noxious stimuli from which escape is impossible, marked alterations to perceptual or motor functions, or treatment that results in self-mutilation or unusual aggression.
  • Grade X: Behavioral needs severely not met, leading to psychotic-like behavior or agonistic interactions which result in severe injury or death, such as maternal deprivation, severe social isolation, or overcrowding.


  • Grade O: No anxiety, fear, pain, or distress.
  • Grade A: Minor discomfort or low-level apprehension or anxiety for short periods.
  • Grade B: Moderate anxiety, fear, pain, or distress for short periods, or minor discomfort or distress for long periods.
  • Grade C: Moderate anxiety, fear, pain, or distress for long periods, or marked anxiety, fear, pain or distress where the suffering ends before it becomes excessive and either the animal completely recovers or dies.
  • Grade X: Severe, inescapable, and unrelieved anxiety, fear, pain, or distress at or beyond the limits of reasonable endurance.

Normally, the fifth domain– anxiety, fear, pain and distress– should be used to determine the welfare effects of an intervention (Mellor and Reid, 1994). However, if the fifth domain is unknown or unusually low, the highest domain should be used (ibid).

The Five Domains model has been used in the past to assess pest control of wild vertebrates (Beausoleil and Mellor, 2015; Sharp and Saunders, 2011). It has several advantages for this purpose, including being systematic, comprehensive, versatile and easy to incorporate new information and information about intensity and duration into (Beausoleil and Mellor, 2015: 38-40). It permits users to rank interventions based on their welfare impact, to identify areas of uncertainty, and to prioritize which mitigation strategies have the largest benefit (ibid: 39-41). The Five Domains explicitly separate physical and affective impacts, while encouraging users to consider multiple kinds of affective impacts (ibid: 40).

However, the Five Domains model also has many limitations (Beausoleil and Mellor, 2015: 41). Some domains may be difficult to assess for certain interventions: for example, it’s hard to know how poisoning affects the animal’s level of hunger (ibid: 41). The Five Domains can give the illusion of more precision than really exists (ibid: 41). It also is no escape from the subjectivity involved in any assessment of animal welfare (ibid: 41-42).

Mellor (2016a, 2016b) articulates three criticisms of the Five Freedoms model. First, the Five Freedoms exclude many kinds of negative affect. Negative subjective experiences can be divided into two categories: sensations that register disruptions in the internal state of the animal and experiences that are related to an animal’s negative cognitive assessment of its life (Mellor 2016b). The former category includes thirst, hunger, pain, nausea, breathlessness, dizziness, debility, weakness, and sickness (ibid). The latter category includes anger, anxiety, fear, panic, boredom, depression, helplessness, frustration and loneliness (ibid). Most of these eighteen sensations are covered, if they are covered at all, in the overbroad category ‘distress’ (ibid).

Second, the Five Freedoms are often understood as “as ‘completely achievable’ and even as ‘rights’” (Mellor 2016a). The rising popularity of animal rights ethical theories and concomitant need for a definitive list of animal rights has caused some activists to use the Five Freedoms for a purpose they really weren’t intended for (Mellor 2016b). It is impossible for any animal to ever be truly free of hunger, thirst, and pain, because hunger, thirst and pain are needed to motivate the animal to eat, drink, and avoid injury (Mellor 2016a). Ideally, animals should not experience the extremes of these negative subjective experiences but should experience them within tolerable limits that promote life-sustaining behaviors (Mellor 2016b).

Third, the Five Freedoms only consider problems and negative experiences; they completely neglect the promotion of positive experiences, which is also necessary for positive welfare (Mellor 2016a). Ideally, animals would thrive, not merely survive (ibid). Thriving means living in an environment that balances predictability and unpredictability; is at a comfortable temperature; offers desirable sites for resting and voiding excrement; allows for enjoyable exploratory and food acquisition behaviors; contains a variety of food with pleasurable textures, smells, and tastes; and permits social bonding activities, including care of young, play and sex (ibid). Welfare assessment should focus on positive affective engagement, which includes both genetic or learned rewarding behaviors and the anticipation, goal achievement, and memory of success associated with those behaviors (Mellor 2016b). Positive affective engagement aligns with the animal’s positive experiences of agency, which means engaging in self-generated, voluntary, goal-oriented behaviors and being in control (ibid).

Mellor (2016a) proposes an alternative Five Provisions model:

  1. Good nutrition. The animal experiences minimal thirst and hunger and enjoys eating.
  2. Good environment. The animal experiences minimal discomfort and exposure and enjoys physical, thermal, and other comforts.
  3. Good health. The animal experiences minimal pain, nausea, breathlessness, etc. The animal enjoys strength, vigor, and physical activity.
  4. Appropriate behavior. The animal is engaged in pleasurable activities with a minimum of threats and unpleasant behavioral restrictions.
  5. Positive mental experiences. The animal experiences comfort, pleasure, confidence, interest and a sense of control.

Botreau et al. (2007) suggested twelve criteria divided into four categories:

Good feeding

  • Absence of prolonged hunger.
  • Absence of prolonged thirst.

Good housing

  • Comfort around resting.
  • Thermal comfort.
  • Ease of movement.

Good health

  • Absence of injuries.
  • Absence of disease.
  • Absence of pain induced by management procedures.

Appropriate behavior

  • Expression of social behavior.
  • Expression of other behaviors.
  • Good human-animal relationship.
  • Absence of general fear.

Aiming more generally, McMillan (2005) offered five elements which play a fundamental role in animals’ quality of life:

  1. Social relationships (in social animals)
  2. Mental stimulation (novelty, play, exploration, challenges)
  3. Health (absence of discomfort and impairments)
  4. Stress (an optimal level; neither too much nor too little)
  5. Control (sense of control over negative events in one’s life)

Fraser (2008) similarly considers animal welfare to have four core values:

  1. Maintain basic health (i.e. sufficient food, water, housing, and medical care)
  2. Reduce pain and distress (i.e. preventing injuries and diseases which cause fear or pain)
  3. Accommodate natural behaviors and affective states
  4. Natural elements in the environment (i.e. outdoor access, sunlight).

Fraser notes that his criteria may trade off against each other (2008: 224-226)- For instance, providing birds with food and warmth during winter (promoting #1) may keep them from migrating (worsening #3), while feeding milk to an animal whose parent is weaning it may prevent pain and distress (promoting #2) but also prevent the animal from developing a varied adult diet (worsening #1). He points out that many people only care about one aspect, but incorporating all four allows for moral pluralism (ibid: 230-232). The fourth criterion, Fraser acknowledges, is on worse empirical ground than the other three, as there has been little research to establish how much animals value natural elements and under what conditions, the effects of the outdoors on health, and to what extent natural environments accommodate natural behaviors (ibid: 251).

Webster argues that animal welfare involves answering three questions (2008: 6):

  1. Is the animal living in an environment consistent with that in which the species has evolved and to which the species is adapted? (Is the animal living a natural life?)
  2. Is the animal able to obtain normal growth, function, and good health, and to sustain fitness in adult life? (Is the animal healthy?)
  3. Is the animal experiencing a sense of mental satisfaction or, at least, freedom from mental distress? (Is the animal happy?)

The answers to these three questions should be accessible through triangulating behavioral, neurobiological, and physiological indicators of suffering (Webster, 20085: 43-45). Behavioral indicators include positive indicators, indicators of coping well with stress, and negative indicators (ibid: 41). Positive indicators are the expression of pleasure and a full range of normal behavior, which can be seen through play, rest, maintenance behavior, and social behavior (ibid: 41). Indicators of coping well with stress include maintenance behavior, startle responses, normal defense behavior (fight or flight), and avoidance of threats (ibid: 41). Negative indicators include stereotypies, self-injurious behavior, apathetic behavior, redirection of behaviors that cannot be performed normally, and body postures and movement indicative of pain or injury (ibid: 43). Although still in its infancy, the science of neurobiology may, combined with behavioral research, allow us to detect otherwise hidden sources of suffering (ibid: 60-62). Physiological indicators are discussed in more detail in the ‘physiology’ section.

One way of assessing the animal’s welfare is through assessing the presence of positive-emotion-associated behaviors (Boissy et al, 2007). Play is a particularly good indicator, because it both is intrinsically rewarding and tends not to occur if the animal is stressed or unhappy (ibid; 387-388). In social species, affiliative behavior such as grooming each other indicates a positive emotional state, although it may also be a response to stress (ibid; 388-389). Exploration is also an excellent indicator, both because it is itself rewarding and because animals tend not to explore if they are frightened (ibid: 389-390). Exploratory behavior, food-related anticipation, preferences, and reward-related operant behavior have been suggested as indicators of positive fish welfare (Martins et al, 2012: 31).

Taylor and Mills (2007) suggest the following domains for assessing companion animal quality of life:

Social/environmental indicators

  • Basic needs (access to food, water, rest area, health care)
  • Non-social environment (access to exercise, object-directed play, exploration opportunities, treats, enrichment; range size; level of restriction; location)
  • Social environment (quality, quantity, and stability of intraspecific and interspecific relationships)
  • Autonomy (independence, control)

Physical/psychological indicators

  • Physical health (sickness, diarrhea, injury, irritation, breathing difficulties, mobility, weight change, behavioral indicators of pain, parasite load, cortisol, immune function)
  • Activity (energy, play, sleep patterns, arousal)
  • Appetite (quality and quantity of food)
  • Self-care (grooming; avoiding injury, soiled living areas, and self-harm)
  • Sociability (interspecific and intraspecific social competence, interest, compliance, obedience, aggression)
  • Intelligence (trainability, cognitive function, perseverence, perception)
  • Temperament (demeanor, mood, emotional stability, compulsivity, introversion/extroversion)
  • Positive behavior (sunbathing, comfort behaviors, play, exploration, interest)
  • Negative behavior (anxiety, distress, abnormal behavior, fear, depression)

Swaisgood (2007) outlines ten motivational theories which have currency among animal-welfare researchers:

  1. Ethological needs: animals are happy when they can perform certain behaviors which typically lead to the acquisition of resources (swimming, digging, exploring, etc.).
  2. Information primacy: animals are happy when they can gather information by searching for hidden food or about novel objects
  3. Mimic nature: animals are happy when their lives are similar to what it would be in nature.
  4. Control/behavioral contingency/choice: animals are happy when their behavior has an effect on their environment and they have control over negative stimuli.
  5. Boredom: animals are happy when they have something to do (even if it isn’t the highly rewarding behaviors associated with the concept of ethological needs).
  6. Lack of sensory stimulation: animals are happy when they experience lots of sensory stimulation.
  7. Stress: animals are happy when they don’t experience chronic stress.
  8. Coping: animals are happy when they have behavioral coping mechanisms to cope with stressors.

The last two motivational theories are about stereotypy (restricted and repetitive behavior performed by mostly captive animals and extraordinarily uncommon in the wild) and are not relevant to this paper. Unlike previously discussed criteria, Swaisgood is explicitly outlining multiple theories of animal welfare, many of which are expected to be falsified after more careful research; therefore, it should not be interpreted as a checklist.

Yeates and Main (2008) discusses ways of assessing positive welfare. They divide positive welfare into three categories (ibid: 296). The first, “pleasures,” consists of enjoyable everyday sensations such as food, play, tactile pleasure, sleep, sex, exercise, and thermal comfort (ibid: 296-297). The second, “engagement,” consists of active involvement with the environment, including relationships with family members and conspecifics, curiosity about a variety of interesting and novel objects, confidence, and optimism (ibid: 296-297). The third, “realization,” consists of the animal meeting their own goals, such as reproduction and learning, and having space and opportunity (ibid: 296-297).

Because they were originally created to assess animals in captivity, these principles may overemphasize the flaws of captivity (i.e. preventing natural behavior) at the expense of flaws of the wild (i.e. predators). This concern is a serious issue for Swaisgood’s framework, since its purpose is to talk about issues facing captive breeding programs. Botreau et al (2007) includes two criteria, #8 and #11, which only make sense in a context in which animals are intensively managed by humans. Similarly, Taylor and Mills (2007) include several indicators, such as trainability, interspecific social competence, and quality of interspecific relationships, which are primarily relevant for companion animals.

Nevertheless, the criteria often overlap and often do apply to wild animals. Criteria for wild animal welfare can be divided broadly into physical criteria and psychological criteria. Physical criteria are likely to be similar for all species: absence of pain, disease, injury, hunger, thirst, excessive heat or cold, and physical impairment. Psychological criteria are likely to be different for different species: a solitary species will not have social needs, and a species that is insufficiently intelligent to have a sense of control over its life has no need for control. Nevertheless, psychological criteria may include any of the following: absence of fear, distress, or other negative affective states; social behavior; access to the outdoors, sunlight, and other aspects of nature; a sense of control over one’s life; mental stimulation through play or exploration; an optimal level of stress, and the ability to cope with it; the ability to meet ethological needs such as digging or swimming; and an appropriate level of sensory stimulation.

Assessment of wild animal welfare through these criteria may be difficult. Human presence may change the behavior of animals. For instance, researcher presence may reduce rates of predation, because the predators are not accustomed to humans (Isbell and Young, 1993). Many animals find the presence of humans stressful and are likely to change their behavior in response to this stress. Very small animals, which constitute the majority of wild species, may be difficult to observe.

The Naturalness Problem

Many people argue that wild animals have a high quality of life because their lives are natural. In a very simplistic form, this is just the naturalistic fallacy. However, in a more nuanced form, I think it is worthy of more examination.

“Natural behavior” is a commonly used phrase in criteria about domestic animal suffering. To a certain extent, this is no doubt because the creators of the criteria have what Fraser (2008) calls a Romantic view of nature: they think that nature is good for its own sake, and therefore domestic animals should be allowed to have as close to a natural life as possible. However, animals legitimately find a lot of natural behaviors rewarding: play behaviors, social behaviors, the fulfilling of ethological needs, and exploration. While some might argue that natural behaviors have no effect on welfare beyond their effect on the physical conditions of the animal, this does not appear to be true on farms: sows and birds with pre-built nests still engage in nest-building behavior, and calves fed milk from a bucket still desire to suckle (Wechsler, 2007). For this reason, “the behavior is natural” may be an acceptable heuristic for “this behavior is rewarding for the animal.”

However, the heuristic fails more for wild-animal suffering than it does for domestic-animal suffering. A lot of the harmful natural behavior is simply ruled out for domestic animals: one is unlikely to decide that the lions at the zoo should be allowed to predate on the giraffes, or that one’s chickens’ welfare would be improved by infecting them with worms. If we remove harmful natural behavior from consideration, then of course the remaining natural behavior is going to be positive.

Many changes in animal behavior are not necessarily symptomatic of worse welfare. As many animal-welfare scholars have argued, being chased by a predator is natural, but it certainly increases the animal’s suffering (Dawkins, 2008). Predator avoidance, coping with bad weather, sickness behavior, and aggression are all common in nature, but likely reduce welfare (Bracke and Hopster, 2006: 79) .Giraffes in captivity typically graze more and lie down more than wild giraffes, which is plausibly because grass is more available year-round and they don’t have to fear predators (Veasey, Waran, and Young, 1996: 16). Captive muskoxen and red deer are much less likely to die from rutting injuries or behavior, a behavior change that very plausibly improves wild animal welfare, even though it removes the ability of these animals to perform a natural behavior (ibid: 17). Conversely, for humans, televisions and air conditioning are unnatural but likely increase humans’ welfare (Bracke and Hopster, 2006: 79). One can imagine similarly unnatural experiences improving welfare for animals.

Many narrower definitions of natural behavior have been provided (Bracke and Hopster, 2006). Some researchers think that natural behavior is morally relevant if it is species-specific, such as grazing in cows and rooting in pigs (ibid: 78). However, many behaviors that are not species-specific, such as walking and turning around, are generally considered to be relevant to welfare, while many indicators of poor welfare, such as tail biting in pigs and feather pecking in poultry, are species-specific (ibid: 79).

Some researchers distinguish between behaviors primarily under external control and behaviors primarily under internal control (Hughes and Duncan, 1988). Behaviors triggered by external stimuli, such as escape behaviors or aggressive behaviors, are a result of the animal’s desire to create a certain state; if the state exists already (e.g. there is no one to flee from), the animal does not experience poor welfare (ibid). Conversely, animals may be motivated by internal stimuli to do certain behaviors, such as (in many species) nest-building or dust-bathing, even if they already have a nest and are already clean (ibid). Since these behaviors are internally rewarding, it is likely to present a welfare problem if the animal is prevented from performing them. (Some behaviors show a combination of internal and external control (ibid), such as sleep in humans.) However, other researchers argue that what matters is not whether the behavior is externally motivated but whether the need is frustrated or satisfied (Bracke and Hopster, 2006: 79).

Some researchers argue that all behavior performed because it is pleasurable should be considered “natural behavior” (Bracke and Hopster, 2006: 79). When a behavior itself is rewarding it is termed an ethological need (ibid: 79). This definition excludes both unpleasant behavior (such as sickness behavior and predator avoidance) and non-motivated reflex behaviors (ibid: 79-80). However, many behaviors performed by humans, such as playing video games, are clearly pleasurable and internally motivated but not natural (ibid: 80). It is conceivable, however, that humans enjoy these behaviors because they fulfill human ethological  needs: maybe some video games are fun because they fulfill human ethological needs for competition, exploration, and creativity.

Perhaps the best definition of natural behavior for purposes of wild-animal welfare assessment is the one provided in Bracke and Hopster (2006). Natural behavior is behavior that animals tend to perform under “a wide range of ‘ad libitum’ natural conditions, because it is pleasurable (i.e. positively motivated) and promotes biological functioning in the environment of evolutionary adaptedness (i.e. the environment in which the animal evolved) (Bracke and Hopster, 2006: 80). This definition includes behaviors such as eating, play, searching for food, mating, locomotion, nest-building, grooming, and rest (ibid: 80).

The study of natural behavior provides welfare-relevant information, because it is essentially a large-scale preference test (see the Self-Report section for discussion of preference tests) in what is often an environment relatively rich in choices (Bracke and Hopster, 2006: 83). For example, some species choose to organize themselves into social groups, while others do not (ibid: 83). Most animals spend much time searching for food; while this could be motivated by hunger, in experimental situations where they have abundant food, animals continue to seek out opportunities to forage (ibid: 83).

It is a mistake to assume that the wild allows animals to perform all their natural behavior. For example, the wild may not necessarily meet behavioral needs for novelty (Veasey, Waran, and Young, 1996: 17). Animals may be unable to mate because of the presence of a dominant animal (ibid: 17). In many species, most animals may not get a chance to perform many natural behaviors (for instance, they may die before they get a chance to engage in mating rituals). Further research should explore how often animals meet their behavioral needs in the wild.

Physiological Measurements

Some researchers argue that we will soon be able to look directly at the brain to figure out how an animal is feeling; however, our understanding of brains has yet to advance to that level (Paul, Harding, and Mendl, 2005). For that reason, we must use indirect measurements. The two systems most commonly studied are the hypothalamic-pituitary-adrenal (HPA) system and the sympathetic nervous system and adrenal medulla (SAM) system (Fraser, 2008: 123). The HPA and SAM systems prepare the body to face certain kinds of demands, and are therefore indicators of certain kinds of suffering (ibid: 123).

Acute Stress

Acute stress is an adaptive behavior which promotes fitness in the short term (Wingfield and Kitaysky, 2002). On a scale of a few seconds to minutes, acute stress brings energy to muscles in use, increases immune function, ceases reproductive physiology and behavior, lessens feeding and appetite, enhances substrate delivery to muscle via enhanced cardiovascular tone, heightens cognition, and increases cerebral perfusion rates and local cerebral glucose utilization (Sapolsky, Romero, and Munck, 2000). Acute activation of the HPA axis promotes a variety of adaptive forms of behavior, such as gluconeogenesis, suppression of reproductive behavior, immune system regulation, increased foraging behavior, escape behavior, night restfulness, and recovery from stressful events (Wingfield and Kitaysky, 2002). In general, acute stress suppresses currently unnecessary behavior and physiological functions and replaces them with behavior and physiological functions appropriate for an emergency (ibid).

HPA system activity is the standard way to measure stress in farmed and laboratory animals, mostly through measuring the levels of cortisol in bodily fluids (Mormede et al, 2007: 317). Cortisol is released during unpleasant events in both nonhuman and human animals, including physical pain, overcrowding, frustration, and fear (ibid: 320). In some species, blood cortisol is a poor indicator of level of distress, because the animal will produce its maximum level of cortisol in any stressful situation (ibid: 321). Cardiac response (i.e. increasing heart rate) is also a sensitive and precise method of measuring stress, and it can be measured through many methods, some more invasive than others (Tarlow and Blumstein, 2002: 441-442).

Injured red deer put down for humane reasons show higher levels of cortisol and β-endorphin than deer who are culled with a rifle, although lower levels than deer who are hunted with hounds (Bradshaw and Bateson, 2000). This suggests that culling is the least stressful way to die among those studied for red deer and that long hunts by human or non-human predators are the most stressful. This may not generalize to other animals, because red deer are a sedentary species that is likely to be particularly distressed by having to run for a long time (ibid). Nevertheless, research on the effects on red deer of being hunted with hounds provides an example of stress-related research being used to advance the welfare of wild animals.

Some scientists argue that our understanding of the nature of the HPA system is too primitive to reach any firm conclusions (Rushen, 1991). Activation of the HPA system may be a response to being tested or handled, which can make responses to stressors seem more similar than they actually are (Dawkins, 1998: 311). While the short-term “alarm” response is the same for all stressors, the medium-term adaptive response to the stressor is often different for different stressors, which means that stress markers do a poor job at assessing medium-term animal well-being (Webster 2008: 45-46).

The HPA system is a preparation for action, not an indicator of distress. Cortisol secretion is increased by exercise as well as enjoyable activities like nursing, mating, and exploration (Fraser, 2008: 120). In addition, the rise in stress hormones after a stressor may be smaller than the changes in stress hormones over a normal day as the animal gears up for activity and then prepares for rest (ibid: 120-121). The HPA system prepares the animal for short-term exertion and does not distinguish between, say, fleeing from a predator and running because the animal has just seen some delicious food. Conversely, animals that respond to predators by freezing may not experience activation of the HPA system under stress, perhaps because they do not need the increased energy and oxygen it provides (Dawkins, 1998: 313). Nevertheless, significant presence of stress hormones is a good indicator of pain and fear, particularly in situations one would naively expect to be painful or stressful (Fraser, 2008: 123).

It is important not to assume that situations that would be stressful for a human are stressful for an animal (Sapolsky, Romero, and Munck, 2000). For instance, king penguins fast for weeks on end as part of ordinary nesting behavior, and do not experience any rise in glucocorticoid levels (ibid). Animals may be less likely to experience stress in a situation if it is a normative life history stage (ibid). However, that generalization is not always true: for instance, many semelparous species such as salmon experience high glucocorticoid levels during mating (ibid).

Context affects responses to stressors. For instance, many animals may be less stressed if they perceive themselves as having control over the stressor (Romero, 2004: 249). Stress responses may increase if the animal’s situation is getting worse and decrease if it is getting better (ibid: 249-250).

Glucocorticoid levels often rise when an animal is not yet in a stressful situation but is in a situation that may turn stressful (Sapolsky, Romero, and Munck, 2000). In the wild, many stressors are very brief: for instance, the median amount of time a hyena chases a zebra is less than a minute, which is far less than the amount of time it takes for glucocorticoids to begin to have an effect on physiology (ibid). Therefore, it is adaptive to prepare when an animal predicts a stressor may come (ibid). Situations that predictably lead to stressors include the following: mating season in seasonal tournament species; having a mate and thus having to drive off other males in non-seasonal tournament species; escalating threats leading to dominance-related aggression in social species; hunger and failed hunts in carnivores; parturition, being at the edge of a social group and visible sickness and injury in prey species; and seasonal and climatic changes (ibid). If we predict that a stressor will lead to lower welfare, then situations that lead up to that stressor will also lead to lower welfare (ibid).

Chronic Stress

Chronic stress is typically defined as a dysregulation of the HPA axis or as allostatic or homeostatic overload (Dantzer et al, 2014). Over time, an animal acclimates to a particular chronic or repeated acute stressor (Romero, 2004: 252). However, the acclimation process enhances the HPA axis’s responses to novel stressors (ibid: 253).  Classic symptoms of chronic stress include higher baseline plasma levels of glucocorticoids (the family of hormones that includes cortisol), increased acute increases in plasma glucocorticoids following a stressor, and increased amount of time taken to return plasma glucocorticoids back to baseline (Dantzer et al, 2014). These symptoms are caused by a decrease in glucocorticoid receptors in key parts of the brain, such as the hypothalamus and the hippocampus (ibid). Classic chronic stress does not always occur in response to repeated or chronic stressors, especially if the stressor is very severe (Romero, 2004: 252). Instead, the animal’s glucocorticoid responses may either not change over time (no acclimation) or become chronically elevated, or the entire HPA axis may shut down, leaving the animal unable to mount an appropriate stress response (ibid: 252).

Some researchers argue that chronic stress presents significantly more welfare implications than acute stress (Sapolsky, 2004). Acute stress is “the salutary responses (be they mediating or suppressive) to noxious stimuli”, while chronic stress “occurs when the natural recovery phase to a noxious stimulus is prevented from occurring” (Sapolsky, Romero, and Munck, 2000). Chronically stressed individuals experience a larger cumulative exposure to glucocorticoids than do individuals who are not chronically stressed, which may induce various pathological effects (Romero, 2004). Chronic glucocorticoid elevation is linked to a variety of health consequences, including reproductive system inhibition, immune system suppression, severe protein loss from skeletal muscle, disruption of second cell messengers, neuronal cell death, and growth suppression (Wingfield and Kitaysky, 2002). The effect appears to be non-linear, with too-low and too-high levels of glucocorticoids both impairing survival (Crespi et al, 2013). In humans, chronic stress is correlated with stress-related diseases and psychological impacts like insomnia, but this is usually not the case with acute stress (Sapolsky, 2004). It is plausible that animals in general do not experience a high level of long-term distress from, say, fleeing a predator. While the situation is acutely stressful at the time, it may not have long-term negative consequences.

Chronic stress may be measured in a variety of ways. In chronically stressful situations, the animal will have a blood level of cortisol similar to a non-distressed animal, although the animal will still behave in a distressed way (Mormède et al, 2007: 321). However, repeated exposures to stressful events do change the functioning of the HPA system, which can be detected through ACTH stimulation tests (ibid: 321-322). Synthesis of ACTH, itself a hormone involved with cortisol synthesis (ibid: 319), is inhibited during chronically stressful situations (ibid: 321); injecting artificial ACTH allows the researcher to route around this change in the HPA system and see how stressed the animal really is.

Baseline glucocorticoid levels may be elevated in chronically stressed animals (Dantzer et al, 2014). These levels can be measured through obtaining blood samples within three minutes of capture, before the stress of capture has had time to elevate glucocorticoid levels (ibid). Salivary glucocorticoid levels in terrestrial vertebrates are believed to reflect baseline glucocorticoid levels, with a twenty-minute lag (ibid). In fish and perhaps other aquatic species, baseline glucocorticoid levels can be measured through determining the amount of glucocorticoids secreted into holding water (ibid). However, since obtaining holding water requires moving fish from one aquarium to another, it may in reality reflect stress-induced glucocorticoid levels instead (ibid). The magnitude of the stress response can be measured either through injecting ACTH (as discussed above) or through exposing the animal to a stressor and then testing glucocorticoids in the same way one tests for baseline glucocorticoids (ibid). Because these methods of measuring chronic stress require capture and restraint, they may not be useful for studying many wild animals (ibid).

The ability to terminate the stress response is tested through injections of dexamethasone, which binds to glucocorticoid receptors and decreases the synthesis of glucocorticoids in normally functioning systems (Dantzer et al, 2014). The duration and magnitude of exposure to glucocorticoids are both equally important, because the biological effect results from hormone-receptor interactions that occur over the course of the entire stress response (Romero, 2004: 250). Poorly regulated negative feedback is a factor in human depression and in problems related to being a subordinate animal in many species, and may cause health problems (ibid: 250-251).

Individuals who experience chronic stress experience higher cumulative exposure to glucocorticoids (Dantzer et al, 2014). Integrated measures of glucocorticoid levels reflect an average of blood glucocorticoids secreted, metabolized, and excreted over a certain duration, and thus are believed to be the most accurate measure of chronic stress (ibid). Effect sizes of human disturbances on chronic stress are consistently larger with integrated measures than with measures of baseline glucocorticoids (ibid). Fecal glucocorticoid metabolites are one of the least invasive methods of measuring chronic stress and can be obtained without capture for many species (ibid). Hair and feathers also contain glucocorticoid metabolites. However, the use of these methods is in its infancy and many methodological concerns have not been resolved (ibid).

Measurements of the effects of chronic stress include glucose levels, free fatty acid and hematocrit levels, immune response, reproductive hormone levels, oxidative stress, telomere shortening rates, physiological stress, and body mass (Dantzer et al, 2014). These measurements are rarely used by scientists studying chronic stress (ibid). Practical problems apply to many of these measurements: for instance, they may require blood draws (ibid). However, one advantage of these downstream measurements is that, even if they are abnormal for reasons unrelated to chronic stress, it may still reflect a welfare problem such as malnutrition.

There are several methodological issues with measuring chronic stress (Dantzer et al, 2014). All methods to measure chronic stress must be validated appropriately for each species, even if they work in another closely related species (ibid). Glucocorticoids fluctuate over the course of the day in all species and fluctuate seasonally in many species. Therefore, samples must either be collected in the same season and time of day, or season and time of day must be included in the analysis (ibid). Sex, age, and reproductive condition affect glucocorticoid levels, so samples must be taken from known animals (ibid). Individuals respond differently to chronic stress, and individual differences in response to stress may make a situation look like it isn’t chronically stressful when it really is (ibid). Fecal glucocorticoid metabolites in particular present several methodological issues: populations that eat different foods or in different seasons cannot be easily compared because these factors alter hormone metabolization and excretion; precipitation, temperature, and decomposition may affect fecal glucocorticoid metabolite levels; preserving the feces in any way other than freezing or failure to homogenize the feces may lead to inaccurate measurements; and it is difficult to compare absolute values across studies because using different antibodies in measuring fecal glucocorticoid metabolites may lead to different results (ibid).

More seriously, there is no consistent profile of chronic stress across species (Dickens and Romero, 2013). Stress-induced glucocorticoid level and HPA axis sensitivity (as measured through ACTH injections) tend to increase in chronically stressful situations, but the variation is so tremendous that it is essentially impossible to predict whether a chronically stressed animal will experience an increase, a decrease, or no change (ibid: 182). Baseline and integrated glucocorticoid measurements are slightly better, but still have considerable variability (ibid: 182-184). Baseline glucocorticoid levels consistently increase if the animal is stressed by lack of food availability, stressful social interactions, captivity, or restraint (ibid: 184). Density led to no change in baseline glucocorticoid levels, and predator exposure and other stressors had no consistent response (ibid: 185). Integrated glucocorticoid measurements appear not to decrease in chronically stressful situations, but may either increase or not change (ibid: 184). There is no consistent response to anthropogenic chronic stressors or to predation, but social stress consistently increases integrated glucocorticoid measurements (ibid: 187). Body weight does consistently decrease in chronically stressful situations (ibid: 182). Publication bias is a serious concern: since scientists expect chronically stressful situations to lead to increases in these measurements, studies that show a decrease or no change may be thrown out as ‘failed’ studies (ibid: 184). Unfortunately, “the overwhelming conclusion from these analyses is that a consistent, predictable, endocrine response to chronic stress, regardless of the protocol used to induce chronic stress, does not exist” (ibid: 188).

Some researchers have argued that wild animals experience acute stress, but generally do not experience chronic stress (Sapolsky, 2004; Boonstra, 2013). It is unclear to me, however, if this is true. Predation produces an ‘ecology of fear’, in which prey animals must alter all their behavior for fear of being eaten (Horta, 2010). (Note that Boonstra (2013)’s research suggests that predation does not cause chronic stress in many, perhaps most, species, including those who fall victim to an “ecology of fear”.) Many animals seem to lack control over important aspects of their lives (whether they will mate, have enough food, be eaten by a predator, etc.), which suggests they may develop learned helplessness, a common psychological outcome of chronically stressful situations in which the animal has little control (Sapolsky, 2004). However, chronic stress leads to far more negative health consequences than acute stress (ibid), and learned helplessness leads to significant fitness costs for animals in situations they can control. One might expect learned helplessness and stress-related health problems to have been selected against.

Several species of animals appear to experience chronic stress. Hares experience high levels of chronic stress when at high risk of predation (Boonstra, 1998). Songbirds experience intermediate levels of chronic stress when they don’t have enough food or they are at high risk of predation, and high levels when both are true (Clinchy et al, 2004). Across vertebrate taxa, exposure to human disturbances such as tourism, noise from machines, and habitat modification increases levels of chronic stress, with a small-to-medium effect size (Dantzer et al, 2014). Seabirds experience chronic stress if there is not sufficient food (Kitaysky, Piatt, and Wingfield, 2007). In captive leopards, fecal glucocorticoid metabolite levels correlated with stress behaviors like spending a lot of time sleeping and hiding, performing stereotypic pacing, and seeming ‘tense’, as well as with plausible welfare issues like being kept near predators, put on public display, or unable to climb (Wielebnowski et al, 2002). Low-status animals in social species often experience chronic stress due to their mistreatment by higher-status animals (Sapolsky, 2004). Socially stressed subordinate fish have a characteristic pattern of depressed food intake, aggressiveness, and locomotor activity, changes in skin coloration, and higher levels of plasma cortisol (Martins et al, 2012: 23). However, only some kinds of animals treat predators as a chronic stress: for instance, elk do not appear to have any chronic stress reaction to wolves (Boonstra, 2013). Chronic stress reactions to predation may be more common if predation is not constant and if the prey animal has an intermediate lifespan of a few years (ibid).

Baseline glucocorticoid levels are higher in species that evolutionarily favor increased reproductive allocation per unit of time, at the cost of growth (Crespi et al, 2012). However, it is unclear to me whether these species experience less chronic stress or are always chronically stressed.

Baseline and stress-induced glucocorticoid levels are typically highest during reproduction (Crespi et al, 2012). Increased baseline glucocorticoid levels may increase effort put into energetically expensive behaviors, such as mating or feeding young (ibid). Baseline glucocorticoid levels tend to be higher in birds who have a large brood or who are putting in more parental effort (Bonier et al, 2009: 212; Crespi et al, 2012). Therefore, chronic stress does not necessarily decrease reproductive fitness. However, it is unclear whether the animal experiences poor welfare. It may be that putting in a great deal of reproductive effort is stressful and reduces the animal’s welfare. However, it may also be that putting in a great deal of reproductive effort is rewarding, and thus chronic stress has the same lack of correlation with well-being that acute stress does.

Wild animals tend to have more sensitive stress responses than domesticated animals (Künzl and Sachser, 1999; Martin, 1978), perhaps suggesting that wild animals tend to be stressed easily and much of the time.

Theoretical Models of Acute and Chronic Stress


Allostasis is the process of achieving stability in the physiological processes necessary through life through change in response to change in environmental or other conditions (McEwen and Wingfield, 2003: 3).  Allostatic load is a term used to refer to the energetic requirements of an animal’s life (Wingfield and Ramenofsky, 2011:105). Allostatic load is an unsatisfactory indicator of welfare because it may be increased by positive situations such as mating, social behavior, and exploration (ibid: 105). The allostatic state is a sustained, altered activity level of a mediator that maintains allostasis, such as a glucocorticoid (McEwen and Wingfield, 2003: 3). Examples of allostatic states include hypertension, altered cortisol rhythms due to sleep deprivation or depression, and elevated cytokines and lowered cortisol in chronic fatigue syndrome (ibid: 3). The allostatic state can be maintained for limited periods if food intake or stored energy allows, particularly if it is a natural part of the animal’s life history (ibid: 3-4). An allostatic state generally leads to a higher allostatic load (ibid: 4).

There are two kinds of allostatic overload (McEwen and Wingfield, 2003: 4). Type 1 allostatic overload occurs when energy requirements exceed energy income and stored energy (ibid: 4). The animal may experience an event which increases the amount of energy they need to live day-to-day, such as bad weather or a change in habitat (ibid: 7). Animals are more vulnerable to type 1 allostatic overload in certain situations, such as social subordination or during winter (ibid: 8). Some animals consistently require a higher energy level than others to survive—for instance, if they have an injury that makes foraging difficult, or a high parasite load—which also increases the risk of type 1 allostatic overload (ibid: 8). Eventually, an emergency life history stage, discussed below, will be triggered (ibid: 8); if it cannot reduce allostatic load, serious pathological effects may result, and eventually the animal will die (ibid: 8). Type 1 allostatic overload can be assumed to reflect poor welfare, particularly if the emergency life history stage fails to reduce allostatic load.

Type 2 allostatic overload occurs when the animal is in allostatic overload but the animal has as much energy as it needs (McEwen and Wingfield, 2003: 4). Environmental conditions lead to a chronic allostatic state, but the energy needed never exceeds the energy in the environment (ibid: 9). An animal may enter type 2 allostatic overload when a predator enters its territory (ibid: 9). Semelparous species like salmon may experience type 2 allostatic overload during reproduction (ibid: 10). Males of lekking species– in which males aggregate and engage in competitive displays, the winner of which gets to breed– may incur injury, suppress their immune systems, and feed less while breeding, and die after a short period as a dominant male (ibid: 10). Some bird species accumulate so much allostatic load during migration that they never do it more than once (ibid: 10). Nevertheless, while type 2 allostatic overload is very common in humans and captive animals, it remains rare in the wild (ibid: 10). Type 2 allostatic overload probably reflects poor welfare, although it is unclear to me whether e.g. males of lekking species also experience poor welfare.

Emergency Life History Stage

The idea of allostatic overload is related to the idea of an emergency life history stage. Some animals experience an emergency life history stage when placed in a stressful situation (Wingfield and Ramenofsky, 2011); potential causes of emergency life history stages include reduced food, infection or injury, prolonged psychosocial stress, and prolonged inclement weather (ibid: 96). A typical emergency life history stage may involve the following steps (ibid: 100):

  1. Cessation of migratory, territorial, social, reproductive, and parenting behavior.
  2. Activation of emergency behavior, such as seeking refuge or leaving the source of the disturbance.
  3. Mobilization of stored energy reserves.
  4. Termination, returning to the original habitat or finding a new habitat, and recovery.

Emergency life history stages may also include gluconeogenesis, the loss of muscle mass, increased foraging behavior, and increased nocturnal restfulness (Wingfield et al, 1998). Emergency life history stages have a characteristic hormonal profile which includes many hormones linked to stress (Wingfield and Ramenofsky, 2011: 101-103). The adaptive value of emergency life history stages may provide an explanation for why animals experience chronic stress in situations in the environment of evolutionary adaptedness (Boonstra, 2013). Animals with particularly short breeding seasons will tend to avoid having emergency life history stages during the breeding season, even if they would have them outside of it (Wingfield and Kitaysky, 2002).

Animals may behave several different ways during an emergency life history stage (Wingfield and Kitaysky, 2002). For instance, some animals “take it”, seeking a refuge to ride out the crisis, either through finding a shelter or through acquiring some alternate, energy-conserving set of physiological and behavioral traits (ibid). Some such animals may go into torpor, try to accumulate extra fat reserves, or utilize stored fat reserves (ibid). Others “leave it”, abandoning their nests and territories altogether (ibid). Still other animals “take it at first then leave it”: for instance, kittiwakes spend more time foraging and less time guarding their nests when under stress, which increases the chance that their offspring will be eaten by a predator, at which point they will escape (ibid).

Interestingly, animals that typically experience more variability in food availability (for example) are less likely to enter an emergency life history stage during a period of short-term food shortage (Wingfield and Kitaysky, 2002). One might argue that this means that animals that evolved in more variable environments have better welfare. Alternately, these animals might have similar overall welfare: animals who evolved in more varied environments experience a small amount of suffering a lot of time, while animals who evolved in less varied environments experience less suffering but stronger suffering when they do.

Unfortunately, identifying animals in an emergency life history stage will not identify all cases of poor welfare. While it may be assumed that all animals in an emergency life history stage have poor welfare, animal species that have chronically poor welfare may not enter an emergency life history stage. In addition, Wingfield’s model does not address responses to short-term stressors such as predator attacks (Romero, Dickens, and Cyr, 2009: 378).

The Reactive Scope Model

A related model of chronic stress is the reactive scope model (Romero, Dickens, and Cyr, 2009). Physiological mediators allow the animal to maintain homeostasis in a particular system, such as the immune system, the HPA axis, the cardiovascular system, the central nervous system, or behavior (ibid: 378-379). For example, the physiological mediators for the cardiovascular system are heart rate, heart rate variability, and blood pressure (ibid: 379).

There are four ranges that physiological mediators can be in: predictive homeostasis, reactive homeostasis, homeostatic overload, and homeostatic failure (Romero, Dickens, and Cyr: 379-380). Predictive homeostasis is the normal range for the mediator, given seasonal and circadian variation; for the cardiovascular system, the predictive homeostasis range covers normal energy needs for the current stage of the animal’s life history (ibid: 379). Reactive homeostasis is an increase above predictive homeostasis that serves to maintain homeostasis: for instance, for the cardiovascular system, the reactive homeostasis range includes energy mobilization and fight-or-flight (ibid: 379). Homeostatic failure occurs when a physiological mediator falls below the normal reactive range: for instance, for the cardiovascular system, the homeostatic failure range includes hypotension, lethargy, and decreased survival (ibid: 379). Conversely, homeostatic overload occurs when a physiological mediator is in a range where it cannot be maintained without causing damage to the animal: for instance, for the cardiovascular system, the homeostatic overload range includes hypertension, myocardial infarction, and muscle breakdown (ibid: 379-380). (Homeostatic overload is analogous to allostatic overload (ibid: 380), discussed above.)

Going into the reactive homeostatic range has a cost, both in direct energy and in lost opportunities to perform other tasks such as basic tissue maintenance (Romero, Dickens, and Cyr: 381). The more often the animal goes into the reactive homeostatic range, the higher the cost they pay (ibid: 381). This “wear and tear” lowers the threshold before going into homeostatic overload (ibid: 381). Therefore, there are two ways to enter a state of homeostatic overload: either the concentration of the mediator extends beyond the normal scope, or the concentration of the mediator stays in the reactive homeostasis range for an extended period (ibid: 381).

The states of homeostatic overload and homeostatic failure clearly reflect poor welfare. However, it is unclear to me whether the reactive homeostatic range should be considered poor welfare as well. It seems to come down to a value judgment. Some might argue that there is little significant suffering when an animal faces challenges they evolved to face and can cope with. For instance, many humans are thrilled and rewarded by activities in their reactive homeostasis range, such as exercise, extreme sports, high-stakes decision-making, and public speaking. Many more humans might suffer while exercising or speaking in public, but not consider themselves particularly unhappy because they have to do those things on a regular basis. However, humans do tend to be unhappy if they’re in homeostatic overload or failure, perhaps because they are working a stressful job they can’t handle (homeostatic overload) or experiencing a depressive episode (homeostatic failure). If we extend this argument to animals, then the serious concern is when an animal is in homeostatic overload or homeostatic failure, facing a challenge with which they cannot cope.

Others might point out that a brief period of hunger or successfully fleeing a predator can be extremely unpleasant, even if they are short-lived. Humans in the developed world are far more likely to enter the reactive homeostatic range by choice (for instance, hunger due to fasting or dieting) than by necessity (hunger due to not being able to obtain food). They are safe from many stressors with are commonplace in the wild, such as having to flee for one’s life. Therefore, generalizing from the experiences of humans in the developed world may be inappropriate. The ability to successfully cope with a problem—for instance, foraging for food when one is hungry—does not necessarily imply that a being doesn’t suffer when they experience the problem.

It may also be that some instances of entering the reactive homeostatic range are negative and some are neutral or even positive. Relying on stress indicators alone may be inappropriate for assessing wild-animal welfare.

Miscellaneous Assessment Strategies

Qualitative Behavior Assessment

Qualitative assessments of animals’ behavior (i.e. “does this animal seem happy?”) have relatively strong inter-rater and intra-rater reliability (Wemelsfelder, 2001). These assessments are also relatively cheap and scalable. Observers can distinguish transport-naive and transport-habituated sheep (Wickham et al, 2012) and cows (Stockman et al, 2011), as well as between drugged and undrugged pigs (Rutherford et al, 2012); in each case, observers describe the animals with words one would generally expect (for instance, transport-naive sheep are described as anxious, and undrugged pigs are described as curious).

However, this may not be an accurate metric of happiness. Qualitative behavior assessments do not correlate reliably with other measures of animals’ welfare (Andreasen et al, 2013), although this may, of course, be due to a flaw in other measures of animals’ welfare. Observer bias affects qualitative observations: people who believe hens are from an organic farm tend to rate them as happier than people who believe hens are from a conventional farm (Tuttyens et al, 2014). Since many people believe wild animals are free and live at peace and in balance with the rest of nature, they may consider wild animals to be happy even when they aren’t. People may interpret behaviors common to a species as natural and therefore good, even when the behaviors probably indicate poor welfare. Qualitative observations may tend to underweight very painful events that don’t happen more than a few times in an animal’s life, particularly those that often result in death (e.g. being eaten alive). In particular, animals that die young will tend to be underrepresented. Prey animals may also conceal signs of poor welfare in order to avoid signaling weakness to predators: for instance, sheep don’t tend to make noise when they’re in pain (Broom, 2003). Moreover, whole-animal assessment studies tend to be of domestic animals—which may have been selected for human-legible emotions—and may thus be an inappropriate way to assess the welfare of wild animals.

Indicators of Distress

Animals may indicate distress in a variety of ways, including: distress calls, particularly separation distress calls (Fraser 2008: 148), alarm calls (ibid: 149) and calls of hunger from dependent offspring (ibid: 149); tonic immobility (ibid: 150); and vigilance, or standing in a position that allows the animal to scan the environment (ibid: 152). Animals may also indicate a positive state of welfare through “all’s well” or play signals (ibid: 159-160). While all of these ways of indicating distress or pleasure are linked to a specific emotion or trigger, and do not provide an overall assessment of the animal’s quality of life, they can be useful to determine (for instance) how much the animal’s mood is affected by fear of predators. Indicators of distress or pleasure are often species-specific, making it difficult to create an overall strategy for assessing diverse species’ pain or pleasure. Species only evolve indicators of distress or pleasure when it is adaptive to do so; as mentioned above, prey animals may refrain from behavior that shows weakness.

Opiate Administration

Rats in pain will self-administer more opiates, but take fewer opiates once the cause of their pain has been treated (Colpaert et al, 2001), suggesting that opiate self-administration is an accurate measurement of animal pain even in prey species.

Cognitive Biases

A very interesting strategy for studying wild animal suffering is use of cognitive biases. Starlings who have been removed from an enriched environment negatively interpret ambiguous stimuli, similar to depressed humans (Bateson and Matheson, 2007); so do rats in unpredictable and stressful conditions (Harding, et al, 2004). Similarly, rats defeated in a fight by another rat and then housed individually (both sources of suffering for rats) showed reduced ability to anticipate a sucrose reward (Von Frijtag et al., 2007). This suggests that one could assess the overall quality of an animal’s life by assessing how “pessimistic” they are.

The advantage of this strategy is that one could get a quick overview of how happy the animal is. In addition, an animal interpreting its environment based on its mood may be more likely to be conscious than one who does not. A wide variety of animal species may exhibit cognitive biases, such as bees (Bateson et al, 2011; Perry, Baciadonna, and Chittka, 2016; but see also Gifura, 2013 for a critique). However, the strategy doesn’t enlighten us about what exactly is making the animals’ lives worse. In addition, what causes animals to be “depressed” may not track perfectly with their quality of life: notably, the starlings were not any more likely to negatively interpret ambiguous stimuli if they had never been in an enriched cage in the first place. In addition, if an animal is captured in order to assess its cognitive biases, it may be depressed solely or primarily because it’s captured.

Mental Illness

Nonhuman animals may experience mental illness. For instance, chimpanzees experience PTSD and depression (Ferdowsian et al, 2011; Bradshaw et al, 2008). In the wild, about 3% of chimpanzees are depressed, while 0.5% have PTSD (Ferdowsian et al, 2011). By comparison, among American adults, 6.7% experience depression in any given year, and 3.5% experience PTSD (Anxiety and Depression Association of America, 2016). While this is evidence that chimpanzee lives are worth living, mental illness may reduce the fitness of an animal, causing them to die more quickly and thus not be available for sampling. It is also unknown how many other animal species experience trauma and mental illness. Elephants may do so (Shannon et al, 2013), but both chimpanzees and elephants are highly intelligent species with complex social structures, so less intelligent and complex species may not.

Fluctuating Asymmetry

Fluctuating asymmetry is a kind of asymmetry which results when the individual does not develop as expected (Tarlow and Blumstein, 2007: 433). Fluctuating asymmetry is commonly used as a measure of developmental instability, particularly because of stress (ibid: 433-434). Therefore, fluctuating asymmetry can be used as a method of measuring how stressful the animal’s environment is. It is very easy to get a large sample size for asymmetry, because you only have to capture the animals once (ibid: 434). However, measurement error is very common, (ibid:434) and some researchers suggest that the ecological significance of fluctuating asymmetry may be a product of selective reporting and publication bias (Palmer 2000:459-468).

Fish Welfare

In fish, there are a variety of useful behavioral indicators of welfare (Martins et al, 2012). Reduced feeding motivation (defined as the time a fish takes to begin feeding) and reduced feed intake both indicate reduced welfare (ibid: 21). Ventilatory activity, the flow of water ventilated over the gills in a unit of time, should be low to moderate (ibid: 21-23). While increased ventilatory activity can be an indicator of positive experiences, sustained high levels of ventilatory activity may indicate a carbon dioxide or oxygen imbalance, oxidative stress, or the presence of a stressor (ibid: 22-23). High levels of ventilatory activity may expose the fish to more contaminants, leading to health issues (ibid: 22). Number of skin lesions, scars, and wounds is an indicator of aggression levels among fish (ibid: 23). Levels of vasotocin and isotocin may be a neurohormonal indicator of aggression levels (ibid: 24).

Individual fish swimming ability can be measured through critical swimming speed, recovery tests, and muscular activity measured via electromyogram or physiological telemetry sensors (Martins et al, 2012: 24). What changes in fish swimming behavior mean for welfare is species- and context-dependent (ibid: 25). Both increased and decreased activity can indicate poor welfare (ibid: 25-26). It may be time-consuming to observe the swimming behavior of individual fish; however, group swimming behavior is also a sensitive indicator of welfare which may be affected by the behavior of as few as one fish (ibid: 26-27). Exposure to negative stimuli may lead to escape behavior (ibid: 27). “Tornados” or “hour glass behavior” of fish swimming quickly in circles indicate an acute stressor such as a predator (ibid: 27). However, similar behavior can also be a positive indicator of welfare and feeding motivation (ibid: 27).

Directions for Further Research

Indicators of Wild-Animal Suffering

The research cited in this paper is either about domestic and captive animal welfare or collected for some reason other than wild-animal welfare assessment (for instance, to understand the effects of stress on reproduction in wild animals). Further study is needed to develop wild-animal-specific indicators of suffering and well-being. In addition to providing empirical evidence about whether the average animal has a life worth living, developing these indicators would allow researchers to measure the effects of interventions intended to promote wild-animal welfare.

Most importantly, research needs to develop “iceberg indicators” which quickly indicate whether all is well (Wathes, 2010). It is time-consuming and resource-intensive to completely assess every aspect of an animal’s well-being (ibid). To know for sure whether an average member of a species has a life worth living, one would need a very large sample, which would make a complete assessment even longer and more resource-intensive. Many wild animals are not habituated to humans and will flee or experience extreme distress upon interaction with humans, and habituating them to humans can often lead to worse welfare for the animal (Orams, 2002/6: 285). Therefore, speedy and non-invasive iceberg indicators are more important for wild animals than they are for domestic animals. Commonly used iceberg indicators in farmed and laboratory animals include body condition, normal behavior, and alertness (Wathes, 2010). Other potential iceberg indicators could include qualitative assessment, indicators of chronic stress, cognitive biases in wild animals, and records of how painful animal deaths are.

Indicators of wild-animal suffering can be divided into two categories: cross-species and species-specific. Cross-species indicators can be applied to any species: for instance, hunger, parasite load, or disease. Species-specific indicators require knowledge of the species to assess: for instance, one might need to know what this species’s social needs are, how this species plays, what counts as a positive reinforcer for this species, and whether qualitative assessment by humans is a good indicator of quality of life for this particular species.

Assessments using cross-species indicators are significantly easier to implement. While most farm animals are members of only a small handful of species, there are many species in the wild, and it is intractable to gain an in-depth understanding of the happiness of each species. Even closely related species may have very different behavioral and physiological responses to stress (Mason, 2010: 716). On the other hand, relying solely on cross-species indicators may leave out important kinds of suffering or happiness, particularly psychological aspects, which often vary based on the species. Further research should explore both cross-species and species-specific indicators.

It may be a good idea to use captive animals for basic research into wild-animal welfare assessment, because animals in captivity are easily accessible and already habituated to humans. However, captive animals may be quite unlike non-captive animals in many important respects.

Indicator Species

There are estimated to be 8.7 million ±1.3 million eukaryotic species on Earth, most of which have never been described (Mora et al, 2011). For this reason, it would be impossible to study the welfare of each species individually. Nevertheless, it is vital to be able to prioritize species for interventions, determine which are the most important causes of wild-animal suffering, and check whether interventions improve welfare.

Analogously to the use of indicator species in conservation biology (Simberloff, 1998), it would be ideal to assess in depth the quality of life of an indicator species (chosen for commonality or typicality) and then generalize from that species to other species in its taxa. In conservation biology, indicator species have sometimes turned out to be very non-representative (ibid), and it seems likely that welfare biology would have a similar problem. For this reason, researchers and wildlife managers should research a variety of likely indicator species and select ones whose welfare seems to correlate well with other species’ welfare in a variety of different situations.


Methodological issues with regards to measuring stress, particularly chronic stress, must be resolved. These issues are discussed in depth in the “chronic stress” section.

It is currently unknown how many animals experience chronic stress. On the basis of little data, some argue that nearly all animals experience chronic stress (Horta, 2010), while others argue that almost none do (Sapolsky, 1994). While researching how common chronic stress is, researchers must avoid unwarranted anthropomorphism; not all situations which are chronically stressful to humans are chronically stressful to animals, and some situations which are not chronically stressful to humans are chronically stressful to animals.

It seems unclear how often acute stress indicates suffering. Some instances of acute stress are clearly pleasurable (sex), some are clearly painful (serious injury), and many are ambiguous. In addition, some researchers argue that forms of acute stress with which the animal is capable of coping should be considered an ordinary part of life, rather than suffering; others disagree. These questions will require philosophical work to resolve and are probably not answerable scientifically.


It would be interesting, although perhaps not urgent or important, to explore which species are capable of being trained to self-report their quality of life. However, this research should probably begin with domestic animals, rather than wild animals, since domestic-animal research is easier and more widely applicable.

Further research into which species can experience mental illness, as well as the rates of mental illness in those species, seems warranted.

It would be extraordinarily useful to develop a physiological and neurological profile of wild-animal suffering in indicator species, so that assessment could be done more cheaply and easily and without using potentially misleading proxy indicators such as acute stress. At this time, however, research has not advanced to the stage where this is possible.

Works Cited

Andreasen, S. N., Wemelsfelder, F., Sandøe, P., & Forkman, B. (2013). The correlation of Qualitative Behavior Assessments with Welfare Quality® protocol outcomes in on-farm welfare assessment of dairy cattle. Applied Animal Behaviour Science, 143(1), 9–17. https://doi.org/10.1016/j.applanim.2012.11.013

Anxiety and Depression Association of America. (2016). Facts & Statistics | Anxiety and Depression Association of America, ADAA. Retrieved from https://adaa.org/about-adaa/press-room/facts-statistics#.

Bateson, M., Desire, S., Gartside, S. E., & Wright, G. A. (2011). Agitated honeybees exhibit pessimistic cognitive biases. Current Biology: CB, 21(12), 1070–1073. https://doi.org/10.1016/j.cub.2011.05.017

Bateson, M., & Matheson, S. M. (2007). Performance on a categorisation task suggests that removal of environmental enrichment induces “pessimism” in captive European starlings (Sturnus vulgaris), 16(S). Retrieved from http://dx.doi.org/

Boissy, A., Manteuffel, G., Jensen, M. B., Moe, R. O., Spruijt, B., Keeling, L. J., … Aubert, A. (2007). Assessment of positive emotions in animals to improve their welfare. Physiology & Behavior, 92(3), 375–397. https://doi.org/10.1016/j.physbeh.2007.02.003

Bonier, F., Moore, I.T., Martin, P.R., & Robertson, R.J. The relationship between fitness and baseline glucocorticoids in a passerine bird. General and comparative endocrinology 163(1-2), 208-213. https://doi.org/10.1016/j.ygcen.2008.12.013

Boonstra, R. (2013). Reality as the leading cause of stress: rethinking the impact of chronic stress in nature. Functional Ecology, 27(1), 11–23. https://doi.org/10.1111/1365-2435.12008

Boonstra, R., Hik, D., Singleton, G. R., & Tinnikov, A. (1998). The impact of predator-induced stress on the snowshoe hare cycle. Ecological Monographs, 79, 317-394. Retrieved from https://era.library.ualberta.ca/files/8p58pd758/EM_79_1998_371.pdf

Botreau, R., Veissier, I., Butterworth, A., Bracke, M., & Keeling, L. J. (2007). Definition of criteria for overall assessment of animal welfare. Animal Welfare , 16, 225–228. Retrieved from http://s3.amazonaws.com/academia.edu.documents/43826052/Definition_of_criteria_for_overall_asses20160317-9280-1l2gvk5.pdf?AWSAccessKeyId=AKIAIWOWYYGZ2Y53UL3A&Expires=1485993648&Signature=My9Lyo4Z0Zz%2F%2BV5KYUwYFUkjHCk%3D&response-content-disposition=inline%3B%20filename%3DDefinition_of_criteria_for_overall_asses.pdf

Bracke, M. B., and Hopster, H. (2006). Assessing the importance of natural behavior for animal welfare. Journal of Agricultural and Environmental Ethics, 19(1), 77-89. https://doi.org/10.1007/s10806-005-4493-7

Bradshaw, E. L., & Bateson, P. (2000). Welfare Implications of Culling Red Deer (Cervus Elaphus). Animal Welfare , 9(1), 3–24. Retrieved from http://www.ingentaconnect.com/content/ufaw/aw/2000/00000009/00000001/art00002

Bradshaw, G. A., Capaldo, T., Lidner, L. & Grow, G. (2008). Building an inner sanctuary: complex PTSD in chimpanzees. Journal of Trauma & Dissociation, 9(1), 9-34. Retrieved from https://www.tandfonline.com/doi/pdf/10.1080/15299730802073619

Broom, D. M. (1988). The scientific assessment of animal welfare. Applied Animal Behaviour Science, 20(1), 5–19. https://doi.org/10.1016/0168-1591(88)90122-0

Broom, D. M. (1998). Welfare, Stress, and the Evolution of Feelings, 27, 371–403. https://doi.org/10.1016/S0065-3454(08)60369-1

Broom, D. M. (2003). Causes of poor welfare in large animals during transport. Veterinary Research Communications, 27 Suppl 1, 515–518. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/14535460

Carey, M. P., & Fry, J. P. (1995). Evaluation of animal welfare by the self-expression of an anxiety state. Laboratory Animals, 29(4), 370–379. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/8558818

Colpaert, F. C., Tarayre, J. P., Alliaga, M., Bruins Slot, L. A., Attal, N., & Koek, W. (2001). Opiate self-administration as a measure of chronic nociceptive pain in arthritic rats. Pain, 91(1-2), 33–45. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/11240076

Council, F. A. W. (2009). Farm animal welfare in Great Britain: Past, present, and future. Farm Animal Welfare Council. Retrieved from https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/319292/Farm_Animal_Welfare_in_Great_Britain_-_Past__Present_and_Future.pdf

Clinchy, M., Zanette, L., Boonstra, R., Wingfield, J. C., & Smith, J. N. (2004). Balancing food and predator pressure induces chronic stress in songbirds. Proceedings of the Royal Society of London B: Biological Sciences, 271(1556), 2473-2479.

Crespi, E. J., Williams, T. D., Jessop, T. S., and Delehanty, B. (2013). Life history and the ecology of stress: how do glucocorticoid hormones influence life history variation in animals? Functional Ecology 27(1), 93-106. Retrieved from https://www.sfu.ca/biology/wildberg/NewCWEPage/papers/CrespietalFunctEcol2013.pdf

Dantzer, B., Fletcher, Q. E., Boonstra, R., and Sherriff M.J. (2014) Measures of physiological stress: a transparent or opaque window into the status, management, and conservation of species? Conservation physiology, 2(1). Retrieved from http://conphys.oxfordjournals.org/content/2/1/cou023.full.

Dawkins, M. S. (1990). From an animal’s point of view: motivation, fitness, and animal welfare.  Behavioral and brain sciences, 13(01), 1-9. Retrieved from http://www.bib.uab.es/veter/expo/benestar/Behanvioral.pdf

Dawkins, M. S. (1998). Evolution and animal welfare. The Quarterly Review of Biology, 73(3) 305-328. Retrieved from http://users.ox.ac.uk/~snikwad/resources/Publication14.pdf

Dawkins, M. S. (2008). The science of animal suffering. Ethology, 114(10), 937-945. Retrieved from https://www.researchgate.net/profile/Marian_Dawkins/publication/227527400_Animal_Suffering_The_Science_of_Animal_Welfare/links/55c09d0708aed621de13cb75.pdf

Dickens M.J. and Romero, L. M. (2013). A consensus endocrine profile for chronically stressed wild animals does not exist. General and comparative endocrinology 191, 177-189. Retrieved from https://www.researchgate.net/profile/Molly_Dickens2/publication/244481043_A_consensus_endocrine_profile_for_chronically_stressed_wild_animals_does_not_exist/links/0deec53bad5c11396f000000.pdf

Ducommon, D. (2016). Rat Reproduction: Mating, Gestation, Birthing, and Growth. Retrieved from http://www.peteducation.com/article.cfm?c=18+1804&aid=889

Ferdowsian, H. R., Durham, D. L., Kimwele, C., Kranendonk, G., Otali, E., Akugizibwe, T., … & Johnson, C. M. (2011). Signs of mood and anxiety disorders in chimpanzees. PLoS One, 6(6), e19855. Retrieved from http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0019855

Fraser, D. (1995). Science, values, and animal welfare: exploring the ‘inextricable connection.’ http://animalstudiesrepository.org/cgi/viewcontent.cgi?article=1036&context=acwp_awap

Fraser, D. (2008). Understanding animal welfare: the science in its cultural context. Wiley-Blackwell.

Gifura, M. (2013). Cognition with few neurons: higher-order learning in insects. Trends in neuroscience, 36(5), 285-294. https://doi.org/10.1016/j.tins.2012.12.011

Harding, E. J., Paul, E. S., & Mendl, M. (2004). Animal behaviour: cognitive bias and affective state. Nature, 427(6972), 312. https://doi.org/10.1038/427312a

Horta, O. (2010). The Ethics of the Ecology of Fear against the Nonspeciesist Paradigm: A Shift in the Aims of Intervention in Nature. Between the Species: A Journal of Ethics, 13(10), 10. https://doi.org/10.15368/bts.2010v13n10.10

Hughes, B. O., & Duncan, I. J. H. (1988). The notion of ethological “need”, models of motivation and animal welfare. Animal Behaviour, 36(6), 1696–1707. https://doi.org/10.1016/S0003-3472(88)80110-6

Isbel, L. A., & Young, T. P. (1993). Human presence reduces predation in a free-ranging vervet monkey population in Kenya. Animal Behaviour, 45(6) 1233-1235. https://dx.doi.org/10.1006/anbe.1993.1145

Kirkden, R. D., and Pajor, E. A. (2006). Using preference, motivation and aversion tests to ask scientific questions about animals’ feelings. Applied Animal Behavioral Science, 100(1), 29-47. Retrieved from http://cmapspublic2.ihmc.us/rid=1GQXCS1XJ-ZNPNFY-GBN/Using%20preference.pdf

Kityasky, A. S., Piatt, J. F., and Wingfield, J. C. (2007). Stress hormones link food availability and population processes in seabirds. Marine ecology progress series 352, 245-258. Retrieved from http://www.int-res.com/articles/theme/m352p245.pdf.

Knierim, U., Carter, C. S., Fraser, D., Gärtner, K., Lutgendorf, S. K., Mineka, S., … Sachser, N. (2001). Good welfare: improving quality of life. In Coping with challenge: Welfare in animals including humans, Dahlem Workshop Report (Vol. 87, pp. 79–100).

Künzl, C., & Sachser, N. (1999). The behavioral endocrinology of domestication: a comparison between the domestic guinea pig (Cavia apereaf. porcellus) and its wild ancestor, the cavy (Cavia aperea). Hormones and Behavior, 35(1), 28-37.

Martin, J. T. (1978). Embryonic Pituitary Adrenal Axis, Behavior Development and Domestication in Birds. American Zoologist, 18(3), 489–499. https://doi.org/10.1093/icb/18.3.489

Martins, C. L., Galhardo, L., Noble, C., Damsgård, B., Spedicato, M. T., Zupa, W., Beauchaud, M., Kulczykowska, E., Massabuau, J.C., Carter, T., Planellas, S. R., and Kristiansen, T. (2012). Behavioural indicators of welfare in farmed fish. Fish physiology and biochemistry 38(1), 17-41.  https://doi.org/10.1007/s10695-011-9518-8

Mason, G. J., & Mendl, M. (1993). Why is there no simple way of Measuring Animal Welfare?, 2(4), 301–319.

McEwen, B.S., and Wingfield, J. C. (2003). The concept of allostasis in biology and biomedicine. Hormones and Behavior 43(1), 2-15. Retrieved from http://library.allanschore.com/docs/Allostasis03.pdf.

McMillan, F. D. (2005). The concept of quality of life in animals. Mental health and well-being in animals, 181-200.

Mellor, D. J. (2012). Animal emotions, behaviour and the promotion of positive welfare states. New Zealand Veterinary Journal, 60(1), 1–8. https://doi.org/10.1080/00480169.2011.619047

Mellor, D. J. (2016). Moving beyond the ‘Five Freedoms’ by updating the ‘Five Provisions’ and introducing aligned ‘Animal Welfare Aims.’ Animals 10(6), 59. http://www.mdpi.com/2076-2615/6/10/59/htm

Mellor, D. J. (2016). Updating animal welfare thinking: Moving beyond the ‘Five Freedoms’ towards ‘A Life Worth Living.’ Animals 6(3), 21. http://www.mdpi.com/2076-2615/6/3/21/htm

Mellor, D. J., and Reid, C. S. W. (1994). Concepts of animal well-being and predicting the impact of procedures on experimental animals. Improving the well-being of animals in the research environment, 3-18. http://org.uib.no/dyreavd/harm-benefit/Concepts%20of%20animal%20well-being%20and%20predicting.pdf

Mendl, M., Held, S., & Byrne, R. W. (2010). Pig cognition. Current Biology: CB, 20(18), R796–8. https://doi.org/10.1016/j.cub.2010.07.018

Mendl, M., & Paul, E. S. (2004). Consciousness, emotion and animal welfare: insights from cognitive science. Animal Welfare, 13(1), 17–25. Retrieved from http://www.ingentaconnect.com/content/ufaw/aw/2004/00000013/A00101s1/art00004

Mora, C., Tittensor, D. P., Adl, S., Simpson, A. G. B., Worm, B. (2011). How Many Species Are There on Earth and in the Ocean? PLoS Biology, 9(8), e1001127. Retrieved from http://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.1001127

Mormède, P., Andanson, S., Aupèrin, B., Beerda, B., Guèmenè, D., Malmkvist, J., … & Richard, S. (2007). Exploration of the hypothalamic-pituitary-adrenal function as a tool to evaluate animal welfare. Physiology & behavior, 92(3), 317-339. Retrieved from https://www.researchgate.net/profile/Isabelle_Veissier/publication/6568895_Exploration_of_the_hypothalamic-pituitary-adrenal_function_as_a_tool_to_evaluate_animal_welfare_Physiol_Behav/links/53d257f00cf220632f3c9218.pdf

Newsome, D., & Rodger, K. (2013). 33. Feeding of wildlife: an acceptable practice in ecotourism? International Handbook on Ecotourism, 436. Retrieved from https://books.google.com/books?hl=en&lr=&id=HfUBAQAAQBAJ&oi=fnd&pg=PA436&dq=+Feeding+of+wildlife:+an+acceptable+practice+in+ecotourism%3F&ots=RzoTA-NcUk&sig=wTboXz2lQb5645CbWQOZTvySGFw

Orams, M. B. (2002/6). Feeding wildlife as a tourism attraction: a review of issues and impacts. Tourism Management, 23(3), 281–293. https://doi.org/10.1016/S0261-5177(01)00080-2

Palmer, A. R. (2000). Quasi-replication and the contract of error: lessons from sex ratios, heritabilities, and fluctuating asymmetry. Annual Review of Ecology and Systematics, 31(1), 441-480.

Paul, E. S., Harding, E. J., & Mendl, M. (2005). Measuring emotional processes in animals: the utility of a cognitive approach. Neuroscience and Biobehavioral Reviews, 29(3), 469–491. https://doi.org/10.1016/j.neubiorev.2005.01.002

Perry, C. J., Baciadonna, L., & Chittka, L. (2016). Unexpected rewards induce dopamine-dependent positive-emotion-like state changes in bumblebees. Science, 353(6307), 1529-1531.

Pianka, E. R. 1970. The American Naturalist, 104(940), 592-597. Retrieved from https://www.researchgate.net/profile/Eric_Pianka/publication/275142242_R-Selection_and_K-Selection/links/5487214f0cf2ef34478ec22e.pdf

Reznick, D., Bryant, M. J., & Bashey, F. (2002). r- and K-Selection Revisited: The Role of Population Regulation in Life-History Evolution. Ecology, 83(6), 1509–1520. https://doi.org/10.2307/3071970

Romero, L. M. (2004.). Physiological stress in ecology: lessons from biomedical research. Trends in ecology and evolution 19(5), 249-255. Retrieved from https://www.researchgate.net/profile/L_Romero/publication/7080797_Physiological_stress_in_ecology_Lessons_from_biomedical_research/links/00b7d526feefccf883000000/Physiological-stress-in-ecology-Lessons-from-biomedical-research.pdf

Romero, L. M., Dickens, M. J., and Cyr, N. E. (2009). The reactive scope model—a new model integrating homeostasis, allostasis, and stress. Hormones and behavior, 55(3), 375-389. Retrieved from https://www.researchgate.net/profile/Molly_Dickens2/publication/222692297_The_Reactive_Scope_Model_-_a_new_model_integrating_homeostasis_allostasis_and_stress/links/0deec51548cd8038c4000000/The-Reactive-Scope-Model-a-new-model-integrating-homeostasis-allostasis-and-stress.pdf

Rushen, J. (1991). Problems associated with the interpretation of physiological data in the assessment of animal welfare. Applied Animal Behaviour Science, 28(4), 381–386. https://doi.org/10.1016/0168-1591(91)90170-3

Rutherford, K. M. D., Donald, R. D., Lawrence, A. B., & Wemelsfelder, F. (2012). Qualitative Behavioural Assessment of emotionality in pigs. Applied Animal Behaviour Science, 139(3-4), 218–224. https://doi.org/10.1016/j.applanim.2012.04.004

Sapolsky, R. M. (2004). Why zebras don’t get ulcers. New York: WH Freeman.

Sapolsky, R. M. Romero, L. M., and Munck, A. U. (2000). How do glucocorticoids influence stress responses? Integrating permissive, suppressive, stimulatory, and preparative actions. Endocrine Reviews, 21(1), 55-89. Retrieved from https://academic.oup.com/edrv/article-lookup/doi/10.1210/edrv.21.1.0389.

Scott, E. M., Nolan, A. M., Reid, J., & Wiseman-Orr, M. L. (2007). Can we really measure animal quality of life? Methodologies for measuring quality of life in people and other animals. Animal Welfare, 16(2), 17–24. Retrieved from http://www.ingentaconnect.com/content/ufaw/aw/2007/00000016/A00102s1/art00004

Shannon, G., Slotow, R., Durant, S. M., Sayialel, K. N., Poole, J., Moss, C., & McComb, K. (2013). Effects of social disruption in elephants persist decades after culling. Frontiers in zoology 10(1), 62. Retrieved from https://frontiersinzoology.biomedcentral.com/articles/10.1186/1742-9994-10-62.

Sharp, T. and Saunders, G. (2011). A model for assessing the relative humaneness of pest animal control methods. Department of Agriculture, Fisheries, and Forestry. Retrieved from http://citeseerx.ist.psu.edu/viewdoc/download?doi=

Simberloff, D. (1998). Flagships, umbrellas, and keystones: Is single-species management passé in the landscape era? Biological Conservation, 83(3), 247–257. https://doi.org/10.1016/S0006-3207(97)00081-5

Spruijt, B. M., van den Bos, R., & Pijlman, F. T. A. (2001). A concept of welfare based on reward evaluating mechanisms in the brain: anticipatory behaviour as an indicator for the state of reward systems. Applied Animal Behaviour Science, 72(2), 145–171. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11278033

Stockman, C. A., Collins, T., Barnes, A. L., & Miller, D. (2011). Qualitative behavioural assessment and quantitative physiological measurement of cattle naïve and habituated to road transport. Animal Production. Retrieved from http://www.publish.csiro.au/?paper=AN10122

Swaisgood, R. R. (2007). Current status and future directions of applied behavioral research for animal welfare and conservation. Applied Animal Behaviour Science, 102(3-4), 139–162. https://doi.org/10.1016/j.applanim.2006.05.027

Tarlow, E. M., & Blumstein, D. T. (2007). Evaluating methods to quantify anthropogenic stressors on animals, 102(3), 429–451. https://doi.org/10.1016/j.applanim.2006.05.040

Taylor, K. D., and Mills, D. S. (2007). Is quality of life a useful concept for companion animals? Animal Welfare, 16(2) 55-65.

Tomasik, B. (2015). The Importance of Wild-Animal Suffering. Relations, (3.2), 133–152. https://doi.org/10.7358/rela-2015-002-toma

Tuttyens, F. A. M., de Graaf, S., Heerkens, J. L., Jacobs, L., Nalon, E., Ott, S.,… & Ampe, B. (2014). Observer bias in animal behaviour research: can we believe what we score if we score what we believe? Animal behaviour, 90, 273-280. Retrieved from https://www.researchgate.net/profile/Frank_Tuyttens/publication/260759467_Observer_bias_in_animal_behaviour_research_Can_we_believe_what_we_score_if_we_score_what_we_believe/links/556aeebb08aefcb861d60b19.pdf

Veasey, J. S., Waran, N. K., & Young, R. J. (1996). On Comparing the Behaviour of Zoo Housed Animals with Wild Conspecifics as a Welfare Indicator, 5(1), 13–24.

Von Frijtag, J. C., Reijmers, L. G., Van der Harst, J. E., Leus, I. E., Van den Bos, R., & Spruijt, B. M. (2000). Defeat followed by individual housing results in long-term impaired reward- and cognition-related behaviours in rats. Behavioural Brain Research, 117(1-2), 137–146. https://doi.org/10.1016/S0166-4328(00)00300-4

Wathes, C. (2010). Lives worth living? The Veterinary record 166(15), 468. https://doi.org/10.1136/vr.c849

Webster, A., Main, D., & Whay, H. R. (2004). Welfare assessment: indices from clinical observation. Animal Welfare , 13(1), 93–98. Retrieved from http://www.ingentaconnect.com/content/ufaw/aw/2004/00000013/A00101s1/art00013

Webster, J. (2008). Animal welfare: limping towards Eden. John Wiley & Sons.

Wechsler, B. (2007). Normal behaviour as a basis for animal welfare assessment. Animal Welfare , 16(2), 107–110. Retrieved from http://www.ingentaconnect.com/content/ufaw/aw/2007/00000016/00000002/art00002

Wemelsfelder, F., Hunter, T. E., Mendl, M. T. & Lawrence, A. B. (2001). Assessing the ‘whole animal’: a free choice profiling approach. Animal Behavior, 62(2), 209-220. Retrieved from http://animalstudiesrepository.org/cgi/viewcontent.cgi?article=1096&context=acwp_asie

Wickham, S. L., Collins, T., & Barnes, A. L. (2012). Qualitative behavioral assessment of transport-naïve and transport-habituated sheep. Journal of Animal. Retrieved from https://dl.sciencesocieties.org/publications/jas/abstracts/90/12/4523

Wielebnowski, N. (2002). Stress and distress: evaluating their impact for the well-being of zoo animals. Journal of the American Veterinary Medical Association, 223(7), 973-977. Retrieved from http://ebusiness.avma.org/Files/ProductDownloads/2002_the_welfare_of_zoo_animals.pdf#page=17

Wielebnowski, N. C., Fletchall, N., Carlstead, K., Busso, J. M., & Brown, J. L. (2002). Noninvasive assessment of adrenal activity associated with husbandry and behavioral factors in the North American clouded leopard population. Zoo Biology, 21(1), 77–98. https://doi.org/10.1002/zoo.10005

Wingfield, J. C. and Kitaysky, A. S. (2002). Endocrine Responses To Unpredictable Environmental Events: Stress or Anti-Stress Hormones? Integrative and Comparative Biology 42(3): 600-609. Retrieved from http://icb.oxfordjournals.org/content/42/3/600.full

Wingfield, J. C., Maney, D. L., Breuner, C. W., Jacobs, J. D., Lynn, S., Ramenofsky, M., & Richardson, R. D. (1998). Ecological Bases of Hormone–Behavior Interactions: The “Emergency Life History Stage.” American Zoologist, 38(1), 191-206. Retrieved from https://www.researchgate.net/profile/Creagh_Breuner/publication/31302841_Ecological_Bases_of_Hormone–Behavior_Interactions_The_Emergency_Life_History_Stage/links/564a170f08ae9cd9c826a017.pdf

Wingfield, J. C., Patrick Kelley, J., & Angelier, F. (2011). What are extreme environmental conditions and how do organisms cope with them? Current Zoology, 57(3), 363–374. https://doi.org/10.1093/czoolo/57.3.363

Wingfield, J. C., & Ramenofsky, M. (2011). Chapter 3 – Hormone-Behavior Interrelationships of Birds in Response to Weather. Advances in the Study of Behavior, 43, 93–188. https://doi.org/10.1016/B978-0-12-380896-7.00003-4

Yeates, J. W. and Main, D. C. (2008). Assessment of positive behavior: a review. The Veterinary Journal, 175(3), 293-300. https://doi.org/10.1016/j.tvjl.2007.05.009