S28.2: Dominance and Territoriality

Evolution and Ecology, University of California, Davis, Davis, California, 95616, USA, fax 530 752 1449, e mail jastamps@ucdavis.edu

Over the years, many workers have attempted to identify and categorise relationships between territorial and dominance behaviour (Wilson 1975; Bernstein 1981; Maher and Lott 1995). One of several definitions of territoriality is an area in which an individual has site-specific dominance (Hediger 1949; Emlen 1957; Leuthold 1977), a definition which emphasises the special salience of location in the outcome of aggressive interactions in territorial animals. Other writers have suggested that social behaviour falls along a continuum of spacing patterns, with territoriality at one end and dominance at the other (e.g. Walther 1977; Oberski and Wilson 1991), or that territoriality is one form of social dominance (e.g. Kaufmann 1983).

One possible reason for the uncertainty about relationships between dominance and territoriality is the lack of information about the process by which territorial animals establish relationships with one another. Thus, the definition of territoriality as site-specific dominance describes the unidirectional aggressive interactions that occur when territory owners detect other individuals within their territory, but does not consider how site-specific dominance is established in the first place. Nor does the simple definition of territoriality as site-specific dominance account for the fact that in many species, territory owners behave differently when encountering neighbours, familiar non-territorial individuals (e.g. floaters, subordinates) or strangers in the same locations. For instance, in species with all-purpose or breeding territories, owners often respond less aggressively to neighbours than to strangers, a phenomenon termed the ‘dear enemy’ effect (Fisher 1954), whereas in species with feeding territories, the reverse is often the case (review in Temeles 1994). In birds, dear enemy effects have been extensively investigated using song playbacks, which show that territory owners respond differently when presented with songs of familiar neighbours versus the songs of strangers, when both types of songs are played from similar positions in relation to territory boundaries (e.g. Baeyens 1981; Cooney and Cockburn 1995; Beecher et al. 1996).

The goal of this paper is to consider the processes by which animals establish dominance relationships and acquire territories, and to consider the similarities and differences between these processes and their outcomes. Many of the ideas in this paper are necessarily somewhat tentative, because many workers have studied the processes by which individuals establish dominance relationships with one another (reviews in Wilson 1975; Chase 1980; Kaufman 1983; Drews 1993), whereas very few have studied the processes by which territorial animals establish social and spatial relationships with one another (Kharitov and Siegel-Causey 1998; Stamps 1994, see also Armstrong et al. 1997; Huntingford and de Leaniz 1997). However, there is a sizeable literature that considers the factors affecting the type and outcome of aggressive contests between unfamiliar opponents in territorial as well as nonterritorial species, and another large literature that considers the social and spatial relationships between territory owners and other categories of residents after territories have been established. The current paper draws from a recent series of empirical studies of the process of territory acquisition (Stamps and Krishnan 1994a,b,1995, 1997,1998), and a recent series of theoretical studies on the role of learning in the establishment of social and spatial relationships in territorial animals (Stamps and Krishnan, in presst, Switzer and Stamps, in prep., Stamps, unpublished manuscript).

At the ultimate level, most workers assume that competition is important in both dominance-structured and territorial societies, and that aggressive interactions between novel opponents reflects the fact that they are currently competing, or are likely to compete in the future, for important commodities (reviews in Wilson 1975; Huntingford and Turner 1987; Archer 1988; Stamps 1994; Maher and Lott 1995; Grant 1997).The term ‘commodities’ is used here to reflect any item which is in short supply relative to the number of competitors, which has the potential to improve individual reproductive success. Commodities include the standard list of ‘resources’, such as food, nest sites, shelter sites, and so forth. In addition, commodities includes potential mates, which differ from resources in that prospective mates can interact in complicated ways that affect the reproductive success of both parties (e.g. Gowaty 1997). Commodities also includes exclusive space, which can affect individual reproductive success for a variety of reasons, in addition to providing a competitor with improved access to the resources or mates within that space. For instance, exclusive space may enhance reproductive success because it reduces the rate of disease transmission (Freeland 1976; Loehe 1995), the chances that predators will detect and/or capture the owner or its young (Arcese et al. 1992), or the probability that a mate will interact socially or sexually with individuals living in adjacent areas (Wagner 1993, 1997; Hoi and Hoi-Leitner 1997).

Another important similarity between many dominance-structured and territorial situations is that a particular individual is likely to repeatedly encounter the same opponents over an extended period of time. Many students of dominance in animals have focused on situations in which individuals repeatedly encounter the same opponents (e.g. Bernstein 1981; Drews 1993), and several of the most useful definitions of dominance focus on the ways that individuals change their behaviour to particular opponents, as a consequence of previous interactions with those individuals (see review in Drews 1993). Of course, these definitions also assume that animals are capable of individual recognition, and can distinguish familiar from unfamiliar opponents that are similar with respect to size, age, sex and other features that differentiate classes of conspecifics.

Many territorial animals also repeatedly encounter the same opponents, because individuals are site faithful, and because they live in areas large enough to contain more than one resident. In this situation, repeated encounters between neighbouring territory owners, or between territory owners and non-territorial residents (e.g. subordinates or floaters) in the same neighbourhood, are the rule rather than the exception. Many territorial animals are also capable of distinguishing familiar from unfamiliar opponents, as evidenced in studies showing that birds respond differently to the songs of strangers versus neighbours played from the same locations (review in Catchpole and Slater 1995), or that mice respond differently to familiar territory owners versus unfamiliar conspecifics based on odor cues (e.g. Gray and Hurst 1997). Individual recognition can be suspected in any species in which territory owners behave in one fashion when interacting with a familiar neighbour, subordinate or floater in a particular location, and in another way when interacting with a stranger of the same class at the same location.

As was noted above, in both dominance-structured and territorial situations, individuals who repeatedly encounter one another are assumed to compete aggressively with one another on at least some of these occasions. In this situation, Pagel and Dawkins (1997) and Matsumur and Kobayashi (1998) have developed theoretical models showing that individuals should adjust their aggressive behaviour over time in a way that improves their access to desired commodities, while reducing the costs of aggressive behaviour used while competing with familiar opponents. While both sets of authors designed their models for species that live in dominance-structured groups, their conclusions also apply to the residents of established territorial neighbourhoods, who repeatedly encounter the same neighbours, subordinates or floaters during the period when they inhabit that neighbourhood. For instance, recent theoretical studies show that optimal patterns of territory defence are quite different when a territory is repeatedly invaded by a single intruder than when it is invaded by a series of different individuals (Switzer and Stamps, in prep). These results imply that selection would favour territory owners who were able to recognize individual intruders, and thus optimise their defensive behaviour when confronted by particular individuals who repeatedly entered their territory.

The aggressive interactions observed when animals compete with one another can be classified in terms of their symmetry, outcome and costs. Symmetry refers to the type of behaviour patterns exhibited by the two participants in an aggressive interaction, where symmetrical (or reciprocal) interactions are those in which both individuals exhibit the same types of behaviour patterns, while asymmetrical (or complementary) interactions are those in which the two parties exhibit different types of behaviour patterns (Hinde 1976, 1995). An example of a symmetrical aggressive interaction is an escalated fight which begins with exchanges of the same displays, proceeds to a wrestling match, and ends with both opponents slowly retreating from one another. An example of an asymmetrical aggressive interaction is a vigorous chase during which the chaser occasionally bites the chasee. Of course, many types of aggressive interactions fall between these extremes. Thus, an interaction might begin with an reciprocal exchange of displays but end with a chase, or begin with a chase but then develop into a fight that ends with both animals withdrawing from the fight-location.

The outcome of aggressive interactions is scored by focusing on the degree of symmetry in the behaviour patterns of opponents at the end of an interaction. Many aggressive interactions end with asymmetrical behaviour patterns, in which one animal chases, supplants, approaches, etc., while the other runs, hides, freezes, avoids or otherwise offers no resistance to its opponent. In this situation, the first individual is said to have 'won' the interaction, whereas the other individual is said to have 'lost' the interaction. Aggressive interactions may also end with symmetrical behaviour patterns. For instance, both opponents may withdraw from one another, the two individuals may alternately chase one another, or an interaction may end with both individuals exchanging the same display with one another. When an interaction ends with symmetrical behaviour patterns, it is said to end in a draw or a tie (e.g. see Boyd and Silk 1983).

Finally, aggressive interactions can be scored according to their potential costs, including energy expenditure, risk of injury, time that could have been devoted to other activities, or elevated risk of predation while engaging in aggressive interactions. Thus, a brief fight involving the exchange of several displays might be more costly than a fight involving a lengthy series of vigorous displays, biting, kicking or other potentially injurious behaviour (Smith and Taylor 1993;Thorpe et al. 1995; Haller et al. 1996;Hack 1997). Generally speaking, interactions that are symmetrical with respect to behaviour patterns are also likely to be equally costly to both participants. Thus, in velvet crabs, Necora puber, escalated fights result in equivalent energetic costs for winners and losers (Smith and Taylor 1993).

An extensive literature on animal contests indicates that when two individuals have their first aggressive interaction with one another, the symmetry, outcome and costs of this interaction vary as a function of morphological and behavioural characteristics of the two opponents. The outcome of contests between novel opponents has been show to be influenced by differences between opponents in body size or weaponry size (Sneddon et al. 1997), familiarity with the contest-location (Stamps and Krishnan 1994a), previous experience with winning or losing contests with other opponents (Chase et al. 1994) or individual differences in aggressive behaviour, such as latency to attack a novel opponent (Verbeek et al. 1996). The symmetry and costs of aggressive interactions have also shown to be affected by differences in morphology and behaviour. For instance, contests between novel opponents of the same size are typically longer and more symmetrical than contests between opponents of different sizes, while contests between two animals who are both familiar with the same area are typically longer and more escalated than contests between an animal who is familiar with the area and a newcomer (reviews in Huntingford and Turner 1987; Archer 1988; Archer and Huntingford 1994). Additional studies have shown that when it comes to predicting the symmetry, outcome and cost of aggressive interactions, particular combinations of morphological and behavioural factors ‘tradeoff’ against one another. Thus, contests between a smaller territory owner and a larger intruder may be longer and more symmetrical than those involving a larger territory owner and a smaller intruder (e.g. Wazlavek and Figler 1989).

When studying species in which individuals repeatedly encounter the same opponents over a period of time, it is important to realize that many of the factors that determine the symmetry, outcome and cost of aggressive interactions are unlikely to change much, if at all, on a day to day basis (Barrette and Vandal 1986; Drews 1993). This is obviously true with respect to variation in morphological traits like body or weapon size, but it is also the case for many of the behavioural characteristics discussed above. For instance, an individual who is very familiar with a particular territory one day is unlikely to forget everything it has learned about that territory the following day, while an individual who meets opponent X after losing all of its fights with other individuals is unlikely to have won most of its fights before it encounters opponent X the following day.

While it is easy to show that some morphological traits exhibit consistent individual differences over periods of time, it is not as easy to show consistent individual differences in behaviour. This is because an extensive literature shows that animals adjust their aggressive behaviour based on previous aggressive experience with the same or other opponents (e.g. Chase et al. 1994, Beaugrand et al. 1996, Sadnabba 1996). In order to investigate individual differences in aggressive behaviour while controlling for effects of prior aggressive experience, one needs to eliminate any possibility of a ‘carry-over’ of aggressive experience from one encounter to the next. One way to do this is to use clonal organisms, in which genetically identical individuals can be tested in a series of aggressive interactions with the members of other clones. For instance, clonal sea anemones compete aggressively over space, and in nature establish stable territorial boundaries with the members of other clones (Francis 1973). Repeated staged encounters between the members of particular clones revealed consistent differences among clones in aggressive behaviour, and in some cases one clone won all of its interactions with opponents from another clone (Ayre and Grosberg 1995; Grosberg and Ayre 1996). Students of aggression in non-clonal animals have approximated the same experimental design by artificially selecting genetically homogenous strains of animals, and then staging repeated encounters between individuals from particular strains. These studies show consistent differences among strains in aggressive behaviour, and as a result, predictable differences in the type and outcome of aggressive interactions across dyads involving opponents from different strains (e.g. Hahn and Schanz 1996; Sandnabba 1996).

If consistent individual differences in morphology, physiology and behaviour affect the symmetry, outcome and cost of aggressive interactions, then in the absence of any learning, the members of a given dyad would be expected to have comparable types of aggressive interactions each time they interacted with one another. For instance, if two animals engaged in an escalated fight ending in a draw on their first encounter, then in the absence of any learning, they would be expected to have another symmetrical, costly aggressive interaction when they next encountered one another. Hence, the question is not why pairs of opponents would be likely to have the same types of aggressive interactions when they encounter one another, but rather why there should be predictable changes in the symmetry, outcome and cost of aggressive interactions when the members of a given dyad repeatedly interact with one another (Bernstein 1981; Drews 1993; Mesterton-Gibbons and Dugatkin 1995).

Changes in the symmetry, outcome and cost of repeated aggressive interactions involving the same pair of opponents are expected if individuals consistently vary with respect to morphological, behavioural or physiological features that affect aggressive interactions, if individuals are capable of recognising one another as individuals, and if individuals are likely to encounter the same opponents many different times. In this situation, one would expect animals to learn about an opponent during aggressive interactions with that opponent, and then adjust their subsequent behaviour so as to improve their access to commodities while reducing the costs of aggressive interactions with that opponent (Bernstein 1981; Van Rhijn and Vodegel 1990; Pagel and Dawkins 1997). Of course, this argument only makes sense if information about an opponent’s competitive ability revealed during the course of aggressive interactions with that opponent.

Two literatures provide insights about the types of information that are likely to be revealed during aggressive encounters with a particular opponent. First, studies of animal contests suggest that aggressive interactions may reveal behavioural or physiological factors that affect the competitive ability of an opponent, including fighting tactics, stamina, strength, persistence, willingness to escalate, or energy reserves (Enquist and Leimar 1983; Franck and Ribowski 1989; Enquist et al. 1990; Archer and Huntingford 1994; Marden and Rollins 1994). Along the same lines, an individual’s motivation to remain in a particular area or group may be revealed by its persistence in returning to that area or group even after being repeatedly attacked there (Francis 1988; Englund and Olsson 1990; Marden and Waage 1990; Stutchbury 1991; Stamps and Krishnan 1995). These are just a few examples of types of information about a particular opponent that may be difficult to assess in the absence of aggressive interactions, but which may be useful for estimating the potential long term costs of repeated interactions with that opponent. Under the assumption that information is accumulated over a series of aggressive interactions with the same opponent, this hypothesis predicts that first encounters between opponents would last longer and involve a wider array of behaviour patterns than later aggressive interactions between those same individuals (e.g. see Getty 1989).

A particularly salient piece of information when an individual interacts with a given opponent is the costs that are likely to incur during subsequent interactions with that opponent. Aggressive social interactions differ from most other types of social behaviour in that the participants deliver aversive stimuli to one another, in the form of bites, kicks, blows, exhausting fights, or potentially dangerous, high speed chases. The classic psychology literature suggests that aversive stimuli delivered during aggressive interactions might directly affect the subsequent behaviour of opponents, by acting as punishment or negative reinforcement (Mackintosh 1983; Anderson 1995).

Stimuli delivered during aggressive interactions can be said to act as punishment if they discourage an opponent from exhibiting behaviour that occurred prior to the aggressive interaction (Estes 1944; Blanchard and Blanchard 1990; Anderson 1995; Clutton-Brock and Parker 1995). For instance, if animal X challenged animal Y to a fight, and was bitten in the course of that interaction, the memory of this bite might reduce the chances that animal X would challenge animal Y when they next met one another. Similarly, if an intruder were attacked by a territory owner immediately after setting foot in a territory, the memory of that attack might reduce the probability of its entering that territory in the future (Stamps & Krishnan, in press).

Aversive stimuli experienced during aggressive interactions can also act as negative reinforcement, if these stimuli encourage the production of behaviour patterns that reduce the chances that an animal will experience additional aversive stimuli in the future (Herrnstein 1969; Bolles 1975, Anderson 1995). For instance, classic psychological studies of escape and avoidance conditioning show that the delivery of shock in an enclosure enhances the production of behaviour (e.g. jumping out of the enclosure) that reduces the probability that the animal will experience shock in the future (e.g. Solomon & Wynne 1953, McAllister & McAllister 1962). Other types of behaviour might be enhanced as a result of previous aggressive experience with a particular opponent. Thus, if animal X were bitten by animal Y, that bite might increase the chances that animal X would give submissive displays to Y during subsequent encounters. In a territorial context, if an intruder were attacked by a territory owner one day, it might be more likely to adopt inconspicuous, ‘sneaky’ behaviour when entering that territory the following day.

In combination, learning about an opponent and learning to avoid additional aversive stimuli help explain a pattern frequently described in the dominance literature, in which pairs of opponents have repeated encounters until all of their interactions have the same symmetry and outcome, and then both parties adjust their behaviour in ways that are likely to reduce the costs of subsequent interactions (Pagel and Dawkins 1997). This pattern makes sense, if initial aggressive interactions between unfamiliar individuals provide the information that allows each individual to assess its opponent, and predict the symmetry, outcome and cost of future interactions with that opponent. Once the members of a dyad have ‘agreed’ on the symmetry and outcome of their aggressive interactions, both can adopt behaviour that reduces the costs of those interactions.

The first point to emphasise when considering the differences between dominant-subordinate and territorial relationships is that both types of social relationships can be established and maintained by the same individual. For instance, many territorial animals share their neighbourhood with non-territorial individuals with whom they interact on a repeated basis (Smith 1978; Arcese 1987; Rohner 1996). In this situation, territory owners establish different types of social and spatial relationships with neighbouring territory owners and familiar non-territorial residents (e.g. see Temeles 1994; Ens 1996). Other species exhibit group-territoriality, such that individuals form a linear dominance hierarchy with the animals with whom they share the territory, while all group members defend the territory against other members of the species (e.g. Woolfenden and Fitzpatrick 1977; Clifton 1990; Jennions and McDonald 1994). In addition, many territorial animals form stable dominant-subordinate relationships or dominance hierarchies when confined in spaces too small for the establishment of several normal-sized territories (Bovbjerg 1956, Stamps 1977, Kodric-Brown 1988).

The main distinction between territorial and dominance-structured situations is that in territorial situations, each individual inhabits a particular area for an extended period of time, and its access to important commodities is affected by the size, share, exclusivity, ‘quality’ and other attributes of this area (Wilson 1975; Stamps 1994). In contrast, in dominance-structured situations, individuals share space with conspecifics on a long term basis. In this situation, access to important commodities in the shared space is affected by an individual’s social relationships with other individuals with whom it shares that area on an ongoing basis. Looking at it another way, in territorial situations animals compete for space, whereas in dominance situations, animals compete for status.

Another major difference between dominance-structured and territorial situations is that symmetrical interactions and relationships are more common in the latter. In dominance-structured situations, repeated encounters involving members of the same dyad are eventually expected to produce predictable, asymmetrical interactions, in which one individual consistently ‘wins’ every encounter with its opponent (Bernstein 1981). Conversely, dyads whose disputes end with a tie or a draw are said to have ‘unresolved’ dominance relationships (Bernstein 1981; Hand 1986). This interpretation follows from the assumption that if two individuals fail to form a stable dominant-subordinate relationship, they will continue to interact aggressively, at high cost to each, whenever they met and compete for mutually desired commodities (Bernstein 1981; Drews 1993; Pagel and Dawkins 1997).

In contrast, in territorial situations, the goal is to achieve exclusive space, not to work out mechanisms to allocate commodities in shared space. One way that prospective neighbours divide space is by engaging in aggressive interactions with one another, after which both participants avoid subsequent interactions with the location of these contests, or with one another (Stamps and Krishnan, 1997, 1998, in press). Symmetrical aggressive interactions may be an important element in this process, because symmetry in behaviour patterns during contests implies that two opponents are well matched with respect to size, fighting ability, persistence and other factors reflecting competitive ability (Schuster et al. 1982). In particular, the formation of stable relationships between neighbouring territory owners may be facilitated by one or more costly symmetrical aggressive interactions, after which the two individuals establish mutually non-overlapping home ranges, and shift to less costly types of symmetrical interactions (Eason and Hannon 1994; Stamps and Krishnan 1997). After territories are established, neighbouring owners frequently maintain co-dominant relationships with one another, where ‘co-dominant’ reflects the fact that both members of the dyad are aggressive, but that neither consistently dominates the other.

Symmetrical social and spatial interactions between neighbours have been reported in many territorial species after territories have been established. In birds, neighbours may exchange the same songs, or songs from a shared repertoire, with one another, in song bouts that end with no clear 'winners' and 'losers' (Payne and Payne 1993; Beecher et al. 1994). Alternately, neighbours may engage in 'parallel walks' along their territorial border (Askenmo et al. 1994; Turpie 1995). Examples from other taxa include wolves, who deposit similar scent-marks on either side of their mutual boundary (Peters and Mech 1975), shrews that mutually avoid one another’s territory (Shillito 1963), or cichlid neighbours who engage in mutual frontal displays that typically end in a 'draw' (Turner 1986). Symmetrical aggressive interactions between neighbouring territory owners often involve animals reciprocally exchanging the same displays at the same rates, or chasing one another back and forth across their mutual boundary, and such interactions frequently end with both individuals retreating from the location of their encounter.