Group Title: comparative behavior of three species of swallows (genus Progne)
Title: The comparative behavior of three species of swallows (genus Progne)
Full Citation
Permanent Link:
 Material Information
Title: The comparative behavior of three species of swallows (genus Progne)
Physical Description: Book
Language: English
Creator: Bitterbaum, Erik Joshua, 1952-
Copyright Date: 1986
 Record Information
Bibliographic ID: UF00102764
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: ltuf - AEK1881
oclc - 15293455

Full Text







This dissertation is dedicated to Dr. John William Hardy,
mentor and friend.


I extend my thanks to all my friends at the Florida

State Museum for making my stay there such a memorable one.

I also wish to thank my friends in the Department of Zoology

of the University of Florida for their help, especially

Drs. Peter Feinsinger, Martha Crump, John Kaufmann, and

Frank Nordlie. I thank Dr. J.W. Hardy, who suggested the

problem, and for his interest and guidance throughout the

course of this study. I am deeply indebted to Dr. Jane

Brockmann for advice and discussion on the project design

and for her valuable suggestions on the manuscript.

Drs. Thomas Emmel, Michael Collopy, and Jonathan Reiskind

read drafts of the manuscript and offered important

criticisms and comments. Drs. Charles Brown and Jeffrey Cox

accompanied me in the field and provided fruitful discussion

of my material. I am indebted to Drs. Joe Wampler and Bob

Bryan of Nebraska Wesleyan University for their considerable

advice on statistical procedures.

I would like to thank Richard ffrench and his family

for all the courtesies extended to me during my stay at

their home in Trinidad. Special thanks are extended to

Texaco Incorporated for granting me permission to study the

Gray-breasted Martin at their oil refinery in Trinidad. The


field work was made considerably easier by the help and

hospitality of Mr. Richard Rothenberg, Mr. Hector Garcia,

Ms. Sarah Sloane, Ms. Rebecca Dellinger, Dr. Tom Webber,

Dr. James Cohen, and Dr. Eugene Morton. I owe a real debt

to many other friends, too numerous to list, who gave me

their unfailing faith and encouragement in difficult periods

during the research and writing that culminated in this


Financial support was made possible by grants from the

Chapman Memorial Fund of the American Museum of Natural

History, the Florida State Museum, the Department of Zoology

of the University of Florida, the Smithsonian Institution,

and Joseph and Evelyn Bitterbaum. Travel funds were provide

by the Society of the Sigma Xi.

Last, but not least, I wish to thank my wife Ellen H.

Burton. Her patience, encouragement, delightful companion-

ship, and assistance in editing the manuscript made the

completion of this study possible.



ACKNOWLEDGMENTS . . . . . . . .. . iii

LIST OF TABLES . . . . . . . . . .viii

LIST OF FIGURES . . . . . . . .. . xi

ABSTRACT . . . . . . . . . . . xvi

INTRODUCTION . . . . . . . . . . 1

Ritualization . . . . . . . . . 2
How Signals Originate . . . . . . . 4
The Information Hypothesis . . . . . . 9
The Manipulation Hypothesis . . . . .. 11
The Honesty Hypothesis . . . . . . .. 13
Predictions . . . . . . . . . 14


METHODS . . . . . . . . . . . . 18

Study Areas . . . . . . . . .. 18
Materials . . . . . . . . . . 19
Description of Behavioral Techniques . . . .21
Criteria for Recognizing Display Units . . .. 24

TERMINOLOGY . . . . . . . . . . . 28

Vocal Displays . . . . . . . . . 28
Visual Displays . . . . . . . ... 29

VISUAL COMMUNICATION . . . . . . . .. 32

Stereotypy of Behavior Patterns . . . .. 33
Discussion . . . . . . . . . . 42
Description of Visual Displays . . . . .. 46

Head Forward Thrust . . . . . ... .48
Gaping . . . . . . . . . 53
Bill-Snapping . . . . . . . . 55
Lunge . . . . . . . . . . 55

High-Up Displays . . . . . . .
Claiming-Reclaiming . . . . . .
White Badge Signal . . . . . .
Fighting . . . . . . . . .

Analysis of Visual Communication . . . .

Sequences in Defense of Territory . . .
Relation of Displays to Subsequent Behavior
Effectiveness of Displays . . . . .
Component Structure of Displays . . .
Signal Value of Postural Components . .
Presentation of the Bill . . . . .
Categories of Behavior . . . . .

Discussion . . . . . . . .


Description of Vocal Displays . . .

. . 100

* . 110

. . 110

Purple Martin . . . . . .
Gray-Breasted Martin . . . .
Caribbean Martin . . . . .


Analysis and Discussion of Vocal Displays

Stereotypy . . . . . . .
Ecology of Communication . . .

TERRITORIALITY . . . . . . . .

Assessment . . . . . . . .
Purple Martin . . . . . . .

Establishment of Territory . . .
Establishment of Territory by Later
Arriving Birds . . . . .
Behavior of First-Year Males . .
Territory Size . . . . . .
Gourd Complex Versus Multiroom Houses

Gray-Breasted Martin . . . . . .

Establishment of Territory . . .
Gray-Breasted Martin Colony Defense .

Caribbean Martin . . . . . . .

. . 140

. . 151
. . 156

. . 161

S. 162
* . 163

. . 163

. . 165
. . 166
. . 167
. . 173

. . 176

. . 176
. 179

. . 180


. 56
. 60
. 62
. 63

. 68

. 68
. 77
. 78
. 80
. 90
. 91
. 92


Asymmetry in Resource Holding Potential ... .181

Asymmetry in Ownership Status . . . . 183
Asymmetry in Age-Related Plumage Character-
istics . . .. ... .. . . . 185
Asymmetry in Aggressive Behavior . . . 186
Asymmetry of Size and Weight on Dominance . 187
Asymmetry in Contests Due to Sex . . .. .190

Discussion . . . .. . . . . . 194

RHP Asymmetries . . . . . . .. .195
Pay-Off Asymmetries . . . . .. . 198
Uncorrelated Asymmetries . . ... . 201

DOMINANCE POSITION . .. . . . . . . 202

Introduction . . . . . . . . ... .202

Data Analysis . .. . . . . . 203

Purple Martin . .. . . . . . 208
Gray-Breasted Martin . . . . . .. .213
Rank in the Hierarchy . . . . . .. .214

Discussion . . . . . . . . . . .218

CONCLUSIONS . . . . . . . . ... . .220

LITERATURE CITED . . . . . . . . .. .227

BIOGRAPHICAL SKETCH . . . . . . . . .. .244



Table Page

1. Comparisons of Agonistic and Nonagonistic Body
and Limb Movement, Final Position, and Duration
of Movements in Seconds . . . . . ... 36

2. Comparisons of Variances of Agonistic and
Nonagonistic Movements . . . . . .. 38

3. Frequencies of Initial Displays . . . ... 69

4. Incidence of Single Displays from Different
Types of Encounters . . . . . . .. 71

5. Proportion of Displays Used as Initial Versus
Terminal Responses in Two-Display Sequences . 74

6. Purple Martins: Display of Signaler and the
Subsequent Behavior of the Signaler and
Receiver . . . . . . . .... .. 75

7. Gray-Breasted Martins: Display of the Signaler
and the Subsequent Behavior of the Signaler
and the Receiver . . . . . . .. 76

8. Simultaneous Occurrence of Behavioral Components
Given by Territorial Male Purple Martins to
Intruders . . . . . . . .... . 83

9. Simultaneous Occurrence of Behavioral Components
Given by Territorial Male Gray-Breasted Martins
to Intruders . . . . . . . ... 84

10. Purple Martins: Behavioral Components of
Displays Versus Subsequent Behavior of Signaler
and of Receiver .. . . . . . . . 86

11. Gray-Breasted Martins: Behavioral Components of
Displays Versus Subsequent Behavior of Signaler
and of Receiver . . . . . . . ... 88

12. Song Rates of Male Purple Martins During
Different External Situations . . . . .. .125


Table Page

13. Percentage Occurrence of Vocalizations During
Agonistic Behavior of Purple Martins . . .. .141

14. Percentage Occurrence of Vocalizations During
Agonistic Behavior of Gray-Breasted Martins 142

15. Percentage Occurrence of Vocalizations During
Agonistic Behavior of Caribbean Martins . . 143

16. Percentage of Several Vocalizations Given in
Various Contexts by Purple Martins . . . 146

17. Percentage of Several Vocalizations Given in
Various Contexts by Gray-Breasted Martins . 147

18. Percentage pf Several Vocalizations Given in
Various Contexts by Caribbean Martins ... .148

19. Comparisons of the Physical Characteristics of
Purple Martin Vocalizations . . . ... .152

20. Comparisons of the Physical Characteristics of
Gray-Breasted Martin Vocalizations . . .. .153

21. Comparisons of the Physical Characteristics of
Caribbean Martin Vocalizations . . . ... .154

22. Number of Male Purple Martins and Their Maximum
and Minimum Territory Sizes with Male's Months
of Arrival . . . . . . . . . 170

23. A Comparison of the Number of Fights at a Gourd
Complex Versus a Multiroom House for Purple
Martin Males . .. . . . . . . 174

24. A Comparison of the Length of Time Spent Loaf-
ing at a Gourd Complex Versus a Multiroom House
for Purple Martin Males . . . . . ... .175

25. Comparison of Whether Prior Occupancy at a
Nesting Site Determines the Winner of the
Encounter . . . . . . . .... .. .184

26. Comparison of the Number of Encounters Between
Adult Purple, Gray-Breasted, and Caribbean
Martins in Which a Bird Dominated as a Result
of Whether or Not it Initiated the Encounter .188

Table Page

27. Comparison of 241 Purple Martin Encounters in
Which a Bird Dominated as a Function of Age and
of Whether it Initiated the Fight . . ... 189

28. Comparison of Dominance with Body Weight and
Body Size in Purple Martins . . . ... .191

29. Comparison of Dominance with Body Weight and
Body Size in Gray-Breasted Martins . . .. .192

30. Number of Supplantings at the Nest Sites
Between Purple Martins at the Gainesville
Country Club (1978) . . . . . . . 204

31. Number of Purple Martin Supplantings Taking
Place Other Than in Front of the Nest hole,
i.e. at Telephone Wires, Colony Rooftops,
Television Antennas . . . . .. . . 205

32. Number of Supplantings Observed at the Nest
Sites Between Gray-Breasted Martins at Pointe-
a-Pierre (1979) . . . . . . . .. 206

33. Number of Gray-Breasted Martin Supplantings
Taking Place Other Than in the Territories of
the Pumphouse Birds, i.e. at Guard Rails,
Pipes, Rooftop . . . . . . . 207

34. A Comparison Between Whether Lower-Ranking
Males Entered Their Nest Holes More Often When
Confronted by a High-Ranking Male . . ... .210

35. The Degree of Association at the Nest Sites
Between Two Measures of Dominance in the Purple
Martin, Rank in the Hierarchy and Number of
Interactions Initiated and Won . . . ... .216

36. The Degree of Association at the Nest Sites
Between Two Measures of Dominance in the Gray-
Breasted Martin, Rank in the Hierarchy and
Number of Interactions Initiated and Won . . 217


Figure Page

1. Schematic outline showing how body and limb
positions were measured . . . . . .. 26

2. Postures associated with inactivity: A. Purple
Martin; B. Gray-breasted Martin; C. Caribbean
Martin . . . . . . . . .. ... . 47

3. Displays associated with threat behavior:
A. Low-intensity Head Forward Thrust (Purple
Martin); B. high-intensity Head Forward Thrust
(Purple Martin); C. low-intensity Head Forward
Thrust (Gray-breasted Martin); D. high-
intensity Head Forward Thrust (Gray-breasted
Martin); E. variation of the high-intensity
Head Forward Thrust (Gray-breasted Martin);
F. low-intensity Head Forward Thrust (Caribbean
Martin); G. high-intensity Head Forward Thrust
(Caribbean Martin) . . . . . . .. 49

4. Displays associated with threat behavior:
A. Gaping; B. Gaping between mates; C. Defen-
sive Gaping; D. Lunge. . . . . . . 54

5. Display associated with appeasement behavior;
A. Withdraw High-Up; B. Alert High-Up;
C. White Badge Signal . . . . . .. 58

6. Fighting interactions between male martins . 65

7. Models of interactions between martins . . 72

8. Gradation of threat displays . . . .. 93

9. Pair-wise comparisons of probabilities that a
Purple Martin signaler will attack, retreat, or
stay for all possible combinations of displays.
Dark blocks indicate a significant difference
at the 0.05 level, test for significance between
two proportions. LIHF= low-intensity Head
Forward Thrust, HIHF= high-intensity Head

Figure Page

Forward Thrust, G= Gaping, BS= Bill-snapping,
L= Lunge, WHU=Withdraw High-Up, AHU= Alert
High-Up . . . . . . . . ... . .94

10. Pair-wise comparisons of probabilities that a
Purple Martin receiver will attack, retreat, or
stay for all possible combinations of displays.
Dark blocks indicate a significant difference at
the 0.05 level, test for significance between
two proportions. LIHF= low-intensity Head
Forward Thrust, HIHF= high-intensity Head
Forward Thrust, G= Gaping, BS= Bill-snapping,
L= Lunge, WHU=Withdraw High-Up, AHU= Alert
High-Up . . . . . . . . ... .. . 95

11. Pair-wise comparisons of probabilities that a
Gray-breasted Martin signaler will attack,
retreat, or stay for all possible combinations
of displays. Dark blocks indicate a signifi-
cant difference at the 0.05 level, test for
significance between two proportions.
LIHF= low-intensity Head Forward Thrust,
HIHF= high-intensity Head Forward Thrust,
G= Gaping, BS= Bill-snapping, L= Lunge,
WHU=Withdraw High-Up, AHU= Alert High-Up . . 96

12. Pair-wise comparisons of probabilities that a
Gray-breasted Martin receiver will attack,
retreat, or stay for all possible combinations
of displays. Dark blocks indicate a signifi-
cant difference at the 0.05 level, test for
significance between two proportions.
LIHF= low-intensity Head Forward Thrust,
HIHF= high-intensity Head Forward Thrust,
G= Gaping, BS= Bill-snapping, L= Lunge,
WHU=Withdraw High-Up, AHU= Alert High-Up . 97

13. Sonograms of Purple Martin vocalizations:
A. series of Cher calls from one individual;
B. one Zweet call; C. series of Zweet calls
from one individual; D. series of Zweet calls
from one individual when a cat was observed;
E. Zweet and Cher calls from one individual;
F. Hee-hee calls from two individuals . .. .112

14. Sonograms of Purple Martin vocalizations:
A. series of Chortle calls from one female;
B. series of Chortle calls from one male;
C. Chortle song; D. Zwrack calls given by
different individuals; E. Rattle call from
one individual; F. Choo calls from a female .118


Figure Page

15. Sonograms of Purple and Gray-breasted Martin
vocalizations: A. Purple Martin Primary song;
B. incomplete Purple Martin Primary song;
C. Gray-breasted Martin Cree calls; D. Gray-
breasted Martin Zwat calls; E. Gray-breasted
Martin Cluck calls; F. Gray-breasted Martin
Krack calls . . . . . . . . 123

16. Sonograms of Gray-breasted and Caribbean Martin
vocalizations: A. series of Gray-breasted
Martin Zurr calls; B. series of Gray-breasted
Martin Cheur calls; C. Gray-breasted Martin
Primary song; D. Gray-breasted Martin Rattle
call; E. Caribbean Martin Peak calls;
F. Caribbean Martin Zwoot call; G. Caribbean
Martin Croot call. . . . . . . . 127

17. Sonograms of Caribbean Martin vocalizations:
A. Primary song; B. incomplete Primary song;
C. series of Wheet calls; D. series of Kweet
calls from one individual; E. Wrack call . . 136

18. The dates on which the minimum territory size
was established at the Purple Martin multi-
room houses . .. . . . . . . . 168

19. Territory size as a function of the number of
Purple Martin males at a multiroom house . .. 171

20. Territory size as a function of the number of
fights at a Purple Martin multiroom house . . 172

21. Territory size as a function of the number of
fights at a Gray-breasted Martin nesting area .178

22. Number of encounters as a function of the sex
of the bird . . . . . . . . ... .193

23. Frequency of participants in encounters as a
function of the distance between nestholes . 211

24. Number of supplantings at a Purple Martin
multiroom house during the breeding season . 212

25. Number of supplantings at the Gray-breasted
Martin nest area during the breeding season . 215


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Erik Joshua Bitterbaum

August, 1986

Chairman: Dr. John William Hardy
Major Department: Zoology

The vocal and visual displays of three species of

swallows (Genus Progne) are described and analyzed struc-

turally and functionally. I examine the way in which

predictions from models based on game theory can help us

understand ethological data about agonistic displays.

Predictions based on game theory are contrasted with the

traditional ethological view that agonistic displays evolved

to transmit information about the probability of attack or

escape. After a signaler's display, both the signaler's

action and the receiver's response are recorded. I conclude

that information about attack is poorly encoded in the

aggressive displays, and suggest that many agonistic

displays may signal that the subsequent behavior of the

displaying bird depends in part on that of the receiver.


I give evidence that most of these aggressive displays are

used in a broad range of situations, suggesting that they

encode general messages and that specificity in communica-

tion depends largely on contextual information.

Social interactions of the swallows I studied are

based upon the mechanism of territoriality. Territoriality

is associated with defense of a critical resource, which in

these birds is a breeding territory. The establishment of

territory is associated with age-related plumage charac-

teristics, residency, and aggression. Dominance hierarchies

in the swallows studied are linear away from the nest sites

and nonlinear at the breeding sites. Thus, the outcome of

an encounter between individuals is site-dependent. These

findings are discussed in the light of recent theoretical

advances in the study of animal threat and fighting behav-



In this study, I investigated the display behavior of

three species of swallows, genus Progne: the Purple Martin

(P. subis), the Gray-breasted Martin (P. chalybea), and the

Caribbean Martin (P. dominicensis). Specifically, the aims

of this study were to (1) describe the vocal and physical

displays, (2) provide information on the function of each

display, (3) suggest how each species's display behavior is

related to its social organization, and (4) examine some

aspects of the ecology of communication.

General accounts of the behavior and life history of

the Purple Martin have been given by Allen and Nice (1952),

Bent (1942), Bitterbaum and Brown (1981), Brown (1975,

1978a, 1979a, 1979b, 1980, 1984), Brown and Bitterbaum

(1980), Finlay (1971, 1976), Gaunt (1959), Jackson and Tate

(1974), Johnston and Hardy (1962), and Rohwer and Niles

(1977, 1979). Previous studies of the behavior of the Gray-

breasted Martin have been published by Beebe, Hartley, and

Howe (1917), Bent (1942), Dyrcz (1984), Eisenmann and

Haverschmidt (1970), ffrench (1976) and Hellmayr (1906).

Little is known about the behavior of Caribbean Martins

apart from comments made by Belcher and Smooker (1937) and

ffrench (1976).

Communication has been defined historically as an

exchange of information between two animals for their mutual

benefit (Geist, 1971; Hailman, 1977; Marler, 1956, 1961;

Miller, 1983; Nugent, 1979; Post & Greenlaw, 1975; Zahavi,

1982). Animal X (the signaler) transmits a signal that

alters the behavior of animal Y (the receiver); however,

neither the signal nor the response constitutes

communication in themselves. Even if one animal signals and

the other responds, there has been no communication unless

the probability of response is altered from what it would

have been in the absence of the signal (Klopfer & Hatch,

1968; Wilson, 1975). Not all signals necessarily provoke an

immediate, specific response from the receiver. A stimulus

may simply change the animal's responsiveness to another

signal, or the stimulus may persist and be perceived after a

delay (Burghardt, 1970; Marler & Peters, 1977; Wilson,



I here discuss how signals may have evolved through

natural selection for effective communication. I mainly

address two questions: (1) What is the evidence that

communication takes place when individuals of three differ-

ent species of swallows interact with others of their own

species, and (2) What is the evidence that particular

behavior patterns of these birds are indeed signals? The

action of a signaler can be an overt behavior, such as a

visual or acoustical display, or a change in the sender's

morphology or physiology (such as coloration of an append-

age), or even an odor associated with an excretion (Smith,


Most signals in animals are transmitted by special

displays. A display is a specific behavior pattern that has

become specialized to modify the behavior of another indi-

vidual (Krebs & Davies, 1981). The process of evolution or

displays from non-display movements is called ritualization.

By definition, a ritualized pattern must have been, at some

time, less functional in communication and not a display

pattern or other specialized communicatory activity.

Previously, ritualization was seen by some workers as

reducing the amount of ambiguity of potentially informative

actions (Darwin, 1872; Morris 1956). In becoming more

stylized and exaggerated, the information content, and

therefore the signal property, of the display was enhanced.

This traditional view of communication argued that it was

possible to determine the information content of a signal

from a knowledge of the response of other animals to that

signal and from knowledge of the environmental and behav-

ioral contexts in which that response was given (Andersson,

1976; Dunham, 1966; Stokes, 1962a).

More recent ideas, however, suggest that displays are

likely to evolve that disguise the real internal state or

intentions of the animal (Charnov & Krebs, 1975; Dawkins &

Krebs, 1978, 1984). This view proposes that signals are

manipulative in function and are used to persuade recipients

to behave in ways that benefit the signaler. Threat dis-

plays, for example, may disguise an animal's intent to flee

by making it appear aggressive (Maynard Smith, 1974). To

view animal communication as potentially deceitful is a

sharp departure from the more traditional view of ritualiza-


This new view leads to a third hypothesis concerning

the evolution of ritualization, which emphasizes that

displays are the result of selection by receivers for

reliable signals. Zahavi (1979) suggests that displays

become more stereotyped and repetitive because stereotypy

provides a uniform background against which subtle differ-

ences between individuals are emphasized. Recipients are

selected to discriminate between signals on the basis of the

signal's reliability as an indicator of an animal's parental

ability, size, strength, or fighting ability. In this

manner a female may assess the courtship feeding or display

of a number of potential mates with their ability to gather

food for her young. Thus, current animal communication

theory is the focus of three conflicting hypotheses that

explain the selective advantage of ritualization:

(1) reduction of ambiguity or information sharing,

(2) manipulation, and (3) reliable or honest signals.

How Signals Originate

Early understanding of ritualization came from Huxley

(1914, 1923, 1966), Lorenz (1950, 1966), and Tinbergen

(1952, 1959) when they realized that displays usually

evolved from other movements that formerly had no signalling

function--the principle of derived activities. According to

these three ethologists, ritualization begins when some

behavior pattern that is functional in another situation

becomes "emancipated" from its original motivational context

and acquires a secondary value as a signal. For example, an

individual bird can begin by recognizing an open bill as an

overt threat. Once this happens, selection will favor

signalers who open their bills in a fighting context because

of the effect on opponents, and the open bill in certain

contexts begins to evolve into a threat display.

Historically, three primary sources of displays were

recognized by ethologists: intention movements, displace-

ment activities, and redirected activities (Andrew, 1956;

Daanje, 1950; Hinde & Tinbergen, 1958). Intention movements

are the preparatory or incomplete movements that often

appear at the beginning of an activity. During the process

of ritualization such movements are altered in a way that

makes their communicative function more effective. Inten-

tion movements tend to become simplified, stereotyped,

repetitive, and exaggerated in form (McFarland & Houston,

1981). They also may acquire morphological support in the

form of additional structures that enhance the

conspicuousness of the movement. One example is found in

the Mandarin Duck (Aix galericulata), which has some of its

wing feathers modified to form a bright orange sail that is

erected during courtship preening (Lorenz, 1941). The

ancestral preening movements are reduced to a quick turn of

the head so that the bill points to the orange sail.

Other displays have evolved from displacement activ-

ities--the "out of context" behavior patterns that often

appear in conflict situations. These movements may reflect

motivational conflict or indecision as the animal vacillates

between, for example, attack and moving away (Hinde, 1970).

Examples include displacement scratching in lovebirds

(Aqapornis species; Dilger, 1960) and domestic chickens

(gallus Domesticus; McBride et al., 1969), which appears to

be derived from activities related to the collection of nest


A third source for the evolution of displays is redi-

rected activities. The behavior pattern is in the right

context but directed at an inappropriate stimulus. When a

male approaches a potential rival with tendencies both to

intimidate and to flee, he may choose neither course of

action at first. Instead he performs a third, seemingly

irrelevant, act. He redirects his aggression toward the

ground or some object nearby such as a blade of grass.

Cullen (1966) and Tinbergen (1952) describe certain displays

of terns and gulls that may have been redirected behavior

derived from attack behavior.

After the appearance of Timbergen's 1952 article,

ethologists began to interpret communication systems by what

came to be known as the conflict theory of the origin of

displays. Subsequent neurophysiological experimentation on

a variety of animal groups failed to provide confirmation of

key elements of the Tinbergen model, and the conflict

hypothesis was modified from its original form (Baerends,

1975; Brown & Hunsperger, 1963; Deluis, 1973; Ewert, 1985;

Greenberg, 1977; Ishii, 1984). Later, Andrew (1963, 1972),

Hinde (1981), and Wickler (1969) interpreted ritualization

as a pervasive, highly opportunistic process that could be

derived from almost any available behavior pattern, anatom-

ical structure or physiological change. Additional sources

of displays have been traced to preening, feather-settling,

autonomic responses, and temperature-regulating movements

(Beer, 1975; McKinney, 1965; Wittenberger, 1981).

Intention movements and displacement activities still

tend to be singled out by many authors as the prime sources

for avian signal movements, but it has been realized for

some time that this is an oversimplification (Caryl, 1982;

Cullen, 1972; Hinde, 1970; Moynihan, 1955; Smith, 1977). In

particular, there seems to be increasing doubt about the

"out of context" interpretation of display origins through

the displacement phenomenon. This whole subject remains

controversial. The concept of displacement activities, once

frequently used loosely, is now widely recognized as unclear

and is to be used cautiously if at all (Hinde, 1970).

In general, the task of integrating the concepts of

motivational conflict and communication processes is still

one of the most difficult confronting ethologists. A major

difficulty in applying motivational concepts is that they

call for increasingly detailed causal analyses and yet they

remain untested until they can be translated into physio-

logical terms. Many researchers have directed the main

focus of their attention away from motivational interpre-

tations toward communication processes and questions of

adaptive function (Andrew, 1972; Beer, 1977; Brown, 1964;

Greenquist, 1982; Nuechterlein & Storer, 1982, 1985;

Robbins, 1983; Tinbergen, 1972; Willis, 1972; Wilson, 1975).

But, as McCleery (1978), McFarland (1971, 1977), Sibly and

McCleery (1985) and Sibly and McFarland (1976) have pointed

out, in order to understand what an animal will do at any

particular moment, it is necessary to take into account

causal and functional considerations simultaneously. For

example, if a bird is hungry but the risk of looking for

food is high because of predators, the bird may wait until

the predators have moved elsewhere. Birds have decision

rules that involve comparing their physiological state

(hunger) with risks in the environment, because these will

influence the animal's calibration of its internal state.

In the future, Sibly, McCleery and McFarland's argument may

lead to important links between causal and functional


In recent years, our understanding of animal displays

has been enhanced by the application of game-theory models

(Bishop & Cannings, 1978; Caryl, 1979, 1981; Hammerstein,

1981; Hines, 1977; Maynard Smith, 1974, 1976; Maynard Smith

& Parker, 1976; Maynard Smith & Price, 1973; Parker, 1974,

1978; Parker & Rubenstein, 1981). Gaming involves the

principle of optimization--an individual should attempt to

minimize its energy expenditures while maximizing its energy

gains. The greatest benefit to fitness can be derived from

a dispute by expending only that energy necessary to win.

The prediction is that when fighting or coercion entails a

large energy expenditure or a significant risk of physical

injury and a loss of reproductive opportunity, individuals

will exhibit less dangerous, more conventional behavior

patterns for settling disputes. These conventions usually

take the form of ritualized displays. Thus, all partici-

pants involved in a dispute will receive benefits from using

displays rather than coercion.

In this dissertation, I am most interested in the use

of communication by birds in conflict situations, specif-

ically in competition for limited resources. Because there

are three hypotheses about the selective advantage of

ritualization, I will discuss the three different models

with specific reference to threat displays.

The Information Hypothesis

According to the information or traditional view of

communication, ritualization is the result of making the

behavior of the signaler more predictable to a recipient by

making available some information about the internal state

or probable next behavior of the signaler (Cullen, 1966;

Dawkins & Kerbs, 1978, 1984; Smith, 1968, 1977). Implicit

in this definition is the requirement that the signal's

message is the information it conveys, and its meaning is

its effect on the receiver's behavior. By communicating,

the signaler gains a certain degree of control over the

actions of the receiver, and the adaptive significance to

the signaler of this type of control is the function of the

signal (Smith, 1977). For example, a bird seeing an

approaching intruder may show a specific display indicating

the high probability that it (the signaler) is, or soon will

be, engaged in attack. Upon seeing the display, the

intruder retreats immediately, indicating that information

was conveyed about what the signaler will do next. Commu-

nication is thus seen as a means by which receivers can

predict the future behavior of signalers and act appropri-

ately. Signalers are selected to inform receivers of their

internal state to make it easy for receivers to predict

their future behavior. Both parties benefit from displays

that are efficient, unambiguous, and as informative as

possible. The information content, and therefore the signal

property, of the action is enhanced when displays become

more stylized and exaggerated.

If, however, displays must be clearcut and unambiguous

to be most effective, increasing the clarity of signals by

increasing their stereotypy may reduce the amount of

information they convey about the signaler. Though signal

complexity is necessary to increase information content, it

increases the probability of misinterpretation at the same

time. For example, consider a bird that uses songs contain-

ing six syllables to transmit messages. Further, suppose

that for each syllable the bird can use any of the six sepa-

rate sounds. One message might consist of the same sound

repeated six times. Another message could consist of one

sound followed by a different sound repeated four times

followed by a third sound. The bird would have the poten-

tial to transmit 66 (or a total of 46,656) different mes-

sages. In avian communication, no bird is believed to have

such a system. Instead, motor patterns are simplified and

the components of displays are often repetitive. Morris

(1956) called this stereotypy (typical intensity).

In "typical intensity" the movement has the same form

over a wide range of motivational states. Morris viewed it

as the product of selection to reduce ambiguity in displays.

Charles Darwin (1872) discussed the principle of antithesis

when he described how threat and appeasement postures had

come to extreme opposites--a threatening dog stands erect,

while a submissive dog crouches or rolls on its back. By

making a movement more recognizable to another individual, a

display becomes more effective in eliciting a response.

The Manipulation Hypothesis

Caryl (1979, 1981, 1982), Charnov and Krebs (1975),

Dawkins (1976, 1982), Dawkins and Krebs (1978, 1984), and

Hammerstein (1981) have criticized the traditional viewpoint

that signals function primarily to provide information.

Instead, they favor a second hypothesis that signals

function to manipulate, prod, coerce, or persuade receivers

to the advantage of the signaler. Dawkins (1976, 1982)

pictured animals as machines designed to preserve and

propagate their genes. As a means to this end, animals may

manipulate both biotic and abiotic objects in the environ-

ment. Dawkins regarded object manipulation as an extended

expression, or part of the extended phenotype, of the

animal's genes. For example, a male bird does not expend

time and energy in physically moving a female to his nest;

instead he may sit at the nest site and sing. The female

may respond to the song by moving toward the male. From the

male's point of view, communication is clearly more effi-

cient than trying to acquire a female by force. Communica-

tion is no longer a cooperative exchange of information

between the signaler and receiver, but a signaler manipulat-

ing a receiver to its own advantage. Usually the signaler

and receiver benefit from the communication, and both will

be selected to maximize their own benefit.

In 1974, Maynard Smith described a model, the War of

Attrition, intended to represent an agonistic encounter that

was settled by display alone. He used an argument from

game-theory to show that individuals would not be expected

to provide truthful information or to respond to information

provided by an opponent because of the way in which this

strategy could be exploited by cheaters. When he considered

how natural selection was likely to act, it was difficult

for him to imagine how selection could favor information-

sharing. Instead, he assumed that there was nothing to

prevent animals from lying about what they would do next.

Moreover, for an animal to declare its intentions early in

an encounter would be like showing one's hand to an opponent

at the beginning of a card game. If two animals were

contesting a resource by means of ritualized displays, it

would not benefit either animal to reveal whether it would

attack or flee until the last possible moment. Thus,

Maynard Smith suggested that typical intensity had evolved

in threat displays to conceal the precise behavior probabil-

ities of the signaler, not to make the signal more easily

recognizable as was postulated by Morris (1956).

The Honesty Hypothesis

A third hypothesis, suggested by Maynard Smith and

Price (1973) and Zahavi (1979), is that ritualization is the

result of selection by recipients for reliable signals.

During evolution a number of different factors could have

encouraged the trend toward species-typical display stereo-

typy. Females would have increased fitness if they used

reliable cues to male fitness as a criterion for mate

selection. If females could detect some consistent species-

unique trait in the male display of their species,

they would avoid investment in non-viable or uncompetitive


Ritualization of display behavior during agonistic

interactions also would be selected if an animal's inten-

tion, degree of motivation, and potential strength (e.g.,

body size) in relation to the viewer's strength were con-

veyed by some set of signals. This information could be

used to predict who would win a physical encounter without

the need to actually fight. Both contestants would benefit

by avoiding energy expenditures and chance of injury.


There have been few empirical tests to distinguish

between these hypotheses. The information hypothesis

predicts that agonistic displays should communicate the

future behavior of the signaler with maximum clarity. The

honesty hypothesis predicts that receivers are selected to

discriminate between signals on the basis of their reliabil-

ity as indicators of an individual's size, strength, age, or

skill, which could affect the receiver's chance of victory

in a contest. In both the information and honesty

hypotheses, receivers would use the information in deciding

how to respond appropriately. The manipulation hypothesis,

in contrast, predicts that signalers will not give away any

information about their motivational state or future inten-

tions. According to this hypothesis, agonistic displays

should be poor predictors of ensuing attack and escape, and

there should be no consistent difference between the behav-

ior of the eventual winner and loser until near the end of

the contest. As a result it is difficult to predict which

individual will win or lose from the displays alone, and

information about intentions is not transmitted. One of the

goals in this study was to test the applicability of the

information, manipulation, and honesty hypotheses to the

information available from martin (Progne spp.) signals.


Making behavioral comparisons between closely related

species emphasizes the differences that have arisen since

speciation in a particular group. Such comparison may not

only reveal evolutionary pathways within a particular group,

but also may elucidate general patterns of evolution of

behavior. The taxonomy of the genus Progne (order Passeri-

formes) is in dispute. Specimens of all members of the

genus are still needed from many parts of their ranges,

since the locations of the wintering areas and migratory

routes are not well known.

Peters (1960) recognized five species of Progne: P.

subis (Purple Martin), with two races, subis and nesperia;

P. chalybea (Gray-breasted Martin), with two races, chalybea

and domestic; P. dominicensis (Caribbean Martin), with

three races, dominicensis, cryptoleuca, and sinaloae; P.

modest (Southern Martin), with three races, modest,

elegans, and murphy; and P. tapera (Brown-chested Martin),

with two races, tapera and fusca.

Hellmayr (1935) considered the Antillean forms

dominicensis and cryptoleuca and the west Mexican form

sinaloae to be races of subis. Although Ridgeway (1904)

recognized the close relationship between Cryptoleuca,

dominicensis, and sinaloae, he gave each specific status.

Zimmer (1955) and the A.O.U. Check-list (1957) also treated

cryptoleuca as a separate species. Zimmer gave dominicensis

specific rank, but also considered the isolated sinaloae

conspecific with dominicensis. Miller et al. (1957) dis-

agreed with Zimmer and his maintenance of sinaloae as a race

of dominicensis separate from subis. Zimmer (1955) pointed

out that treatment of populations of dominicensis as races

of subis seemed unwarranted if chalybea was maintained as a

distinct species, because dominicensis was intermediate in

characters and distribution between subis and chalybea.

Allan Phillips (1959, and pers. comm.) supported Zimmer.

Phillips collected specimens of sinaloae while working in

northern Mexico. He assigned them to dominicensis, stating

they bore no important resemblance to subis.

The relationships of the remaining two species of

Martins, modest and tapera, have been little studied.

However, the A.O.U. Check-list (1983) recommends making them

two species, the extralimital P. m. elegans and P. m.

modest. Mayr and Short (1970) consider subis, chalybea,

dominicensis, tapera, and modest a superspecies, and when

more is known about their relationships in the areas of

contact all might eventually be treated as subspecies of

subis, although this seems unlikely to me.


Study Areas

Field studies of the Purple Martin (PM), were conducted

in residential backyards at Gainesville, Alachua County,

Florida. I watched PMs almost daily from 15 February to

9 June 1977, from 24 February to 3 August 1978, from 21 Feb-

ruary to 14 March 1979, and from 8 February to 9 June 1980.

Over 2100 hours of observation were made. The principal

colonies occupied 16 man-made martin houses of different

sizes and designs totalling 168 nesting compartments. The

most studied colony was at the Gainesville Country Club.

This house contained eight compartments and was occupied by

six pairs of martins in 1977, seven pairs in 1978,

five pairs in 1979, and seven pairs in 1980.

The primary study site for Gray-breasted Martins (GM)

was the Trinidad-Texaco Inc. oil refinery adjacent to the

town of Pointe-a-Pierre, Trinidad. GMs nesting in holes in

broken steel pipes were studied from 18 April to 22 June

1979, for about 500 hours of observation. Additional

observations of GMs nesting among fissures in a cliff face

on the northeast coast of Galera Point, Trinidad, totaled

17 hours and were made during June 1979. I collected

supplemental data on GMs from July to August 1976 and 1977

at a secondary site in downtown Acapulco, Gerrero, Mexico.

The center of the city had several blocks of three- to

five-story buildings where birds nested in drainage pipes

and under the eaves of buildings.

I observed Caribbean Martins (CM) daily at a 2.7 hect-

are tract on the northwest side of Pigeon Point, Tobago,

from 23 June through 16 July 1979, totalling about 180 hours

of observation. The vegetation was palm trees (Euterpes

spp.) in grassy clearings with a few dense patches of native

broad-leaved trees and shrubs. CMs nested in dead palms in

holes excavated by Red-crowned Woodpeckers (Melanerpes


I observed all three species of martins during the

following five three-hour time periods: Dawn-0900, 0900-

1200, 1200-1500, 1500-1800, and 1800-dusk. Night behavior

was monitored once a week with a flashlight until birds



Aluminum PM houses in Florida were lowered in telescop-

ing poles, and doorpanels were raised to record nest data.

Ladders were used to reach PMs nesting in gourds and wooden

houses, and nest contents were observed with a flashlight.

A small, movable dental mirror attached to a fishing pole

was used with a flashlight to observe the contents of GM

nests in broken pipes. Nest contents of GMs nesting in palm

trees accessible by ladder were examined with a flashlight

and a dental mirror.

An observation blind was not used. My presence did not

obviously affect the birds' behavior as long as I was 10 to

20 meters away. Field notes were either written directly or

dictated into a portable cassette recorder for later tran-

scription. Line drawings and diagrams were made from field

notes, frames of 8mm movie film, or 35mm still photographs.

Super-8 movie film totaling 1350 meters was taken at

16 frames per second using a GAF ST602 movie camera and was

analyzed using a Bell and Howell 1623Z stop-action projec-


Tape recordings of the vocalizations were made using a

Nagra #III open-reel recorder at 19 cm/s or with Superscope

C-105, Uher CR-134, and Sony Model TC-55 cassette recorders

at 4.25 cm/s. Recorders were coupled with a Sennheiser 805

or a K2U-ME80 unidirectional condenser microphone, a Dan

Gibson E.P.M. Model P-200 parabolic microphone, or a Realis-

tic super cardioid condenser microphone. Vocalizations were

analyzed using a Kay Elemetrics 7029A Sona-graph and/or a

Spectral Dynamics Real Time Analyzer (Hopkins et al., 1974)

Model SD 301C-C. All tapes made during this study were

deposited in the Bioacoustics Archives of the Florida State

Museum, Gainesville, Florida.

Color-marking consisted of spreading enamel model

airplane paint on PM and GM feet, remiges, and retrices.

The color-marking technique had little observable effect on

bird behavior other than the birds occasionally pecking at

their feet to remove the paint. I was unable to color-mark

CMs, but natural variations in the plumage pattern of the

ventral surface served to identify a number of birds.

All PM males with entirely dark-blue feathers (i.e., at

least two years old) were called "adults." One-year-old

males having gray ventral parts with only a few dark blue

feathers on the crown, throat, and belly were called

"yearlings." Birds in both classes were sexually mature. I

did not try to separate the very similar adult and yearling

females. Because GMs lack obvious sexual dimorphism, sex

was determined from behavioral characters associated with

copulation, courtship, and/or territorial defense. All CM

males with blue-black feathers and conspicuously contrasting

white feathers from mid-breast to vent were called "adults."

I did not separate adult and yearling CM females, which were

duller than adult males and had extensive grayish brown

feathers on the upper breast and sides. Yearling CM males

were intermediate in plumage between the adult sexes. In

this study, any references to adult or yearling individuals

or pairs are with respect only to the age of the males (as

in Brown, 1978a).

Description of Behavioral Techniques

I considered PM male arrival dates as the earliest

dates after which I regularly saw the males at the colony.

An arrival date was not necessarily the date of territory

establishment or the date of pair bond formation. During

this study GMs and CMs were already at the breeding sites

when I arrived. I also determined the proportion of time

that males spent on the territory singing and defending

(including silent periods of alert perching), and the

proportion of time that they spent preening, loafing (inac-

tivity exceeding 30 seconds), foraging, and flying near the

colony. Physical displays were named according to their

characteristic motor patterns. Vocal displays were named

according to their onomatopoeic sounds. A signaler-receiver

reference system was adopted, e.g., signaler directs behav-

ior pattern A to receiver, which can respond with behavior

pattern B. The method used to determine display function

was to describe (1) the components of the display, (2) the

circumstances in which the display occurred, (3) the behav-

ior accompanying the display, and (4) the behavior shown by

both participants immediately before and after the display

was given.

Social behavior was monitored by use of focal-group

sampling (Altmann, 1974). Interactions were recorded by

noting (1) time of occurrence, (2) sex and identity, when

known, of both the individual that initiated the interaction

and the recipient, (3) the sequence and kinds of behavioral

patterns used by both individuals, (4) the behavior immedi-

ately following the interaction, and (5) the behavior of

both individuals immediately preceding the interaction.

I studied the patterns of social dominance at and near

the nesting sites. The frequency and outcome of those

encounters, which illustrated a dominant-subordinate rela-

tionship, were recorded on matrix tables according to which

animal was successful (columns) and which one was defeated

(rows). Criteria of subordination included the turning away

of lateral body presentation, avoidance, or fleeing of a

martin relative to the approach of another individual.

Scores for dominance and subordinance were calculated for

each animal as the ratio of its wins or losses, respec-

tively, in aggressive interactions to the total number of

encounters in which it was engaged. This provided a basis

for ranking the animals for dominance or subordinance. The

discerned dominance structure, therefore, was considered an

agonistic hierarchy (Deag, 1977).

Three classes of participants in agonistic encounters

were considered: (1) encounters between males (male-male),

(2) encounters between females (female-female), and

(3) encounters in which a male directed an act toward a

female (male-female). Female-directed acts toward males

were not included because of their low frequency. Rates of

agonistic interactions for each sex class were calculated by

dividing the number of agonistic encounters that occurred in

a specific sex class by the number of pair-hours of observa-

tion for that class. The number of pair-hours of observa-

tion was a measure of the time available to any pair of

animals for potential interactions and was calculated for

each class following the method of Struhsaker (1967).

Criteria for Recognizing Display Units

The study of ritualized behavior patterns involves an

ethological description of each pattern, tests to show that

the pattern is effective in communication, and measurements

and comparisons of the pattern with other patterns to

suggest its origin. However, variation in form of display

behavior raises several difficult problems for observers who

wish to describe display behavior and to determine the

number of display units in a species' repertoire.

In practice, two types of behavioral classifications

may be distinguished: one involves description by motor

pattern and the other involves description by consequence

(Hinde, 1970). In the first method, an attempt is made to

describe the actual pattern of muscle movements made by the

animal with little reference to the effects of the behavior

on the environment. Terms such as "wing flick" and "bill-

snapping" are of this type. When classifying by conse-

quence, the observer records the effect of the behavior on

the environment. Thus, one may use categories of behavior

such as an animal "staying put" or "retreating." In this

case, one emphasizes the important consequences and

directedness of behavior. In this study, the two types of

description intergrade, though each extreme can be quite

useful in the appropriate circumstances.

Questions about how much stereotypy is typical of

displays and how much stereotypy is necessary to justify

calling the behavior a display are under increasing dis-

cussion in ethology (pers. comm. Brockmann, 1984; Schleidt,

1974). Some degree of stereotypy seems necessary in order

for displays to be distinguishable from other stimuli

competing for the attention of the receiver. Indeed, one

secondary purpose of this study is to measure how much

evolutionary modification is required before a behavior

pattern is called a signal.

I made position measurements from 35mm photographs and

enlarged drawings traced from the projection of individual

frames of 8mm film. Two types of movements were measured:

(1) movements of the head, bill, trunk, wings, tail, and

legs during agonistic interactions, and (2) the same move-

ments when birds were not engaged in agonistic interactions.

I measured speed of movement as the number of frames of film

between the start of a movement and the completion of that

movement. Postures were measured as diagrammed in Figure 1.

Degrees of movement of the body or limbs were recorded as

the difference between starting and final positions. From

these measurements, I calculated the mean and variance of

final postures and speed of movements for behavior patterns

executed during agonistic and nonagonistic interactions. I

used the coefficient of variation (CV) to measure how

"fixed" or stereotyped a behavior, particularly a display,

was relative to other behavior patterns (Schleidt, 1974).

Figure 1. Schematic outline showing how body and limb
positions were measured.

Significant differences between CVs were determined using
the "c" test statistic (Dawkins & Dawkins, 1973):

c = (CV1-CV2) where Scv=CV
2 2
Scv +Scv2 2N
1 2

The probability associated with c was obtained from the

table for the distribution of t. Using this method, I could

test for a significant difference between the CVs for

behavior patterns used in different contests. I also

measured certain additional changes in behavior patterns:

(1) changes in speed of execution of movements, (2) freezing

of movements, (3) exaggeration of some components of move-

ments, (4) omission of components of movements, and

(5) decreases in variability of movements.


Vocal Displays

Shiovitz (1975) refined the definitions of Bondesson

and Davis (1975) in an attempt to standardize bird vocaliza-

tion terminology. Unfortunately, the classification scheme

proposed by Shiovitz was not adequate to fully describe

Progne vocalizations; hence, I have employed the following


Note: Any short, uninterrupted sound represented by a

continuous trace on a sonogram.

Syllable: A sound consisting of a note or group of

notes; the basic structural unit of a song,

separated from adjacent syllables by a silent

period of 0.02-0.20 seconds.

Terminal Syllable: A syllable terminating a song,

unlike syllables making up the main phrase.

Call: A vocalization represented by a single discrete

figure on a sonogram.

Song: A unique combination of syllables in a

stereotyped sequence, often shared by several

members of the population. The distinction

between song and call in this study conforms to

that of other workers (see Armstrong, 1963).

Phrase: A natural section of the song; a grouping of

syllables in a characteristic temporal sequence.

Song Type: Used interchangeably with theme; a fixed

sequence of phrases characterizing a population

or group of populations.

Trill: A consecutive series of similar syllables

repeated serially at about 4 to 17 per second,

producing a rattling sound.

No attempt was made here to describe in quantitative detail

all parameters of all song types. Instead, I have empha-

sized those general features shared by most song types since

they apparently identify the songs as those of Progne.

Visual Displays

In the recent literature of animal communication

certain terms have come to mean different things to differ-

ent investigators (Caryl, 1979, Dawkins & Krebs, 1978;

Hinde, 1981). For that reason I have defined and will use

the following terms.

Intention Movement: These are incomplete movements

and precursory movements.

Act or Action Pattern: A set of observable activities

not analyzable into separate occurring compo-

nents (Russell et al., 1954).

Signal: The behavior (e.g. posture, display,

vocalization) transmitted by the signaler.

Display: Any stereotyped behavior pattern (such as

vocal or visual stimuli), alone or in combina-

tion, that serves a signal function to other


Ritualization: The process by which a behavior pattern

becomes stereotyped and acquires a signal


Signaler: An individual which transmits a signal.

Receiver: An individual whose probability of behaving

in a particular way is altered by a signal.

Context: The setting in which a signal is transmitted

and received.

Communication: Action on the part of an organism

(signaler) that alters the pattern of behavior

in another organism (receiver).

Territory: A restricted area defended by an animal

against intruders, usually for breeding pur-


Dominance Hierarchy: A social ranking, formed through

agonistic behavior, in which individuals are

associated with each other so that some have

greater access to resources than do others.

Agonistic Behavior: Any behavior associated with

threat, fighting, and retreat.

Approach: Movement of one or both martins toward each

other, but with no attempt to strike with the


Threat: Indicates that the signaler could act


Supplant: Replacing an individual at a given place.

In this action, one martin moves to the place

where a second martin is perched, but does not

follow the supplanted individual.

Attack: Movement toward an individual that culminates

in physical contact between individuals.

Fight: When a receiver retaliates to a signaler, both

martins attack.

Chase: Pursuit of a second, moving individual.

Retreat: Behavior in which one bird moves away from

another in response to an attack or display.

Stay: Neither attacking nor retreating, but remaining


Inactive: A bird resting and not engaged in obvious

interactions for 30 seconds or more.


This section describes visual communication during the

breeding season in the three species of martins that I

studied. Detailed description of behavior is a prerequisite

for the analysis of its function. No matter what the aim of

the ethological study, all such studies must start by

splitting behavior into categories, even if only to enable

the observer to decide which he should include and which he

will ignore. If all behavior patterns were species-typical

and invariant fixed action patterns, this task would be an

easy one. In birds, many behavior patterns vary consid-

erably in form and intensity, making classification more

difficult. By specifying precisely what an animal does and

does not do, one can begin to cut down the number of tenable

hypotheses about underlying function.

The first part of this study was designed to obtain

some measure of stereotypy and to compare ritualization in

these three martin species. The second part contains

descriptions of behavioral acts. Some of these acts have

been described previously (Allen & Nice, 1952; Brown, 1978a,

1978b; Johnston & Hardy, 1962), but are described again to

bring out the important features emphasized in this study.

I will concentrate on those behavioral acts that have

acquired a special function in social communication and are

considered displays. A few displays are considered in some

detail if they have not been studied by other researchers in

the past, and in a few cases I have separated some complex

displays into simpler parts. The third part deals with the

actual process of communication. Communication involves

concepts such as the context in which the signal occurs, the

message "intended" by the signaler, and the meaning of the

signal for the recipient (Smith 1965).

Stereotypy of Behavior Patterns

Some degree of stereotypy is necessary in order for

displays to be distinguished from other stimuli competing

for the attention of the receiver. A researcher has to deal

with the question of how much stereotypy is required to

constitute the criteria for calling a behavior pattern a

signal. The coefficient of variation (CV) has been used to

measure how "fixed" or stereotyped a behavior, particularly

a display, is relative to other behavior patterns (Schleidt,

1974). The CV is the standard deviation expressed as a

percentage of the mean (CV=SDxlOO/x; Hazlett, 1972). The

greater the CV, the greater the variability in the data set.

Bekoff (1977) discussed whether it might be possible to

define some cut-off point between the categories of "fixed"

and "variable." He was asking whether it would be possible

to set up a standard CV against which a variety of data

could be compared. For example, it might be the case that

one could recognize a behavior pattern and differentiate it

from another behavior pattern when the CV is less than 30%

(Schleidt, 1974). On the other hand, birds are undoubtedly

better perceivers of their own conspecific behavior than are

humans, and a 30% variability may be too high. According to

Barlow (1977) and Wiley (1973), the maximum values of CV

that might be allowable in order to refer to a behavior

pattern as stereotyped are undecided and relatively

arbitrary. Since the communicative value of many displays

varies with context, guidelines for the use of CV measures

also would be difficult to formulate. Therefore, it seems

premature to attempt to define a cut-off point between

"fixed" and "variable," when further quantitative research

could answer questions dealing with motor pattern variation.

To obtain some measure of stereotypy, I took motion

picture films of nonagonistic movements and agonistic

interactions of martins. The position measurements were

taken as outlined in Figure 1. At this stage of the

analysis, I did not classify movements into the display

categories, since this would have biased the results (see

section on description of displays). That is, some of the

movements of the body and limbs that occurred in agonistic

situations were not clear "head presentation" or "body

extension" displays. Nevertheless, all movements were

included in the analysis, making the data more confusing

than if cleaned up by exclusion of patterns that "did not

fit." In Table 1, the means, standard deviations, and

coefficients of variability of the final position and

duration of movements of the different motor patterns are

prestned. In Table 2, the differences in the variances of

the movements were tested using the c statistic for the

final position and duration of movements.

As shown in Table 2, the final posture of the body for

movements executed during agonistic interactions was signif-

icantly less variable (more stereotyped) than the final

posture of the body during nonagonistic interactions such as

during locomotion or loafing. The speed of movement also

was significantly less variable for agonistic body move-

ments. The final posture of the head during aggressive and

appeasement interactions was significantly less variable

than the posture of head movements performed during non-

agonistic situations. For the tail movements, the differ-

ences in position variance were significant, while

differences in duration of tail movements were not signif-

icant. I found the final position of the wing to be less

variable in aggressive interactions than the final position

of wing movements involved in nonagonistic situations. The

final position of the wings during appeasement interactions

was significantly less stereotyped. There was no difference

in the speed of wing movement during either agonistic or

nonagonistic situations. It was important that the movement

with the highest degree of stereotypy, that of the bill,

Table 1. Comparisons of Agonistic and Nonagonistic Body
and Limb Movement, Final Position, and Duration of
Movements in Seconds.

Final position
(degrees) Duration
N Mean SD CV Mean SD CV

Purple Martin
Body-nonagonistic 50 42.2 11.2 26.5 6.7 2.1 31.3
Body-aggressive 31 22.9 0.84 3.6 2.5 0.08 3.2
Body-appeasement 35 90.3 6.4 7.1 2.8 0.29 10.3
Head-nonagonistic 50 70.5 28.6 40.5 1.3 0.67 50.3
Head-aggressive 31 39.2 5.1 13.0 0.66 0.05 7.5
Head-appeasement 35 86.7 6.4 7.3 1.0 0.11 10.8
Bill-nonagonistic 50 75.4 32.3 42.8 2.4 0.79 31.9
Bill-aggressive 31 28.6 1.5 5.2 0.78 0.12 15.3
Bill-appeasement 35 112.5 4.5 4.0 0.85 0.09 10.5
Tail-nonagonistic 50 26.0 18.9 72.6 3.1 1.6 50.6
Tail-aggressive 31 30.6 4.8 15.6 0.81 0.06 7.4
Tail-appeasement 35 24.1 3.9 16.2 1.2 0.09 7.2
Wing-nonagonistic 50 41.8 17.5 41.8 2.7 0.89 32.9
Wing-aggressive 31 36.8 5.3 14.4 0.34 0.05 14.7
Wing-appeasement 35 47.4 7.0 14.7 0.79 0.08 10.1

Gray-breasted Martin
Body-nonagonistic 32 40.1 11.2 16.7 5.3 1.8 33.9
Body-aggressive 17 20.7 0.94 4.5 2.4 0.09 3.7
Body-appeasement 21 92.6 7.3 7.8 2.6 0.23 8.8
Head-nonagonistic 32 68.4 21.5 31.4 1.7 0.71 41.7
Head-aggressive 17 38.4 4.9 12.7 0.72 0.08 11.1
Head-appeasement 21 86.5 7.2 8.3 1.3 0.16 12.3
Bill-nonagonistic 32 77.1 29.3 38.0 2.1 0.85 40.4
Bill-aggressive 17 27.0 1.4 5.1 0.66 0.11 16.6
Bill-appeasement 21 110.3 6.1 5.5 0.91 0.07 7.6
Tail-nonagonistic 32 27.2 18.6 68.3 4.4 2.5 56.8
Tail-aggressive 17 31.5 5.3 16.8 0.78 0.06 7.6
Tail-appeasement 21 23.7 4.1 17.3 1.7 0.08 4.7
Wing-nonagonistic 32 42.9 19.2 44.8 2.5 0.70 28.0
Wing-aggressive 17 37.8 5.0 13.2 0.41 0.05 12.1
Wing-appeasement 21 45.2 7.8 17.3 0.67 0.07 10.4

Caribbean Martin
Body-nonagonistic 12 39.7 16.1 40.5 5.9 2.2 37.2
Body-aggressive 5 22.6 0.79 3.4 2.4 0.07 2.9
Body-appeasement 7 87.5 6.8 7.7 3.0 0.29 9.6
Head-nonagonistic 12 70.6 27.1 38.3 1.5 0.68 45.3
Head-aggressive 5 39.8 4.8 12.1 0.85 0.09 10.6
Head-appeasement 7 85.5 5.7 6.6 1.4 0.17 12.1


Table 1. Continued.

Final position
(degrees) Duration
N Mean SD CV Mean SD CV

Bill-nonagonistic 12 74.9 31.3 41.7 2.1 0.83 39.5
Bill-aggressive 5 27.3 1.7 6.2 0.98 0.14 14.2
Bill-appeasement 7 113.6 5.6 4.9 0.63 0.05 7.9
Tail-nonagonistic 12 27.5 15.6 56.7 4.2 2.1 50.0
Tail-aggressive 5 32.1 5.2 16.2 0.56 0.08 14.2
Tail-appeasement 7 25.3 4.3 16.9 1.5 0.16 10.6
Wing-nonagonistic 12 40.9 13.2 32.2 2.6 1.1 42.3
Wing-aggressive 5 34.6 4.5 13.0 0.48 0.07 14.5
Wing-appeasement 7 45.2 6.6 14.6 0.89 0.13 14.6

Table 2. Comparisons of Variances of Agonistic and Non-
agonistic Movements.

Final Position Duration
c ratio p c ratio p

Purple Martin
Aggressive-nonagonistic 3.46 <0.05 4.06 <0.05
Appeasement-nonagonistic 3.17 <0.05 3.71 <0.05
Aggressive-nonagonistic 2.59 <0.05 2.68 <0.05
Appeasement-nonagonistic 3.04 <0.05 3.21 <0.05
Aggressive-nonagonistic 2.35 <0.05 2.83 <0.05
Appeasement-nonagonistic 2.97 <0.05 3.17 <0.05
Aggressive-nonagonistic 2.86 <0.05 0.37 <0.05
Appeasement-nonagonistic 3.54 <0.05 1.23 <0.05
Aggressive-nonagonistic 3.72 <0.05 1.17 <0.05
Appeasement-nonagonistic 0.87 <0.05 1.34 <0.05

Gray-breasted Martin
Aggressive-nonagonistic 3.27 <0.05 3.34 <0.05
Appeasement-nonagonistic 3.09 <0.05 2.96 <0.05
Aggressive-nonagonistic 2.78 <0.05 3.94 <0.05
Appeasement-nonagonistic 3.35 <0.05 3.82 <0.05
Aggressive-nonagonistic 4.02 <0.05 2.72 <0.05
Appeasement-nonagonistic 3.12 <0.05 3.69 <0.05
Aggressive-nonagonistic 3.43 <0.05 0.32 <0.05
Appeasement-nonagonistic 2.91 <0.05 0.75 <0.05
Aggressive-nonagonistic 2.74 <0.05 1.06 <0.05
Appeasement-nonagonistic 1.23 <0.05 0.74 <0.05

Caribbean Martin
Aggressive-nonagonistic 3.25 <0.05 2.76 <0.05
Appeasement-nonagonistic 2.95 <0.05 2.62 <0.05
Aggressive-nonagonistic 4.21 <0.05 2.67 <0.05
Appeasement-nonagonistic 3.54 <0.05 3.12 <0.05


Table 2. Continued.

Final Position
c ratio p

c ratio p


3.72 <0.05
2.65 <0.05

3.06 <0.05
3.39 <0.05

2.86 <0.05
1.37 <0.05

2.83 <0.05
3.16 <0.05

0.95 <0.05
0.46 <0.05

1.02 <0.05
1.53 <0.05

also was the movement that effected the greatest change in

the behavior of the recepient martin in all three species

(see section on analysis of displays).

In addition to data on variability of movement, the

films also provide information on other differences between

display movements and nondisplay movements in Progne. As

noted by McKinney (1975), exaggeration of one component of a

movement may occur during ritualization. When comparing the

forward movement of the body during agonistic and nonagon-

istic situations, the body was greater than the normal

position in both final position (230 from the substrate

compared to 420 average for nonagonistic movements) and

degrees of arc traversed from the perpendicular (average 67

for aggressive movements, average 480 for walking move-

ments). However, the range of nondisplay movements was

greater than that of display movements. Similarly, the

average final positions of the head and bill were greater

for aggressive display movements than for nondisplay move-

ments (Tables 1 and 2). During appeasement displays, the

body, head, and bill were raised laterally to a position

1150 15 up from the dorsal-ventral axis of the bird's body.

The final position of the body, head, and bill also were

significantly higher in degrees from the substrate in

appeasement situations than for movements involved in

loafing or alert watching (Tables 1 and 2).

Another characteristic of ritualized movements men-

tioned by Morris (1966) was the "freezing" of components of

the movements. For example, in the movements concerned with

walking, the number of frames of film showing the PM body,

head, and bill in the final position was very small (18-

25 frames). In contrast, the head was held in the forward

position between 2.7 and 3.5 seconds (49-63 frames), while

the body was held in the forward position between 4.2 and

6.4 seconds (75-115 frames) for aggressive displays. During

appeasement displays, the head was held away from an

opponent between 3.4 and 4.7 seconds (61-85 frames), while

the body was held in the outward position between 4.9 and

7.1 seconds (88-128 frames).

Morris (1966) noted changes in the speed of execution

of movements as another characteristic of ritualization.

For all three species studied, the agonistic movements of

the body, head, and bill were performed more rapidly than

general locomotory movements of those limbs (Tables 1 and

2). For example, during PM aggressive interactions, the

head was raised forward at 0.66 0.05 seconds compared to

1.33 0.67 seconds during nonagonistic movements (c=2.59,

p < 0.05; Table 2). It is important to note that there was

no significant difference in the speed of movement for wing

and/or tail movements. This fact suggested that movement to

the final posture was not as important as the final posture

itself. Similar results were found for GM and CM (Table 1

and 2).


There are few, if any, concepts that are more central

to the development of modern ethological ideas than that of

the fixed action pattern (for reviews see Barlow, 1977;

Eibl-Eibesfeldt, 1975; Krebs & Davies, 1981; Schleidt,

1974). Stereotyped motor patterns in vertebrates usually

serve either for communication or for locomotion. Moreover,

Schleidt (1974) suggested that actions that occurred at a

very high frequency and/or were adapted to deal with uniform

situations or objects such as pecking, biting, swallowing,

chewing, digging, and preening also should be highly stereo-

typed. Although locomotor behavior exhibited little varia-

tion within short time periods (at least in vertebrates), it

demonstrated considerable plasticity over longer intervals.

In contrast, displays that subserved communication were

nearly invariant over much of an animal's adult life span.

Since the idea of fixed action patterns generated

considerable controversy during the 1950s and subsequent

decades (Alcock, 1975; Lehrman, 1953, 1970; Lorenz, 1965),

one might be surprised to find that there have actually been

very few detailed quantitative analyses of actions that have

been categorized as fixed (Barlow, 1977).

Some displays were clearly very fixed when a single

individual was studied. Wiley (1973) measured the time

between the first swish and the first snap in 45 successive

repetitions of the strut display of an individual sage

grouse (Centocercus uropasianus). The average interval was

1.55 seconds and the standard deviation was 0.011 seconds

giving a CV of 0.7%. Stamps and Barlow's (1973) study of

Anolis lizard behavior showed that display behavior can be

extremely stereotyped. Measurements of the durations of

elements in the signature bob display of individual lizards

yielded CV's ranging from 2.5 to 9.5%.

Variation between individuals tended to be greater. As

Barlow (1977) pointed out, the displays of the goldeneye

(Bucephala clangula) studied by Dane, Woldott, and Drury

(1959), and often quoted as essentially invariant, mostly

showed CV's of 10 to 20%. However, the bowspirt, nodding,

and ticking had CV's of 17.14%, 72.22%, and 26.3%, respec-

tively. Figures taken from Dane and van der Kloot (1964)

for the highly stereotyped head-throw display yielded a CV

of 7.2%. Three measures of duration taken from the signa-

ture bob of Anolis by Stamps and Barlow (1973) gave CV's

ranging from 11.5 to 39% while those for three morphological

features were 10 to 12%. Hazlett (1972) studied the varia-

bility of movements of the chelipeds and ambulatory legs of

spider crabs (Microphyrs bicornutus) and found CV's never

less than 10%. Nonetheless, the displays were sufficiently

different from the non-display behaviors for both humans and

spider crabs to tell them apart. These attempts to quantify

the concept of the fixed action pattern demonstrated that,

although some displays showed relative fixity, ideas about

invariant stereotypy were not supported.

In the present study, the agonistic display movements

of martins were shown to be more stereotyped than the

non-display movements from which presumably they may have

evolved. Of course, this conclusion, even when based on

repeated observation of single-frame analysis film, still

remains rather subjective. The posture and the duration of

movements, however, were the best criteria of stereotypy or

variability that could be measured objectively.

If body and limb positions were assumed to be an index

of stereotypy, it was encouraging to find that the various

filmed actions were very uniform. I found marked stereotypy

when comparing different males while using position or

duration measurements of displays. For example, the

31 observations of body position during an aggressive

display had a mean of 2.990 from the subtrate and a standard

deviation of only 0.84. However, there were some body

movements (such as those of the wing and tail) which

appeared in a relatively constant form, but which varied in

duration. A detailed analysis showed that this variation

was produced by a difference in the speed of the movement at

times, while at other times by the variable length of the

pauses at the beginning and end of the active part of the

movements themselves. If the pauses were disregarded, the

duration the body was held in an aggressive display became

5.7 seconds with a standard deviation of 0.09 seconds.

Therefore, the measurements the movements' duration con-

firmed the impression of stereotypy. Thus, display

movements were distinguishable from no display movements by

exaggeration of movement (body movement forward to the

horizontal plane), exaggeration of one component of the

posture (open bill), and by the development of stereotypy.

In summary, there appear to be extremely few, if any,

behavior patterns that are absolutely fixed (e.g., CV = 0%

or even less than 1.0%). Indeed, according to Eibi-

Eibesfeldt (1975), the implication of absolute morphological

rigidity of displays was unfortunate and not originally

intended. It would be misleading to concentrate solely on

attempting to "prove" stereotypy or "fixity" without dealing

with patent variability. Various factors may favor vari-

ability in the characteristics of displays. For example,

variable aspects may be anti-monotony devices (Hartshorne,

1973), since too much stereotypy may cause the receiver to

habituate to the signal before responding in a manner

appropriate for the sender. Barlow (1968) suggested that it

might be more correct to drop the word "fixed" and substi-

tute the word "modal" (to refer to the most frequently

occurring form of a given act). The term modal action

pattern was meant to embody the idea that, while not

strictly fixed, these behavior patterns could be identified

in a reliable statistical way. Barlow's point is an impor-

tant one when one's main interest is in choosing categories

for data collection rather than in a particular theoretical

framework. Quantitative data from this study supported

Barlow's suggestion.

The frequent occurrence of behavior patterns having the

following characteristics can be used when describing modal

action patterns: (1) they are species-typical, (2) the

component movements that make them up occur together either

simultaneously or sequentially with a high degree of pre-

dictability and (3) they are repeatedly recognizable. In

addition to the terminological problems, it is important to

stress that the "meaning" of the value of CV's will

undoubtedly vary from species to species and from act to


Description of Visual Displays

The following is a list and description of the visual

displays seen during the course of this study. Similarities

between the displays of the three species suggested homol-

ogous relationships and permitted a parallel categorization.

The names of the displays were chosen to be as descriptive

as possible without implying motivation or function to the

behavior patterns.

However, before a discussion of agonistic behavior can

be undertaken, the normal body position of martins must be

described so that it may be compared with the principal

display patterns (Figure 2). All three martin species sat

in an upright position during nonagonistic situations. The

neck was withdrawn, the bill was held approximately horizon-

tal, and the feathers were relaxed. The wings were held

close to the body and the tail was stationary.

Figure 2.

S C.

Postures associated with loafing or inactivity:
A. Purple Martin; B. Gray-breasted Martin;
C. Caribbean Martin.

Head Forward Thrust

In all three species of martins, the Head Forward

Thrust display and its variants were found to be the chief

displays preceding attack. In the most common posture, a

bird assumed a horizontal crouch with the plumage sleeked

and the bill pointed at the opponent. This display preceded

an attack and appeared in territorial defense during the

approach of a potential intruder.

Purple Martin. The display was divided into two cate-

gories, the low-intensity Head Forward Thrust and the

high-intensity Head Forward Thrust (Figure 3A dnd 3B). In

the low-intensity Head Forward Thrust, the closed bill was

directed horizontally toward the opponent (92% of film

records, 8% open bill; n=29), the neck was partially

extended (96%, 4% retracted), and the legs were flexed (98%)

with the body held in a horizontal position (94%, 6%

upright). The feathers of the crest, breast, and back were

sleeked (92%, 8% crest raised), with the wing and tail

feathers continually flicked (100%). This posture was

maintained for at least 4 seconds (x = 4.9 seconds of field

records; n=63). The response of an intruder was to advance

no further (81% of field records; n=218) or leave (19%).

Cher (p. 111) and Chortle (p. 116) calls often accompanied

this display (70% of field records; n=227).

Since a bird giving this display most often remained

stationary (65% of field records, attacked 31%, retreated

4%, n=379), the display may be interpreted as suggesting a





Figure 3.

Displays associated with threat behavior:
A. low intensity Head Forward Thrust (Purple
Martin); B. high intensity Head Forward Thrust
(Purple Martin); C. low intensity Head Forward
Thrust (Gray-breasted Martin); D. high intensity
Head Forward Thrust (Gray-breasted Martin);
E. variation of the high intensity Head Foward
Thrust (Gray-breasted Martin); F. low intensity
Head Forward Thrust (Caribbean Martin); G. high
intensity Head Forward Thrust (Caribbean Martin).

weak tendency to attack. Thirty-six percent of all agon-

istic encounters ended with the winner showing low-intensity

Head Forward thrust (n=153). Sixty-four percent showed no

visual display (n=153).

In the high-intensity Head Forward Thrust, the bill was

held closed (98% of film records, 2% open bill; n=24) in

line with the opponent (100%), and the body was always

horizontal (100%). The signaler sleeked the feathers of its

crest and body (98%, 2% just crest), extended its neck (91%,

9% retraced), and moved forward in the Lunge, striking the

target bird on its head or body (38%) if it failed to

retreat (62%).

In 94% of film records (n=18) of high intensity encoun-

ters that preceded the Lunge, birds held the wings away from

the side with the primaries backward and with the wing and

tail feathers stationary (6% of the time wings against body

and tail and wing feathers flicked), and gave Hee-hee calls

(p. 87). In 79% of the field records (n=379), the high-

intensity display was followed by the opponent retreating or

remaining stationary (21% returned the attack).

Low-intensity Head Forward Thrust was directed by males

toward both males (83% of field records; n=441) and females

(17%). However, high-intensity Head Forward Thrust by males

was never directed toward females. Both head Forward

displays occurred 89% of the time at the boundaries of

territories when resident males were less than 20 cm apart

(n=217). The displays ceased 97% of the time when one bird

returned to the center of his territory or entered his

nesting compartment.

If two neighboring males showed either of the head

forward displays later in the season while young were being

fed, only 17% of these encounters ended in aggressive

interaction (n=133). Most displays (83%) ended with the

birds remaining motionless and at least one bird bill-wiping


Gray-breasted Martin. GM assumed low- and high-

intensity Head Forward Thrust displays similar to those of

PM, but in GM these displays were more elaborate. In the

low-intensity Head Forward Thrust, the displaying individual

oriented on a perch so that the axis of its body ran paral-

lel with the surface on which it was perched (87% of film

records, 13% upright; n=29), the neck was extended (93%,

7% retracted), and the wings and tail were flicked upward

repeatedly (95%, 5% remained still), making the bird appear

larger than normal (Figure 3C). Moreover, the feathers of

the cheek and crest were raised (96%, 4% sleeked). This

component of the display was noticeably absent in the PM low

intensity display posture.

At high intensity, the GM held the body parallel to the

surface and thrust the head and bill forward below the level

of the back (94% of film records, 6% other body variations;

n=16), held the wings just free of the flank feathers, and

partly spread the primaries below the line of the tail (88%,

12% above the tail line; Figure 3D).

Zurr calls (p. 97) were given by the birds in low-

intensity thrust and Zwat calls (p. 96) were given during

high intensity thrust. High-intensity Head Forward Thrust

in the GM, as in PM, was never directed toward females.

A slightly different variation of the high-intensity

Head Forward Thrust seen in the PM was observed in the GM

(Figure 3E). During this display variant, the neck was

retracted (88% of film records, 12% extended; n=32), the

legs extended (91%, 9% retracted), the bill opened (94%,

6% closed), the body feathers sleeked (100%), and the

carpals raised as the wrist was rotated outward (100%).

Birds that gave the carpal raised variation always attacked

opponents (n=32).

Caribbean Martin. As in the PM and GM, the Head

Forward Thrust in the CM varied in intensity and was divided

into two categories (Figure 3F and 3G). In the

low-intensity Head Forward Thrust, the closed bill was

directed toward the opponent (99% of the field records,

1% directed away; n=32), the neck was not extended (84%,

16% extended), the legs were flexed (85%, 15% retracted),

and the body was held in a horizontal posture with the wings

and tail continually flicked (100%).

In high-intensity thrust, the bill pointed straight out

(93% of field records, 7% bill to side; n=47), and the neck

was stretched (100%). At times, the wings were rotated

(12%) as in the carpal raised variation of the GM. On the

part of the attacking bird, the low intensity thrust was

accompanied by Wheet calls (p. 101), whereas the high

intensity thrust was accompanied by Kweet calls (p. 103).


Gaping in all three species was initiated by any

conspecific's movement close to the territory boundary. The

body posture was very similar to that of the Head Forward

Thrust (Figure 4A), except that the bill was held wide open

for two or three seconds (PM: x=2.8 sec., n=31; GM:

x=2.6 sec., n=30; CM: x=3.4 sec., n=13). The body was

inclined forward from the pelvis (91% of the field records,

9% more upright; n=60), and the neck extended forward

(100%). The wings were held by the side of the body (85%,

n=60), but were sometimes spread out (for balance) if the

martin moved forward (15%). Both sexes used the Gape. The

Gape may be silent or accompanied by a Rattle call (p. 91)

in the PM and a Rattle call (p. 100) in the GM. CM appear

to gape silently.

In all three species, mutual Gaping between mates was

common early in the breeding season (Figure 4B). This

behavior was observed mainly at the nest hole when a male or

female landed near its mate. One bird often took the offen-

sive while its mate was defensive. In offensive Gaping, the

martin leaned forward with the most weight on its phalanges.

It sometimes sidled forward in small steps or remained

motionless. In defensive Gaping, the body was held back,

Figure 4. Displays associated with threat behavior:
A. Gaping; B. Gaping between mates;
C. Defensive Gaping; D. Lunge.

the neck withdrawn; the bird sometimes moved backward or

turned aside while returning the Gape (Figure 4C).


Bill-snapping in all three species consisted of loud

clicking produced as the mandibles were forcibly brought

together. This display was given from the head Forward

Thrust (72%, n=25) or Lunge (28%). The body was held

horizontally (89%, n=25) or slightly downward (11%) and the

bill was snapped as an audible reinforcement of threat.

Bill-snapping was used by martins in their territories

(87%, n=25), but occasionally occurred outside the terri-

tories (for example, when a martin was approached too

closely while resting on a perch, 13% of the time). It was

directed at strangers moving past the territory (10%, n=20)

or at neighbors moving about in adjacent territories (90%,

n=20). It also was directed at interspecific intruders,

such as House Sparrows (Passer domesticus, n=31) and Star-

lings (Sturnus vulgaris, n=13). Bill-snapping also was used

in conjunction with Gaping in both interspecific inter-

actions and interactions between members of a mated pair.

This display was possibly a ritualized intention to bite.


In all three species, when a high-intensity Head

Forward Thrust did not dislodge an opponent, the attacker

frequently proceeded directly at the opponent. The

high-intensity Head Forward Thrust led to the Lunge in 58%

of field records for the PM (42% remain inplace, n=207), 32%

in the GM (68% remain inplace, n=193), and 47% in the CM

(53% remain inplace, n=12). The most conspicuous feature of

the Lunge was the forward movement. In this display, the

bird moved toward the stimulus object with a sudden rush and

then stopped, with head and neck outstretched, and the bill

directed at the opponent and slightly agape (100%,

Figure 4D). At the extension of the Lunge the wings were

twisted upward over the tail, and in 36% (n=309) of all

observations the mandibles were closed, producing a snap

that was audible at distances greater than 10 m (64%

produced no sound).

The Lunge was an attack that was generally sufficient

to make an opponent flee (Tables 6 & 7). If the opponent

failed to retreat, a fight resulted. The Lunge merged into

a fight in 19% of field records in the PM (81% the opponent

retreated, n=201), 11% in the GM (89% opponent retreated,

n=194), and 15% in the CM (85% opponent retreated, n=16).

High-Up Displays

High-Up displays were considered to be retreat behavior

because they were directed away from an attacking bird and

were given by a retreating bird. The High-Up postures were

divided into two broad categories: Withdraw High-Up and

Alert High-Up. Alert and Withdraw High-Up displays were

performed by both sexes.

The Withdraw High-Up display included the turning away

or lateral body presentation, avoidance, or fleeing of a

martin relative to the approach of another individual. The

feathers of the body and head were erect, the head was held

above the horizontal plane, the bill pointed downward, and

the neck held backward (Figure 5A). During a territorial

encounter the bird showed Withdraw High-Up followed by

staying (PM: 56%, n=380; GM: 61%, n=220; CM: to few

records) or retreating (PM: 44%, GM: 39%, CM: to few

records) or retreating (PM: 44%, GM: 39%, CM: to few

records), but in all three species the birds never attacked.

Purple Martin. A PM may show Withdraw High-Up from an

opponent in several different ways depending on the

agonistic encounter. A retreating martin performed one of

several behavior patterns, which included rocking back on

the tarsi while retracting the neck (93% of film records,

7% extended the neck; n=31), leaning to one side (78% of

film records, 12% straight ahead, n=31), turning the head or

the head and body away from the aggressor (84% of film

records, 16% toward the opponent, n=31), sidling away, and

flying away from the aggressor. Sidling, sideways movement

either toward or away from another individual, usually took

place during periods when the martins were sitting on the

birdhouse or on a telephone wire. In some cases the head

feathers were raised for a few seconds during sidling





Figure 5. Displays associated with appeasement behavior:
A. Withdraw High-Up; B. Alert High-Up; C. White
Badge Signal.

I saw the Withdraw High-Up posture given in the follow-

ing contexts: (1) by a female when sitting on the house in

the presence of a male other than her own mate; (2) by a

bird shifting its position when another bird had landed

nearby; (3) between members of a pair when they were near

each other and either bird moved toward the other; and

(4) when one bird was threatened by another bird with a Head

Forward Thrust.

Gray-breasted Martin. In the GM the Withdraw High-Up

was not as elaborate as in the PM, and often appeared as an

alert posture seen in many birds immediately before they

ever flew in alarm, after a supplanting attack, or in birds

that had just avoided an attack.

The Withdraw High-Up of the GM involved raising the

feathers (82% of film records, 18% relaxed; n=ll), hunching

the body (91%, 9% upright, n=ll), and assuming a generally

rounded appearance. This also was in many respects the

opposite of the Head Forward Thrust, and suggested that the

bird was avoiding an aggressive encounter (see section on

analysis of displays).

Caribbean Martin. Withdraw High-Up in the CM fre-

quently involved a turning away from the dominant individual

(88% of field records; 12% turned halfway; n=27). The bird

stood with the body vertical (92%, 8% not upright, n=27),

neck extended and bill pointed upward (96%, 4% bill pointed

down, n=27), so that the breast was exposed and the head was

held rigid. The tail was not spread (96%). Several times

males in boundary encounters perched on palms or thatched

roofs with their bills turned away from their neighbors.

Alert High-Up postures were given in the presence of

alarming or novel stimuli, such as those eliciting loco-

motion. In all three species, when a bird suddenly detected

a predator, a loud harsh noise, or a sudden change in

movement of another animal or object, the activity in which

it was involved immediately ceased. During the Alert

High-Up the head was raised or directed toward the stimulus

and the neck extended out (Figure 4B).

In the Alert High-Up posture the bird stood erect with

compressed plumage and with the head and bill in line with

the rest of the body. The wings were held close to the body

and the legs became very straight. The birds remained in

this exaggerated posture for about four seconds, longer than

in any other display (PM: x=4.3 seconds, n=132; GM:

x=5.1 seconds, n=75; CM: x=4.5 seconds, n=23). Once the

alarm stimulus was know, birds of all three species flicked

their wings and tails and raised their crest feathers for

the length of the disturbance. Frequently the PM, GM, and

CM uttered Cher (p. 83, Cheur (p. 94), and Zwoot (p. 100)

calls respectively.


Males of all three species had a display which attrac-

ted females and advertised territory ownership. Claiming-

Reclaiming involved a flight whereby the male attracted his

mate or potential mate to the nest site (Johnston & Hardy,

1962). PM males flew from the breeding area in a wide arc

(20-70 meter diameter, n=278), which could vary in flight

path, circle size, duration, and flight posture. Besides

following a circular course (10-30 meter diameter, n=21),

CMs also flew a figure-8 (n=9) or a double lap circle

(n=15). GMs showed both circular flights (25-60 meter

diameter, n=17) and short direct flights (n=29) in which a

male came close to striking a sitting female before return-

ing to the breeding site. On these 29 occasions, GM males

flew so close to their mates that the female was forced to

either retreat or fly from the perch. GM females responded

to these two kinds of flight encounters by either following

the male to his nest hole (n=26), moving to another section

of steel pipe (n=16), or leaving the area (n=4).

A Claiming-Reclaiming display ended abruptly in all

three species with the male returning to the nest site to

repeatedly enter and emerge from the nest hole (PM:

x=2.6 bouts of emerging and entering, n=38; GM: x=2.7,

n=29; CM: x=1.3, n=8). This behavior ceased when the male

thrust his head out of the entrance hole and sang several

songs in succession (PM: x=2.1 songs per bout, n=238; GM:

x=1.8, n=25; CM: x=2.8, n=39). At the end of the song in

PMs and GMs, but not in CMs, the bill was opened and the

yellow mouth lining was flashed in stark contrast to the

rest of the bird's head. CMs did not end songs with an open

bill (see discussion under vocalizations).

White Badge Signal

Badges are characteristics of an animal's appearance

that have been modified to be informative "adornments"

(Rohwer 1982). Badges are more persistent than displays,

because displays last only as long as the behavior is

sustained. Both sexes of all three species of martins had a

patch of white feathers on the anterior flank that normally

was concealed by the dark feathers of the middorsal region

of the spinal tract (Johnson and Hardy 1962, this study).

These bold plumage markings can be seen at the level of the

tertial feathers on either side of the midline whenever

birds preened the head or mantle (Figure 6C). Size and

shape of the white badge varied considerably among individ-

uals of the three species. The white patch was highly

organized, contrasted with the dark background feathers, and

could be displayed or covered at will; thus, Johnston and

Hardy (1962) hypothesized that in the PM the white badge

functioned as a social signal associated with preening. All

my observations supported this hypothesis for the CM and GM.

My observations for these species showed that martins

displayed the badge only after preening (65%, n=85) or

during sunning postures (35%, n=85).

The white badge also appeared to signal staying inplace

behavior. Birds showing the white badge, as a class,

remained stationary much more frequently than birds that did
not exhibit the badge (x2=26, df=32; p < 0.05). The badge

also functioned as an appeasement signal by inhibiting

agonistic encounters and permitting close contact among
loafing birds (x =23, df=24; p < 0.05). Several aspects of

martin behavior supported this interpretation: (1) when the

badge was uncovered, birds were preening or had completed

preening and were loafing or sunning (n=85); (2) partici-

pants were positioned in parallel or at angles of less than

90 degrees where there was little prolonged direct front

orientation or approach (87%, 13% frontal orientation,

n=85), and (3) at no time was the badge revealed in any

agonistic situation (100%, n=85).

Similar signaling badges that function to inhibit

behavioral interactions have been described for young Water

Rails, Rallus aquaticus (Lorenz 1952), White-necked Ravens,

Corvus cryptoleucus (Johnston 1958), and numerous Accipters

(Hafner & Hafner, 1977).


Fighting in all three species usually was associated

with defense of a breeding site or mate-guarding. Fighting

was most frequent among males early in the breeding season.

At these times exploration of nest sites resulted in much

movement in the breeding areas, and males trespassed while

courting or watching another male's mate. Fighting among

females occurred only during pair formation, when a female

crossed a territorial boundary. Fighting did not result in

noticeable body damage, and in most cases was of short

duration (although see Allen & Nice, 1952; Bent, 1942;

Brown, 1977).

Purple Martin. The circumstance for an encounter that

led to a fight was trespassing by a nonresident male onto

another male's territory. Of 256 encounters between

adjacent territorial males, 249 started when a male

approached a neighbor's mate as she was perching at the

birdhouse or gathering nesting material. The remaining

seven encounters resulted from males exploring the birdhouse

and in which attraction to a female did not seem involved.

When a male trespassed onto another bird's territory,

the owner adopted an agonistic posture directed toward, or

attacked, the moving male. The following examples of

encounters between adult males White and Yellow at the

Gainesville Country Club nesting colony are given to illus-

trate some of the sequences of events that fighting can

typically take. The flow diagram in Figure 6 is not com-

plete, but is intended only as a suggestion of the course

agonistic behavior may take early in the breeding season.

Encounters of more than two birds were difficult to follow

or record accurately and were therefore not included.

Male White was standing on the porch outside his nest

compartment in an upright position. Male Yellow was in

flight returning to the house and was about 10 m from the

house. On noticing Yellow, White started calling in low-

intensity Head Forward Thrust directed toward approaching

male Yellow (n=51). When Yellow landed about 10 cm away,


:1 i.

.-J L C


9. -3.

^ n

- 5

both birds bill-thrusted toward each other and then turned

away, presenting their sides. If however, Yellow approached

White, White usually responded in one of several ways:

(1) White flew away (n=5). This was sometimes followed by

chasing behavior (n=2), or Yellow landed in the spot just

vacated (n=3); (2) White entered his nest compartment (n=4)

or shifted his position on the perch a few centimeters away

and gave a low-intensity Head Forward Thrust. After a few

seconds, the posturing ended and the birds remained perched

a few centimeters apart; (3) White remained perched, but did

not posture; Yellow came in and attacked (n=17). A fight

ended with White flying off and Yellow chasing after him

(n=14). On only three out of 17 occasions did Yellow fly

off leaving White in the same position; (4) White remained

perched, and immediately showed high-intensity Head Forward

Thrust (n=15). Yellow then did one of three things: (a)

attacked-with a fight ensuing (n=3), (b) retreated and

perched more than 10 cm away (n=10), or (c) withdrew less

than 16 cm away with both birds giving Withdraw High-Up

(n=2); and (5) White flew up and met Yellow in the air

(n=3). During these fights both birds were facing each

other, striking with their bills, wings, and feet until they

struck the ground, whereupon the fight ended.

Once nesting began, if the encounter took place in

front of the nest compartments, fights seemed to follow

two less elaborate patterns. A resident adult male some-

times landed on the perch in front of his nesting room, in

which case he either advanced and immediately engaged the

intruder in fight (61%, n=129); or the resident male entered

his room, and poked his head out and called or sang (39% of

these observations). The bird then either emerged to perch

in front of the nest room or remained stationary in the nest

room aperture.

Gray-breasted Martin. In Trinidad where nearest pairs

of GM often nested up to 20 m apart, male GM flew directly

at their opponents, forcing the attacked bird to move, on

59 occasions. The attacked bird responded in one of three

ways: (1) it flew immediately away, with the attacker

flying after and chasing it (n=24), (2) it shifted its

position a few feet away (n=26), or (3) it flew up and met

the attacker in mid-air (n=9).

Two forms of fighting were observed in GM. One was the

aerial fight in which two males ascended about 10 m from the

perching pipes, each had a foothold on the other combined

with feather-pulling and wing-beating. The birds would then

descend, not separating until striking the ground or enter-

ing the water, at which point all fighting ceased (n=37).

The other type of fight was a Lunge on the pipe perches in

which the two birds pecked each other's heads and body areas

until one individual broke off in retreat (n=68). A winner

adopted an erect posture, calling as it sat on the spot

vacated a few seconds before. The bird then began preening

to rearrange feathers that may have become displaced during

the attack.

Caribbean Martins

Fights of the CM did not differ appreciably from those

described for PM and GM, except that fighting occurred at

greater heights around nest sits in palm fronbs. In Tobago,

I recorded details of 16 intraspecific fights and saw a few

other short agonistic episodes. Less fighting in CM than in

GM and PM may reflect the greater CM minimum interest

distance (> 30 m as opposed to < 20 m in the GM and 10 cm in

the PM). Physical opportunities to interact decreased as

the distance between birds increased.

Analysis of Visual Communication

Sequences in Defense of Territory

The behavior of individual martins consisted of both

individual acts and sequences of acts. A behavioral act was

a display only if it conveyed a signal to another animal and

was specifically adapted for that purpose (see Hinde, 1970;

Krebs & Davies, 1981; Tinbergen, 1952). Only PM and GM

provided sufficient data to permit me to analyze their

visual displays.

In both species, any of the seven displays listed in

Table 3 could occur as an initial response in an interaction

with another martin. Most interactions were brief

(x=2.8 seconds; n=289 field records) and consisted of only

one or two displays. The most frequent initial responses

were low- and high-intensity Head Forward Thrust (PM males:

Table 3. Frequencies of Initial Displays.

(Proportion of total encounters)
Display PM GM PM GM
Male Male Female Female

Low intensity Head Forward 0.25 0.24 0.32 0.43

High intensity Head Forward 0.14 0.18 0.16 0.21

Gaping 0.09 0.09 0.03 0.07

Bill-Shaping 0.04 0.03 0.02 0.01

Lunge 0.08 0.06 0.07 0.01

Alert High-Up 0.15 0.19 0.14 0.14

Withdraw High-Up 0.25 0.20 0.26 0.13

Total probability 1.00 1.00 1.00 1.00

Total encounters 1537 1089 86 134

39%, n=1537 field records, females 46%, n=86; GM males:

42%, n=1089 field records, females: 64%, n=134), and Alert

and Withdraw High-Up (PM males: 40%, n=1537 field records,

females: 40%, n=86; GM males: 39%, n=1089 field records,

females: 27%, n=134). The probability of the occurrence of

these responses was increased when the conditions of an

encounter were specified (Table 4). Low-intensity head

Forward Thrust was the most frequent response to a martin

pausing near the territory (PM: 51%, n=291 field records;

GM: 39%, n=257) or to a martin flying by the breeding site

(PM: 28%, n=279; GM: 23%, n=284). Alert High-Up was the

most frequent response to a distant disturbance (PM: 52%,

n=279; GM: 61%, n=284). Withdraw High-Up occurred most

commonly in encounters with neighbors who retaliated in

response to an initial challenge (PM: 54%, n=214; GM: 43%,

n=173). High-intensity Head Forward Thrust, Gaping, and

Bill-snapping occurred less frequently and were responses to

intruders entering or pausing near the territory (See

Table 4). Gaping was followed by a Lunge when the intruder

approached to within bill-striking range. The Lunge also

was used by some martins against neighbors pausing near the

territorial boundary.

The time elapsed since the last encounter and movement

of the stimulus influenced the initial response of martins

(Figure 7A-E). When an intruder approached slowly, a

territorial martin responded with low-intensity Head Forward

Thrust (69%, n=135 field records) or Gaping (31%, n=135







4J. t


0 tC


















4- (

r- o
o o
* C
0 0








* d







O 0
o o
o o

0 C

C n
o *e

O o

0 o

o o



I o
o0 C


0-I -0

o o

0 0

I *r

J r
*4 *.

() 4






01 -

IA .
U S. m
Q- 00. O--- c -

--' 4-- 2 I-- ,- "
3- S- : 1

tiL =1 L a C

LlC L0 C L C-
**--o 0 c W c c

. 3 41 5 ? 4j :3 '0 4-
1 faO 4. S0 S 0

i| 0o S| M

(U F- c C

cm La.
4- 4n s
r ^^ *r
v ~Ea p,
5~L cs 5~ cs" c

field records; Figure 7A). If the intruder approached sud-

denly, a resident male responded with high-intensity Head

Forward Thrust (86%, n=84 field records) and Bill-snapping

(14%, n=84 field records; Figure 7B). Low-intensity Head

Forward Thrust was followed by high-intensity Head Forward

Thrust when the intruding martin moved closer (Figure 7C).

High-intensity Head Forward Thrust was followed by Lunge

when the rival martin moved away only slightly or slowly

(Figure 7D).

Sometimes the same type of approach elicited different

reactions. Generally this could be attributed to the

previous stimulation of the territorial martin. A male who

had been undisturbed for some time might show only low-

intensity Head Forward Thrust (Figure 7A), whereas a male

who had just repelled a martin at a territorial boundary

reacted more aggressively to the approach of an intruder

(Figure 7E).

Gaping, Bill-snapping, and Withdraw High-Up tended to

occur at the end of sequences rather than at the beginning

(Table 5). Specifically, these displays occurred at the end

of sequences in which Lunge had not been elicited; thus,

they did not precede attack. High-intensity Head Forward

Thrust also occurred at the end of sequences but was

followed by attack when a neighbor approached or entered the

territory (See Tables 6 & 7). Low-intensity Head Forward

Thrust and Alert High-Up occurred at the beginning of

sequences rather than at the end. Both were seldom used

Table 5.

Proportion of Displays Used as Initial Versus
Terminal Responses in Two-Display Sequences.

Position in Sequence


Display Begin % End % N Begin % End % N

Low intensity
Head Forward 61 39 131 74 26 124

High intensity
Head Forward 34 76 104 39 61 93

Gaping 18 82 34 22 78 36

Bill-snapping 6 94 28 11 89 20

Lunge 17 83 59 6 94 32

Alert High-Up 81 19 93 92 8 85

High-Up 25 75 142 13 87 69





















C 0 o Co
' cN N m co
r-1 N m

o m r- r- in

T.. rf0 0n 00 0
-1 q:T

-1 oo CN in 0 0N o


CM .0 0 C '>) 0
S Nr N 0 C0
-4 -4 CMN

I*H 4- r3
- O I-rt

(1 0 4-) 0

O I cl

0a *H a
4 e (3
0~ 5


a .0

a) 4-)
H- -
F< 3



C >





3 O



H o Co ) m () N m
0o N r- co co

r- 0) ) u un -f a%
-l n (N 4 N a -

(N N 0 (N r Co









14m m 0 I 0 T
00 ( Y )> 0 01 O0

co u N uN co C
Lfr ) r- LO co I O
- 0 ,(N -4I 01

uin -- o I M (1 0o













tZ C4 s -1 0 0 0 o

n N Cov C CD
m (N 0-

m mn ( CW
o m .0 i-i CN

>.,ro 4-1 r

-Ho Ito
r4 -H
(O 0)a

C -,-I

Or -i-ia
.rl ffir

-H O

-P -

0) 4-)
-I -,-




0 )







0 m (N w C

-1 (N (

within the stationary attacking range, and thus were seldom

followed by attack.

Relation of Displays to Subsequent Behavior

This analysis was designed to determine if one could

predict what a bird would do depending upon its display.

Using an approach developed by Stokes (1962a, 1962b) and

Balph (1977), I recorded the frequency of the seven displays

and the subsequent reaction of the bird (Tables 6 & 7).

Interactions were considered to have three phases: attack,

when one bird drove the other away; staying, when the two

birds did not move away; and retreat, when a bird moved away

from another in response to an attack.

Tables 6 and 7 give the probabilities of particular

displays leading to attack, retreat, or stay. In this

study, low-intensity Head Forward Thrust indicated that a PM

would subsequently attack 31% of the time (n=379 field

records) and a GM 36% of the time (n=262 field records), but

there was an even higher probability of staying (PM: 65%,

n=379 field records; GM 63%, n=262 field records). In the

PM, high-intensity Head Forward Thrust suggested a transi-

tion between low-intensity Head Forward Thrust and Lunge

since it was associated with probabilities of attack (58%,

n=207 field records) and of staying (38%, n=207 field

records) that were intermediately between those for low-

intensity Head Forward Thrust and Lunge. When high-

intensity Head Forward Thrust occurred in the GM, however

the predominant tendency was to stay (68%, n=193 field

records). A GM attacked from the Gaping position 52% of the

time (n=142 field records) and from the Bill-snapping

position 25% of the time (n=76 field records). Lunge had

the highest probability of being followed by attack in both

species (PM: 100%, n=120 field records, GM: 100%, n=62

field records). Alert High-Up was most often associated

with staying (PM: 78%, n=233 field records; GM: 89%,

n=216 field records), but also had a low probability of

attack (PM: 12%, n=233 field records; GM: 8%, n=216 field

records). A martin showing Withdraw High-Up never attacked

(n=380 field records) and was likely to retreat (44%,

n=380 field records) if approached more closely.

On the few occasions when recipients retaliated (PM:

22 observations, n=1537 field records; GM: 25 observations,

n=1089 field records), the signaler then responded with

continued attack (PM: 73%, n=22 field records; GM: 72%,

n=25 field records) or retreat (PM: 27%, n=22 field

records; GM: 25%, n=25 field records). Most agonistic

encounters were of the attack-retreat kind and once ended

were rarely repeated (PM: 98%; 2% were interactions that

continued, n=280 field records; GM: 96%; 4% were inter-

actions that continued, n=84 field records).

Effectiveness of Displays

To determine the effectiveness of the displays, I

measured the valence of a given display to the behavior of a

rival martin at a territorial boundary. Most encounters

were initiated not because intruders trespassed onto the

territory, but because of their movement near the boundary.

In both the PM and GM, the assumption of a given posture by

one member of an interacting pair was followed by a number

of behavior patterns in the other martin. Analysis of the

distributions of following acts showed highly significant

changes in the probability distribution of acts dependent

upon the preceding act. That is, execution of displays did

change the behavior of recipient individuals.

Three displays (low-intensity Head Forward Thrust,

Gaping, and Bill-snapping) elicited stay or retreat of

intruders (Table 6 and 7). When a PM or GM gave low-

intensity Head Forward Thrust, the recipient subsequently

stayed 81% of the time (PM: n=379 field records; GM:

n=262 field records) and rarely attacked (PM: 2%, n=379

field records; GM: 5%, n=262 field records). For Gaping

(PM: n=142 field records; GM: n=103 field records) and

Bill-snapping (PM: n=76 field records; GM: n=33 field

records) in both species the probability of subsequent

staying of the recipient was 73% or better and the likeli-

hood of being attacked by the recipient was 6% less. The

two responses most effective in driving intruders away were

the high-intensity Head Forward Thrust and the Lunge

(Table 6 and 7). Lunge had the greatest probability of

eliciting retreat in the recipient (> 95% in both species;

PM: n=120 field records; GM: n=62 field records), and the

lowest probability of being followed by retreat from the

signaler (< 5%).

Alert (PM: n=233 field records; GM: n=216 field

records) and Withdraw High-Up (PM: n=380 field records; GM:

n=220 field records) had the lowest probability of eliciting

retreat (< 11%) and the greatest probability of eliciting

stay (> 82%). They also were the best indicators of sub-

sequent retreat by the signaler (Tables 6 and 7). Since

Withdraw High-Up was effective in eliciting staying from

intruders, it was surprising that most attacks from rival

martins occurred after this display. The higher probability

of attack may be due in part to conditions associated with

high density in the breeding area and the resultant forced


Competent Structure of Displays

In order to determine quantitatively if certain behav-

ioral components were associated with one another, I

assessed the responses of other martins to those components.

I recorded 312 encounters for the PM and 239 encounters for

the GM that were used in tabulating the results of agonistic

interactions. Each display consisted of a combination of

different components, of which the following were recorded.

(1) Body position: only two categories were

distinguished--the martin stood either upright or

horizontal. An upright body position was characteris-

tic of reduced aggressiveness and increased retreating

behavior. A horizontal position indicated a tendency

to retreat by opponents.

(2) Body feathers: the body feathers of the

breast, back and flanks were erected, sleeked, or

relaxed. Major feather erection occurred only when the

martins were retreating, or showing movements of

intention to retreat. Sleeked feathers were most often

seen when a male approached or threatened a potential


(3) Crest feathers: the feathers of the head

were raised or lowered to varying degrees. There

appeared to be considerable variation in the extent of

crest erection during all of the above displays.

(4) Bill position: the bill was either opened or

closed. In some threat displays the bill was opened

wide, as in the Gape and Lunge displays. In retreat,

or retreat intention movements, the bill was usually


(5) Wing position: the wings were held against

the body when martins were loafing. When martins were

disturbed, the wings were repeatedly flicked away from

the body (PM: x=2.4 cm, n=15 film records; GM:

x=2.3 cm, n=7 field records). The wings were extended

(PM: x=l.8 cm, n=7 film records; GM: x=1.8 cm,

n=3 film records) when martins were moving forward in a

Lunge and when moving away in Withdraw High-Up.

(6) Tail position: the tail was held stationary

and kept in line with the body when martins were

loafing. The tail was flicked up and down in a verti-

cal plane when martins were startled or disturbed.

(7) Orientation: a martin faced the rival or

turned at any angle away from it. Facing the rival

increased the probability of attack. Body positions

turned at some angle away from the rival were associ-

ated with an increased probability of retreat.

Of the threat displays recognized in this study, each

had a different component structure. The first step in this

analysis was to record the frequency of occurrence of the

seven different behavioral components and to determine the

degree of correlation between these components (Tables 8 and

9). For the seven behavioral components there were 21 pos-

sible two-factor combinations. I found a correlation

(p < 0.05) between components in 19 out of 21 two-factor

combinations for the PM and 20 out of 21 two-factor combina-

tions for the GM. Thus when a martin raised its body

feathers, its crest was generally raised at the same time,

and a martin with the bill open almost always faced its

rival. Even more striking was the observation that some

components rarely occurred with others. In PMs and GMs a

horizontal body position rarely occurred with body feathers

erect, crest erect, or during facing away (15 observations,

0a C' -r I

rei x< 3 0a X O
-) ,< 1.a -5C

-1X (a -7

;f 0ifO
l/^ Ll1

r-i in


1 II


oH Z
41 U

) U



0 Q)


4(1 0

o0 L

. H *

o o
4- 0

m 4



-P 0

a -)



1 :


.-4 0.
U n]




..3 -4

e 0

4 a






>. -I UJ

oo c

o*- V-4


CO :



Ll n

v~ v

'a -4

~n 'a a' 'a a
3 CS 'a ~ I

'a O-' X0. 'a en o XO' -4XC x o -.XC a' f i
CJ ;.Cfl ;:en Xen C- 0'.- Cl
-. n ;: r- c.i - I n |

rC- C~ -o '3 -7 Oa -' -7r o 7 I
"-n -i 7, 10 lfl 10 1 I 1


" 0
c-i 0a


C-a -C'

C- -3
C- '0


r --

n 'O


C-a Cl

m In

f^> '5i




-- 4
CT^ i?
1 C0


o ':0
' f


0 r rJ l C 0 I c

.- I C-I

0 D !

in r' i eI



I t

e n r rC I -T
9j u-l 0 I '0

-4 -
.3 CS C-f~ :0 I
'0 -7 oa 1 -7 4

X -

I m
I o

0 I

0) r ri
j -Y a
(u -O 55
X001 -4i S.

If^ L U tl LU
a4'- U -
W m
>~,. '.- I-a 1 (n u
'0 V/ 0 [j r '
0 0 IS S
CO 10 U u

) V
O '-l 0 *- 5 L
U Lr 0 0 y 1a
f 00 w O O 1

0 r
.I ~O M r-r -4 4- C4.S l
-I 0 '-4 :9 U : c
-t-4 -r4 i-4 '0 40 O 1
cO 3a 3 E- H L. L o


o 0


l~a CSi


-7 m
m o0


n (,

in I-

m a'

e -

o -t

o r-4

0 0n


CO <

N -3 CO 1 In 0r -i -t L -t 0 1t I I | 0
- 7 0 -T i 'I 0 7 -0 C r- 0 I I 0
14I -

- O ; i 3 n 3' X 0 < O f I
a ;x: cO ,' -. r-- I m I



0 0

H -H--I

E-i Hn



C 0


o M
*H r

4) -44
C "( 0

0 ,41

.C .
r-q 4-4

0 0
0 a)

o x

O cmo

I. -4 H
0 z

C (U
(0tr -
*i-i(Q kr

Cf 4




m3 0

b> u

mu U
Li w



"3 -W-

41 --4



0 U

c =


-4 a

3-' 0

3 IA



*^ 0


a' CL
^ u

*i'l a 4


U ^

U 0



'Li N


3' b-4




03 3

- o

0 -


0 00
0 '0i

=0 r-~

04 --

CO 0

I (0~
I -




o0 n 0 I

- X.

D0 0 3 0 Co
C' '0 0- -7 -e

'0 '- 0 0 I 0
'0 -. 'n I *


-t I 7 r0- I

'n 30


-7 -7r

203 I
-4 I
xf i


I -


(4 0 00 --7 04

' 0 O '0

X x


I '

Lit I

C/l 'o U
ci Li '

-cw Cc Lo u ) t
SWU o. a 0 0
1) u .0 0 U r
a-Li a at CnI a 3

w u) ul U w Ul
0 0'-a

Su 'a ca :a a a -

> >.

- 3 a

C C aaa:
4 -u u"I W
U '3 a< Cl
C&, m O

m n


r-. 0
-7 '0



cn n
m mr

'0 04-

.4 01 'j '1' 0 m 0- ) o co '0
I Tr r -4 7 r -4

- CJ


0 -T

-4 a,

m r-

0 Ca,

-7 -7

in oo

a) Qr

--0 c'

- cr

C a
0 .0
N 0

a a
-r-< -i-

0 0.
3: 3

o 0

n=201). Only wing and tail flicking combined with body

position in the PM and wing flicking combined with crest

position in the GM did not show any significant degree of

correlation between these components.

The next step in the analysis was to determine if one

could predict what a bird would do from its posture.

Following the same approach used for analyzing the displays,

I recorded the frequency of the seven different behavioral

components and the degree of correlation between these

components and the subsequent reactions of the birds.

Tables 10 and 11 show that a bird's behavioral posture

indicated different probabilities that it would stay,

attack, or retreat. Only five components permitted pre-

diction of subsequent action by a male PM (Table 10). Bill

open (as in Lunge) had a 92% (n=52 field records) chance of

being followed by attack. Tail flick, which was associated

with moving forward, had a 72% chance of being followed by

attack (n=74 field records). Body feathers erect, crest

erect, and facing away indicated high probabilities of

retreating. In the GM (Table 11) only bill open permitted

prediction of the outcome of an encounter with any degree of

accuracy. For all other components of behavior the

probability of subsequent attack, retreat, or stay was

usually 60% or less in both species (Tables 10 and 11).

Another approach was to compare the probability of

action when a given component was present or absent. Thus

when a PM behavior changed from body feathers sleeked to

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - Version 2.9.9 - mvs