PREDATION
AND
GUIANAN
ANTIPREDATOR
BEHAVIOR
COCK-OF-THE-ROCK
AT
LEKS
PEPPER W. TRAIL •
Sectionof Neurobiology
and Behavior,CornellUniversity,Ithaca,New York 14853USA
ABSTRACT.--I
documentedpredatorattacksand antipredatorbehaviorat four displaysites
of Guianan Cock-of-the-Rock(Rupicolarupicola)in Suriname from 1980 to 1985. Most observationswere madeat a large lek that averaged55 territorial males,with supplementaldata
on displaygroupsof 1, 2, and 6 males.Males at the main lek were attackedby 6 speciesof
raptors,1 mammalspecies,and 1 snakespecies.A total of 56 attacksand 2 kills by raptors
were observed(0.22attacks/day).Two kills of displayingmalesby the snakeBoaconstrictor
were
also observed.
Male Cock-of-the-Rock
were very wary. I observed832 spooks(completeor partial flushes)
at the Main Lek, more than 90%of which were apparentlyfalsealarms.Males at the smaller
display siteswere much lesslikely to spook.The raptor attackrate was inversely correlated
with group size, supportingthe hypothesisthat socialantipredatorbehavior reducesrisk for
displaying males. There was no evidence that peripheral or young males were especially
vulnerableto predators.
Snakepredationrepresentsa previouslyunrecognizedcostof lek displayfor tropicalbirds.
The Cock-of-the-Rockfailed to exhibit any effectiveantipredatorbehavioragainstsnakesand
may rely for protectionon the infrequencywith which snakeslocateleks. Received
14 July
1986, accepted11 March 1987.
THEselectiveimportance of predation at leks
has been long debated.Becausemost vertebrate
leks are conspicuoustraditional aggregations,
they presumablyattractboth residentand transient predators.The combined attention of the
lek males, however, may make it difficult for a
predator to approachundetected. Most authors
have concludedthat males displaying on leks
are safer from predators than are males at solitary displaysites,at leastin open habitats(Lack
1968, Hjorth 1970, Oring 1982). The enhanced
safety of social display has been proposed as a
factor that favors the evolution
of leks and the
preference of females for clustered males in a
number of prairie grousespecies(Koivisto 1965;
Lack 1968; Hjorth 1970; Wittenberger 1978,
1979).
Few data are available on predation at leks.
Long-term studiesof Greater Prairie-Chickens
(Tympanuchus
cupido;Bergeret al. 1963), Sharptailed Grouse(Tympanuchus
phasianeIIus;
Oring
(Pipra erythrocephala;Lill 1976) revealed extremely low ratesof attemptedpredation,while
higher rateswere found for populationsof Sage
Grouse (Centrocercus
urophasianus;
Wiley 1973,
Hartzler 1974) and Eurasian
(Tetrao tetrix; Koivisto 1965).
Black-Grouse
I observed predation and antipredator behavior during a 6-yr (1980-1985) field study of
the lek-breeding Guianan Cock-of-the-Rock
(Rupicolarupicola).This large (200-230 g) cotinga inhabits tropical rain forest in the Guianan
region of northern SouthAmerica.The brilliant
orange plumage and active display of adult
malesmake Cock-of-the-Rockleks conspicuous
to humans,and presumablyto predatorsaswell.
I studieda color-bandedpopulation of Rupicola
at the Raleigh Falls-Voltzberg Nature Reserve
of Suriname (Trail 1985a-c, Trail and Koutnik
1986). In this wilderness area Cock-of-the-Rock
• Presentaddress:Department of Ornithology and
Mammalogy,CaliforniaAcademyof Sciences,
Golden
are exposed to a large natural assemblageof
predators,including raptors,felids, and snakes.
I describeand analyze Cock-of-the-Rockantipredator behavior, and then use data on predation at different-size display sitesto examine
hypothesesconcerningthe role of predation in
lek evolution. Specifically, I tested if lek be-
Gate Park, San Francisco, California
havior
1982), White-bearded Manakins (Manacus manacus; Lill 1974), and Golden-headed Manakins
94118 USA.
496
made
males more
or less vulnerable
to
The Auk 104: 496-507. July 1987
July 1987]
497
Predationat Cock-of-the-Rock
Leks
predation, and if a male's position on the lek
correlated with his exposureto predation.
STUDY AREA AND METHODS
TheRaleighFalls-VoltzbergNatureReserve(4ø40'N,
56ø10'W)is a 560-km2 area of lowland tropical rain
forest surrounded by the much larger undisturbed
forestsof the CoppenameRiver drainage(Fig. 1). The
study area was centeredaround the Voltzberg, a 240m-high granite dome located 5 km southeastof the
river. Large boulders on the forested slopesof the
Voltzberg provide the sheltered sitesfemale Cock-ofthe-Rockrequire for their mud nests.A large population
of Cock-of-the-Rock
is found
in this area in
associationwith the abundant nest sites, and I located
3 leks and 1 solitary male display site.
Male Cock-of-the-Rock
maintainlek territoriesyearround, but active courtshipand mating occuronly
from late Decemberto April. The majority of matings
are performedin Januaryand February.I madedaily
observations
from approximately0645to 1630during
December1980to April 1981,January-February1982,
January-March 1983, January-February 1984, and
January 1985.
My primary researchsite was the Main Lek, which
comprisedan averageof 55 territorial malesper year
from 1981 to 1984 (range 52-61). Display territories
of malesat this lek were smalland denselyclustered.
Each territory consistedof a cleared display court,
usually0.8-1.2m in diameter,and the adjacentperches to a height of 1.5-2.0 m over the court. The mean
nearest-neighbordistance between courts was 1.4 _+
1.26 m (n = 96 courts).
I observedactivity from a blind located approximately 6 m from the nearestcourts.With the exception of a small number of the most peripheral territories (-<5/yr), all courts could be monitored
simultaneously.
Wheneverthe Cock-of-the-Rock
gave
alarm calls or spooked from the lek I recorded the
time, the approximateproportion of malesinvolved,
and the apparent cause.If a predator was seen, I recorded its behavior
until
it left the area.
To estimatemean daily group size at the Main Lek
I conductedfull-day censusesat least once a week
through each season.On these days I counted all
visible males at half-hour
intervals
from 0700 to 1700
(21 censuses/day).
I calculated42 weekly meansfor
group size spanningthe period January-April 19811984.Thesewere comparedwith the frequencyof
spooksand predation (see Results).
Female visitation at the Main Lek occupied an average of 28% of the time from 0700 to 1700 daily.
During female visits males displayed on their courts
(describedby Trail 1985c).During the remainder of
the day territorial malestypically perchednear their
courts,from just above the ground to 3 m. Nonterritorial malesusuallyremainedin the subcanopy8-
SURINAME
Fig. 1. Suriname, South America. The Raleigh
Falls-VoltzbergNature Reserveis indicated by the
asterisk.The dot indicatesthe capital, Paramaribo.
12 m above the lek. The dispersionof males in the
vicinity of the Main Lek thus included a densecluster
of maleslow over the displaycourts,with a scattering
of other malesin the subcanopyand at varying distancesaround the lek periphery.
The interior of tropical rain forest is characterized
by structurallycomplexand stratifiedlayersof vege-
tation and by low light levels.To assess
visibility at
the height of the perched males at the Main Lek, I
useda squareof pink plasticflagging with an area of
25 cm2.Flaggingwas mounted on polesat the height
of 2 m and was moved progressivelyaway from a
samplecourt in eachof the four main compassbearings. An observer standing on the court noted the
distanceat whichthe flaggingwashiddencompletely
by foliage. For 15 courts,representingevery area of
the lek, the mean visibility was 15.7 _+ 1.21 m (n =
60 measurements,range 4-30 m). I alsomeasuredthe
height from the ground to the lowest foliage above
eachof 30 courts,againrepresentingall areason the
Main Lek. The mean height was 3.3 + 1.75 m (n =
30, range 1.0-8.0 m). These measurementssuggest
that visibility was sufficientlyrestrictedat the Main
Lek to make detectionof predatorsdifficult. During
periods of display, the calls of the males could be
heardfor severalhundredmetersthroughthe forest,
and predators(exceptsnakes)presumablyfirst detectedthe Main Lek by sound.
In addition to intensive studies at the Main Lek, I
498
PEPPER
W. TRAIL
[Auk,Vol. 104
entire group. Alarm calling consistedof directing the ringing "hey" call (Fig. 2) toward a particular stimulus, and was performed by Cockof-the-Rockof all age and sex classes.Males
occasionallygavealarm callswithout spooking
when large mammalsvisited the lek. This responsewas given to humans, to felids (jaguar,
0
Time (sec)
0.5
Fig. 2. Two alarm calls (the "hey" vocalization) of
Guianan Cock-of-the-Rock, recorded at the Main Lek
in 1981.
observed3 otherdisplaysites:the Creek Lek, the Nest
SlopeLek, and the SolitaryMale site.The Creek Lek
consisted
of 6 territorial
males and was located
in
dense liana forest near a small stream approximately
2 km southeastof the Voltzberg dome. The almost
impenetrable vegetation of this lek wasa marked contrast to the more open Main Lek. All-day observations
were made 1-2 timesper week at the Creek Lek during January-March1983.
The Nest Slope Lek consistedof only 2 territorial
males(briefly joined by a third in 1981) and was locatedon the boulder-coveredslopeof the Voltzberg.
This was the center of Cock-of-the-Rocknesting activity, and 6 nestswere locatedwithin 50 m of this
display site. The Nest Slope Lek was approximately
300 m from the Main Lek, and the chorusing of Main
Lek maleswas heard there easily. The vegetationon
the rocky nestslopewasmoreopen than at the Main
Lek, with widely spacedtrees and almost no lianas
or understorypalmsor shrubs.The forestin this area
Felisonca;puma, F. concolor;ocelot, F. pardalis),
and, rarely, to other mammals(e.g.brocketdeer,
Mazama spp.; and tamandua, Tamanduatetradactyla).
At the Main Lek spookswere the most common response to startling stimuli. Given the
risk of surpriseattackin the rain forest,the most
appropriate reaction to threat appeared to be
headlongflight, often precededby a few abrupt
"hey" calls.Following a falsealarm malestypically returnedto their territorial perchesat once
(in less than 30 s) and resumed their previous
activities.
! recorded 832 spookson the 254 observation
days of the 1981-1985 breeding seasons(3.28
spooks/day),spanning the months DecemberApril. The rate of spooksvaried through the
breeding season.The mean number of spooks
perdaywassignificantlylessin December(2.! _+
1.9, n = 23 days),March (1.6 + 2.1, n = 19 days),
and April (0.8 _+0.8, n = 10 days) than in Jan-
uary (4.0 + 2.8, n = 107 days) and February
(3.3 _+2.5, n = 95 days),the height of the breeding season(Duncan's multiple-range test, 247
df, P < 0.05). Frequenciesof spookswere not
significantlydifferent in Januaryand February.
This seasonalpattern in number of spooksparresembled the Cock-of-the-Rock
lek in the Kanuku
alleled a decline in group size following the
Mountains, Guyana, describedby Gilllard (1962) and peak of female visitation in January and FebSnow (1971). The Nest SlopeLek wasobservedat least ruary (Trail 1984).
once a week in the 1981 and 1982 breeding seasons.
I eliminated seasonaleffectsby considering
The Solitary Male display site was also locatedon
the rocky eastern slope of the Voltzberg, approximately 200 m from the Nest Slope Lek. The display
observationdaysonly during Januaryand February 1981-1984in the analysesof the frequency and causesof spooks(Table 1). Only one of
court of this male was located in the bottom of a small
basin (ca. 30 x 15 m) completelyclosedin by 10-20- the differencesbetween yearswas statistically
m-high boulders.Five nestswere locatedon the ad- significant.Spookscausedby fights and chases
jacentboulders.The SolitaryMale site was observed were significantly more frequent in 1983 and
at least once a week in 1981 and 1982.
1984than in the two previousyears(Duncan's
multiple-range test, 6 df, P < 0.05). This was
RESULTS
associated
with intense competitionfollowing
the disappearanceof the most successfulmale
Antipredator
behavior.--MaleCock-of-the-Rock in 1983 (Trail and Koutnik 1986).
More than half of the spooksat the Main Lek
were extremelywary at leks. Antipredator besimhavior took two forms: spooksand alarm call- had an unknown cause: Cock-of-the-Rock
ing. Spookswere headlong flights off the lek ply flew explosivelyaway from their perches
by subgroupsof neighboringmalesor by the without previous alarm calls and without an
July1987]
Predation
at Cock-of-the-Rock
Leks
499
T^I•I•E1. Causesof spooksobservedat the Main Lek, Januaryand February 1981-1984. Mean number of
spooksper day were 3.4 (1981), 3.7 (1982), 4.5 (1983), and 3.0 (1984).
Cause
Observed attacks
Possible attacks
Cock-of-the-Rock alarm calls
Harmless animals
Cock-of-the-Rock
fightsand chases
Falling vegetation
Unknown
Total
na
1981
1982
1983
1984
Total (%)
6
7
22
20
21
11
33
17
18
7
30
42
5
3
13
21
50 (6.7)
28 (3.8)
98 (13.2)
100 (13.4)
I
8
2
5
8
6
10
0
21 (2.8)
19 (2.6)
117
181
88
177
143
254
81
133
54
48
56
44
429 (57.6)
754 (100.0)
202
Number of observationdays.
evidentstimulus.The magnitudeof thesespooks and abrupt, and apparently startled the Cockvaried from just a few malesto as many as half of-the-Rock into spooking. The movements of
the birds on the lek. All "cause unknown"
monkeys (Cebusapella,Saimirisciureus,
Saguinus
spooksappearedto be falsealarms,and the birds midas)through the lek triggered spooksin 11
returned immediately and resumed their pre- cases,and 6 more were causedby the soundof
viousactivities.Complete,lekwide spookswere small mammals running on the ground. The
rare and never occurred without
an obvious
flight sounds of unidentified birds caused 19
cause,in most casesan actual predator attack. spooks,including 3 in responseto large birds
Documented
attacks on Cock-of-the-Rock
ac(probably BlackVultures, Coragyps
atratus),discountedfor only 50 spooks,lessthan 7% of the play diving abovethe canopy.The flight of Redtotal. In an additional 28 spooksI saw no pred~ billed Toucan(Ramphastos
tucanus),Gray-frontator, but the behavior of the males indicated
ed Dove (Leptotilarufaxilla),Black-facedHawk
melanops),and Barred Forest-Falthey had detectedone in the area.In thesecases (Leucopternis
a minimum
of one-third of the males left their
con (Micrasturruficollis)through the lek trigterritories;they remained off the lek for min- gered 1 spookeach.None of thesespeciesattack
A small number
of
utes, not seconds,and the spookwas followed adult Cock-of-the-Rock.
by sustainedalarm calling. I classifiedthesein- spookswere triggered either by Cock-of-thestancesof severe and sustainedalarms as "pos- Rock fights (21 cases,2.8% of the total) or by
sible attacks."
the soundsof branchesbreakingor treesfalling
Cock-of-the-Rock "hey" calls triggered 98 (19 cases,2.6%).
spooks(13.2%).In 89 of thesecasesthe causeof
This analysis emphasizes the wariness of
at the Main Lek. At most 12%
the callswasunknown and the spooksappeared Cock-of-the-Rock
to be false alarms. In 5 instances the cause was
of the spookswere responsesto potentially danthe presenceof a mammalianpredator(2 jaguar, geroussituations.This is the sum of actual at1 puma,2 ocelot)at the lek. The pattern of alarm tacks,possibleattacks,and the 9 casesin which
callinggiven to a felid wasdistinctive.The alarm felids passednear the lek. All remaining spooks
began beyond the lek and usually continued were false alarms, as indicated by the males'
for many minutes, gradually shifted location, immediate resumption of normal behavior.
Predation at the lek.--Cock-of-the-Rock
at the
and finally faded in the distanceasa few males
followed
the cat off the lek. In 4 instances I
Voltzberg are exposedto three classesof predheard this pattern of alarm calling without ators:birds, mammals, and snakes.The study
seeingthe cause.In thesecasesa felid presum- area supportsa diverse community of 23 species
ably passednear the lek out of my view.
of diurnal raptors and 3 speciesof owls. AlOne hundred spooks(13.4%) were associated though owls attackTemperateZone lek grouse
with the callsof nonpredatoryanimals,includ- (Bergeret al. 1963,Ruschet al. 1972),they probing 5 bird species(Monasaatra,Cacicus
cela,Dap- ably are not important predatorson lek birds
trius americanus,Ortalis motmot,Querula purpu- in the tropics, where dawn and dusk are brief.
rata) and 3 mammal species (Saguinusmidas, Crested(Lophostrix
cristata)and Spectacled(PulDasyprocta
sp.,Sciurus
sp.).Thesecallswere loud satrixperspicillata)owls occurred at the Voltz-
500
PEPPER
W. TRAIL
TABLE2. Raptor attacksobservedat the Main Lek
during 254 observationdays,January-April 19811985. Lined Forest-Falcon, Barred Forest-Falcon,
Black-facedHawk, and BlackHawk-Eaglewere also
seen at the Main Lek, but did not attack the Cockof-the-Rock.
No.
days
with
Species
Ornate Hawk-Eagle
(Spizaetusornatus)
Collared
Total
no.
No.
attacks attacks kills
males spooked from the lek and stayed away
for severalminutesto give persistentalarm calls.
Suchbehavior never accompaniedknown false
alarms (e.g. in responseto nonthreatening animals or falling limbs). The total rate of raptor
attackswas thus 0.22 confirmed and 0.36 possible attacksper day at the Main Lek.
Compared with other lek birds, Cock-of-theRock are subject to a high rate of attempted
predation. Lill witnessed no attacks on either
8
8
2
White-bearded Manakins in 835 h (Lill 1974) or
Golden-headed Manakins in 2,000 h of obser-
7
28
0
3
3
0
I
1
0
I
1
0
0
Forest-Falcon
(Micrastursemitorquatus)
[Auk, Vol. 104
1
1
Unidentified
14
14
0
vation (Lill 1976). His study site on Trinidad,
however, lacked the bird-eating Micrasturand
Accipiterhawks.Wiley (1973)observed8 Golden
Eagle (Aquilachrysaetos)
attackson SageGrouse
during 86 mornings of observation(0.09/day).
Bergeret al. (1963)witnessedonly 3 raptor kills
in 4,745 mornings of observation at Greater
Prairie-Chicken leks, but did not specify the
Total
35
56
2
number of actual attacks. Koivisto (1965) ob-
Bicolored
Hawk
(Accipiterbicolor)
White
Hawk
(Leucopternis
albicollis)
Great Black-Hawk
(Buteogallus
urubitinga)
Crested Eagle
(Morphnusguianensis)
served 17 visits by Northern Goshawks(Accipitergentilis)to EurasianBlack-Grouseleksin Finberg,but neitherwere seenat Cock-of-the-Rock land during approximately 195 mornings of
observation(0.09/day). Oring (1982) reported
display sites.
Sixspeciesof raptorswere seento attackCock- only 1 raptor kill seen on approximately 600
of-the-Rockduring this study(Table 2). The un- mornings of observation at Greater Prairieidentified raptors(Table 2) were probably Col- Chicken and Sharp-tailed Grouse leks. Amlared Forest-Falcons (Micrastur semitorquatus), mann (1959), Moran (1966), Blus (1967), and
Slaty-backedForest-Falcons
(M. mirandollei),
or Moyles (1979) provided additional descriptions
Bicolored Hawks (Accipiterbicolor),similar-size of raptor predation at leks, but without inforspeciesthat can be difficult to distinguish.Four mation on attack rates.
All raptors that attacked Cock-of-the-Rock
additional potential predators,the Harpy Eagle
(Harpia harpyja), Black-and-white Hawk-Eagle maderapid rushesonto the lek, apparentlyfrom
(Spizastur
melanoleucus),
BlackHawk-Eagle(Spi- not far beyondthe periphery. The larger species
zaetustyrannus),and Slat-y-backedForest-Fal- (Morphnus,Spizaetus,Buteogallus,Leucopternis)
con, occur in the area but were not seen to attack
made fast, shallow dives onto the lek, while the
Cock-of-the-Rock.
smaller Micrasturand Accipiterflew in with rapid wingbeats. In most casesthe males spooked
I observed56 attacksby raptors at the Main
Lek. I defined
an attack as an instance in which
off the lek with
loud
alarm
calls before
I saw
a hawk flew rapidly onto the lek and triggered the raptor. After making their strikes the rapa sustained spook. This definition excludes tors typically perched on or near the lek (4-10
flights throughthe lek by small,nonthreaten- m up) for only a few minutes, while males coning raptors(e.g. Lined Forest-Falcon,Micrastur cealed in the subcanopyfoliage gave continu-
gilvicollis).
Strikesat Cock-of-the-Rock
were ob-
ous alarm
served in 75% of attacks. In the remainder
The malesquickly returned following attacks
by Micrasturand Accipiter,and even resumed
display while thesehawks were perched on the
no
strike was seen, either because of limited visi-
bility or becauseof evasivebehavior by all of
calls.
the males. Two of the 56 attacks resulted in kills,
lek itself.
for a success rate of 3.6%.
Forest-Falcon
On one occasion
remained
an adult
Collared
on the lek for more than
There were an additional 35 possibleraptor 2 h, during which time it made 15 unsuccessful
attacksduring the 254 observationdays.In these strikes.These small raptors probably were cacasesno raptor was seen, but all or almost all pable of subduing a Cock-of-the-Rockonly by
July1987]
Predation
atCock-of-the-Rock
Leks
501
surpriseand were not treatedasa seriousthreat
while in view. Similar responsesby lek grouse
to relatively nonthreateningraptors were reportedby Bergeret al. (1963)and Hartzler (1974).
In contrast, Cock-of-the-Rock
reacted with
panicto Ornate Hawk-Eagles.This speciesmade
both of the observedraptor kills at the Main
Lek. Ornate Hawk-Eaglesare reported to feed
3O
MALES
on both mammals and birds (Brown and Ama-
don 1968)but probablyspecializeon large birds
(Lyon and Kuhnigk 1985). In both successful
attacksthe hawk-eaglesdived and struckmales
on or near the ground. Both kills were instantaneous.In one casethe hawk-eagle pinned its
prey between the buttressesof a tree, possibly
stunning itself in the process,as it remained
motionlesswith outflung wings for at least 30
s after the strike. It remained perched at the
baseof the tree for 2 min beforeflying off easily.
This hawk-eagle plucked its Cock-of-the-Rock
on a fallen tree trunk approximately 30 m be-
PER CENSUS
Fig. 3. Correlation between male group size and
frequencyof spooksat the Main Lek. For eachof the
42 weeksof observationat the Main Lek from January
throughApril 1981-1984,themeannumberof spooks
per day and the mean group size were calculated.
Group size was determinedby countsof all visible
malestaken every half hour from 0700 to 1700 on 13 censusdays/week.
I observedonly onepossibleattackby a mammal. Shortly after dawn, when it was still very
dim in the forest,an ocelotran to the edge of
yond the peripheryof the lek. In the othercase the lek, sprangonto a tree trunk, and looked
the hawk-eagleapparentlycarriedits prey away toward the displayingmales.The Cock-of-thefrom the lek beforefeedingon it. After all Or- Rockspookedwell beforethe ocelotcouldhave
nate Hawk-Eagleattacks,the male Cock-of-the- attacked, and no pounce occurred. The other
Rock remained very wary for hours and re- felids passedthrough the lek without exhibitpeatedlyspookedwithout apparentcause.
ing hunting behavior,but causedthe birds to
There was probably a third kill by raptorsat fly into the forest subcanopyand give alarm
the lek. On 18 February 1983 there was a lek- calls. Gilliard (1962) also reported an ocelot at
wide spookfollowed by more than 90 min of a Cock-of-the-Rock lek, but did not observe an
alarm calling. This coincidedwith the disap- attack. Ocelots exhibit behavioral specializapearanceof the bird that had been the top male tionsfor bird hunting (Ewer 1973)and are prob(in termsof number of copulations)for the pre- ably the mammalthat posesthe greatestthreat
vious 3 yr and was third ranked in 1983. He to displayingbirds.Jaguarsand pumasprey prihad been seen earlier that morning, and had marily on medium-sizeto large mammals,and
never previouslybeen absentfrom the lek for occasionallyon large ground birds suchas cua full day. His disappearance,coincidingwith rassows(Ewer 1973, Koford, 1983, pers. obs.).
the massivespook, suggestsstrongly that he The tayra,a largemustelid,is omnivorous(Janzen 1983), but probably cannot capture fastwas killed by a predator, probably a raptor.
The impact of mammalian predatorsat leks flying birds.
I found no published records of snake preis poorly understood.Mammals may account
for more kills of displayinggrouseat leksthan dation on lek-breeding birds, and snakesare
do the more frequently observed raptors not likely to be important predators on any
(Hartzler 1974, Maxson et al. 1979).
TemperateZone lek species.In the Neotropics,
The potential mammalian predatorsobserved
at Voltzberg were jaguar, puma, ocelot, and
tayra (Eira barbara).Margay (Fellswiedii), little
spotted cat (F. tigrinus),and jagarundi (F. yagouaroundi)
probablyoccurin the areabut were
not seen.Jaguarswere seenat the Main Lek 5
however, there are numerous speciesof bird-
eatingsnakes,includingthe bird snake(Pseustes
poecilonotus),
the tree boas (Coralluscaninusand
C. enydris),the boa constrictor(Boaconstrictor),
the tigerrat snake(Spilotes
pullatus),
andthe ferde-lance (Bothrops
atrox).These speciesoccur in
times, puma once,and ocelottwice. Tayraswere
the rain forests of Suriname, and I saw the latter
never seenat the Main Lek, although one was
three speciesat the Voltzberg.I saw a boa con-
seen at the Nest Slope Lek.
strictor
kill
two male Cock-of-the-Rock
at the
502
PEPPER
W. TRAIL
[Auk,Vol. 104
Main Lek, and a third male was almostcertainly and that overall predation rate was not correkilled by the same snake. These observations lated with group size. To assessthe effect of
are described in detail below.
display group size in Cock-of-the-Rock,I exOn 12Ddcember
1980I spotteda maleCock- amined the frequency of spooksand predation
of-the-Rocklying on the forest floor, wrapped at the Main Lek asgroup size declinedseasonin the cryptically patterned coils of a boa con- ally and comparedthe Main Lek with the three
strictor. The snake was approximately1.75 m smallerdisplay sites.
long and had alreadyswallowedthe bird'shead.
There was a highly significant positive corThe bird's band combination
confirmed that the
relation between number of spooksper day and
dead male was the owner of the adjacentcourt, display group size at the Main Lek (Fig. 3). In
0.5 m away. He apparently was struck while contrast,the number of actualattacksappeared
displaying on or perched near his court. The to be related inversely to group size. Attacks
boa finished swallowing the Cock-of-the-Rock were observed 3 times more often when the lek
in approximately15 rain, then climbed out of was small that when it was large (Fig. 4), alsight into a small tree 3 m from the edge of the though these differenceswere not statistically
dead male's court. No mobbing or alarm calls significant (X2 = 4.18, 3 df, P > 0.10). Thus, the
were directed at the snake, and a few males
increasedincidence of spooksfor large group
remained perched within 3-4 m of the boa as sizescannot be explained by a greater number
it ate its victim.
of predator attacks.Instead, it presumably reOn 11 January 1981 I found the dead male's suited from the transmission of individual false
numbered metal band and one of his color bands
alarmsthrough the group. The flushing of one
embeddedin a massof digestedbone beneath male, even in the absenceof apparent threat,
the tree into which the boa had disappeared. frequently precipitateda majorspook.For large
Under the same tree I found another metal band,
group sizes there were many potential alarm
belongingto a male that had not been seenin instigators, resulting in frequent spooks,most
the current field season.This male may have of which appearedto be falsealarms.For smallbeen an earlier victim
of the boa. The band
er groups the males were exposedto fleeing
could not have been removed from a living conspecificslessoften and appearedlesswary.
bird, and it is unlikely that raptors or mam- The ratio of total spooksto actual attacksinmalian predatorswould consumetheir prey on creaseddramaticallywith group sizeat the Main
the lek itself, given their observedintolerance Lek. For very small groups (< 10 males, n = 9)
of Cock-of-the-Rock
alarm calls.
it was 2:1, for small groups (10-19.9 males,n =
The boa killed another male the following
18) it was 7:1, for medium-size groups (20-20.9
day, 12 January,at 1317.There is no doubt that males,n = 36) it was 13:1,and for large groups
the same snake was involved,
size and coloration
as it was the same
and attacked
males on the
same site only 4 weeks apart. The strike must
have occurredonly secondsbefore I noticed it,
asthe bird was still struggling feebly. The time
(->30 males, n = 15) it was 50:1. The increased
incidence of false alarms representsa cost of
socialdisplaythat could imposean upper limit
on group size.
I also examined the effect of display group
size by comparingthe incidenceof spooksand
surrounding males did not react to the snake's predation at the Main Lek and the three smaller
activity, and neither they nor the victim gave displaysites(Table3). During Januaryand Febalarm calls. The victim was the new owner of
ruary 1981-1983, we observedthe Creek Lek (6
the court where both kills occurred. To read his
males)for 11 days,the Nest SlopeLek (2 males)
bands I had to prod the coils of the boa, and for 12 days, and the Solitary Male display site
this disturbance caused the snake to drop its (1 male) for 6 days. On these dayssimultaneous
prey. It apparently then left the area, as there observations were made at the Main Lek, allowing matchedcomparisonsbetween sites.
was no further evidence of its presence.
More spooksoccurredat the Main Lek than
The effectof displaygroupsize.--For a groupdisplaying frog subject to intense predation the other sites, both in absolute numbers and
pressure,Ryan et al. (1981) showed that pre- in number of dayswith spooks.The only spook
dation risk was inversely correlated with in- observedat a small site accompaniedan Ornate
creasedchorussize for individual calling males Hawk-Eagle attack. This confirms the positive
from
strike
to death
was less than
1 min.
The
July1987]
Predation
at Cock-of-the-Rock
Leks
503
T^BLE3. Incidence of spooksand attacksduring simultaneousobservationdaysat the Main Lek and
smallerdisplaysites.Main Lek = 55 territorialmales,
Creek Lek = 6, Nest Slope Lek = 2, and Solitary
40
Male
site = 1. Observations
were
made from
0700
to 1700 on each day. "Probable attacks"includes
all casesin which an attackingpredatorwas seen
or inferred
from
Cock-of-the-Rock
Main /
>, 20--
ß
behavior.
Main /
Nest
Creek Slope
No. observationdays
Days with spooks
Days with confirmedattacks
Days with probable attacks
Total spooks
10--
Very Small MediumLarge
Small
Main /
Soil-
tary
!!
11/0
1/0
3/0
56/0
!2
12/1
0/1
4/1
37/1
6
6/0
2/0
4/0
29/0
Total confirmed attacks
1/0
0/1
2/0
Total probable attacks
5/0
4/1
5/0
males might be morevulnerable to predators
than peripheral males.
Cock-of-the-Rockleks are irregular in shape
becauseof the thick vegetationof the rain forest.The Main Lek was composedof three lobes
Fig. 4. Frequencyof attacksat the Main Lek in
of
courtsin a relatively open area,surrounded
relation to male group size. For each censusday the
daily mean group size was determined and assigned by thicker understoryvegetation.Males with
to the appropriate size class:very small (<10 males, territoriesin the thickervegetationwere on the
n = 9 days), small (10-19.9 males, n = 18 days), me- marginsof the lek and rarely receivedfemale
dium (20-29.9 males,n = 38 days),or large(•>30 males, visits. During the courseof the study 40 courts
GROUP
SIZE
n = 13 days).The frequencyof dayswith attackswas
then calculatedfor each group size class.
were occupied in this peripheral area, compared with 108 courtsinside. There was no evidence of predation on the peripheral males,
relation between group size and spooksdocu-
whereas 5 males on 4 of the more central courts
mented
were takenby predators(the ownershipstatus
of the remaining victim was unknown).
for the Main
Lek. Because of fewer ob-
servationdaysat the smallersites,I was unable
to analyze preciselythe risk per male at the
different-sizesites.The long-term persistence
of small displaysitesand the occupancyof territoriesthere by the sameindividuals for at least
3 breedingseasons
indicatespredationwas not
intense enough to preclude display groupsof
various
sizes.
I determined the density of malesin the area
of eachcourtby countingthe territoriesfalling
inside
a 5-m-diameter
circle
centered
on that
court. Two of the courts where predation oc-
curredwere in the areaof greatestmale density
on the lek (0.9 courts/m2). The other two courts
were in areaswith only slightly lower clusterThe effectof maleattributes.--Afinal question ing (0.5 and 0.6 courts/m2).Thesedensitiesare
concerningpredationon leksis whether males very high comparedwith almostall other lek
differ in their vulnerability basedon territory birds (seeTrail and Koutnik 1986:table 1). These
position, age, or both. Wiley (1973) suggested data reject the hypothesis that males in lowthat peripheral male SageGrouseshould be tak- densityor peripheralareasof leks are particuen by predatorsmore often than central males, larly vulnerable to predators.
Successful,older male Cock-of-the-Rock were
but he observed too little predation to test the
hypothesis.Hartzler (1974) found 5 predator- equallylikely to be killed by predatorsasyoung
killed male Sage Grouseduring his study, of and unsuccessful males. Males attain full adult
which 3 were adults with central territories, 1 plumageat 3 yr of age,generallydefend courts
was a peripheralmale of unknown age,and 1 for the first time as3- or 4-yr-olds,and matefor
was unidentified.
He concluded
that central
the first time at 4 or more years of age. I saw
504
PEPPER
W. TP,
AIL
no subadult 1- or 2-yr-old male attacked by
predators. All 6 Cock-of-the-Rock for which
there was direct or indirect evidence of predation were banded in adult plumage before
the first breeding seasonof the study (1980).
[Auk,Vol. 104
test whether lek display is indeed safer than
solitary display for males of a given species.
Cock-of-the-Rock at the Voltzberg inhabit
rate of 0.24 (n = 90), vs. a rate of 0.20 for males
undisturbed rain forest and are exposed to a
natural predator community. The predation rate
on males at the Main Lek was comparableto
the highest level previously reported for lek
birds (Hartzler 1974), and the significanceof
this predation pressurewas emphasizedby the
well-developed antipredator behavior of lek
males. The regular occurrence of predation at
all avian leks studied in relatively undisturbed
habitats (Koivisto 1965, Luresden 1968, Wiley
1973, Hartzler 1974, this study) suggeststhat
the importance of predation asa sourceof mortality on leks has been underestimated. This
does not mean, however, that lek-displaying
malesare more vulnerable to predatorsthan are
conspecificsthat display solitarily. Males on
large leks could gain antipredatorbenefitsfrom
three consequences
of group display:better detection and mobbing of predators,lower predator successthrough the flock-confusioneffect,
and lowered individual risk through the self-
that did not mate (n = 171, G = 0.718, not sig-
ish-herd
The agesat death of the four confirmedpredation victims were: ->3 yr (1 male), ->5 yr (2
males), and ->6 yr (1 male). The male whose
band
was found
beneath
the boa's tree was ->4
yr old, and the male that disappeared immediately after the intense panic was ->6 yr old.
Four of the males were killed after mating
was under way for the year, and two of them
mated before being killed (on average,only 33%
of territorial malesmatedeachyear; Trail 1984).
One victim
was the most successful
male on the
Main Lek for the three breeding seasonsbefore
his death and occupied a court in the densest
part of the lek. Overall, the yearly mortality
rates of successful
and unsuccessful
court
own-
ersdid not differ significantly.Basedon returns
of banded
individuals,
males that mated in a
given seasonexperiencedan annual mortality
nificant).
effect.
The benefitsof increasedvigilance in groups
have been documented for a wide variety of
organisms(Bertram 1978) and appear to apply
DISCUSSION
well to Cock-of-the-Rock.Displaying males at
To understandthe role of predation, detailed the Main Lek were alert, spooked readily, and
studiesof both the behavior of predators(e.g. were rarely attackedsuccessfully.I found a sigSchaller 1972,Spencerand Zielinski 1983) and nificant positivecorrelationbetweengroup size
the antipredator behavior of prey (e.g. Greene and frequency of spooks,and a suggestivebut
et al. 1978) are required. Such necessarilyde- nonsignificant inverse correlation between
scriptiveaccountsare basicto the development group size and frequency of attacks.If the oband testing of models of selection, and as such served decline in attack rate with increasing
are an indispensiblepart of modern evolution- group size is genuine, it suggeststhat raptors
ary biology (Greene 1986).
regard the largestdisplay groups as unprofitIt is generally acceptedthat males are safer able targets. I was able to judge group size at a
displayingon leksthan alone(Lack1968,Oring distanceby the volume of calling at the Main
1982). Some consider this to have been an im- Lek, and raptors may do so as well.
portant force in the evolution of leks (Hjorth
Snakesare the only predatorsfor which Cock1970; Wittenberger 1978, 1979). Others argue of-the-Rock lack effective antipredator behavthat predation is too infrequent or variable ior. The Cock-of-the-Rock were particularly
among lek speciesto be a general explanation vulnerable to the boa constrictor because of their
for lekking (Lill 1974, Borgia 1979, Bradbury terrestrial display, tightly clustered courts,and
and Gibson 1983). The available data have two extreme site tenacity. Once a snakeencounters
limitations. First, almost all studies have been
a lek it may be able to kill a significant proconductedin extensivelyman-modifiedareas, portion of displaying males. The stability of
leks in the face of this threat
where the predator community has been re- Cock-of-the-Rock
ducedin diversityand abundance.Second,there suggestseither that the feeding patterns of
is a lackof datacomparingpredationon display snakes are inconsistent with the methodical regroups of different sizes. These are needed to moval of all membersof a group or, more likely,
July1987]
Predation
atCock-of-the-Rock
Leks
that snakesrarely find leks (becauseof the rarity
of large snakes or leks, or both, or becauseof
the snakes'limited searching radius). This potentially significanthazard of lek display in the
tropics is thus apparently mitigated by the rarity and limited mobility of snakes,rather than
by direct or indirect benefits of group defense.
The increasedwariness of males in large display groups resulted in a greatly increasedfrequency of false alarms.The significanceof this
cost is difficult
to assess but could involve
three
different currencies:energy, display time, and
female preference. False-alarm spooks were
usually brief and involved most of the males
on the lek. Thus, the time consumed by false
alarms was probably small and equally distributed among the males. The energetic cost of
headlongflight off the lek severaltimesa day
could be more important, but the risk of ignoring an alarm was apparently greater. Finally, if falsealarmsare frequentenoughto disrupt
female courtship visits on a regular basis,they
couldimposean upper limit on lek size through
female preferences. Females could bypass the
largest leks to mate at smaller groups where
falsealarms interrupt courtshiplessfrequently.
More data on display groups of different sizes
are needed to test this hypothesis.
A second possibleantipredator function of
leksis the confusionof the predatorby multiple
fleeing prey, as demonstrated in fish schools
505
ritories on leks. I found no data to support this
hypothesisin Cock-of-the-Rockor other lek
species.Territory positiondid not correlatewith
predation risk in Cock-of-the-Rock.None of the
males taken by predators owned peripheral
courts, and two were
resident
in the most
densely packedarea of the lek.
Male Cock-of-the-Rockat leks perform active
behaviors,includingsurveillance,mobbing,and
evasion,that provide effective defense against
avian and mammalian predators. They appear
to lack such defenses against snakes, perhaps
becausethese potentially significant predators
rarely encounter leks. It has been stated that
forest
environments
are unfavorable
for
the
evolution of leks becausepredator detection is
difficult in thick vegetation (Wiley 1974; Wittenberger 1978, 1979). This hypothesiswas pro-
posedto explainwhy mostforestgrousedisplay
solitarily, while most grouse of open habitats
lek. Forest-dwelling birds in many other families form leks, however, including manakins
(Lill 1974, 1976; Foster 1981), hummingbirds
(Snow 1974,Stilesand Wolf 1979),and cotingas
(Snow 1982,this study).The effectiveantipredator behaviorexhibited by Cock-of-the-Rockin
rain forest suggeststhat correlations between
mating systemsand habitats are unlikely to be
due to predation.
ACKNOWLEDGMENTS
(Neill and Cullen 1974). The shrill alarm calls
and explosiveflight of dozensof malesscattering from the Main Lek in a spook could have
made target selection difficult for predators,
contributing to the very low attacksuccess
rate.
It may be relevant that the drab females often
remained on the lek during a spook. These females apparently relied on escapingthe attention of predators in the confusion created by
the colorful fleeing males. I never saw a predator attack a female.
Finally, males that display in groups could
benefit from the "selfish-herd" effect (Hamilton
1971, Wittenberger 1979, Ryan et al. 1981). Individuals can reduce their predation risk by increasedclustering,in effectseekingcoverwithin a group (Hamilton 1971). Care must be taken,
however, when extending this argument from
mobile groups of animals to the immobile arrangementsof territories on leks. Wittenberger
(1979) suggestedthat femalesmay reducetheir
predation risk by favoring central males or ter-
I thank
the Suriname
dation for Nature
Forest Service
Preservation
and the Foun-
in Suriname
for their
cooperationthroughout the fieldwork. Assistancein
the field was ably provided by D. Clark, L. Dallago,
G. Farley, L. Kellogg, B. McCafiery, and G. Tabor. S.
T. Emlen gave valuable advice at all stagesof the
project,and discussionswith H. W. Greene, D. L.
Koutnik, and J. B. Phillips greatly improved the
manuscript,as did the suggestionsof A. H. Brush,K.
Winnett-Murray,G. Caldwell, and an anonymousreviewer.I thankL. F. Baptistafor generouslyproviding
spaceand facilitiesduring the writing of this paper.
The fieldwork wassupportedby grantsfrom the NationalGeographicSociety,the NationalScienceFoundation,the NationalAcademyof Sciences,
the George
Harris Foundation,the Frank Chapman Fund, and
Sigma Xi.
LITERATURE CITED
/kMMANN,
G.A. 1959. Sharp-tailedGrousepredation
by Goshawk.J. Wildl. Mgmt. 23: 110-111.
BERGER,D. D., F. HAMERSTROM, & F. N. HAMERSTROM.
506
PEEPER
W. TRAIL
1963. The effect of raptors on Prairie Chickens
on booming grounds. J. Wildl. Mgmt. 27: 778791.
[Auk,Vol. 104
LILL, A. 1974. Social organization and spaceutilization in the lek-forming White-bearded Manakinß M. manacustrinitatus Hartert. Z. Tierpsy-
BERTRAM,
B. C.R. 1978. Living in groups:predators
and prey. Pp. 64-96 in Behaviouralecology,an
evolutionaryapproach(J. R. Krebsand N. B. Davies, Eds.). Sunderland, Massachusetts, Sinauer
Assoc.
BLus,L.J. 1967. Goshawk predation on Sharp-tailed
Grouse
in the Nebraska
sandhills.
Wilson
Bull.
79: 449.
chol.
36: 513-530.
1976.
Lek behavior
in the Golden-headed
Manakin, Pipraerythrocephala,
in Trinidad, West
Indies.
Adv.
Ethol.
No.
18.
LUMSDEN,
H.G. 1968. The displaysof the SageGrouse.
Ontario Landsand Forests,Res.Dept. No. 83: 194.
LYONßB., & A. KUHNIGK. 1985. Observations on nest-
BORGIA,G. 1979. Sexual selection and the evolution
ing Ornate Hawk-Eagles in Guatemala. Wilson
Bull. 97: 141-147.
of mating systems.Pp. 19-80 in Sexualselection
and reproductive competition in insects(M. S. MAXSON, S. J., L. W. ORING, &: G.-A. MAXSON. 1979.
Blum and N. A. Blum, Eds.). New York, Academic
Red fox capturesSharp-tailed Grouse at the lek.
Press.
Prairie
BRADBURY,J. W., &: R. M. GIBSON. 1983. Leks and
mate choice.Pp. 109-138 in Mate choice (P. Bate-
son,Ed.).Cambridge,EnglandßCambridgeUniv.
Press.
BROWN,L., & D. AMADON. 1968. Eagles,hawksßand
falcons of the world.
New York, McGraw-Hill
1973. The carnivores. Ithaca, New York,
Cornell
Univ.
FOSTER,
g.S.
Press.
1981. Cooperative behavior and social
organizationof the Swallow-tailedManakin (Chiroxiphiacaudata).Behav. Ecol. Sociobiol.9: 167177.
GILLIARD,E.T. 1962. On the breeding behavior of
the Cock-of-the-Rock(Aves, Rupicolarupicola).
Bull. Amer.
Mus.
Nat.
Hist.
124: 31-68.
of lower vertebrates (M. E. Feder and G. V. Lauder, Eds.). ChicagoßUniv. Chicago Press.
ßG. M. BURGHARDT,B. A. DUGAN, &: A. S. RAND.
Predation
and the defensive
behavior
of
green iguanas.J. Herpitol. 12: 169-176.
HAMILTON,W. D. 1971. Geometry for the selfish
herd. J. Theor. Biol. 31: 295-311.
HARTZLER,
J.E. 1974. Predationand the daily timing
of SageGrouseleks. Auk 91: 532-536.
HJORTH,I. 1970. Reproductive behavior in Tetraonidae with special reference to males. Viltrevy 7:
183-596.
JANZEN,D.H.
Chicken by a Goshawk.Auk 83: 137.
MOYLE$,D. L.J. 1979. Cooper'sHawk attacksSharptailed Grouse.Raptor Res. 13: 120.
NEILL,S. B. ST.J.,&:J.M. CULLEN.1974. Experiments
ators. J. Zool. 172: 549-569.
ORING,L.W. 1982. Avian mating systems.Pp. 1-92
in Avian biology, vol. 6 (D. S. Farner and J. R.
King, Eds.).New York, AcademicPress.
RUSCH, D. H., E. C. MESLOW, P. D. DOERR, &: L. B.
KEITH. 1972. Responseof Great Horned Owl
populationsto changingprey densities.J. Wildl.
Mgmt. 36: 282-296.
RYAN, M. J., M.D.
GREENE,
H. W. 1986. Natural history and evolutionary biology. Pp. 99-108 in Predator-preyrelations:perspectives
and approachesfrom the study
1978.
11: 124.
on whether schoolingby their prey affectsthe
hunting behavior of cephalopodand fish pred-
Book Co.
EWERßR.F.
Natur.
MORAN, R. J. 1966. Predation on a Greater Prairie
TUTTLE,&: L. K. TAFT. 1981. The
costsand benefits of frog chorusing behavior.
Behav.
Ecol. Sociobiol.
8: 273-278.
SCHALLER,
G.g. 1972. The Serengeti lion. Chicagoß
Univ. Chicago Press.
SNow, B. K. 1974. Lek behavior and breeding of
Guy'sHermit hummingbird,Phaethornis
guy.Ibis
116: 278-297.
SNowßD.W. 1971. Noteson the biologyof the Cockof-the-Rock(Rupicolarupicola).J. Ornithol. 112:
323-333.
1982. The cotingas.IthacaßNew York, Cornell
Univ.
Press.
SPENCERß
W. D., & W. J. ZIELINSKI.1983. Predatory
behavior of pine martens. J. Mammal. 64: 715717.
1983. Lira barbara.Pp. 469-470 in Costa
Rican natural history (D. H. Janzen,Ed.). Chicago,Univ. ChicagoPress.
KOFORD,C. B. 1983. Felisonca.Pp. 470-471 in Costa
Rican natural history (D. H. Janzen, Ed.). Chicago,Univ. ChicagoPress.
KolVISTO,I. 1965. Behavior of the Black Grouse Ly-
rurustetrix during the spring display. Finnish
Game Res. 26: 1-60.
LACK,D. 1968. Ecologicaladaptationsfor breeding
in birds. LondonßChapman and Hall.
STILES,
F. G., & L. L. WOLF. 1979. Ecologyand evolution of lek mating behavior in the Long-tailed
Hermit hummingbird.Ornithol. Monogr. No. 27.
TRAIL,P. W. 1984. The lek mating system of the
GuiananCock-of-the-Rock:
a field studyof sexual
selection. Ph.D. dissertation, Ithaca, New York,
Cornell
Univ.
1985a. Territoriality and dominance in the
lek-breeding Guianan Cock-of-the-Rock. Natl.
Geogr. Res. 1: 112-123.
1985b. Courtship disruption modifies mate
July1987]
Predation
atCock-of-the-Rock
Leks
choice in a lek-breeding bird. Science227: 778780.
1985c. The courtship behavior of the lekbreedingGuiananCock-of-the-Rock:a lek's icon.
Amer.
--,
Birds
39: 235-240.
the
lek
in
the
Guianan
1974. Evolution of socialorganization and
life-history patterns among grouse.Quart. Rev.
Biol. 49: 201-227.
WITTENBERGER,
J. F. 1978. The evolution of mating
systemsin grouse.Condor80: 126-137.
& D. L. KOUTNIK.1986. Courtshipdisruption
at
507
Cock-of-the-Rock.
Ethology 73: 197-213.
WII,E¾,R. H. 1973. Territoriality and non-random
mating in SageGrouse,Centrocercus
urophasianus.
Anim. Behav.Monogr. 6: 87-169.
1979. The evolution of mating systemsin
birdsand mammals.Pp. 271-349 in Handbookof
behavioralneurobiology.Vol. 3, Socialbehavior
and communication (P. Marler and J. Vanden-
berg, Eds.).New York, Plenum Press.