- Email: [email protected]
Explanations of hierarchy structure
Short Communications hierarchy formation: deciding which are important is a matter for empirical testing in particular cases. Descriptive analyses such as Chase has published so far cannot provide such answers. P. J. B. SLATER Department of Zoology and Marine Biology, University of St Andrews, Fife KY16 9TS, Scotland.
References Beeman, E. A. 1947. The effect of male hormone on aggressive behaviour in mice. Physiol. Zool., 20, 373405. Chase, I. D. 1982a. Behavioral sequences during dominance hierarchy formation in chickens. Science, N.Y., 216, 439440. Chase, L D. 1982b.Dynamics of hierarchy formation: the sequential development of dominance relationships. Behaviour, 80, 218-240. Chase, I. D. 1985. The sequential analysis of aggressive acts during hierarchy formation: an application of the 'jig-saw puzzle' approach. Anim. Behav., 33, 86-100. Clutton-Brock, T. H., Greenwood, P. J. & Powell, R. P. 1976. Ranks and relationships in highland ponies and highland cows. Z. Tierpsychol., 41,202~16. Jakobsson, S., Radesater, T. & Jarvi, T. 1979. On the fighting behaviour of Nannacara anomala (Pisces, Cichlidae) males. Z. Tierpsychol., 49, 210-220. Shawcross, J. E. & Slater, P. J. B. 1984. Agonistic experience and individual recognition in male Quelea quelea. Behav. Proc., 9, 49-60. (Received 15 October 1985; revised 18 November 1985; MS number: AS-360)
Explanations of Hierarchy Structure in reply to Professor Slater's comments I shall review some general studies of individual differences and hierarchy structure, discuss his alternative model, and finally propose some constructive future research to help resolve our differing viewpoints. The jigsaw puzzle approach has two goals: (1) describing the behavioural dynamics (the proximate causes) of dominance hierarchy formation and (2) explaining how hierarchies develop their structural form, e.g. become highly linear. The approach is achieving some promising initial success in meeting the first goal; researchers using it have found considerable regularity in behavioural sequences during dominance encounters across several species (Chase 1980, 1982a, b, 1985; Mendoza & Barchas 1983; Barchas & Mendoza 1984; Eaton 1984). The approach is also meeting the second goal. A variety of behavioural patterns are possible in hierarchy formation, and applications of the approach have demonstrated that in several
1265
species those patterns which ensure transiti'dty and thus promote linear structures are most common (Chase 1980, 1982a, b, 1985; Mendoza & Barchas 1983; Barchas & Mendoza 1984; Eaton 1984). Professor Slater and I appear to agree on the descriptive value of the jigsaw puzzle approach but part company over my proposal that the behavioural patterns themselves cannot be explained by differences among the individuals making up the groups. While there is ample documentation of correlations between individual differences and dominance ranks, the question here is whether the correlations are sufficiently high to account for the types of hierarchy structures commonly found in animals. A body of research reviewed in my papers (Chase 1974, 1980, 1982a, b, 1985) suggests the answer is no. Some of this research is theoretical, examining models both explicit and implied in the animal behaviour literature through which individual differences might produce hierarchy structures (Landau 1951; Chase 1974). The conclusions from this work are that severe conditions, in terms of extremely high correlation coefficients and skewed probabilities for winning encounters, are required to generate relatively linear hierarchy structures and that available data do not meet the requisite conditions. The other portion of this research is experimental. Some of it compares the constancy in dominance hierarchies across two situations: (1) in a group before and after a short period of separation among members (Guhl 1975), (2) in a round-robin tournament and when the group meets altogether (King 1965), and (3) in a group by itself and when members have been added serially in reverse rank order to another group (Bernstein & Gordon 1980). If individual differences were strongly predictive of rank, then the correlation between an individual's place in the first and second hierarchies would be very high. Instead the correlations are usually only low to moderate. Other research compares the structure of real triads, where all three members meet at the same time, with 'constructed' triads where they meet serially only as pairs (Chase 1982b). The relationship in both kinds of triads would be expected to be transitive with a clear top, middle and bottom-ranking animal if individual differences were highly predictive of rank. Instead, real triads always formed transitive relationships, but constructed ones were o f t e n intransitive with no clear ranks among individuals (Chase 1982b). This body of theoretical and experimental research needs to be assessed in any exploration of the role individual differences play in explaining hierarchy structures. Professor Slater's alternative model to account
1266
Animal Behaviour, 34,4
for my empirical findings of high occurrence of Double Dominance and Double Subordinance patterns in chicken triads made several assumptions. First, all scores on the distribution of dominance ability are equally likely. Second, the animals have some way of sensing the relative ranks of their own and others' scores before encounters begin. Third, the first dominance encounter is always between the highest and lowest-scoring members and the highest always wins. Fourth, the middle-scoring animal always waits until the highest and lowest-scoring individuals finish their encounter, and then the middle-ranking individual defeats the lowest-ranking one or the highest ranker defeats the middle ranker. Professor Slater is correct in saying that this model would closely match my empirical results by predicting that all triads would show either Double Dominance or Double Subordinance patterns. But is this model reasonable? Is the distribution of dominance ability a flat one with all scores equally likely? Can animals always assess their own scores and those of others before encounters begin, and are these scores, in contrast to the results of the theoretical and experimental research reviewed above, highly predictive of dominance ranks? Is there a prescribed order for pairwise encounters in triads determined by dominance scores with middle-ranking animals always standing by? In order to have greater utility, a model such as this should be generalized to larger'groups and there are a variety of ways in which this could be done. But larger group size places more severe requirements on individual difference explanations (Landau 195I; Chase 1974), and very elaborate prescriptions for the order of pairwise encounters with guaranteed wins would be required. However, Professor Slater correctly points out that I did not collect data on individual differences which could be used to test his model and, so far as I am aware, the proper data are not available in published form. Furthermore, Professor Slater also correctly indicates that the jigsaw puzzle approach does not give information about individual-level mechanisms of hierarchy formation but 'begs' (please read 'raises') a lot of interesting questions about them. Although there is obviously some heat concerning jigsaw puzzle versus individual difference approaches to explaining hierarchy structures, I should like to suggest that we work toward generating more light. To this end I propose a programme of research in different species and sizes of groups in which we gather data on individual differences before hierarchy formation, individual-level mechanisms of dominance relationships, and behavioural dynamics, such as those indicated in the
jigsaw puzzle approach, which occur during hierarchy formation. The individual difference measures could cover a broad range of variables depending upon the the species and the situation, and individual-level mechanisms could relate to those variables, as well as things like bystander effects which are associated with events during hierarchy formation. I should point out, however, that the complexities of groups raise difficulties for discovering mechanisms that are not presented in more usual investigations where only one or two individuals are involved. After such a programme of research, we may still have differing points of view, but we shall know much more about the interrelations among individual differences, individual-level mechanisms, and behavioural dynamics of hierarchy formation, and that will ultimately make our discussions of this important area richer and more sophisticated. IVAN D. CHASE* Department of Sociology, State University of New York at Stony Brook, Stony Brook, New York 11974-4356, U.S.A. * Present address: Department of Sociology, William James Hall, Harvard University, Cambridge, MA 02138,' U.S.A.
References
Barchas, P. R. & Mendoza, S. D. 1984. Emergent hierarchical relationships in rhesus macaques:an application of Chase's model. In: Social Hierarchies: Essays toward a Sociophysiologieal Perspective (Ed. by P. R. Barchas), pp. 81 95. Westport, Connecticut: Greenwood Press. Bernstein, I. S. & Gordon, T. P. 1980. The social component of dominance relationships in rhesus monkeys (Macaea mulatta). Anita. Behav., 28, 1033-1039. Chase, I. D. 1974. Models of hierarchy formation in animal societies. Behavl Sci., 19, 374-382. Chase, I. D. 1980. Socialprocess and hierarchy formation in small groups: a comparative perspective. Am. Sociol. Rev., 45, 905-924. Chase, I. D. 1982a. Behavioral sequences during dominance hierarchy formation in chickens. Science, N.Y., 216, 439-440. Chase, I. D. 1982b.Dynamics of hierarchy formation: the sequential development of dominance relationships. Behaviour, 80, 218-240. Chase, I. D. 1985. The sequential analysis of aggressive acts during hierarchy formation: an application of the 'jigsaw puzzle' approach. Anim. Behav., 33, 86-100. Eaton, G. G. 1984. Aggression in adult male primates: a comparison of confined Japanese macaques and freeranging olive baboons. Int. J. Primatol., 5, 145-160. Guhl, A. M. 1975. Social behavior of the domestic fowl. In: Social Hierarchy and Dominance (Ed. by M.W.
1267
Short Communications
Schein), pp. 156-201. Stroudsburg, Pennsylvania: Dowden, Hutchinson & Ross. King, M. G. 1965. The effect of social context on dominance capacity of domestic hens. Anirn. Behav., 13, 13~133. Landau, H. G. 1951. On dominance relations and the structure of animal societies: I. Effect of inherent characteristics. Bull. Math. Biophys., 13, 1-19. Mendoza, S. D. & Barchas, P. R. 1983. Behavioral processes leading to linear status hierarchies following group formation in rhesus macaques. J. Hum. Evol., 12, 185-192. (Reechoed 6 December 1985; revised 8 December 1985; MS. number: As-374)
Listening Behaviour and Song Learning in Zebra Finches A number of estrildid finch species have been reported to show 'listening' or, as it sometimes has been called, 'peering' behaviour (Morris 1958; Immelmann 1962; Immelmann & Immelmann 1967). Immelmann (1962) describes listening as follows: during the singing of one male, another male perches beside him, stretches his neck until his head is close to the bill of the singing male and freezes in this posture. Listening has mainly been reported for adult birds and in several species it is shown by both sexes. One of the most intensively studied estrildid finch species is the zebra finch. Morris (1958) mentions this species as one in which listening 'definitely does not occur' and indeed studies on the singing and song development of zebra finches never reported its occurrence (e.g. Immelmann 1969; Sossinka et al. 1975; Price 1979; B6hner 1983; Eales 1985). However, the results presented below show that under certain conditions, zebra finches do show listening and suggest a connection between listening and song learning. The data presented here originate from a study on the development of species recognition in zebra finches, which involved intensive observations of large numbers of juvenile zebra finches between 0 and 55 days of age, together with their zebra finch or Bengalese finch foster parents (Ten Cate 1982, 1984). A total of four zebra finch (Z) pairs, four Bengalese finch (B) pairs, 33 male Z-female B and 54 female Z male B pairs were observed once every 5 days. The observations lasted 12 h (normal pairs) or 4 h (mixed pairs). Apart from the systematic data collected on specific behaviour patterns (for details see Ten Cate 1982, 1984), notes were made on other behaviour patterns that captured the attention of the observers. Among these was the following behaviour: a male foster parent starts singing, and immediately a juvenile bird flies up to the male, perches near him, focuses on him,
stretches his neck towards him and remains frozen in this posture during the singing of the foster father. This posture was not observed in any other context. Since the description of this behaviour seems identical to the behaviour described above as listening in other estrildid finches, it can be concluded that juvenile zebra finches do show listening behaviour. Notes concerning listening were not made systematically nor by all observers involved in the studies. The analysis presented here deals only with behaviour that fits fully to the description given above. Notes mentioning only that a juvenile zebra finch flew up to a singing male were not considered as representing listening, although it might have occurred at that time. For these reasons, the present data are likely to underestimate the occurrence of listening. In the following analyses, the observations are grouped according to the species of the foster father, since listening was shown towards males only. Table I summarizes the observations of listening. Listening was seen in juvenile birds only. With the exception of one observation between 20 and 25 days of age, it always occurred after day 35. Once recorded for an individual, listening was usually also recorded at later ages. The highest recorded frequency within a 4-h observation period was 10 times. Nearly all males showed the first indication of singing (subsong) at an age of about 25-35 days, i.e. usually 10 or more days before listening was observed. Notes of the occurrence of listening towards day 55 sometimes mention that the song of the young male resembled that of the foster father, as might be expected from studies of song learning in zebra finches. The song output of the foster fathers was analysed during the observation periods between 30 and 55 days of age. For each foster father, 20 h of observations were available (five periods of 4 h),
Table I. Occurrence of listening behaviour Species of foster father Z
B
Pairs in which young showed listening 1/37 11/58 Juvenile males showing listening --/38 14/56 Juvenile females showing listening 1/17 1/36 The first number is the number of pairs or individuals in which listening was observed; the second number is the total number of pairs or individuals observed.
References Beeman, E. A. 1947. The effect of male hormone on aggressive behaviour in mice. Physiol. Zool., 20, 373405. Chase, I. D. 1982a. Behavioral sequences during dominance hierarchy formation in chickens. Science, N.Y., 216, 439440. Chase, L D. 1982b.Dynamics of hierarchy formation: the sequential development of dominance relationships. Behaviour, 80, 218-240. Chase, I. D. 1985. The sequential analysis of aggressive acts during hierarchy formation: an application of the 'jig-saw puzzle' approach. Anim. Behav., 33, 86-100. Clutton-Brock, T. H., Greenwood, P. J. & Powell, R. P. 1976. Ranks and relationships in highland ponies and highland cows. Z. Tierpsychol., 41,202~16. Jakobsson, S., Radesater, T. & Jarvi, T. 1979. On the fighting behaviour of Nannacara anomala (Pisces, Cichlidae) males. Z. Tierpsychol., 49, 210-220. Shawcross, J. E. & Slater, P. J. B. 1984. Agonistic experience and individual recognition in male Quelea quelea. Behav. Proc., 9, 49-60. (Received 15 October 1985; revised 18 November 1985; MS number: AS-360)
Explanations of Hierarchy Structure in reply to Professor Slater's comments I shall review some general studies of individual differences and hierarchy structure, discuss his alternative model, and finally propose some constructive future research to help resolve our differing viewpoints. The jigsaw puzzle approach has two goals: (1) describing the behavioural dynamics (the proximate causes) of dominance hierarchy formation and (2) explaining how hierarchies develop their structural form, e.g. become highly linear. The approach is achieving some promising initial success in meeting the first goal; researchers using it have found considerable regularity in behavioural sequences during dominance encounters across several species (Chase 1980, 1982a, b, 1985; Mendoza & Barchas 1983; Barchas & Mendoza 1984; Eaton 1984). The approach is also meeting the second goal. A variety of behavioural patterns are possible in hierarchy formation, and applications of the approach have demonstrated that in several
1265
species those patterns which ensure transiti'dty and thus promote linear structures are most common (Chase 1980, 1982a, b, 1985; Mendoza & Barchas 1983; Barchas & Mendoza 1984; Eaton 1984). Professor Slater and I appear to agree on the descriptive value of the jigsaw puzzle approach but part company over my proposal that the behavioural patterns themselves cannot be explained by differences among the individuals making up the groups. While there is ample documentation of correlations between individual differences and dominance ranks, the question here is whether the correlations are sufficiently high to account for the types of hierarchy structures commonly found in animals. A body of research reviewed in my papers (Chase 1974, 1980, 1982a, b, 1985) suggests the answer is no. Some of this research is theoretical, examining models both explicit and implied in the animal behaviour literature through which individual differences might produce hierarchy structures (Landau 1951; Chase 1974). The conclusions from this work are that severe conditions, in terms of extremely high correlation coefficients and skewed probabilities for winning encounters, are required to generate relatively linear hierarchy structures and that available data do not meet the requisite conditions. The other portion of this research is experimental. Some of it compares the constancy in dominance hierarchies across two situations: (1) in a group before and after a short period of separation among members (Guhl 1975), (2) in a round-robin tournament and when the group meets altogether (King 1965), and (3) in a group by itself and when members have been added serially in reverse rank order to another group (Bernstein & Gordon 1980). If individual differences were strongly predictive of rank, then the correlation between an individual's place in the first and second hierarchies would be very high. Instead the correlations are usually only low to moderate. Other research compares the structure of real triads, where all three members meet at the same time, with 'constructed' triads where they meet serially only as pairs (Chase 1982b). The relationship in both kinds of triads would be expected to be transitive with a clear top, middle and bottom-ranking animal if individual differences were highly predictive of rank. Instead, real triads always formed transitive relationships, but constructed ones were o f t e n intransitive with no clear ranks among individuals (Chase 1982b). This body of theoretical and experimental research needs to be assessed in any exploration of the role individual differences play in explaining hierarchy structures. Professor Slater's alternative model to account
1266
Animal Behaviour, 34,4
for my empirical findings of high occurrence of Double Dominance and Double Subordinance patterns in chicken triads made several assumptions. First, all scores on the distribution of dominance ability are equally likely. Second, the animals have some way of sensing the relative ranks of their own and others' scores before encounters begin. Third, the first dominance encounter is always between the highest and lowest-scoring members and the highest always wins. Fourth, the middle-scoring animal always waits until the highest and lowest-scoring individuals finish their encounter, and then the middle-ranking individual defeats the lowest-ranking one or the highest ranker defeats the middle ranker. Professor Slater is correct in saying that this model would closely match my empirical results by predicting that all triads would show either Double Dominance or Double Subordinance patterns. But is this model reasonable? Is the distribution of dominance ability a flat one with all scores equally likely? Can animals always assess their own scores and those of others before encounters begin, and are these scores, in contrast to the results of the theoretical and experimental research reviewed above, highly predictive of dominance ranks? Is there a prescribed order for pairwise encounters in triads determined by dominance scores with middle-ranking animals always standing by? In order to have greater utility, a model such as this should be generalized to larger'groups and there are a variety of ways in which this could be done. But larger group size places more severe requirements on individual difference explanations (Landau 195I; Chase 1974), and very elaborate prescriptions for the order of pairwise encounters with guaranteed wins would be required. However, Professor Slater correctly points out that I did not collect data on individual differences which could be used to test his model and, so far as I am aware, the proper data are not available in published form. Furthermore, Professor Slater also correctly indicates that the jigsaw puzzle approach does not give information about individual-level mechanisms of hierarchy formation but 'begs' (please read 'raises') a lot of interesting questions about them. Although there is obviously some heat concerning jigsaw puzzle versus individual difference approaches to explaining hierarchy structures, I should like to suggest that we work toward generating more light. To this end I propose a programme of research in different species and sizes of groups in which we gather data on individual differences before hierarchy formation, individual-level mechanisms of dominance relationships, and behavioural dynamics, such as those indicated in the
jigsaw puzzle approach, which occur during hierarchy formation. The individual difference measures could cover a broad range of variables depending upon the the species and the situation, and individual-level mechanisms could relate to those variables, as well as things like bystander effects which are associated with events during hierarchy formation. I should point out, however, that the complexities of groups raise difficulties for discovering mechanisms that are not presented in more usual investigations where only one or two individuals are involved. After such a programme of research, we may still have differing points of view, but we shall know much more about the interrelations among individual differences, individual-level mechanisms, and behavioural dynamics of hierarchy formation, and that will ultimately make our discussions of this important area richer and more sophisticated. IVAN D. CHASE* Department of Sociology, State University of New York at Stony Brook, Stony Brook, New York 11974-4356, U.S.A. * Present address: Department of Sociology, William James Hall, Harvard University, Cambridge, MA 02138,' U.S.A.
References
Barchas, P. R. & Mendoza, S. D. 1984. Emergent hierarchical relationships in rhesus macaques:an application of Chase's model. In: Social Hierarchies: Essays toward a Sociophysiologieal Perspective (Ed. by P. R. Barchas), pp. 81 95. Westport, Connecticut: Greenwood Press. Bernstein, I. S. & Gordon, T. P. 1980. The social component of dominance relationships in rhesus monkeys (Macaea mulatta). Anita. Behav., 28, 1033-1039. Chase, I. D. 1974. Models of hierarchy formation in animal societies. Behavl Sci., 19, 374-382. Chase, I. D. 1980. Socialprocess and hierarchy formation in small groups: a comparative perspective. Am. Sociol. Rev., 45, 905-924. Chase, I. D. 1982a. Behavioral sequences during dominance hierarchy formation in chickens. Science, N.Y., 216, 439-440. Chase, I. D. 1982b.Dynamics of hierarchy formation: the sequential development of dominance relationships. Behaviour, 80, 218-240. Chase, I. D. 1985. The sequential analysis of aggressive acts during hierarchy formation: an application of the 'jigsaw puzzle' approach. Anim. Behav., 33, 86-100. Eaton, G. G. 1984. Aggression in adult male primates: a comparison of confined Japanese macaques and freeranging olive baboons. Int. J. Primatol., 5, 145-160. Guhl, A. M. 1975. Social behavior of the domestic fowl. In: Social Hierarchy and Dominance (Ed. by M.W.
1267
Short Communications
Schein), pp. 156-201. Stroudsburg, Pennsylvania: Dowden, Hutchinson & Ross. King, M. G. 1965. The effect of social context on dominance capacity of domestic hens. Anirn. Behav., 13, 13~133. Landau, H. G. 1951. On dominance relations and the structure of animal societies: I. Effect of inherent characteristics. Bull. Math. Biophys., 13, 1-19. Mendoza, S. D. & Barchas, P. R. 1983. Behavioral processes leading to linear status hierarchies following group formation in rhesus macaques. J. Hum. Evol., 12, 185-192. (Reechoed 6 December 1985; revised 8 December 1985; MS. number: As-374)
Listening Behaviour and Song Learning in Zebra Finches A number of estrildid finch species have been reported to show 'listening' or, as it sometimes has been called, 'peering' behaviour (Morris 1958; Immelmann 1962; Immelmann & Immelmann 1967). Immelmann (1962) describes listening as follows: during the singing of one male, another male perches beside him, stretches his neck until his head is close to the bill of the singing male and freezes in this posture. Listening has mainly been reported for adult birds and in several species it is shown by both sexes. One of the most intensively studied estrildid finch species is the zebra finch. Morris (1958) mentions this species as one in which listening 'definitely does not occur' and indeed studies on the singing and song development of zebra finches never reported its occurrence (e.g. Immelmann 1969; Sossinka et al. 1975; Price 1979; B6hner 1983; Eales 1985). However, the results presented below show that under certain conditions, zebra finches do show listening and suggest a connection between listening and song learning. The data presented here originate from a study on the development of species recognition in zebra finches, which involved intensive observations of large numbers of juvenile zebra finches between 0 and 55 days of age, together with their zebra finch or Bengalese finch foster parents (Ten Cate 1982, 1984). A total of four zebra finch (Z) pairs, four Bengalese finch (B) pairs, 33 male Z-female B and 54 female Z male B pairs were observed once every 5 days. The observations lasted 12 h (normal pairs) or 4 h (mixed pairs). Apart from the systematic data collected on specific behaviour patterns (for details see Ten Cate 1982, 1984), notes were made on other behaviour patterns that captured the attention of the observers. Among these was the following behaviour: a male foster parent starts singing, and immediately a juvenile bird flies up to the male, perches near him, focuses on him,
stretches his neck towards him and remains frozen in this posture during the singing of the foster father. This posture was not observed in any other context. Since the description of this behaviour seems identical to the behaviour described above as listening in other estrildid finches, it can be concluded that juvenile zebra finches do show listening behaviour. Notes concerning listening were not made systematically nor by all observers involved in the studies. The analysis presented here deals only with behaviour that fits fully to the description given above. Notes mentioning only that a juvenile zebra finch flew up to a singing male were not considered as representing listening, although it might have occurred at that time. For these reasons, the present data are likely to underestimate the occurrence of listening. In the following analyses, the observations are grouped according to the species of the foster father, since listening was shown towards males only. Table I summarizes the observations of listening. Listening was seen in juvenile birds only. With the exception of one observation between 20 and 25 days of age, it always occurred after day 35. Once recorded for an individual, listening was usually also recorded at later ages. The highest recorded frequency within a 4-h observation period was 10 times. Nearly all males showed the first indication of singing (subsong) at an age of about 25-35 days, i.e. usually 10 or more days before listening was observed. Notes of the occurrence of listening towards day 55 sometimes mention that the song of the young male resembled that of the foster father, as might be expected from studies of song learning in zebra finches. The song output of the foster fathers was analysed during the observation periods between 30 and 55 days of age. For each foster father, 20 h of observations were available (five periods of 4 h),
Table I. Occurrence of listening behaviour Species of foster father Z
B
Pairs in which young showed listening 1/37 11/58 Juvenile males showing listening --/38 14/56 Juvenile females showing listening 1/17 1/36 The first number is the number of pairs or individuals in which listening was observed; the second number is the total number of pairs or individuals observed.