Behaviour 125 (3-4) 1993, © E.
J.
Brill, Leiden
THE CONCEPT AND DEFINITION OF DOMINANCE IN
ANIMAL BEHAVIOUR
by
CARLOS DREWS1) (Department of Zoology, Cambridge University, Downing Street, Cambridge CB2 3EJ,
U.K.) (With 1 Figure) (Ace. 7-VII-1993)
Abstract
The concept of dominance has contributed greatly to our understanding of social structurc in animals. Over the past three decades, however, a variety of concepts and definitions of dominance have been introduced, leading to an ongoing debate about thc usefulness and mcaning of the concept. Criticisms aimed at one definition of dominance do not neces sarilly apply to other definitions. Existing definitions can be structural or functional, refer to roles or to agonistic behaviour, regard dominance as a property of individuals or as an attribute of dyadic encounters, concentrate on aggression or on the lack of it, and be based either on theoretical constructs or on observable behaviour. Thirteen definitions of domi nance are reviewed, and their usefulness assessed with respect to their descriptive value. The predictive and explanatory values of definitions are specific to the questions asked in each particular study and are not considered as criteria to judge the usefulness of the dominance concept. By virtue of its high descriptive value, the original definition of dominance by SCHJELDERUPp-EBBE (1922, Z.Psychol. 88: 226-252) emergcd as the basis to formulate a structural definition with wide applicability and which reflects the essencc of the concept: Dominance is an attribute of the pattern of rcpcated, agonistic interactions between two individuals, characterized by a consistent outcome in favour of the same dyad member and a default yielding response of its opponent rather than escalation. The status of the consistent winner is dominant and that of the loser subordinate. Dominance status refers to dyads while dominance rank, high or low, refers to the position in a hierarchy and, thus, depends on group composition. Dominance is a relative measure and not an absolute property of individuals. The discussion includes reference to the heritability of dominancc, application of dominance to groups rather than individuals, and the role of individual recognition and memory during agonistic encounters.
Introduction Dominance is a keyword in most studies of social behaviour in gregarious animals and has contributed greatly to our understanding of social struc ture, Tables listing dominance studies in a diversity of animal taxa are 1) I thank S.K.
ELTRI~
G.W. NORTON, and P.C. LEE for valuable discussion.
284
CARLOS DREWS
provided by DEWSBURY (1982) and GAUTHREAUX (1978). Despite the fact that dominance is such an important and widely used concept, there is still no agreement regarding its meaning. SCH]ELDERUPP-EBBE (1922) introduced the concept of the peck-order, later to be called dominance, into the behavioural sciences. He vividly described his observations on the peck-order of domestic fowl, the dynamics of the hierarchy, and the determinants of rank. \VILSON (1975) summarized different types of domi nance hierarchies found in animals. HArm (1986) provided an ample discussion of the different categories found among dominance relation ships and rigorously defined the terms commonly used in studies of dominance. Other conceptual discussions about dominance can be found in VAN KREVELD (1970), FEDIGAN (1982) and DUNBAR (1988). A lively debate has underlined the application of dominance in behav ioural studies over the past three decades (e.g. BERNSTEIN, 1981). GARTLAN (1968) criticized the different interpretations of dominance because these often led to ambiguity. He further criticized the fact that the term dominance is often used without a definition or else is arbitrarily redefined to fit the findings. ROWELL (1974) pointed out that the defini tion of dominance is elusive. Investigators are far from a consensus on how to define and measure dominance (BERNSTEIN, 1981), leading to disparate opinions regarding its function and evolution. BERNSTEIN (1981) concentrated on primates but his review of the problems associated with the concept of dominance has general application. His article is accom panied by comments from various investigators and thus provides an overview of the dominance debate and lack of concensus at that time. One of the central problems of the dominance debate has been that critics have not always explicitly specified which definitions they advocate and which they criticize. This has kept the debate going since the criti cisms may have been justified with respect to a particular definition but not to others. In this paper, an overview of the diversi ty of definitions of dominance found in the animal behaviour literature is presented and the usefulness of each definition assessed with respect to furthering our understanding of social structure and behaviour. Elimination of some definitions and finding the common denominator of others lead to a synthesis illustrating the essence of dominance in behavioural studies. This synthesis is condensed in the definition of dominance proposed in the concluding section.
THE CONCEPT AND DEFINITION OF DOMINANCE
285
Categories of d.ornfnanc e definitions
The terms winner, loser, submissive, subordination and aggression are part of the specialized vocabulary surrounding the dominance concept. These terms are defined and used here as follows: Winner: the contestant that expresses consummatory behaviour according to its initial goal (BAENNINGER,
1981;
HAND,
1986). Winner/loser refers to the outcome of one single contest.
Submissive: submissiveness is an active response to an aggressive action by another individ
ual (HANBY, 1976) in which yielding or surrendering are displays given by the loser 1986). Submissive behaviour refers to single contests. Subordination: to lose consistently certain conflict encounters refers to a pattern based on several contests.
(HAND,
(HAND,
1986). Subordination
Aggression: actual attacks, threats of attack or encroachments; signals which suggest that
attack may occur can be called aggressive
(HAND,
1986).
Dominance definitions can be separated into those based on empirical observations and definitions based on theoretical constructs (GAGE, 1981, HINDE & DATTA, 1981). These have been referred to as "data-language" and "theory-language" definitions respectively by HINDE & DATTA (1981). GAGE (1981) distinguishes two approaches to the study of domi nance: in the first, the investigator formulates in theory-language the conditions which must be met by a useful application of the term domi nance, then derives a testable hypothesis and lets the results of empirical inquiry support the definition of dominance or else lead to its rejection. In the second approach, the investigator assumes the existence of domi nance, asserts an operational definition, and proceeds to demonstrate the utility of the term and definition. This latter approach is similar to the data-language definitions mentioned by HINDE & DATTA (1981), in which a particular behavioural interaction is given the name "dominance inter action". More broadly, dominance is used to describe a particular kind or set of observable interactions. The dominance definitions based on empirical observations can in turn be subdivided into structural and functional ones (BERNSTEIN, 1981). Structural dominance definitions describe the pattern of an observable type or set of interactions without referring to their function, whereas functional definitions describe domi nance from the point of view of its apparent function. The distinction between definitions based on empirical observations and theoretical postulates, as well as between structural and functional definitions is used in this discussion to broadly characterize the definitions
286
CARLOS DREWS
of dominance prescnted below. A further distinction between definitions is whether they describe agonistic or non-agonistic dominance relation ships (BERNSTEIN, 1981). First, the definitions of dominance based on empirical observations are presented. These are ordered from broad definitions to more specific ones and, as such, they are not mutually exclusive. Then, a definition based on theoretical constructs is given. Some definitions are included which are not explicitly referred to in the literature, but which help in the process of categorization. Coxrrrrron definitions of d omdnanoe Definitions based on empirical observations. 1. "Privileged role": animals in a role which subjectively represents a privileged position with respect to others are dominant. This definition is illustrated by WILSON'S (1975) description of the dominance hierarchy of food transfer from forager bees (subordinates) to nurses (dominant). There is no overt aggression in the relationship and the subordinate bees acquire dominant status as they grow older. This is a non-agonistic dominance definition. The remaining definitions include agonistic behaviour as part of the dominance concept. First, definitions (2.-4.) are listed which consider dominance to be an attribute of the individual. In subsequent definitions (5.-13.) dominance is considered a relative measure, an attribute of dyadic encounters or relationships and not a property of individuals. 2. "Reproductive status": reproductive dominance in social insects refers to differences in reproductive status such that the reproductive individual is the dominant and the non-reproductive ones are subordinates (WILSON, 1975). This structural definition could be regarded as a case of the "privileged role" definition. The usc of overt aggression by the reproduc tivcs, howcvcr,justifies a separate mcnt ion of this definition. Wh~ll two or more paper wasp (Polistes sp.) females start a nest together, one becomes the egg layer while the others become workers. The queen establishes her dominant position and controls the other wasps by direct aggression (\VEST, 1967; EBERHARD, 1969). Here, dominance refers to the reproduc tive condition achieved and not to the pattern of aggressive interactions.
THE CONCEPT AND DEFINITION OF DOMINANCE
287
If, however, the investigator bases dominance status on the pattern of aggression, then reproductive dominance is a functional definition derived from the "dominance is aggressiveness" definition (see below).
3. "Dominance is aggressiveness": a dominance hierarchy is the set of sustained aggressive-submissive relations among a group of animals which coexist within one territory (WILSON, 1975). WILSON (1975) is concerned with " ... dominance behaviour mediated by aggression and inferentially based upon natural selection at the level of the individual. .. ". The occurrence of aggressive contests over resources among any social group of animals is indicative of dominance relationships. The expression of the dominance system IS based on individual variation III aggreSSIveness. 4. "Dominance IS a trait that conveys rank": a dominant individual routinely wins encounters with significantly greater than chance proba bility against a variety of conspecific opponents. This occurs because that individual has more of a trait called dominance (or aggreSSIveness or assertiveness) than do the losers (BAENNINGER, 1981). 5. "Winner is dominant, loser is subordinate": dominance status IS a synonym for winner and loser used to describe the outcome of any single dyadic encounter, regardless of whether or not the contest is escalated. 6. "Successful combatant": dominance ranks are established on the basis of dyadic contests resembling a competition. Generally each dyad meets on one or few occasions in which the decisive contest takes place. CLUT TON-BROCK et at. (1979) established the dominance rankings of male red deer on the basis of fighting success in the rutting season. Similarly, the ranks of elephant seals are determined through the outcome of escalated fights in the breeding season (LEBOEUF, 1974). The observed pattern of interactions resembles HAND'S (1986) aggressive dominance definition, in which the winner uses or seems willing to use the most aggression. Dominance scores based on overall fighting success belong to this defini tion (e.g. BOYD & SILK, 1983, ENS et al., 1990). Another example are lekking species, in which the contest is sometimes decided on the basis of acoustic or visual displays and may include escalated fights. This defini
288
CARLOS DREWS
tion neither implies that escalation is avoided nor that individuals recog nize each other and incorporate past agonistic experiences into contest decisions. 7. "Dominance IS lack of aggressiveness": dominance refers to the "peaceful" nature of encounters between animals, in which escalated fights do not take place and conflicts are resolved using non-agonistic assessment or submissive behaviour. VESSEY (1981) advocates a use of the term dominance that conveys its likely function, namely control of resources and minimum energy expenditure through the use of displays to control the behaviour of the receiver. Learning from past encounters and individual recognition are not necessary conditions under this definition. Typically, an individual recognizes a feature in the opponent which indicates superiority (e.g. size, weaponry, quality of display) and submits without a fight. High predictability of outcomes seems necessary under this definition of dominance. Dominance status is assigned after a single contest. 8. "Consistent winner of agonistic contests": when A consistently wins agonistic encounters against B then A is dominant and B subordinate. This definition is characterized by the asymmetry in the outcome of a series of contests of any type and implies that A and B meet repeatedly. It neither implies that high levels of aggression are avoided nor that individ ual recognition and memory of other contests play a role in contest decisions. 9. "Consistent winner in a given context": dominance is a descriptive term for individuals that consistently win dyadic encounters in a given context (HAND, 1986). A dominance relationship is described for each competitive context. Dyadic relationships can be pure or mixed, since they may exhibit different kinds of relationship for different conflict contexts (HAND, 1986). This definition emphasizes the role of context in determining the outcome of interactions (e.g. LEE, 1983). 10. "Priority of access to resources": dominance is by definition priority of access to resources (VAN KREVELD, 1970; MORSE, 1974; SYME, 1974; WILSON, 1975; GAUTHREAUX, 1978; WOLFE, 1984; WAGNER &
THE CONCEPT AND DEFINITION OF DOMINANCE
289
GAUTHREAUX, 1990). Por-r- & DEVORE (1979) define dominant/ subordi nate as the winner/loser in contests over resources. The winner is the contestant which displaces the other or appropriates the opponent's resource (e.g. CLUTTON-BROCK et al., 1979). Avoidance of escalation dur ing contests is not a requisite for dominance. It refers to obtaining what one wants by winning a fight or without having to fight for it (RALLS, 1976). The functional, central statement of dominance is that particular individuals in social groups have regular priority of access to resources in competitive situations (CLUTTON-BROCK & HARVEY, 1976). Dominance describes an asymmetry in the outcome of such contests. High ranking individuals can be identified by observing the outcome of contests over resources or by identifying the holders of resources directly (e.g. the male with the largest harem is by definition the alpha male). II. "Peck-order" after SCHJELDERUPp-EBBE (1922): if A pecks at Band B never or seldom reciprocates then A is dominant to Band B is subordi nate to A (SCHJELDERUPp-EBBE, 1922; ALLEE, 1938). A is the despot and B is typically afraid of it, adds SCHJELDERUPp-EBBE (1922). According to the original descriptions of the way in which dominance relationships develop (SCHJELDERUPp-EBBE, 1922; ALLEE, 1938), dominance implies that individuals recognize each other. Dyadic dominance status is assessed on the basis of one agonistic interaction type, namely pecking, and refers to its consistent, unidirectional pattern over time. 12. "Modified peck-order definition": a slight modification of the "peck order" definition is to include more than one agonistic interaction type in the assessment of dominance status. BARRETTE & VANDAL (1986) rephrased the "peck-order" definition of dominance summarizing its essence: dominance is an attribute of a relationship between two individ uals, whenever an asymmetry in the outcome of agonistic interactions is measured. If a dominant seldom has to fight to supplant a subordinate, the subordinate is repelled without a need to escalate, then a true domi nance relationship exists. HAND'S (1986) social dominance relationships between individuals who meet repeatedly corresponds to this definition. In agonistic dominance relationships the directionality of the agonistic encounters is not dependent upon location, as opposed to territoriality (KAUFMANN, 1983). In order to explain this dominance pattern it has been
290
CARLOS DREWS
argued that one member of the dyad in question consistently submits at the onset of any encounter as a function of the past history of interactions with the other member (BERNSTEIN, 1981). Consequently, it may be assumed that the subjects discriminate between different opponents. A consistent change in the response pattern of two individuals in agonistic conflicts, following their first contest, indicates that a relationship between the subjects has been established. The defeated member of the dyad submits or emits a terminating response in subsequent encounters with the other member without escalation. On the basis of such a change from symmetric to asymmetric initial responses in subsequent agonistic interactions, dyads with a dominance relationship can be differentiated from those without, since in the latter case each encounter is contested before yielding. The investigator should not assume that a dominance relationship exists in each dyad, because unresolved or egalitarian rela tionships may exist (BOYD & SILK, 1983; HAND, 1986). In summary, the "modified peck-order definition" is characterized by i) asymmetry in the outcome of diverse agonistic interactions, ii) avoidance of escalated encounters mediated by the subordinate's deferring behaviour, and iii) influence of past encounters on subsequent responses to opponents. The latter point implies that discrimination on the basis of individual identity is a necessary element of dominance relationships.
Dominance definition based on theoretical constructs. The theoretical approach to the study of dominance was introduced by HINDE (1978) and GAGE (1981) who postulate that dominance must represent a collection of observable actions, the validity of which can be empirically verified. The inter-correlation between the observables deter mines the definition and usefulness of the term or else justifies its rejection. 13. "Intervening variable": dominance is postulated in theory as an intervening variable between independent variables and a set of depen dent variables which are intercorrelated and covary equally in the major ity of dyads (HINDE, 1978; HINDE & DATTA, 1981). Experience, for exam ple, influences the direction of agonistic interactions via the intervening variable "dominance". A dominance/ sub ordinance relationship implies
THE CONCEPT AND DEFINITION OF DOMINANCE
291
one in which aggression is reduced or (overtly) absent. The pattern of asymmetry in the various interactions, including non-agonistic ones, may vary according to the situation. HINDE (1978) uses the term "agressive dominance" as a subcategory of dominance, to which other, non-agonis tic, interactions may be related. If aggression and non-agonistic behav iour are correlated, then both can be considered to be under the influence of dominance. The "intervening variable" definition is also used by BAENNINGER
(1981),
CANDLAND
&
HOER
(1981), and
GAGE
(1981), among
others.
General attributes of dOininance definitions
There is general agreement about some potentially confounding terms which should not be equated to dominance. Dominance does not imply leadership (ALLEE, 1938), and, thus, it is not to be confounded with "control" (HINDE, 1978): e.g. in baboon (Papio cynocephalusi consortships, the female may be controlling many aspects of the male's behaviour such as the direction of movement, while the male remams dominant to the female. The association between dominance and aggression differs widely between the different definitions of dominance (Table 1). Dominance is generally associated with conflict resolution during agonistic encounters. The "privileged role" definition of dominance is an exception, in which status assignement is independent of any agonistic encounters. Fig. 1 shows the minimum set of options encountered on the way to a definition of dominance which refers to agonistic behaviour. Each possible pathway in the flowchart leads to a different definition. 'When agonistic encounters are the basis to determine the directionality of the relationship, definitions of dominance differ with regard to the expression of overt aggression shown during conflict resolution. In some definitions overt aggression is the diagnostic feature of a dominance system. Other definitions do not discriminate between non-aggressive and escalated dominance interactions and pool all agonistic encounters to determine status. Yet other definitions explicitly limit the applicability of the dominance concept to those instances in which conflict resolution does not include overt aggression or escalated fights (Table 1). In "peck order" definitions, overt aggression by the dominant individual may be
292
CARLOS DREWS TABLE 1. Attributes of common definitions of dominance
Ernpdr-ic a! observation definitions
Attributes
Non-agonistic behaviour: Dominance as a role description: - "privileged role"
S, P, LD, NT
Agonistic behaviour as part of definition:
S/F, (AG), LD, NT
- "reproductive status"
Dominance as a property of the individual:
S, AG, LD, IH
- "dominance is aggressiveness"
- "dominance is a trait that conveys rank"
Dominance as an attribute of dyadic encounters:
- "winner is dominant, loser is subordinate"
- "successful combatant"
- "priority of access to resources"
- "dominance is lack of aggressiveness"
Consistent outcome of dyadic interactions: - "consistent winner of agonistic contests" - "consistent winner at a given context" Escalation usually avoided:
- original "peck-order"
- "peck-order" modified
this study ->
- essence of dominance
S, AG, LD, IH NT S, AG, LD S, AG, LD F, (AG), (ASY), HD S, AG, P, HD S, AG, ASY, HD S, AG, S, AG, HD S, AG, HD S, AG,
ASY, HD ASY, IR, P, ASY, IR, P, ASY, P, HD
Theoretical construct definition:
- "intervening variable"
AG, NI, HD/LD
S = structural definition, F = functional definition, AG = pattern of agonistic interactions reflects dominance, ASY = asymmetric outcome of contests within a dyad, TR = individual recognition involved, IH = dominance inheritable, NI =dominance not inheritable, P = "peaceful" (no overt aggression during conflict resolution), brackets ( ) = not a necessary condition. LD = low descriptive value, HD = high descriptive value.
the start of an interaction but contests are generally not escalated, given the default yielding response of the subordinate. These definitions are included among those in which conflict resolution is usually a "peaceful" event. While some definitions consider dominance to be an attribute of the individual ("dominance is aggressiveness" or "a trait that conveys rank" definitions), others view dominance as an attribute of dyadic relationships or of single interactions. Under definitions which do not regard domi nance as a relationship between individuals, dominance status within dyads is assigned on the basis of the outcome of a single contest or the subjects are ranked according to the ratio of individuals defeated to individuals lost to. Conversely, under those four definitions which regard dominance as an attribute of a relationship between two individuals,
THE CONCEPT AND DEFINITION OF DOMINANCE
293
IBASIS: AGONISTIC BEHAVIOUR ~
~
DEFINITION AS THEORETICAL CONSTRUCT
F1JNCTIONAL
STRUCTURAL
I
V
ESCALATED CONTESTS INCLUDED
OR EXCLUDED FROM DEFINITION
v
OUTCOME OF ONE CONTEST IS SUFFICIENT
v
CONSISTENT OUTCOME FROM SEVERAL CONTESTS BETWEEN SAME INDIVIDUALS
v CONTEXTS
EACH CONTEXT TREATED SEPARATELY
OBSERVED PATTERN OF INTERACTIONS CALLED DOMINANCE RELATIONSHIP
DOMINANCE DEFINITION
Fig. I. Flowchart illustrating points of decision when dominance definitions related to agonistic behaviour are formulated.
294
CARLOS DREWS
dominance status is assigned on the basis of the outcome of several conflicts, if a consistent asymmetry in favour of one dyad member is observed (Table I). The dyadic asymmetry in contest outcome is a possible, but not necessary element of the "priority of access to resources" definition. The asymmetry in the direction of the outcome of contests arises from one individual consistently winning and the other consistently losing consecutive encounters. Ideally, status assignment should only be made in those dyads in which the asymmetry is statistically significant over a set period of time. A dominance relationship is just one possible relationship between two individuals. There are unresolved and egalitarian relation ships in which there is no clear asymmetry in the outcome of contests (HAND, 1986). A dominance relationship between individuals, thus implies that the same individuals meet and resolve conflicts on several occasions (HINDE & STEVENSON-HINDE, 1976) in a clearly unidirectional pattern. The word "relationship" is ambiguous in the context of dominance definitions. A dominance relationship can be understood as a mental experience of the individuals concerned. This view has been criticized by ALTMANN (1981) and is associated with those definitions of dominance which imply that individual recognition operates during interactions (e.g. "peck-order" definitions). A simpler use of the term "relationship" refers strictly to data, to the interaction between the role of winner and loser among two individuals during a series of agonistic encounters. The latter use corresponds to the "consistent winner of agonistic encounters" and "the consistent winner at a given context" definitions. Some advocates of dominance as an attribute of dyadic relationships have argued that dominance presupposes individual recognition (e.g. WYNNE-EDWARDS, 1962; ROWELL, 1974; HINDE & STEVENSON-HINDE, 1976). Past conflicts may influence the outcome of future encounters (e.g. mice Mus sp.: GINSBURG & ALLEE, 1942; rhesus monkeys Maraca mulatta: ROSE et al., 1972, 1975). Individual recognition, however, is not a neces sary condition to explain those cases in which one individual consistently submits to another, familiar individual without an escalated fight taking place. At the onset of each conflict the opponents may assess one another on the basis of a morphological feature or display which correlates well with competitive ability (KODRIC-BROWN & BROWN, 1984) regardless of
----------
THE CONCEPT AND DEFINITION OF DOMINANCE
295
past experiences with the same opponent. This pattern of "peaceful" conflict resolution corresponds to the "dominance is lack of aggressive ness" definition. Individual recognition and memory of past encounters is an important ability when such predictive cues are not available and fighting ability can only be assessed in escalated contests or inferred from observation of the opponent's behaviour towards other, known members of the social group. In conclusion, individual recognition is not a necessary condition to produce the behavioural pattern observed under any of the dominance definitions, although advocates of the "peck-order" definition envision dominance as an identity relationship between two individuals. The hypothesis that individuals use individual recognition in their assessment of contest success has to be tested in each particular case.
Domdrrarrce and heritability
Dominance can only be inheritable when it is a property of individuals (Table 1). Several breeding studies claim to have selected particularly dominant individuals or to show that dominance status has an inheritable component (e.g. Nlo0RE, 1990; DEWSBURY, 1990; and references therein). An investigator can successfully breed for dominant individuals on the basis of the "dominance is aggressiveness" definition. Aggressiveness has been shown to have an inheritable component in several species (e.g. chickens, Gallus domesticus: GUHL et al., 1960; dogs, Canisfamiliaris: SCOTT & FULLER, 1965; sticklebacks, Gasterosteus aculeatus: BAKKER, 1986; sil vereyes, Zosterops lateralis: KIKKAWA et al., 1986; fruit flies, Drosophila melanogaster: HOFFMANN, 1988). On the other hand, breeding experiments suggest that there is a genetic component to the determinants of rank, when dominance is defined as an attribute of relationships (DEWSBURY, 1990). Explicit definition of dominance and careful formulation ofthe conclu sions of breeding experiments are essential to avoid misunderstandings about the heritability of dominance. Claims regarding the heritability of dominance traits are bound to evoke criticism by the advocates of domi nance as an attribute of conflict resolution and of dominance as a descrip tion of role. Dominance as an attribute of an interaction or of a relation ship between two individuals is a relative measure which cannot be
296
CARLOS DREWS
selected for (GARTLAN, 1968; ROWELL, 1974; BERNSTEIN, 1981; BARRETTE, 1987). Natural selection operates on absolute, competitive properties of individuals and not on relative dominance status. Dominance ranks depend on the group composition and are, therefore, not an inheritable property of an individual either (DUNBAR, 1988). Dominants and subordi nates behave differently, but these are not genetically distinct roles, or genetically fixed alternative strategies, since subordinates are fully capa ble of behaving as dominants given the appropriate circumstances for a change in status (MCGUIRE et al., 1984). Furthermore, most individuals in a hierarchy are simultaneously subordinate to some and dominant to other members of the group. The issue of the heritability of dominance illustrates how a given definition of dominance may have applications which are incompatible with other definitions of the term. Descriptive, predictive and explanatory value of dominance
One limitation of definitions based on observations rather than theory is that these are not open to empirical investigation. If another definition is proposed, there is no methodology to determine which of the definitions is appropriate and the question remains one of semantics rather than science (GAGE, 1981). Nevertheless, criteria such as the descriptive, pre dictive and explanatory value can be used to differentiate between definitions. The descriptive value of dominance. Dominance has a high descriptive value when its definition refers to something for which no other single term existed before and the term is useful when it can be used to differentiate that something from something else. In addition, dominance can be used as a descriptive tool to test the possibility of merging, and hence simplifying, categories of behaviour (CANDLAND & HOER, 1981). Thus, the descriptive value of a dominance definition is high if it serves efficiently the process of categorization. The descriptive value is an absolute property of a definition, irrespective of the theoretical context in which it is used. Contrary to some contentions (BERNSTEIN, 1981; HINDE & DATTA, 1981), the descriptive value of a dominance definition may justify keeping the definition without it being predictive or explanatory. Neither the predictive nor the explanatory
THE CONCEPT AND DEFINITION OF DOMINANCE
297
value are inherent absolute qualities of the definition but rather a func tion of the theoretical context in which the concept is used and of the questions asked. In assessing the usefulness of the dominance concept, one can first estimate its descriptive value, then test empirically whether the definition is supported by the data, and lastly assess the explanatory value of using the concept for a particular question. In the following definitions, dominance has a low descriptive value because it is used as a synonym of other, already existing terms, such as aggressiveness, winner, or an arbitrarily chosen status or role: "domi nance is aggressiveness", "winner is dominant, loser is subordinate", "successful combatant", "reproductive status", and "the privileged role" definition. The use of dominance following these definitions does not help to describe, predict or explain any phenomena better than do its syn onyms. SYME (1974) argued, with respect to context specific asymmetries in the outcome of agonistic interactions, that if each group has a large number of "dominance orders" then the concept is effectively useless. An asymmetry in the outcome of particular interactions is not a sufficient justification to introduce a dominance concept, either as a descriptive tool or an explanatory mechanism (DREWS, 1973; SYME, 1974). The "dominance is a trait that conveys rank" definition is used to collectively refer to all those behavioural characteristics of an individual which are decisive in the outcome of an agonistic conflict. Dominance in this sense has a high descriptive value. Dominance definitions which include consistency in the direction of the outcome of dyadic contests and/ or lack of aggressive escalation during conflict resolution, also have a high descriptive value in that they describe formerly unrecognized phenomena. In general, the functional definition of dominance, i.e. "priority of access to resources", describes the access to resources mediated by ago nistic contests. Por-e & DEVORE (1979) refer more specifically to the pattern of context specific access to resources. This definition equates dominance with non-egalitarian access to resources, and consequently has a high descriptive value since there was no other term to dennote that phenomenon. Dominance in theory language is a hypothetical construct to investi gate the common denominator of certain categories of behaviour (CAND LAND & HOER, 1981), and is similar to constructs such as "stress",
298
CARLOS DREWS
"motivation" and "drives" (MAXIM, 1981). Dominance is a useful con cept because it may enable simplifying our description: by using domi nance as an intervening variable between independent and dependent variables, we would reduce the number of links needed to describe the interaction between both sets of variables (HINDE, 1978). Economy in description is a first step in explanation (HINDE, 1978). When agonistic rank has a high predictive value in determining the outcome of other dyadic relationships, it is useful to abstract dominance as a description of some aspects of social structure (DEAG, 1977; HINDE, 1978). The descrip tive value of dominance as an intervening variable is a function of the inter-correlation between different interaction types and can be tested empirically. In female primates, for instance, well inter-correlated vari ables could be "supplants", "grooming", and "submissive", possibly with a weaker link to "priority of access" (HINDE & DATTA, 1981). We know, however, that there is lack of consistency even within agonistic categories since context is likely to influence the pattern observed (Por-r & DEVORE, 1979; LEE, 1983; HAt'lD, 1986). If the dependent variables are unlikely to be well correlated then, following GAGE (1981), the intervening variable definition can be rejected. The usefulness of dominance, however, does not depend on a correlation between dominance order and other aspects of social behaviour. Dominance is still useful in the description of a particular type of social structure, in which by some criterion of bossing related to aggression, the individuals can be arranged in a hierarchical order (HINDE, 1978).
The predictive value of dominance. Dominance has a high predictive value when the direction or outcome of a given interaction can be accurately predicted on the basis of past observations of dominance behaviour. Dominance is, however, a variable which remains dynamic and, thus, predictability needs neither to be perfect nor permanent; ontogeny alone guarantees that physical abilities vary (BERNSTEIN, 1981). "Prediction" (sensu BERNSTEIN, 1981) alludes to the degree of confidence with which the statistical asymmetry in dyadic contests predicts the outcome of any given subsequent conflict within the same dyad. Cardinal dominance ranks have been calculated on the basis of the probability of winning contests (BOYD & SILK, 1983). Hence,
THE CONCEPT AND DEFINITION OF DOMINANCE
299
dominance is useful as a shorthand term to indicate that the outcome of an agonistic interaction between two individuals is predictable with some practical level of certainty (ROWELL, 1974). "Prediction" can also be used to refer to the reliability with which the statistical asymmetry in dyadic contests predicts the direction of other, non-agonistic interactions outside the context of dominance. The predictive value of a dominance definition is only of interest, therefore, when the investigator uses dominance relationships to antici pate the behaviour of individuals. The level of accuracy of the prediction is not an attribute of the definition itself but of the behaviour of the study animals, and the usefulness of such a prediction depends on the questions asked in each particular case. Accurate prediction alone can be a heuris tic value of a dominance definition without it being explanatory. This approach has yielded contrasting results, for some authors claim that dominance does predict a wide variety of social interactions (e.g. RICH ARDS, 1974; SEYFARTH, 1976, 1980; CHENEY, 1978) and others point out the lack of correlation between different measures of dominance (GAR TLAN, 1968; SYME, 1974; BERNSTEIN, 1981; BRAIN, 1981). Variation in the definitions of dominance contributes to the discrepancy in such results.
The explanatory value of dominance.
It has been contended that the concept of dominance as an explanation of behaviour requires that it should actually predict priority of access to resources at least sometimes (RICHARDS, 1974; DEAG, 1978). This require ment, though, implies a priori that dominance influences the pattern of access to resources or else that priority of access to resources be part of the definition of dominance. Predictions about patterns of resource access, however, are not necessary for the concept of dominance still to be explanatory. Dominance has a high explanatory value when the introduc tion of the term and measurements of dominance reduce the magnitude of formerly unexplained variance in the data. The explanatory value arises from finding which variables depend on dominance since, in this way, dominance improves our understanding of social organization (BERNSTEIN, 1981). It is difficult, nevertheless, to define which is an independent measure of dominance and which are its dependent vari ables (BERNSTEIN, 1981). The explanatory value can be an inherent qual
300
CARLOS DREWS
ity of theoretical definitions of dominance (see next paragraph) but is not a necessary property of structural definitions. It is rather a function ofthe theoretical context in which the concept is used and of the questions asked by the investigator. With regard to theoretical definitions of dominance, HINDE (1978) argued that the concept of dominance as an intervening variable has explanatory value if the following conditions are satisfied: (I) the investi gation is concerned with more than one dyad, (2) each dyad has a multiplex relationship involving comparable complementary interactions in which one bosses the other, and (3) the pattern of inbalance in those interactions is similar in the different dyads. The usefulness of dominance increases with the variety of correlations among dependent variables (HINDE & DATTA, 1981). In contrast, CANDLAND & HOER (1981) argued that dominance as an intervening variable is a descriptive tool which serves the process of categorization and is not explanatory. Dominance as an intervening variable does not make any inferences about the direction of causality in the observed correlations among dependent variables. Moreover, dominance as an intervening variable is, by definition, not an empirical variable. Consequently, it cannot be incorporated into an empirical model to attempt to improve the fit between predictions and observations. Thus, dominance as an intervening variable can be a pow erful descriptive tool but is of low explanatory value with respect to questions about the nature of the observed behaviour.
The (in)adequacy of dOIninance definitions Much of the controversy around the concept of dominance has arisen from the variety of definitions and usages given to the term. Most of the definitions of dominance are structural and as such, they are all justified on semantic grounds (GAGE, 1981). Parsimony, however, dictates that there should be an upper practical limit to the number of new definitions created for the same term and that their justification should depend on their descriptive power. Ideally, for the concept to be more useful and to reduce ambiguity, all users of dominance should refer to one and the same phenomenon. In the following discussion I evaluate the adequacy of the dominance definitions listed above and synthesize the essence of a unitary dominance concept for behavioural studies. The criteria used to
THE CONCEPT AND DEFINITION OF DOMINANCE
301
categorize the definitions into appropriate and less appropriate are their respective descriptive values and their affinities to the original definition of dominance as described by SCH1ELDERUPp-EBBE (1922). The original definition is not only taken as a reference because it happens to be the first definition, but also because it has a high descriptive value and called our attention to a distinct behavioural phenomenon. It is noteworthy that in view of the controversy around dominance definitions, SADE (1981) suggested to look back at the early literature on dominance, which is clear and straightforward in the definitions and use of the concept (e.g. ALLEE, 1938; COLLIAS, 1944). Most definitions of dominance listed above, including the original "peck-order" definition, have in common that they refer to agonistic behaviour. Dominance should be reserved to its already widely estab lished use for patterns of interaction, in which the direction of comple mentarity is initially established through aggression, andf or in which aggressive behaviour is potentially present (HINDE, 1978). Its usefulness should not be diluted with other cases of complementarity in interactions, e.g. as referred to by WILSON (1975) with respect to food passage in honey bees. For this reason, as well as because of their low descriptive value, the "privileged role" and the "reproductive status" definitions of dominance are inadequate. Also, those definitions which refer to dominance as a synonym of aggressiveness, winnerfloser or "successful combatant" are inadequate on the basis oftheir low descriptive value. Dominance should not be defined as a synonym of any other already existing term. BAENNINGER (1981) understands dominance as a trait which combines several different intercorrelated variables (see HINDE, 1978), and explic itly defines dominance as a property of individuals. Similarly, WILSON'S (1975) mention of dominance as equivalent to aggressiveness implies that dominance is a variable, individual trait. Definitions of dominance as a trait of individuals ("dominance is aggressiveness", "dominance is a trait which conveys rank") are incompatible with those definitions which regard dominance as an attribute of dyadic encounters or relationships, including SCH1ELDERUPp-EBBE'S "peck-order" definition. Discontent with the "dominance is aggressiveness" definition has been emphasized by several authors. It is generally agreed that dominance refers to the patterning of interactions between individuals and not to the amount of aggressive behaviour that they show (e.g. JAY, 1965;
302
CARLOS DREWS
KAUFMANN, 1967; CHALMERS & ROWELL, 1971; ROWELL, 1974; SYME, 1974; BERNSTEIN, 1976; 1981; HINDE, 1978). Rather, aggression can be viewed as an expression of dominance (BERNSTEIN, 1976). Whilst domi nance may depend on aggression or potential aggressiveness, it does not necessarily correlate with them (LOCKWOOD, 1979; BERNSTEIN, 1981). In a principal factor analysis of wolf (Canis lupus) behaviour, the absolute level of agonistic behaviour was not part of the dominance factor detected (LOCKWOOD, 1979). The results of a dominance and communication analysis in children support the theoretical conception of dominance, that emphasizes the relationship between individuals rather than the coercive behaviour of some group members (CAMRAS, 1984). Although aggressiveness is known to vary between individuals (e.g. CUHL et al 1960), the levels of aggression observed in natural agonistic contests do not necessarilly reflect these differences. The game theoretical approach to the study of aggressive competition (e.g. PARKER, 1974; CARYL, 1980; MAYNARD-SMITH, 1982; ENQUIST & LEIMAR, 1990) suggests that all individuals have the potential to express aggression, but that the aggression observed is a circumstantial product of the cost/benefit ratio of using it. Hence, the directionality of aggression commonly used to assign dominance status is not necessarilly related to an aggressive prop erty of the individuals concerned, but rather to the fact that some individ uals defer upon being attacked because their chances of winning a given contest are low. A definition of dominance based on the outcome of several dyadic conflicts relates more to SCHJELDERUPp-EBBE'S (1922) "peck-order" definition than the "dominance IS aggressiveness" definition. In the previous paragraphs, those definitions of dominance were con sidered inadequate which were not related to agonistic interactions at all, had a low descriptive value because they referred to dominance as a synonym of an already existing term, and/ or described dominance as an inherent attribute of individuals rather than of interactions between them. Those definitions which refer to dominance as an attribute of conflict resolution patterns (Table I) are now discussed, bearing in mind that SCHJELDERUPp-EBBE'S definition is characterized by the asymmetry in the outcome of dyadic encounters and the fact that conflicts are generally resolved by the deference of the loser without escalation. The only defini tions which regard asymmetry and lack of escalation as necessary ele
THE CONCEPT AND DEFINITION OF DOMINANCE
303
ments of dominance are the original "peck-order" definition and the "modified peck-order definition". Definitions which consider one of these conditions as necessary include the "dominance is lack of aggressiveness" (lack of escalation), "consistent winner of agonistic contests" (asym metry), and the "consistent winner at given contexts" definition (asym metry). Asymmetry and lack of escalation mayor may not be given in the conflicts considered in the "priority of access to resources" definition, which, thus, departs more noticeably from the "peck-order" definition. The "priority of access to resources" definition of dominance has been subject to criticism (e.g. SVME, 1974; BERNSTEIN, 1981; BARRETTE & VAN DAL, 1986). Attention on advantages to dominants leads to frame a definition of dominance based on priority of access to incentives, while ignoring structure in favour of consequences (BERNSTEIN, 1981). Such functional definitions are problematic when different strategies yield similar benefits (examples in BERNSTEIN, 1981). If dominance is measured on access to only some, specific incentives it is then of no explanatory value to understand the principles of social organization. The use of the "priority of access to resources" definition is oflimited value to the study of dominance, because dominance more generally refers to all conflict resolution instances, while not all conflicts are over resources (HAND, 1986). Further, studies using the "priority of access to resources" definition of dominance sometimes claim to describe an effect of dominance on resource access, a circular statement which arises from measuring dominance on the basis of the pattern of resource access (e.g. GAUTHREAUX, 1978; SVME, 1974; ApPLEBV, 1980). This functional defini tion is, hence, inadequate to study whether high dominance rank per se is advantageous. The interpretation of an aggressive order must be restricted to aggressive behaviour unless it can be correlated with a quantitative demonstration of general priority of access to resources (SVME, 1974). In addition, when measuring dominance on the basis of access to incentives, there is sometimes the problem of identifiying what are incentives (BERNSTEIN, 1981). The constraints of the "priority of access to resources" definition illustrated above, nevertheless, do not make it less valid with respect to the "peck-order" definition. The "prior ity of access to resources" definition, however, is concerned with a particular kind of interaction, namely contests over resources, irrespec tive of a dyadic consistency in their outcome and of the degree of
304
CARLOS DREWS
escalation involved. For this reason, this definition is incompatible with respect the dominance concept of SCH]ELDERUPp-EBBE (1922). A similar argument applies to the "intervening variable" definition, which also lacks affinity with the original definition of dominance. Few authors have objected to the "peck-order" definition. HINDE & DATTA (1981) argued that if dominance/ subordinance is assessed in terms of only one dependent variable (here: pecking), then dominance is merely descriptive and has no explanatory value. Dominance would be merely a way of describing the data in other words. I argued above that the explanatory value is not a property of the definition but of the context in which it is used. Clearly, the original definition was concerned with only one kind of interaction, namely pecking, which would limit its appli cability to other taxa. The "modified peck-order definition" includes other complementary agonistic interactions in an analogous way to SCH]ELDERUPp-EBBE'S treatment of pecking. This modified definition also derives its high descriptive value from referring to dominance as a consis tent pattern in dyadic interactions whereby deference rather than esca lated fights are the rule, a phenomenon which deserves a description on its own. Memory and individual recognition are implicit in the "peck order" definition as proximate mechanisms to explain the deference behaviour. These mechanisms may apply in some cases but need not be necessary for the consistent deference behaviour to be observable. Mech anisms and function should not form part of the definition and used instead within hypotheses concerning the causation of dominance relationships.
The feature recognition criterion and dODlinance relationships
Broad definitions of dominance do not preclude the study of dominance across different sizes, sexes and age categories, and even between groups, populations or species (e.g. GAUTHREAUX, 1978). SEYFARTH (1981) argued, however, that it makes little functional sense to lump together all age-sex classes in studying dominance, because we know about the variety of different selective presures acting on each class. If the outcome of contests can be predicted by location (territoriality) or a relational attribute which correlates with the direction of outcomes (e.g. size, age, sex) then it is not useful to describe each of them as a special type of agonistic dominance
THE CONCEPT AND DEFINITION OF DOMINANCE
305
1981). Such a view tacitly implies that dominance relation ships exist as long as the investigator does not identify one single reliable predictor of contest outcome. From this perspective, dominance relation ships reflect the investigator's ignorance about the identity and interrela tion of determinants of conflict outcomes and dominance would cease to exist when these are understood. This example emphasizes the impor tance of explicitly specifying exactly how dominance is defined (or per ceived) by each investigator. However, if dominance is viewed as the consistent asymmetry in the outcome of contests between two individuals whereby the subordinate generally defers to the dominant without escala tion, then identifying the determinants of this pattern is a necessary step towards explaining it and no identified determinant should invalidate the phenomenon. In a rigorous application of the "peck-order" definition of dominance, memory of past encounters and individual recognition determine the behaviour in dominance dyads. Such a definition implies a relationship between each of two individuals and thus, it refers to dominance as a much more specific phenomenon than just asymmetrical outcomes in non-escalated contests within a dyad. When observed agonistic patterns are equivalent to those expected by a peck-order dominance, we need to know whether identity of the opponent is used to predict the outcome of a contest, before a peck-order dominance relationship is inferred. This question can be approached by witnessing the first contests between unfamiliar individuals and then observing the corresponding change in agonistic response over time. In stable social groups, changes in status suggest that identity recognition may be used by opponents instead of predictive morphological features when such features do not change as frequently as dominance status. ALTMANN (1981) regards dominance relationships as cognitive experi ences which should be reserved to the higher primates and humans. He argues that as long as it is not shown that individuals react to a relation ship, rather than to agonistic behaviour past or present or correlations thereof, parsimony dictates that relationships need not be inferred. If a general feature of the opponent (like sex or age) is used at the onset of each encounter to decide whether the contest should be escalated or avoided then past experience of the opponent and its identity would be irrelevant. The feature-recognition criterion is probably used during most (BERNSTEIN,
306
CARLOS DREWS
dominant/ submissive interactions between individuals which meet only once, as explained in the "dominance is lack of aggression" definition. BERNSTEIN (1981) regards the dominant individual as taking an active part in the dyadic dominance relationship when it acknowledges immedi ate submission rather than attacking. There is no need to postulate an identity relationship on that ground, if the dominant individual follows the simple rule of not attacking when the opponent submits immediately (a behavioural feature). Por-r- & DEVORE (1979) discuss the adaptive value of stopping further attack once the opponent has submitted, in contexts in which a relationship between the two individuals is irrelevant.
Individuals, sex, groups, or species: the units of do mfrrarrce The concept of dominance has been mostly utilized to characterize the behaviour of individuals. It has been also applied to the relationship between categories such as sexes, groups, and species. The working hypothesis in studies of intersexual dominance (e.g. RALLS, 1976; DESROCHERS et al., 1988) is that one sex may exclude the other from limiting resources. Such an exclusion is referred to as dominance, resem bling the "priority of access to resources" functional definition. Domi nance is not to be interpreted as the agent causing resource monopoliza tion because dominance is measured on the basis of the outcome of all agonistic encounters, most of which may be over resources. Dominance in this context is descriptive, not explanatory. Intergroup dominance is similar to the dominance relationships in dyads of familiar individuals, because in intergroup contests often the entire group behaves as a unit which is known to the opponent group from previous encounters. Intergroup dominance in wild rhesus monkeys has been described on the basis of supplants, avoidance, or fights between groups (SOUTHWICK et al., 1965). This usage of the term corresponds, at the group level, to the "consistent winner of agonistic contests" definition. Interspecific dominance after MORSE (1974) is the priority of access to resources that results from successful attacks, fights, chases or supplanting actions present or past. It represents one mechanism of interference competition. For example, gulls (Larus sp.) dominate curlews (Numenius arquata), while curlews dominate small waders (ENS et al., 1990). ~WAGNER
THE CONCEPT AND DEFINITION OF DOMINANCE
307
& GAUTHREAUX (1990) refer to MORSE'S (1974) definition as the definition
of dominance in general. GAUTHREAUX (1978) discussed the ecological significance of interspecific dominance following the "priority of access to resources" definition. Other claims that the concept of dominance can be applied to groups, populations and species (VVILSON, 1975) have referred to the "winner is dominant, loser is subordinate", or the "dominance is lack of aggression" definitions. Such definitions were shown above to be inadequate at the level of individuals, and the same reasoning applies to their application to units of higher order. A further problem of applying the concept of dominance to interactions outside the familiar social group is that this use is not compatible with definitions of the term that imply a relationship between the interactants. Advocates of the latter definition argue that dominance hierarchies are only possible in small groups because individual aquaintance is a prereq uisite (WYNNE-EDWARDS, 1962; ROWELL, 1974). In the usage of domi nance beyond the confines of the familiar group, the individual identity of the opponent and past agonistic experiences with that particular individ ual are often irrelevant to the decisions of the contestants. In interspecific dominance interactions the dyadic conflict is probably settled by recogni tion of species specific attributes of the opponent. The crucial criterion to justify the use of dominance for higher order categories is not the choice of category, but rather the definition of dominance used. When dominance is used beyond the level of dyadic interactions between single conspecifics there is the danger that analogies of dominance will be confused with homology (BERNSTEIN, 1981).
The essence of d orrri'n arrce i synthesis and definition
Despite a great variety of dominance definitions, it is generally agreed that dominance refers to agonistic behaviour. Several definitions of domi nance were rejected on the grounds of their low descriptive value. The original definition of dominance by SCHJELDERUPp-EBBE (1922) described a distinct phenomenon in chicken, which is regarded here as the basis for a more general concept of dominance with broader applicability. The high descriptive value of this dominance concept justifies its use; predic tion and explanation are applications of dominance, which originate in the particular field of interest of each investigator. The essence of domi
308
CARLOS DREWS
nance which emerges from the above discussion IS summarized in the following, structural definition of the term: Dominance is an attribute of the pattern of repeated, agonistic interac tions between two individuals, characterized by a consistent outcome in favour of the same dyad member and a default yielding response of its opponent rather than escalation. The status of the consistent winner is dominant and that of the loser subordinate. Operational definitions of dominance, based on the above definition, can be formulated in each particular study according to the choice of subjects and the theoretical framework used. Although this definition concen trates on dyadic interactions (because these are readily observable), it does not preclude the possiblity that the outcome is influenced by the nearby presence of other individuals, which may tacitly or overtly support one of the contestants (see KAWAI (1958, 1965) for an introduction to the concept of dependent rank). A semantic clarification complements this definition: dominance status and dominance ranks are different measures (HINDE, 1978; BERNSTEIN, 1984; BARRETTE & VANDAL, 1986; DUNBAR, 1988). This distinction applies to the definitions of dominance which refer to the pattern of dyadic encounters (Table 1). Dominance status refers to a relationship between two animals and depends on an animal's attributes compared to those of the other, while rank refers to the relative position within a group and, hence, depends on group composition. Dominance status refers to the status of one individual within a given dyad and can be either dominant or subordinate according to the direction of the statistically significant asymmetry in the outcome of several contests. Dominance rank refers to the position of one individual in a dominance hierarchy and can be expressed at an ordinal level either numerically, in greek letters (a = top rank, Q = bottom rank, SCHJELDERUPp-EBBE, 1922) or qualitatively as high or low, but not as dominant or subordinate. In general, dominance ranks are calculated after assessment of domi nance status in all possible dyads of the group. The resulting hierarchy can be linear, but does not necessarilly have to be so (ApPLEBY, 1983). Dyadic dominance/ subordinance relationships and the dominance hier archy are not necessarily related and the first does not imply the existence
THE CONCEPT AND DEFINITION OF DOMINANCE
309
of the other in a social group (HINDE, 1978). The limitations of ordinal ranks, in which by definition the distance between adjacent ranks is the same, have lead some investigators to develop indices which reflect the more realistic "dominance distances" that separate animals in a hier archy (e.g. BOYD & SILK, 1983; ARCESE & LUDWIG, 1986; ZUMPE & MICHAEL, 1986; DREWS, in prep.). With respect to the above definition, dominance has been viewed from different perspectives. At a general level dominance is seen as a descrip tive tool (e.g. CANDLAND & HOER, 1981). ALTMANN (1981) and SEYFARTH (1981) maintain that dominance relationships and dominance hierarchies have no function and are not ends in themselves or cognitive experiences of the animals, but simply shorthand, structural descriptive terms used by the observer. BERNSTEIN (1981) argued that dominance can be regarded as a relative measure used to illustrate one attribute of a dyadic relation ship. SMUTS (1981) sees dominance relationships as a useful estimate of an individual's ability to influence the behaviour of another for its own benefit, through that individual's power to inflict physical injury. Por-t- & DEVORE (1977) view agonistic dominance as an unavoidable product of cost/benefit considerations during competitive interactions. They, conse quently, frame dominance under the theory of games for aggressive competition. Other investigators concentrate on the fact that escalated fights are generally avoided and regard dominance as a dimension ofthe communi cation system (e.g. CHANCE, 1956; MAXIM, 1981). The subordinate may signal defeat before the dominant completes the attack or immediately when intention to attack is signaled by stereotyped movements or vocaliz ations. This is seen as evidence that a communication system has evolved (SADE, 1981). The lack of agreement in the way that investigators perceive dominance is evident, and preference for one interpretation or the other is bound to remain arbitrary. The structural definition proposed above simplifies the debate around the dominance concept because it makes no assumptions about the men tal experiences of the subjects or about those of us investigators who use the concept. This definition provides a frame for future studies on domi nance which can deal with more specific aspects, such as the role of memory and individual recognition, as well as the identification of deter minants and consequences of dominance status.
310
CARLOS DREWS
Literature cited
ALLEE, W.C. (1938). The social life of animals. - W.vV.Norton, New York. ALTMANN, S.A. (1981). Dominance relationships: the Chcshire cat's grin? - Behav. Brain Sci. 4, p. 430-431. ApPLEBY, M.C. (1980). Social rank and food access in red dcer stags. - Behaviour 74, p. 294-309. - - (1983). The probability of linearity in hierarchies, - Anim.Behav. 31, p. 600-608. ARCESE, P. & LUDWIG, D. (1986). Improving estimates of dominance based on ratios. - Condor 88, p.106-107. BAENNINGER, R. (1981). Dominance: on distinguishing the baby from the bathwat er. - Bchav. Brain Sci. 4, p. 431-432. BAKKER, T.C.M. (1986). Aggressiveness in sticklebacks (Gasterosteus aculeatus L.): a bchav iour-genetic study. -- Behaviour 98, p. 1-144. BARRETTE, C. (1987). Dominance cannot be inhcrited. - TREE 2: 251 - - & VANDAL, D. (1986). Social rank, dominance, antler size, and access to food in snow bound wild woodland caribou. - Behaviour 97, p. 118-146. BERNSTEIN, LS. (1976). Dominance, aggression and reproduction in primate societies. J. theor.Biol. 60, p. 459-472 - - (1981). Dominance: The baby and the bathwater. - Behav. Brain Sci. 4. p. 419-457. - - (1984). The adaptive value of maladaptive behavior, or you've got to be stupid in order to be smart. - Ethol.Sociobiol. 5, p. 297-303. BOYD R. & SILK, J.B. (1983). A method for assigning cardinal dominance ranks. Anim.Behav. 31, p. 45-58. BRAIN, P.F. (1981). The concept of dominance also has problems in studies on rodents. Behav. Brain Sci. 4, p. 434-435. CA~lRAS, L.A. (1984). Children's verbal and nonverbal communication in a conflict situa tion. - Ethol.Sociobiol. 5, p. 257-268. CANDLAND, K.D. & HOER, J.B. (1981). The logical status of dominance. - Behav. Brain Sci. 4, p. 436-437. CARYL, P.G. (1980). Escalated fighting and the war of nerves: gamcs theory and animal combat. - In: Perspectives in ethology 4-. (P.P.G. BATESON & P.H. KLOPFER, eds). Plenum Press, New York, p. 199-224. CHALMERS, N.R. & ROWELL, T.E. (1971). Behavior and female reproductive cycles in a captive group of mangabeys. -- Folia primatol. 14, p. 1-14. CHANCE, M.R.A. (1956). Social structure of a colony of Macaca mulatta. BritJ.Anim.Behav. 4, p. 1-13. CHENEY, D.L. (1978). Interactions of immature male and female baboons with adult females. - Anim.Behav. 26, p. 389-408. CLUTTON-BROCK, T.H. & HARVEY, P.H. (1976). Evolutionary rules and primate societies. - In: Growing points in ethology (P.P.G. Bateson & R.A. HINDE, eds). The University Press, Cambridge, p.195-237. - - , ALBON, S.D., GIBSON, R.M. & GUINESS, F.E. (1979). The logical stag: adaptive aspects of fighting in red deer (Cervus elaphus L.). - Anim.Behav. 27, p. 211-225. COLLIAS, N.E. (1944). Aggressive behavior among vertebrate animals. - Physiol. Zool. 17, p.83-123. DEAG,J.M. (1977). Aggression and submission in monkey societies. - Anim.Behav. 25, p. 465-474. - - (1978). The adaptive significance of baboon and macaque social behaviour. - In: Population control by social behaviour. (P. EBLINGAAND & D.M. STODDART, eds). Institute of Biology. London, p. 83-ll3. DESROCHERS, A., HANNON, SJ. & NORDIN, K.E. (1988). Winter survival and tcrritory acquisition in a northern population of black-capped chickadees. - The Auk 105, p. 727-736.
THE CONCEPT AND DEFINITION OF DOMINANCE
311
DEWSBURY, D.A. (1982). Dominance rank, copulatory behavior, and differential reproduc tion. - QRev.Biol. 57, p. 135-159. - - (1990). Fathers and sons: genetic factors and social dominance in deer mice, Per omiscus maniculatus. - Anim.Behav. 39, p. 284-289. DREWS, D.R. (1973). Group formation in captive Galago crassicaudatus: notes on the domi nance concept.- Z. Tierpsychol. 32, p. 425-435. DUNBAR, R.UvL (1988). Primate social systems. - Croom Helm, London & Sydney. EBERHARD, MJ.W. (1969) The social biology of polis tine wasps. - Miscellaneous Publica tions, Museum of Zoology, University of Michigan, Ann Arbor 140, p. 1-10 I. ENQUIST, M. & LEIMAR, O. (1990). The evolution of fatal fighting. - Anim.Behav. 39, p. 1-9. ENS, B., ESSELINK, P. & ZWARTS, L. (1990). Kleptoparasitism as a problem of prey choice: a study on mudflat-feeding curlews, Numenius arquata. -- Anim.Behav. 39, p. 219-230. FEDIGAN, L. (1982). Primate paradigms: Sex roles and social bonds. - Eden Press, Montreal. GAGE, F.H. (1981). Dominance: measure first and then define. - Behav. Brain Sci. 4, p. 440-441. GARTLAN, j.S. (1968). Structure and function in primate society. - Folia primatol. 8, p. 89-120. GAUTHREAUX, S.A.jr. (1978). The ecological significance of behavioural dominance. - In: Perspectives in ethology 3. (P.P.G. BATESON & P.H. KLOPFER, eds). Plenum Press, New York, p. 17-54. GINSBURG, V. & ALLEE, W.C. (1942). Some effects of conditioning on social dominance and subordination in inbred strains of mice. -- Physiol. Zool. 15, p. 485-506. GUHL, A.M., CRAIG,j.V. & MUELLER, C.D. (1960). Selective breeding for aggressiveness in chickens. - Poultry Sci. 39, p. 970-980. HANBY, j. (1976). Sociosexual development in primates. - In: Perspectives in ethology 2. (P.P.G. BATESON & P.H. KLOPFER, eds). Plenum Press, New York, p. 1-67. HAND, J.L. (1986). Resolution of social conflicts: dominance, egalitarianism, spheres of dominance, and game theory. - QRev.Biol. 61, p. 201-220. HINDE, R.A. (1978). Dominance and role - two concepts with dual meaning. j.soc.bioI.Struct. I, p. 27-38. - - & DATTA, S. (1981). Dominance: an intervening variable. - Behav. Brain Sci. 4, p. 442. - - & STEVENSON-HINDE,J. (1976). Towards understanding relationships: dynamic stabil ity. - In: Growing points in ethology. (P.P.G. BATESON & R.A. HINDE, eds). Cam bridge University Press, New York, p. 451-479. HOFFMANN, A.A. (1988). Heritable variation for territorial success in two Drosophila melanogaster males. - Anim.Behav. 36, p. 1180-1189. JAY, P. (1965). The common langur of North India. -- In: Primate behavior, field studies of monkeys and apes. (1. DEVORE, ed.). Holt, Rinehart and Winston, New York, p. 197-249. KAWAI, M. (1958). On the system of social ranks in a natural group ofJapanese monkeys, (parts I & II). - Primates I, p. 111-148 (in Japanese). - - (1965). On the system of social ranks in a natural group ofJapanese monkeys: Basic rank and dependent rank. - In: japanese monkeys: A collection of translations (K. IMANISHI & S.A. ALTMANN, eds). Emory University Press, Atlanta, p. 66-86. KAUFMANN, J.H. (1967). Social relations of adult males in a free-ranging band of rhesus monkeys. - In: Social communication among primates. (S.A. ALTMANN, ed.). Univer sity of Chicago Press, Chicago and London: p. 73-98. - - (1983). On the definitions and functions of dominance and territoriality. - Biol.Rev. 58, p. 1-20.
312
CARLOS DREWS
KIKKAWA, L., SMITH,J.N.M., PRYS-JONES, R., FISK, P. & CATTERALL, C. (1980). Determi nants of social dominance and inheritance of agonistic behavior in an island popula tion of silvereyes (;:;:,osterops lateralisi. - Behav.Eco!.Sociobio!. 19, p. 165-169. KODRIC-BROWN, A. & BROWN,J.H. (1984). Truth in advertising: the kinds of traits favoured by sexual selection. - Am. Nat. 124, p. 309-323. VAN KREVELD, D. (1970). A selective view of dominance- subordination relations in mam mals. - Genetic Psycho!. Manogr. 81, p. 141-173. LEBoEUF, B.]. (1974). Male-male competition and reproductive success in elephant seals. - Am. Zoo!' 14, p. 163-176. LEE, P.C. (1983). Context specific unpredictability in dominance interactions. - In: Primate social relationships. (R.A. HINDE, cd.). Blackwell Scientific Publications, Oxford, p. 35-44. LOCKWOOD, R. (1979). Dominance in wolves: useful construct or bad habit? - In: The behaviour and ecology of wolves. (E. KLINGHAMMER, ed.). Garland Press, New York, p. 225-244. MAXIM, P.E. (1981). Dominance: a useful dimension of social communication. - Behav. Brain Sci. 4, p. 444-445. MAYNARD-SMITH, J. (1982). Evolution and the theory of games. - Cambridge University Press, Cambridge. MCGUIRE, M.T, RALEIGH, M.J. & BRAMMER, G.L. (1984). Adaptation, selection, and benefit-cost balances: implications of behavioral-physiological studies of social domi nance in male vervet monkeys. - Etho!.Sociobio!. 5, p. 269-277. MOORE, A.J. (1990). The inheritance of social dominance, mating behaviour and attrac tiveness to mates in male Nauphoeta cinerea. - Anim.Behav. 39, p. 388-397. MORSE, D.H. (1974). Niche breadth as a function of social dominance. - Am. Nat. 108, p. 818-830. PARKER, G.F. (1974). Assessment strategy and the evolution of fighting behaviour. j. theor.Bio!. 47, p. 223-243. Por-r-, J.L. & DEVORE, 1. (1979). Aggressive competition and social dominance theory: synopsis. - In: The great apes. (D.A. HAMBURG & E.R. MCCOWN, eds). The Ben jamin/Cummings Publishing Co., Menlo Park, p. 317 -338. RALLS, R. (1976). Mammals in which females arc larger than males. - QRev.Bio!. 51, p. 245-276. RICHARDS, S.M. (1974). The concept of dominance and methods of assessment. Anim.Behav. 22, p. 914-930. ROSE, R.M., GORDON, TP. & BERNSTEIN, 1.S. (1972). Plasma testosterone levels in the male rhesus: influences of sexual and social stimuli. - Science, 178, p. 643-645. - - , BERNSTEIN, LS. & GORDON, T.P. (1975). Consequences of social conflict on plasma testosterone levels in rhesus monkeys. - Psychosom. Medicine, 37, p. 50-61. ROWELL, TE. (1974). The concept of social dominance. - Behav. Bio!. II, p. 131-154. SADE, D.S. (1981). Patterning of aggression. - Behav.Brain Sci. 4,p. 446-447. SCHJELDERUPp-EBBE, T. (1922). Be itrage zur Sozialpsychologie des Haushuhns. Zeitsch.f.Psycho!. 88, p. 226-252. SCOTT, J.P. & FULLER, J.L. (1965). Genetics and the social behavior of the dog. University of Chicago Press, Chicago. SEYFARTH, R.M. (1976). Social relationships among adult female baboons. Anim.Behav., 24, 917-938. - - (1980). The distribution of grooming and related behaviors among adult female vervet monkeys. - Anim.Behav. 28, p. 798-813. - - (1981). Do monkeys rank each other? - Behav. Brain Sci. 4, p. 447-448. SMUTS, B. (1981). Dominance: an alternative view. ~- Behav. Brain Sci. 4, p. 448-449. SOUTHWICK, C.H., BEG, M.A. & SIDDIQI, M.R. (1965). Rhesus monkeys in North India. In: Primate behavior. Field studies of monkeys and apes. (L DEVORE, cd.). Holt, Rinehart and Winston, New York, p. 111-159.
THE CONCEPT AND DEFINITION OF DOMINANCE
313
SYME, GJ. (1974). Competitive orders as measures of social dominance. -- Anim.Eehav. 22, p. 931-940. VESSEY, S.H. (1981). Dominance as control. - Eehav.Erain Sci. 4: 449. WAGNER, SJ. & GAUTHREAUX, S.A. Jr. (1990). Correlates of dominance in intraspecific and interspecific interactions of song sparrows and white-throated sparrows. Anim.Eehav. 39, p. 522-527. WEST, MJ. (1967). Foundress associations in polistine wasps: dominance hierarchies and the evolution of social behavior. - Science 157, p. 1584-1585. WILSON, E.O. (1975). Sociobiology. The new synthesis. - Harvard University Press, Cambridge. WOLFE, L.A. (1984). Female rank and reproductive success among Arashiyama E Japanese macaques (Macacafuscata). - Int. J. Primatol. 5, p. 133-143. WYNNE-EDWARDS, V.C. (1962). Animal dispersion in relation to social behaviour. - Oliver and Eoyd, London/Edinburgh. ZUMPE, D. & MICHAEL, R.P. (1986). Dominance index: A simple measure of relative dominance status in primates. - Am. J. Primatol. 10, p. 291-300.