Foraging and reproductive ecology in a community of bats in Panama


Material Information

Foraging and reproductive ecology in a community of bats in Panama
Physical Description:
vi, 122 leaves : ill., map ; 28 cm.
Bonaccorso, Frank J
Publication Date:


Subjects / Keywords:
Bats -- Panama -- Canal Zone   ( lcsh )
Bats -- Feeding and feeds   ( lcsh )
Bats -- Reproduction   ( lcsh )
Zoology thesis Ph. D
Dissertations, Academic -- Zoology -- UF
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis--University of Florida.
Includes bibliographical references (leaves 115-119).
Additional Physical Form:
Also available online.
General Note:
"This study was submitted as partial fulfillment of the Ph.D. degree...".
General Note:
General Note:
Statement of Responsibility:
by Frank Joseph Bonaccorso.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 03887505
lcc - QL737.C5 B66
System ID:

Table of Contents
    Title Page
        Page i
        Page ii
        Page iii
    Table of Contents
        Page iv
        Page v
        Page vi
        Page 1
        Page 2
    Study area
        Page 3
        Page 4
        Page 5
    Materials and methods
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
    Phenology and food resources
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
    Species diversity
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
    Resource partitioning
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
        Page 94
        Page 95
        Page 96
        Page 97
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
        Page 103
        Page 104
        Page 105
        Page 106
        Page 107
        Page 108
        Page 109
        Page 110
        Page 111
        Page 112
        Page 113
        Page 114
    Literature cited
        Page 115
        Page 116
        Page 117
        Page 118
        Page 119
    Appendix 1. Seasonal variation in captures of each bat species on Barro Colorado Island in 1973
        Page 120
        Page 121
    Biographical sketch
        Page 122
        Page 123
        Page 124
Full Text






This work is dedicated to Clark Sanford, Julie Wiatt, and Bill

Biven, my field assistants, who endured a year of damp weather, irritating insects, bat bites, tough pork chops, and numerous other tropical hardships, yet shared the enumerable joys we encountered. We learned much together of tropical forests and ourselves arid are better people

for it.


This study was funded by NSF Grant GB-36068 to Dr. J. H. Kaufmann, NIH Biomedical Sciences Grant No. RR7021-07 from the Division of Sponsored Rsearch of the University of Florida to Dr. S. R. Humphrey, and the Environmental Sciences Program of the Smithsonian Tropical Research Institute. The Florida State Museum and Smithsonian Tropical Research Institute provided logistical support.

Dr. B. K. McNab, both in his writings and classroom discussions, induced and encouraged the "germplasm" of interest which launched me into the study of the ecology of tropical bat communities. Drs. S. R. Humphrey, J. H. Kaufmann, E. Leigh, N. Smythe, A, F. Carr, D. H. Hirth, and T. C. Emmel unselfishly took time to provide constructive guidance. Dr, Robin Foster verified my seed identifications and cultured in me a deep appreciation for tropical plant ecology, Clark Sandford, Julie Wiatt, Bill Biven, and Janet Hall faithfully assisted with fieldwork and laboratory preparations under trying conditions. The creative talents of Nancy Halliday and Sylvia Scudder have rendered the illustrations. Finally, I wish to thank the scientists, students, and visitors coinciding with my residence on BCI, as well as the Smithsonian staff, for bringing encouragement, friendship, intellectual atmosphere, and volleyball to an isolate field station and making 1973 the most pleasant and memorable year I have experienced.




AKNOWLEDGEMENTS ................................................... iii

ABSTRACT .......................................................... v

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

STUDY AREA ........................................................ 3

MATERIALS AND METHODS ............................................. 6

Mathematical Formulae ........................................ 10

PHENOLOGY AND FOOD RESOURCES ...................................... 12

SPECIES DIVERSITY ................................................. 20

RESOURCE PARTITIONING ............................................. 27

Canopy Frugivore Guild ....................................... 31
Groundstory Frugivore Guild .................................. 50
Scavenging Frugivore Guild ................................... 62
Nectar-Pollen-Fruit-insect Omnivor6 Guild .................... 65
Sanguivore Guild ............................................. 70
Gleaning Carnivore Guild ..................................... 73
Slow-Flying Hawking Insectivore Guild ........................ 80

REPRODUCT ION ...................................................... 87

Canopy Frugivore Guild ....................................... 87
Groundstory Frugivore Guild .................................. 94
Scavenging Frugivore Guild .................................... 97
Nectar-Pollen-Fruit-Insect Omnivore Guild .................... 97
Sanguivore Guild ............................................. 97
Gleaning Carnivore Guild..................................... 97
Slow-Flying Hawking Insectivore Guild ........................ 98

CONCLUSIONS ....................................................... 101

Species Diversity And Phenology ............................... 101
Foraging And Reproductive Strategies ......................... 104

LITERATURE CITED ................................................... 115

A PPEND IX ........................................................... 120

BIOGRAPHICAL SKETCH ................................................ 122


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



Frank Joseph Bonaccorso

August, 1975

Chairman: Dr. John H. Kaufmann Major Departmet nt: Zoology

Resource partitioning, reproduction, species diversity, and community structure in a forest community of 35 bat species were studied on Barro Colorado Island, Panama Canal Zone. Sixteen months of fieldwork w ere conducted between July 1971 and August 1974. Over 2,800 bats were captured, banded, and released with data collected on food habits.. activity cycles, habitat selection, reproductive timing, and morphological feeding adaptations for each species. Information on the seasonality and abundance of fruit, flower, and insect resources

used by bats also was collected.

The diversity of tropical lowland bat communities in any one

habitat changes significantly on a seasonal basis. Fluctuating levels of food resources require that many species utilize different habitats and foraging strategies through a year. Competitive interactions, predator avoidance, and climatic fluctuation further influence the


foraging strategies of each species, Tropical bat faunas can be broken down into feeding gUilds on the basis of general food habits and method of food procurement. Within the most complex guilds, such as the canopy frugivore guild, food resources are partitioned in time and space and by size and quality, Within the simplest guilds food resources are partitioned primarily by food particle size. The most important mechanism of resource partitioning separating similar species is food particle size. Some species complexes appear to be limited not by absolute amount of food but by the distribution of those resources in a few concentrated patches accessible to only a limited number of individuals at a given time. Reproduction coincides with high levels of available food resources within each feeding guild.



Mention of the words "tropical forest" among ecologists typically triggers visions of species rich communities, complex competitive interactions, and relatively stable environmental conditions. Indeed, faunal lists in the tropics are large,and food webs are intricately complex. It is also true that organisms inhabiting tropical latitudes usually are subjected to less extreme environmental fluctuations than are their counterparts in temperate or polar regions. However, it is too infrequently e-mphasized that even species in tropical forests must possess behavioral flexibility to counter and survive climatic and biotic environmental change. There are two major reasons for this general lack of insight. Firstly, few detailed studies of tropical organisms have spanned periods of Lcveral years or even seasons. And secondly, the behavioral responses of tropical species to environmental fluctuations are often quite subtle. Whereas temperate animals commonly exhibit

obviou-s and dramatic reactions to seasonal change such as hibernation or long distance migration, tropical species may only need to switch food types or microhabitats, or briefly halt reproduction. Nevertheless, genetic and behavioral flexibility are requisites for survival for most tropical as well as temperate species.

Tropical bats are particularly worthwhile subjects for studies of diversity, competitive interaction, and response to environmental fluctuation because of their individual abundance and the complex taxonomic


and ecological communities they form. About 100 species of bats occur in each of the small countries of Central America (Hall and Kelson, 1959). It is common to find 30 to 50 species in one macrohabitat measuring a few square kilometers in area. For example Barro Colorado Island (15 sq km), Panama, currently supports populations of at least 35 species.

Among tropical bat species, few are known or suspected to reproduce year round or to specialize on constantly abundant food resources. The common vampire bat, Desmodus rotundus, is one notable exception (Wimsatt and Trapido, 1952). Instead, most bats, even in equatorial regions, are seasonally polyestrous or monestrous in reproduction (Baker and Baker, 1936; Mutere, 1970; Fleming, 1973) and make seasonal shifts in food habits (Wilson, 1971b; Fleming et al., 1972; Heithaus et al., 1975).

The objective of this dissertation is to delineate adaptive strategies used by tropical bats that enable them to survive fluctuating environmental conditions and coexist with numerous similar species in complex communities. The field work represented herein documents seasonal changes in diversity, mechanisms of resource partitioning, and reproductive timing through one complete year and portions of two other



The primary research site was on Barro Colorado Island (BCI).

Barro Colorado lies within freshwater Lake Gatun, in the Panama Canal Zone, at 9 0 10 North latitude and 79 0 51'1 West longitude. This field site was selected because it has a rich bat fauna, relatively undisturbed mature moist forest, modern living and laboratory facilities, and reference collections of animals and plants. A secondary site was located on the mainland opposite BCI at the base of Buena Vista Peninsul a.

The climate of this lowland area of Panama is warm and humid with a seven-month wet season and a three-month dry season. Dry season

months, January through March, each receive less than 60 mm of rain. Wet'season months, May through November, typically receive in excess of 250 ,mm of rain. April and December are months of transition between dry and wet seasons and receive amounts of rain that vary considerably from year to year. Thus in years when April and December are very dry, the dry season may last for five months. Average annual rainfall since 1926 has been 2,820 mm (Smythe, 1974). Monthly sums of rainfall for 1973 are shown in Table 1.

During night time sampling of bats, relative humidity under the forest canopy never fell below 80 percent. Measurements were made at

2 m above ground with a sling pyschrometer. Daily temperatures on the forest floor fluctuate from a mean minimum of 22.1 0 C to a mean maximum of 28.0 0 C with no significant seasonal variation (Smythe, 1974).



m co
4J r"
>- (14
E 00
V) m



ru r
0 "0
ra L- <

u M C14

ro m co

0 cli

E c 00
E co



fo m

E M _0 fo
C: Ln
0 CL 4

0 4-J

fu ra -0
4- _r 4- .- 0 c 4-j c E

0 cc E 4


Barro Colorado is in the Iropical Moist Forest life zone (Holdridga, 1967). This 15 km2 island is covered with mature forest that is over 60 years in age. The only human disturbance to the vegetation' osults from re-cutting forest trails and maintaining a small laboratory clearing, and an undetermined amount of illegal poaching. Further details on the geology, climate, biology, and history of the island are given by Kaufmann (1962) and Foster (1973).

Seventeen sampling stations were located in an approximately 2 km

central strip of Barro Colorado Island and one station was on Buena Vista Peninsula (Fig. 1). Habitats sampled during the study are classified as mature forest (14 stations on BCI), creeks (3 stations on BCI), and second growth (I station on Buena Vista). The mature forest has a completely closed canopy and is a minimum of 60 years old in all places. Some tracts within the forest have been undisturbed for 400 years (Robin Foster, pers. comm.). The creek stations are lined w;th rich .shrub growth and the creek bed receives direct sunlight. The second growth habitat at Buena Vista is approximately 20 years old and consists of thick shrub growth and scattered small trees that form a discontinuous canopy.

Ex.cept on rare occasions v/hen nets were damaged by tree falls or vandalized by poachers, each sampling station consisted of four or six

6 x 2 m mist nets and one or two Tuttle harp traps (described in Tuttle, 1974) set across permanent trails. Nets were set in pairs at 100 m intervals, with one of each pair at ground level (0 to 3 m) and the other at subcanopy and lower canopy level (3 to 12 m). Early in the study nets were rigged in the canopy as high as 25 m above ground, but use of these nets was soon discontinued because few bats were captured in themwhich seemed to reflect a lack of much flight activity in the canopy levels. Harp traps were usually set at ground level in low, narrow tunnel-like passages created by the vegetation and trails. At a few stations where



E (a



(U S





........ ..............

.............. .

.. . . . . . . .. ... .

.. ..............
.. ..........
... ...........

....... .............

....... ..

. ................

..... ..... ....

... ............

C 5 ------ ......

. . . . . .



..... ...... ..

.... .........


.. .. .. .. .. . .. . .. .. .
... . . . . . . . . . . . . . .
. .. . . . . . . . .


. . .. . . . . . . . .
. . . . . .

. . . . . . .

1, -7.

. . . . . .. . . . . .. . . . . . . .... . . .

. . .. . . . . . . .. .

.... . . . . . . .

... .........
... .................

. . . . . . . . . . . . . . ... . . . . . . . . .

. . . . . . . . . . . . . . ...... . . . . . .

. . . . . . . . .. :-X .: X
. . . . .

. . . . . . ..

.. . . . .

. . . . . . . . . .
. . . . . . .
. . . . . .
. . . . . . . . . .. . . . .
. . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .

. . . . . ... . . . . . . . .
....... . . . .... .. . . . . . . .

... . . . . . .

. . . . .

. . . . . . .
. .. . . . . .

. . . . . . . . . . . . . . . . .


... .....

.... .. .............. ..

. . . . . . . . . .

. . . .. ... .
. . . . . . .. . . . . .
. . . . .

... . ......

............... .. ..


.. ......................
. . . . . . . . . . . . . . .

.......... ......

............... .....

........ ..

. . . . . . .. . ..

I C:


the vegetation permitted, harp traps were rigged in subcanopy level

"tunnel s".

Nets and traps were open from sunset to sunrise 67 times between

11 January and 31 December 197/3. On 28 other nights during that period sampling was conducted for less than a full night. The total sampling during 1973 involved 4,376 net-hours, 1,213 trap-hours, and 2,324 captured bats.

In 1971, 3147 net-hours of sampling during a pilot study yielded 282 bats between 20 June and 18 August. In 1974, 454 net-hours of additional sampling yielded 278 bats between 10 June and 17 July. No harp traps were available during these times for effective sampling of small insectivorous bats.

Because Crespo et a]. (1972) and Morrison (1975) have demonstrated that vampires and fruit bats avoid flying during intense moonlight, whole-night samples were taken only between the last and first quarters of the moon. Only such whole-night samples were used to calculate species diversityand activity cycles were taken during phases of the lunar cycle that do not produce enought light to influence bat flight activity.

Nets and traps were, checked at least twice every hour for the purpose of removing bats. Whenever possible, checks were made more frequently to prevent bats from chewing out of nets. Upon removal from a net or trap each bat was placed in an individual cloth bag. Usually within an hour after capture the bats were banded and released at the sampling, station. The following data were recorded for each individual: species, hour of capture, capture location, sex, age class, reproductive condition of females, food in feces or mouth, weight, and forearm length.


Age classes were distinguished as follows. Infants were unable to fly and were encountered only when carried by the mother. Juveniles were able to fly but still had the infant pelage. Subadults had the adult pelage but were smaller in weight than adults and were reproductively immature. Adults possessed both adult pelage and weight.

Pregnancy, lactation, and reproductive inactivity of adult females were determined by palpation. Additionally, females could be distinguished as nulliparous or post-lactating by examining the condition of the teats.

Fecal pellets obtained from individual animals were placed in separate glassine envelopes for laboratory identification of food species. Fruits and pollens in fecal pellets were identified to species by comparing unknowns with seeds, pulp fibers and pollen grains in a reference collection assembled by the author. Pellets were collected from insectivores but remain unidentified because the hard parts of arthropods eaten by bats are masticated into tiny fragments that are difficult to identify. Pollen on the fur was collected by swabbing with a gelatin described by Beattie (1971). The pollen-containing gelatinwas then melted on slides for microscopic identification. Frequently, animals were captured with whole fruits held in the mouth. Additional information on food habits was gathered by placing plastic sheets under two roost trees of Carollia perspecillata to gather discarded fruits and fecal matter.

Mathematical Formulae

(I) Species Diversity, HI pi loge Pi, where pi is the number of the ith species divided by sample size (Shannon and Weaver, 1949).


(2) Equitability, E = HI/Hmax, where Hmax is the natural logarithm of the number of observed species (Sheldon, 1969).

(3) Niche breadth, loge B = -2,Pi loge Pi, in which the functions are the same as described in Equation 1. Values approaching zero indicate narrow niche breadths and specialists. Values approaching one indicate wide niche breadths and generalists (Levins, 1968).
2 2
(4) Niche overlap, CX = 2E Xi Yi/E Xi + Yi where Xi is the proportion of the ith food species in the diet of bat species X, and Yi is the proportion of the ith food species in the diet of bat species Y (Morista, 1959). I follow Zaret and Rand (1971) in considering species with overlap values greater than 0.6 to be critically similar in terms of food overlap.


Most of the bat species on Barro Colorado depend largely on fruit, flowers, or insects as food resources. Only a few species feed on the flesh or blood of vertebrates or non-insect invertebrates, The abundance and diversity of fruits, flowers, and insects in Central America, even in moist and wet forests, strongly fluctuate on a seasonal time scale (Foster, 1973; Smythe, 1974; Frankie et al., 1974).

Pollen and nectar on Barro Colorado are available to bats as reliable food sources only in the dry season, and only four species of flowering plants are known to be used by bats (Table 2). Two common species, Ochroma ]a.opus and Pseudobombax septenatum, flower from mid-December to mid-March. While these two species are in bloom nectar and pollen are very abundant, The other two pollen types used by bats remain unidentified. One of these is known only from February-March sampling and the other from August-September.

Fruits from 45 plant species were found to be eaten by bats on the island (Table 2). Nineteen of these species were trees, 11 were shrubs, four were vines, four were epiphytes, and seven are unknowns, Mature fruits of the species eaten by bats are available all year. There are times, however, when few fruits of only a few species are available. During 1973 a maximum of 19 fruiting species was available from midMarch to mid-April, and a minimum of 6 species was available in NovemberDecember (Table 2). Two of the fruits available in November-December, 12



Ln rQ)


-0 Ln

Ln 0_ 4-J
m u
M 4-1 CD

C 4-J CL

2 = v


-0 4J

in -P

r_ C Cl.
o c
4 --14 (D LU

0 0
(D 4 LA m I ro LiLn ILn V)

c 0 (0 4-J V)
Ln <

2 c) 4 t4- 4-J
CL 11 M
4-- F= 4J
0 ru

U Q)

E V)
c :3 .
4 4J
rc ro C> ro c Ln
co Q)
Ll SQ) 0)
Ln E 0 4
4J V) Ln cu LO
C: a) 0 x C: ro
0 -0 T all cc ro E
z 0 C- CD m C: E 4- 0-1 -3 M
0 E C14 C-4 Ln 0 01 0
4- ra 0 C: V) c C: Ln
Ul Ln ro C) 0 ro
E 0 4-J
4 41 0 -0 0 Ln -0 0 0 CD Z) cu 0) 0- rD
ro C: :3 C: c 0 cil >-I c L. 4 0 L-0 u u CO 0) z e u 0 ro
CU. 0 u V) rZ : U: I L I LZ 0- LM
F- -0 4- CD Cl- ::D V)


0 z




4-J Lij LLJ uj LO LU


:3 C: fa
oj I
Ln 0 Ln

U') (n u u E
ro E U) (o ro (n
a) 41 :3 U') u Lr*%
Q) (a
fu (U u 4-J E (a 4 CL <
E tn Ln :3 E T 4-)
4-J Q) S- ro
a x o >- E u m ro I- m V)
0 u Q) 4- E ro 7 U
u Q) =3 -c- C). Ln
V) V) I- D 0-1 0 > ME 10 c (D o o
4-J I- u Q) U I (D .2 3. 3: a)
4 ) 0 ro c 0 M Ln Q) 0 0 E M 4j
u rD E 1 m V) c c 7 w
ru V) Lr) -P LO
o Z- V, 0 0
< I
CL L.) < V) CL C:






ra I It 11

Ln V) Ln V) Ln V) (n V) Ln tn

CL- Ln


-a t4
c E E 0
D Z3
n E
c ro ru 41 41 m -T C) -:T C> a)
0 :3 0 fu m 0
u 0) 07 :: C %D C)
0- 0 u CD
(D -0 V) c c C: c c c c c co
&- L- 4J 3: 3: 3:
0 M a) o o o o o o
u u E Q) c c r_ c c m ro
0- CL CL 0- \1 _z -Y :e :e ,4 4-1 >
c c C: c c c 0 ru
C, CL I CL 0-- Q-- :D =D ZD :D =D Z)

Ficus insipida and F. yoponensis, were very scarce, but Spondias radlkoferi and S. mombin were quite abundant.

Most of the plant species producing fruits eaten by bats produce

ripe fruits for periods of only one to four months. Only three species, F. insipida, F. obtusifolia, and F. yoponensis, have ripe fruits available nine or more months per year. Individuals in the populations of these fig species fruit asynchronously once or twice per year. F. insipida and F. y v.ponensis populations show three major fruiting peaks and troughs each year (Morrison, 1975).

The plant genera CeLropia, Spondias, Vismia, and Pier each have two or more bat-dispersed species that set fruit in sequential time periods (Table 2). There are 10 species of pipers on BCI eaten by bats. Though no one of these species is available for more than a few months, two or more species have ripe fruit throughout the year. Heithaus et al. (1975) report that pipers are important bat fruits in Costa Rica and that several species are available in similar sequential series. Snow (1965) reports that 18 species of the bird-dispersed genus Miconia are sequentially available throughout the year in Trinidad.

The biomass and numbers of nocturnal insects caught in light traps

in Barro Colorado forest over a three-year period were reported by Smythe (974). Though these samples represent all nocturnal flying insects, not just those eaten by bats, they provide a useful index of the abundance and fluctuation of the potential food resources for insectivorous bats through the year (Fig. 3). The light trap collections show that nocturnal insect biomass in the early wet season is as much as eight times that of the biomass at the end of the wet season and during the dry season (Fig. 2). Large insects (> 5 mm in length) were responsible

C 0)

() Q 0 -C

E 0 V)

Q 4-J


0 .- W

4-J 4-J

E 0 4- S
0 U)

0 0.- E
41 (o (a M
4-J c cC) 0 V)


V) j

C14 :3
0 Q)







........ .......

.~ 0
*2. ... ... .. ... ... ... .. ...........

.3:: ....

.. .. .. .. .. .

~ r7IAALe) .. .. .. .. ..
..... ..... ....

.... ... ... ... .... ...

............ z

.........K. L

..... .....

.... .... ... ... .... .... ...
........ .... .. .... .... ...

.... .... ...

........ ...................

... .. ... .. ... .. .. ....... .. ..


for this seasonal change in biomass, with Isoptera, Diptera, and Lepidoptera among the orders eaten by bats that have particularly dramatic population increases in the wet season. By contrast, small insects k 5 mm in length) were abundant throughout the year.


Considerable variation occurred in the bimonthly measures of species diversity (Table 3). All three diversity indices, H', E, and SN, were at maxima during the dry-to-wet transition sampling period. The three diversity measures then declined in each of the next three bimonthly periods to a minimum in the late-wet season. Whereas 27 bat species were present in the study area in the dry-to-wet transition, only 19 species were sampled in the late-wet season. During this same interval H' dropped from 2.33 to 1.52 and equitability from 0.707 to 0.515. In the wet to dry transition period the diversity values began to increase. Dry season values were very similar to wet-dry transition values,but SN increased from 22 to 25 in this period.

The diversity values were lowest in the mid- and late-wet season

samples because 13 of the species common in the dry season became noticeably rare or absent (indicated by asterisks in Appendix 1) from the study area in one or both bimonthly periods. These include seven insectivorous species (47%g of the total insectivorous species), five frugivorous species (42%), and one nectarivorous species (50%). These species appear to move to other habitats precisely at the time when bat-dispersed fruits and nocturnally flying insects, the two most important food resources for bats in this community, become relatively scarce on the study site (Fig.

2 and Table 2).

Of the seven insectivorous bat species that move out of the

mature forest late in the wet season, six are fol iage gleaners and one 20


0- 0 rru .- (D r co
Q) u Cn 4 U' co
u C l
c -C V) >
ro u c 0 C C
-0 m (U
-0 0 r-u 0 4J
CI- 4 m
(0 0) 0 CD m
_rZ 4

fu -W
:3 0

4 fO m
-0 4 c"I 0-%
4-J W co U-1
Ck C


C14 0-\
-0 L- LLJ m CD* C)
E (D ::3 n


L) CL Q) E 41 0 >
(D ru

4-j c u n (a
.- ro >U) U

> ru

W 0 I- 0 u 0 Lr\ m --T
V) 0 Lr) ru C


0 c (U
co U) c ro
fu u -C > a) j a C a _0
Ln 0 E
co E 4 co
C 4 Lij V) CL
0 rD Ln ro
> 4-) E Stn cn a) :3 :3
0 4-J 4-J
n 4-J a M
ru C Ln ra
0) 0 (v ro W L5
u u LO ro
-0 a) =3 a) 4-J 4-1
(D a- 0- 0- ro 0
(n Lij V) ca F-


is an aerial hawker. The foliage gleaners prey chiefly on large insects such as cicadas, grasshoppers, and roaches (Wilson, 1971b; this study). Smythe (1974) has shown that large nocturnal insects (greater than 5 mm in length) become quite scarce on this same study area beginning in the middle of the wet season, whereas small insects (smaller than 5 mm in length) are relatively constant in abundance through the year. It appears that the foliage-gleaning bats find food resources in tile study area sufficiently depressed in the latter part of the wet season that they move out of the area. On the other hand, small insects remain an abundant food resourceand aerial feeding bats remain in the mature forest of Barro Colorado all year. Peropteryx kappleri, an aerial feeding bat, is an exception as it does move out of forest habitat in the late wet season.

Of the two bat species that are primarily nectarivorous in this

community, Glossophaga soricina switches from pollen and nectar to fruits and insects in the early wet season and then moves out of the area in the middle wet season, not to return until bat flowers appear in the dry season. Phyllostomus discolor stays in the area the entire year, subsisting on fruits and insects during the wet season.

Among the fruit-eating species that seasonally move in and out of the Barro Colorado mature forest is Vampyressa pusiIla, the smallest species of the 13 frugivorous bats on BCI. V. pusilla is a feeding specialist on small fig fruits. This bat left the study area when few individuals of its most important food species, Ficus yoponensis, were producing fruits in July-October and returned in November when mature F. Yponensis fruits were again abundant. While specific reasons why other frugivorous species moved in and out of the study site are unclear,


their absence in September-November corresponds with the annual period when few food plants are producing fruit (Table 2).

The abundance of individual bats on the study area, as measured by the number of bats captured per sampling-hour, showed a pattern of seasonal change markedly different from the diversity measures pattern. Bats were captured in greatest numbers relative to sampling effort in the dry--to-wet transition, 0.425 bats/hr, and in the late wet season,

0.423 bats/hr (Table 3). The large numbers of bats captured in the dryto-wet transition sampling reflect large populations. Food resources were abundant then; females of most species were in the latter stages of lactation; and juveniles were entering the flying population and learning to forage. These latter two activities are among the most energetically demanding in mammalian life cycles (MigUela, 1969; Studier et al.,

1973), and the timing of these costly activities seems geared to a period of food abundance.

The late wet season peak in capture rate does not solely reflect large numbers of individuals on the study area. Though a number of species had at this time moved out of the young forest, some of the remaining frugivorous species were recruiting juveniles into the flying population from the second birth pulse of the year. Probably much of the high capture rate is attributable to intense foraging activity necessitated by low food supplies.

The diversity values for insectivorous and frugivorous species,

when computed separately, change similarly through the seasons (Table 4). For both groups, species diversity is high from mid-November through mid-July and low the remainder of the year. Fruit bat diversity rises and falls in time as does the diversity of fruit (see Table 2). However,


M 4J ru C) CA co rl_ C)
1 01% m
4-J Lt) I
0 >
ro 0
n L> rD

0 m cI co
a) 4 ON Lr C. co
0 Q) 41 _zr -::Tru 00) C
U) tn

ro 0
4-J a
0) 0 CD rl_ co Lr% 00
-1 4-J LO m I-- C) r LC"\
m Lf'% c"l C11 Lrl%
_0 CIL L) C
t- E >, ra ro u

73 cli _zr
E !v 4-) .43- m

r vi c Ll- %0
Q) 0 -0 S.- -0 Q) 0
U Ln 0 >
LO u
0 Q)
L- 4-1 0 LL Ul)
4- 0
: ir. 4-J ra C14 m 00
C) 00
>, Ln I rl_ %lo
S- M (D

Y) Ln E

u u ro
0-1 0 -0
CL 0- [: t-- C)
V) Q) ro Lr\
C: 4-J (U rlru u C
0 $- u
-I ro 0 0)

> 0

> 4-J 11)
m V) -0 m LLJ

0 E

Ln M 4L

> 4-J E V) >

0 -C V) m
Ul) Ln
4 ro 0) Q)
-0 ro C: 41 4a) 0 u ro u u
(A Z 0) 4-) (L) Z; cl, 41
0- tT 0- 0 CL C7 CL 0
V) LU cn F- V) LU V) F-

insectivore bat diversity sharply rises four months before insect biomass explosively increases (see Fig. 2). The latter anamoly may be due to ineffective harp trap placement and particular under-representation of the abundant species, Pteronotus_ parnelii, during the first months of field work. This would cause the dry season diversity value to be higher than it should be. Most species of fruit bats remain in the BCI mature forest habitat throughout the year. On the other hand, the species number of insect-eating bats in the dry-to-wet transition is nearly double that of the late wet season because of the movement of foliagegleaning species in and out of the forest.

A measure of annual variation is achieved by comparing the diversity of frugivorous species in June, July and August of 1971, 1973, and 1974 (Table 5). In all three years there is a consistent trend toward lower diversity as the wet season progresses. However, the magnitude of the diversity values varies from year to year. This indicates that some annually variable factor or complex of factors, possibly food availability, predation, or reproductive success, influences fruit bat species diversity. Insectivore diversity is not compared because harp traps for effective sampling were available only in 1973.




4 U 4-J


S.- 0-) Ln
L. -0 a) Ln
4- 4 E
o In ru E


D :3 E 0
c Ln _r4
Q) 0
CID ff) -z3- (D
a) j
0 Ln 4-1
Q- :3
Ln (0 0-)

(D (U

4J fu rE

0 4J
-0 Ln 4-J
m a (T -:J- 110 Q) 0
E W 0- 4-J
ru ca.
V) E

-C rZ

Ln D

> u 4-1
ca a)
_0 Q
$- CTI
U) CD -0 C:
m 1-0 C) 0 U) .0) a) cu (D C) C) C)r-L >, Q) E uV) L. ro
ca (n
L- 0- LrN Lf\ o &- Ln
__ 00 CD Q)
C: 41 J-0 4 tn

4J 0 > 4-J
(D ca
Ln E
c 4J CU
ru 0
> Ln U') L- Ln
co 0 0)

ro o >
D .- LO 4J
C 4 Q)
c 0 >
< LU) =1 43)
Q) u >
m cr)

E 0
ro 0


Thirty-five species of bats were found to coexist on Barro Colorado Island in 1973. Thirty-one species were captured in diversity samples and four additional species were seen in flight or at roosts. Noctilio leporinus, N. labialis, and Molossus molossus restricted their flight activities to habitats that were not sampled--the shallow inlets of the lake (Noctilio) and above the forest canopy (Molossus). The fourth species not captured in the diversity samples, Vampyrum spectrum, is a top carnivore and may be represented by very-few. individuals on the island. A pair of V. spectrum was netted by A. L. Gardner and D. E. Wilson on 5 January 1973. I saw a single animal in June 1973 flying at dawn. No other sightings of V. spectrum were reported in 1973.

A first step at understanding how 35 species of bats coexist on

this small island can be made by dividing the fauna into feeding guilds. Feeding guilds will be distinguished on the basis of two parameters-general 'food type and method or place of food procurement. It will be assumed that little or no competition for food resources occurs between members of different feeding guilds, though they may compete for roosting space. The bats on Barro Colorado may be divided into nine feeding guilds, each of which contains one to nine species.

Justification for the placement of species into specific feeding guilds will be provided in succeeding sections. For the moment, the feeding guilds are defined as follows:

(1) "Canopy frugivores" -- forage mostly on fruits that grow in



the trees of the canopy and subcanopy level of the forest, above 3 m from the ground.

(2) "Groundstory frugivores" -- forage mostly on fruits of shrubby groundstory plants, 0 to 3 m above ground level.

(3) "Scavenging frugivores" -- feed mostly on over-ripe fruit.

(4) ''Nectar-pollen-fruit-insect oinivores" -- forage for pollen and nectar from flowering trees when available in the dry season and then switch to a fruit and insect diet in the wet season.

(5) "Sanguivores' -- feed only on the blood of mammals and birds.

(6) "Gleaning carnivores'' -- forage for small animals (arthropods or vertebrates) that are perching or moving on vegetation or on the ground.

(7) "Slow-flying hawking insectivores'' -- forage for flying insects in small openings beneath or in the forest canopy or over streams.

(8) "Fast-flying hawking insectivores'' -- forage for flying insects above the forest canopy or in very large open spaces.

(9) "Piscivores" -- forage for fish or aquatic invertebrates at or just above the surface of lakes and large streams.

The distribution of mean body weights for each bat species on

Barro Colorado by guild is plotted in Figure 3. Three guilds contain a single species, and I expect that each of these species is sufficiently unique to preclude serious interspecific competition for food. The species within the groundstory frugivore, canopy frugivore, and piscivore guilds increase in body weight with a geometric progression factor of about 1.3 to 1.8 (with one exception in the canopy frugivore guild). We might expect that the species within each of these guilds exploit very similar types of food, captured in very similar manners, and that

Figure 3. Mean body weights of bat species by feeding guilds. (Dashed lines separate members of different families that belong to the same feeding guilds.) 1. Carollia castanet, 2. C. perspecillall,
3. Vamgyressa pusilla, 4, Thiroderma trinitatum, 5. Artibeus Ehaeotis, 6, Vamgyrops helleri, 7. Chiroderma..Vijjosum, 8. Vamp Wes caraccioloi,
9. Artibeus jamaicensjj, 10. A, lituratus, 11. Centurion yLfj, 12. Glossophaga soricina, 13. Phylloltomus discoloL, 14, Phyl]oderma stenops, 15. Desmodus rotundus, 16. Micronycteris meaplotis, 17. M, brachyotij, 18. Mimon crenulatum, 19. ronycteris hirsute, 20. TrachoRs cirEhosus, 21. Tonatia sylvicola, 22. T. bidens, 23. Phyllostomus hantatu 24, Vampyrum spectrum, 25. Rho2eessa tumida, 26. Myotis nigicans, 27. S19ccopteryx leptura, 28. Centronycteris maximilliani, 29. S2ccqpteryx bilineata 30. Peropteryx kappleri, 31. Pteronotuj sugpurensis, 32. P. Larnellii, 33. Molossus f 34. Noctilio labially, 35. N. leporinus


41 CC"



0 co



r<" cli


4n V)
Lu LU Lu uj Lu w
cx w- c cd oc
o o 0 o b2 0
> > > >
3 -z -Z u I
D cc X cz w In
rx m z (x a: i
< 00 0
< > >

PO 0 :7-- LU w
z 0 z
< <
0 In
fle z


food resources are partitioned largely by particle size as predicted by the theoretical reasonings of Hutchinson (1959), McNab (1971a and b) and May (1973). These authors postulate that similar species may avoid competition for food by differing in body weights by a factor of at least 1.3 (May, but McNab and Hutchinson used the figure 2.0), each species specializing in food particles proportional to its body weight (and to the linear dimensions of its food handling apparatus, e.g., tooth row length, gape size, tongue length, etc.).

The gleaning carnivore and slow-flying hawking insectivore guilds

each contain some species that are very similar in size to other species of their guilds. We might expect that such species feed on similarsized food particles of very different taxa or of similar taxa from different microhabitats.

The nectar-pollen-fruit-insect omnivore guild contains species very different in size; this may partly result from a recent extirpation of Lonchophylla robusta, a species intermediate in size between Glossophaga soricina and Phyllostomus discolor. The sexual dimorphism in body weights of P. discolor is another complicating factor. Canopy Frugivore Guild

Body size

Eight species, all in the subfamily Stenoderminae of the Phyllostomatidae, constitute the canopy frugivore guild on Barro Colorado. These eight species range from 8.1 to 69.3 g in mean body weight (Table 6). There is a mean increment of 1.44 between the body weights of adjacently sized animals among the seven species designated as fig feeding specialists in Table 6. Artibeus phaeotis, a feeding generalist, and Chiroderma trinitatum are nearly the same in size.


Table 6. Weights rn grams of canopy frugivore bats on Barro Colorado.

Sample Sample Wgt / Wgt* Remarks
Species Mean size size Ig sm
(OR) (S.D.) n

V. pusilla 8.1 0.6 22 -- Fig specialist

C. trinitatum 12.3 1.2 7 1.53 very rare fig specialist

A. phaeotis 13.0 1.2 30 -- food generalist

V. helleri 16.2 2.2 8 1.31 very rare fig specialist

C. villosum 22.4 2.1 13 1.38 fig specialist

V. caraccioloi 36.0 2.3 27 1.61 fig specialist

A. jamaicensis 47.2 3.4 30 1.31 fig specialist

A. lituratus 69.3 5.6 30 1.47 fig specialist

mean ratio of weight increment=1.44

weight of larger species divided by weight of smaller species in the pair compared.


Food selection

All eight canopy frugivore species feed primarily on fruits of

large canopy and subcanopy trees, in particular figs of the genus Ficus. Over 60% of the annual diet (by frequency of occurrence in fecal matter) of seven of these bat species consists of fig fruits (Table 7); these include C. trinitatum, but not A. phaeotis of the same weight. A. phaeotis depends on figs for 30% of its diet. Five species of Ficus, all of which are green colored at maturity, are eaten and dispersed by these stenodermines on Barro Colorado. Fig species that produce large fruits are preferred by large batsand fig species that produce small fruits are preferred by small bats.

Figs form the bulk of the diet of Artibeus jamaicensis throughout most of the year. However, during the latter part of the wet season and the very beginning of the dry season mature fig fruits are very scarce (Morrison, 1975). At this time A. jamaicensis turns more heavily to other fruits and pollen (Table 8). The relative importance of pollen in the diet of A. jamaicensis is grossly underestimated here because my sampling schedule did not coincide with the two weeks in late December and early January when figs were very scarce and flowers were very abundant. Similar seasonal switches in diet also probably occur for A. lituratus and V. caraccioloi but the data are weak. No conclusions can be made from the scant data on the smaller species of canopy frugivores with respect to seasonal switches in diet.

Unlike the fig specialists, A. phaeotis eats a more even distribution of many types of fruits (Tables 7 and 8) with no one species strongly dominating the diet. Throughout the year figs are a minor component of the diet, while other fruits are very important in certain


cn UN m % o 110 r- ul C)
M C,4

0 4> 0

0) Ln
c _0 0
ru 0 (a &- _0 m -0
0 0 0
4- c C W 0- rZ 00 0 Q) m w
0 0. c- c- Lo c- V) Ln U)
c E c 0 0 -a rZ 0 c
(U ro 0- C
u 0
4-J CL n
cu U) U) L LZ L L
0 L
4J 0

-W 4-J
cu C:
-0 m
I- 7L
0 4-J
Ln -0
0 0
E 0
u 04d) L- 0
Cl. 4U'l fu Ln Ln v) Ln Ln (n to Ln
4-J L- =1 =3 =1 =3
(U 0 u u u u u u u u

ro cu L.L- Ll- U- LL- L.L- Li- U- Ucr,
m 4-- 0 0 ONO 010 NO ON ONO ONO ONO
S- 0 C14 CD CD r-I C) 00 Ll%
%.D C)
4-J fU cu
C 4-J

u 0

0 0) E
0 Cl4- u
ro 44-J 0
0 >a u
E c

4-J cr
(A E 0
0 :3 V) 0
4-1 Ln E 0 c 4m .7 :3 .3 W 4,j
f 4 0 U
u 0 (U 0 U
W c CU m m
0- Ln ro E 4-1

C.L 4-J CL, -C > u

ro fu


Table 8. Bi-monthly samples of important food species in the diets
of Artibeus and Vampyrodes. Sampling periods begin at

Food Species Jan-Mar Mar-May May-Jul Jul-Sep Sep-Nov Nov-Jan

Artibeus jamaicensis

Ficus spp. 18 25 25 35 21 17

Cecropia spp. 3 3

Spondias spp. 1 8 6

Quararibea 9

Pollen 1

Total feeding samples* 20 30 32 39 35 37

Artibeus lituratus

Ficus spp. 2 4 .2 1 4

Spondias spp. I I I

Pollen 1

Total feeding samples* 3 4 5 2 7

Artibeus phaeotis

Ficus spp. 2 1 2 1 3

Cecropia spp. 1 5

Spondias spp. 8

Total feeding samples* 5 6 9 1 11

Vampyrodes caraccioloi

Ficus spp. 1 2 4 3 1

Pollen 2

Total feeding samples* 3 3 4 3 2

*Includes genera of lesser importance not shown here.


months. Cecropia exim-ia is an important food item from July to September, as is Spondias radlkoferi in November to January.

Feeding niche breadths based on food species by frequency of occurrence in the diet are presented in Table 9. Large niche breadth values represent food generalists and small values food specialists. A. phaeotis stands alone at the generalist extreme of this index. A. jamacensis has an intermediate position between the generalist and the extreme specialists. The remaining six species are bunched as extreme specialists. Hereafter, all members of this guild will be referred to as 'fig specialists" except for the feeding generalist, A.phaeotis.

Niche overlap in food species is compared in Table 10. The highest values of overlap in canopy frugivores occur between species most similar in size (diagonal left edge of Table 10). A. phaeotis overlaps little with all the fig specialists, except for V. helleri, which is similar in size. The high values of overlap between many of the fig specialist species indicate that some mechanism other than selection of food species must be operable to reduce behavioral interference and/or interspecific competition for food in this guild.

Several types of evidence strongly suggest that food is a limiting factor for fruit bats on Barro Colorado, at least during some parts of the year, The biomass of fruit and the number of species of fruiting trees fluctuate quite drastically on a seasonal basis (see Phenology section). During the late wet season fruit availability is low~ and an increased, proportion of captured fruit bats have empty stomachs (83% in Oct-Nov) as compared to times of fruit abundance (71% in Mar-Apr).


Table 9. Feeding niche breadths of canopy frugivores.

Bat Species Number of genera of Number of species of Niche Breadth*
known food plants known food plants (loge B)

V. pusilla 2 4 0.94

C. trinitatum 3 4 1.33

A. phaeotis 10 12 2.10

V. helleri 2 3 1.01

C. villosum I 3 1.01

V. caraccioloi 4 5 1.04

A. jamaicensis 9 16 1.61

A. lituratus 5 7 1.33

*Sample sizes for calculating niche breadths are as in Table 7.


Table 10. Feeding niche overlaps (CA ) among species of the
canopy frugivore guild

<< I> I>


rt 0

A. phaeotis .215** .354*and .615** .241** .465** .485** .518**

V. pusilla .968* .452 .796 .679 .272 .152

C. trinitatus .893* .743 .644 .209 .412

V. helleri .798* .886 .852 .452

C. villosum .727* .200 .310

V. caraccioloi .994* .962

A. jamaicensis .983*

A. lituratus

Denotes species most similar in body weight ** Denotes overlap with the feeding generalist


Also by the late wet season several species of fruit bats have temporarily moved out of the study areaand the remaining individuals and species spend a greater part of their nightly time budgets in foraging (see Species Diversity section). Even when fruit is very abundant in terms of total biomass, it is concentrated in a limited amount of space, the few trees fruiting at any moment, and may still be a limiting factor for population size.

Vertical stratification

Handley (1967) and Harrison (1962) demonstrated a vertical stratification of flight activity in Neotropical bat species, with most canopy frugivores preferring upper levels of the forest. On Barro Colorado, pusilla, A. phaeotis, C. villosum, V. caraccioli, and A. lituratus were captured with highly significant frequency in the nets and traps set above 3 m (Table 11). V. helleri and C. trinitatum also were captured most frequently in subcanopy-canopy levels, but sample sizes for these species are small, and frequency differences are not statistically significant. A. jamaicensis is the only species of the guild to show a significant preference for activity at the groundstory level, yet 42% of the captures of even this species were in the upper levels of the forest. Though most of its food items grow in the upper levels of the forest, A. jamaicensis may fly close to the ground to avoid predators. On the other hand this behavior may be an artifact of the human management of the forest, with this species opportunistically finding it more efficient to fly along cleared trails than to repeatedly detect and avoid vegetation at higher levels.


Table 11. Vertical stratification of canopy frugivore species on
Barro Colorado. Statistical significance indicates preference for one of the two vertical strata.

Bat Species No. of bats captured at No. of bats captured at
ground level, 0 to 3 mm subcanopy levels, 3 to 12 mm

V. pusilla 5 25**

C. trinitatum 2 4

A. phaeotis 36 56*

V. helleri 3 6

C. villosum 4 24**

V. caraccioloi 4 30**

A. jamaicensis 467** 326

A. lituratus 23 66**

Significant by Chi Square Test (P <.05).

** Highly significant by Chi Square Test (P < .01).

Yates Correction for Continuity is used on all tests of samples
with N < 200 (Sokall and Roh Ifl969).


Habitat selection

Comparison of netting samples from the young open forest of Buena Vista and the closed canopy forest and creek habitats of Barro Colorado provide a measure of species preferences for three habitats (Fig. 4). As a group the fig specialists are much more common in the closed forest and creeks lined by closed forest than in the shrubby open forest where few mature trees of their preferred food species are found. A. phaeois. and A. jamaicensis are common to very abundant in all three habitats, as would be expected from their more generalized food requirements. None of the extreme fig specialists are common on Buena Vista Peninsula. Fedn behavior

Canopy frugivores usually carry fruits by mouth from fruiting trees to night feeding roosts (Goodwin and Greenhall, 1961; Morrison, 1975). On BCI Morrison found that the night feeding roosts of A. -jamaicensis are frequently several hundred meters away from the fruiting trees where they.are picked. Only when feeding on the large fruits of Dipteryx panamensis did Artibeus feed on fruiting trees. All four most common canopy frugivore species were observed to carry whole or partially eaten fruits in flight. These animals presumably were transporting food items to a night feeding roost for consumption. Whether the less common species in the guild use night feeding roosts is unknown.

The fruits carried in flight by fruit bats vary in weight from less than 1 g to about 20 g. Most bats carry fruits that weigh 20 to 40% of their own body weight. Table 12,lists the range in weights of some fruits eaten by stenodermine bats. There is considerable variation in the weights among and within species for these fruits (even in fruits from the same individual tree).





CL _0 (n

4-J CD

4- (U

c (A

c u

-0 C)cu V) a) 0 > 4-1 4-J L.

_:T x

CD 4-J U- 0

0 -C


CN- .CN.

0 co % t

C~) u LUJ

r*( CN


.. .. .....



('4 >.. . . . .


. .. .... . . . .

. ... . . .
. . . . .
. . . . 0. . .

'0 CIO

cvfl 101 3Jfl


Table 12. Wet weights in grams of some fruits eaten by bats on
Barro Colorado Island that were collected beneath
fruiting trees.

Plant Species Fruit Weights
Range Mean S. D. N

Ficus insipida 7.1-11.4 9.1 1.5 10

Ficus obtusifolia 14.2-19.0 17.0 2.5 3

Ficus yoponensis 1.5- 5.6 3.1 1.1 12

Anacardium excelsum 4.2- 6.2 5.1 0.7 7

Calophyllum longifolium 9.3-17.7 14.7 3.3 7

Dipteryx panamensis 18.0-26.3 22.3 3.6 5

Spondias radlkoferi 8.6-13.0 o10.6 1.4 9

Quararibea asterolepis 4.9- 6.3 5.45 0.6 4

Astrocaryum standleyanum 17.0-20.5 18.8 1.8 4

Piper cordulatum 0.5- 2.0 1.2 0.5 15

Food particle size plays an important role in the partitioning of food resources among similar species in many types of animals (e.g. Diamond, 1973; Brown and Lieberman, 1973) and may be particularly important for fruit bats because of the behavior of carrying fruits in flight to feeding roasts. According to the theory of optimal foraging strategy (Schoener, 1969), each bat should attempt to maximize the amount of food it harvests per unit of time and thus select the largest food particles it can efficiently find and handle. The weight that a bat can carry in flight without seriously impeding manuverability probably sets the upper limit on food particle size for these animals.

Figure 5 shows that there is a highly significant correlation (by F distribution, P <.01) of fruit weight with bat weight for fruits carried into nets by the three largest species of bats in the canopy frugivore guild. Most of the points in this figure represent Ficus insipida fruits, the most important food species in the diet of all three bat species. Thus even though these three bats have high overlap in food species (Table 10), they are able to specialize on food particle sizes proportional to their body weights. The smaller canopy frugivore species probably do the same thing, but no data are available.

Each species in the canopy frugivore guild has a distinct cycle of flight activity. The three largest species, V. caraccioloi, A. Jamaicensis, and A. lituratus, each have their greatest peaks in activity at different times of the night (Fig. 6). Since'all three of these species feed largely in the same individual trees in the course of the night, the offsetting cycles of activity probably function to minimize interspecific aggression from crowding at the resource trees, especially when resources are concentrated in a few trees per night. Reduced




V) 0 4J

(t) 4J

0 4-J


LA 4-J .Ln

4- u 4
0 E c q


O)o 0) s- 0 L- U m u a) m c L

cu Ll- c



Ln 10

C%4 a

co co LO
04 o a LO



< co

c w U

E C3
0 zi
CL -n
0 0


IH013/v\ iim-A

4-) M 4J






-C a) 'j..)

u 0


0-0 LL- u

Q) U)

49 co
cn CN. co

.2 .0 46 o u a co u 0 0



LU tn





crowding at resource trees presumably is of importance in permitting more efficient feeding and in making these bats less obvious to the many kinds of arboreal and aerial predators that eat bats (Humphrey and Bonaccorso, 1975).

A similar pattern of offsetting major activity peaks should be expected in the small canopy frugivores, all of which feed heavily on Ficus yoponensis and F. popenoaei. Figure 7 shows that V. pusilla is most active in the first two hours after sunset, and C. villosum is most active later in the night. Paucity of data prevents comparison of the other small fig specialists.

A. phaeotis, the feeding generalist, has a much more even distribution of activity through the night than any other species (Fig. 7). Many of the fruits eaten by A. phaeotis are not eaten by other stenodermine bats and it need not compromise its activity cycle to avoid crowded resource trees.

Groundstory Frugivore Guild

Body size

Two species in the subfamily Carollinae of the Phyllostomatidae

constitute the groundstory frugivore guild on Barro Colorado. They are Carollia castanea and C. perspicillata. These have mean body weights of 12.4 and 17.9 g, thus differing in body weight by a factor of 1.44 (Table 13).

A few individuals of a third species of the genus Carollia,

C. subrufa, were captured and banded by R. K. LaVal in 1972 on Barro Colorado (pers. comm.). In 1973 and 1974 I recaptured some of LaVal's banded C. perspicillata and C. castanea, but I have not encountered any of the C. subrufa he marked. It is difficult to distinguish


4J r





(A fu


CT) =3
L- Ln
__ Q)
4 -
u W Q) L- CL

4-) 0_ a) ru su 0


U 4

:3 CL cr 0






Z 0 -p.


9 -~ Go

CL so .*9*** ,@,- ?


qo~~~ IN **.3dC


Table 13. Weights of groundstory frugivore bats on Barro Colorado

Bat species Mean Standard Sample Weight of larger species
) deviation size divided by weight of
smaller species in the
pair compared

C. castanea 12.4 1.8 30 -C. perspicillata 17.9 1.8 30 1.44


C. perspicillata and C. subrufa by field characters,and it is possible that I lumped a few individuals of C. subrufa with C. perspicillata because I was not aware that C. subrufa was present on BCI. I believe C. subrufa is very rare on Barro Colorado, and lumping a few of them with C. perspicillata would influence the data on this latter species to a very minor extent.

Food selection

C. castanea and C. perspicillata are food generalists in that they eat a fairly even distribution of a large number of kinds of fruits and have large niche breadth values (Tables 14, 15, and 16). Though no one food species dominates their diet in any one season or over a long portion of the year, eleven species of the shrubby plant genus Piper (Piperaceae) constitute the bulk of the diet of C. castanea and nearly one-third of the diet of C. perspicillata. Ten species of pipers were identified in the fecal samples from C. castanea and nine species from C. perspicillata. At least one species of piper is available with mature fruit in every month of the year on Barro Colorado (see Table 2). C. castanea eats pipers all year long, but no pipers were evident in the diet of C. perspicillata from mid-September through mid-November. C. perspicillata appears to feed exclusively on subcanopy and canopy fruits in the late wet season. Particularly important are Solanum hayseii, Quararibea asterolepis and Cecropia exima. Other fruiting trees are important food species along with pipers at other seasons. Though fruiting shrubs dominate the diet of C. castanea, fruiting trees are somewhat more important than shrubs for C. perspicillata.

In addition to the fecal samples from captured animals, food habits data for C. perspicillata were obtained by monitoring droppings below


Table 14. Food species in the diet of C. castanea as determined from frequency of occurrence of seeds in fecal samples. Sampling periods begin at mid-month.

Plant species Jan-Mar Mar-May May-Jul Jul-Sep Sep-Nov Nov-Jan Total


Piper aequale 3 3 3 6 15

P. cordulatum 8 1 9

P. reticulatum 1 3 1 5

P. marginatum 3 1 1 1 6

P. carrilloanum 1 1 2

Piper 109 1 1 2 4

Piper 114 1 1

Piper 120 1 1 2

Piper 122 3 3

Piper 150 2 2


Carludovica palmata 1 1

Solanum hayseii 2 2

Markea panamensis 9 1 1 11

Vismia 1 2 1 3

Brosimum bernadettae 1 1

Dipteryx panemensis I I

Aechmeia tillandsoides 1 1


Unknown 104 1 1

Unknown 123 1 1


Table 15. Food species in the diet of C. perspicillata as determined from frequency of occurrence of seeds in fecal samples. Sampling periods begin at mid-month.

Plant species Jan-Mar Mar-May May-Jul Jul-Sep Sep-Nov Nov-Jan Total


Piper aequale 1 I

P. cordulatum 10 1 II

P. reticulatum 7 7

P. marginatum I1 1 3

Piper l09 4 1 1 6

Piper 114 1 1 2

Piper 116 1 1

Piper 120 2 1 3

Piper 150 2 2


Carludovica palmata 1 1

Solanum hayseii 1 6 1 1 9

Markea panamensis 1 4 5

Vismia 1 4 5 9

Vismia 2 1 1 2

Cecropia exima 1 1 1 3

Brosimum bernadettae 3 3

Quararibea asterolepis 6 6

Dipteryx panamensis 7 1 8

Cassia undulata 3 3


Table 15, continued,

Plant species Jan-Mar Mar-May May-Jul Jul-Sep Sep-Nov Nov-Jan Total

Unknown 101 1 1

Unknown 103 1 1

Unknown 104 1 1

Unknown 125 4 4

Unknown 127 2 2

Insects 5 1 6




-C 0

0 Ln U I


0 E
0 fo

4- 40
Ln 4-J
U) 4 a)
a) c Ln .- m i;
> Ln -o 00 -zr 4-)
0) 0
:3 0 4C

C: u
M Ln

=3 u
0 0 Ln
t JI-i
cm cc rL
4- Q) 0
0 C CL
(D W
Ln 0)-o u
0 00 0
4- 0 0 4

S- C

_n 0 u
E c 0

"a cr

LL 4-P V)
(n L. Ln m 0 0
u CL

Ln m
0 0
4-J L- Lm ro m M
F- co L.) u -11,


two day roosts of this species. Both roosts were in hollow Anacardium excelsum trees. With the exception of A. excelsum all common food items identified from the day roost droppings appeared as important food items in the fecal samples from captured bats during the same bimonthly periods (Table 17).

Why did not A. excelsum ever show up in the fecal material from captured bats? Probably A. excelsum is the only tree species that commonly serves as both a day roost and an important food resource for C. perspicillata. These bats need only fly to the canopy of the roost tree, pick a fruit, and carry it back inside the roost to eat it. The bats would usually digest and excrete the fruit before flying away from the roost to forage for other fruits; thus little chance would exist for this pulp to show up in netted animals. The Carollia colonies in A. excelsum trees consisted only of 6 to 8 bats and each colony probably had access to more A. excelsum fruits when in season than they could eat.

Anacardium excelsum is the only fruit I know to be eaten by bats

on Barro Colorado that is not dispersed. Not only are the fruits carried within the hollow parent tree, but it is the single, large seed and not the fruit' pulp that is eaten. However, a few seeds probably are dispersed when dropped by mistake.

Overlap between the diets of the two Carollia is moderate in terms of food species. A CX value of 0.584 is obtained from lumping the food habits data from fecal samples from the entire year. Food overlap was very high in May-July sampling, CA = 0.798. This latter value, as well as the annual value of overlap, would be considerably smaller if it were possible to correct for Anacardium eaten in roost trees by C. perspicillata. Even though roosts were not monitored, it is unlikely that


Table 17. Frequency of occurrence of food species in the diet of C. perspicillata as determined from fruit droppings and seeds below day roosts. Sampling periods begin at mid-month.

Plant species Jan-Mar Mar-May May-Jul Jul-Sep Sep-Nov Nov-Jan*Total

Anacardium excelsum 5 83 39 127

Piper cordulatum 39 35 9 83

P. reticulatum 6 6

Piper 109 6 6

Solanum hayseii 1 1 2

Vismia 1 2 2 4

Quararibea asterolepsis 5 15 20

Cassia undulata 5 1 6

Unknown 155 12 12

Unknown R-l 3 3

* No data.


C. castanea eats much of this fruit, as it is larger than all other important fruits in the diet of C. castanea. Habitat selection

Of the three habitats sampled, the Carollinae were most common in the second growth forest and least common in the mature forest, as are their most important food plants. C. castanea accounted for 21.7% of all bats captured in the second growth forest, 2.7% of the bats in the creeks, and 1.4% of the bats in the mature forest (Fig. 6). C. perspicillata constituted 15.8%, 16.0%, and 5.4% of the bat individuals captured in those habitats. Whereas many species of pipers grew abundantly in the sunlight of the open canopy second growth and along the creeks (though less so along creeks), only one species, P. cordulatum, was abundant in the shade of the mature forest. Vertical stratification

C. castanea and C. perspicillata were both captured more frequently at ground level than at upper levels of the forest (C. castanea = 20 ground level, 14 upper levels; C. perspicillata = 50 ground level, 34 upper levels), but the difference was not statistically significant. Both species feed on plants of ground and canopy levels. Known groundstory fruits make up 78.4% of the diet of C. castanea and 38% of the diet of C. perspicillata. During seasons when C. perspicillata is feeding mostly on canopy fruits, it also is captured more frequently in high nets and traps.

Feeding behavior

Carollia castanea and C. perspicillata both have been captured

carrying fruits in the mouth and presumably use night feeding roosts as do canopy frugivores. Some fruits are carried back to the day roost


for consumption as already discussed. The use of day roosts as feeding places by C. perspicillata is mainly a phenomenon related to feeding on one fruit, Anacardium excelsum, as is evident from the dominance of this fruit below day roosts and the decrease of dropped fruits and seeds when A. excelsum is not in fruit (Table 18). It is likely that temporary night feeding roosts are used by these bats to avoid making the day roosts conspicuous to predators and to reduce flight distances between foraging forays.

The flight activity of Carollia through the night is presented in

Figure 8. Both species show major peaks of flight activity in the first hour of darkness. This is much earlier than the start of most canopy frugivores' flight activity and is probably due to the groundstory becoming dark about an hour before the canopy level of the forest. Cycles of flight activity in groundstory frugivores are bimodal or trimodal as are those of canopy frugivores. Two or three such bouts of diel feeding activity also have been observed in fruit bats by Brown (1968) and LaVal (1970) and seem characteristic of bats that feed on foodstuffs that are not efficiently assimilated. Scavenging Frugivore Guild

Body size

Centurio senex (Stenoderminae, Phyllostomatidae), the wrinkle-faced bat, is the sole member of the scavenging frugivore guild. A lactating female weighed 22 g and a pregnant female weighed 27 g. No other weights are available from Barro Colorado for this species, nor are there any useful data on vertical stratification or habitat selection.




0 L

cu 4)


_rCD :3

4-J Q) 4)

a- CL (a 0
4- L
0 0


W >LL- o

00 c
Q) 0 %- S

0 U- :


64 CN







0 LO 0


Food species selection and feeding behavior

From several morphological features, particularly the small teeth and narrow esophagus, Paradiso (1967) concluded that C. senex probably feeds on a "soft fruit or fluid diet". The small teeth, narrow esophagus, and lack of facial hair (like vultures) on this bat are suggestive of its possibly feeding on over-ripe or decaying fruit. Hence I tentatively designate it a "scavenging frugivore". The amount of rotting fruit on the forest floor is incredibly large on Barro Colorado and potentially could provide an abundant food resource for such a bat. Goodwin and Greenhall (1961) mentioned finding fruit pulp in stomachs of C. senex from Trinidad. Of the individuals that I captured on Barro Colorado one defecated an unidentifiable fruit pulp and the other carried a fresh Spondias radlkoferi fruit in its mouth. At this time it can neither be confirmed nor disproved that Centurio is a scavenging frugivore. Though it is very similar to Chiroderma villosum in body size I haye no doubt that this anatomically unusual bat is ecologically quite different from any other frugivorous species on Barro Colorado with respect to food habits.

Nectar-Pollen-Fruit-Insect Omnivore Guild Body size

The nectar-pollen-fruit-insect omnivore guild (hereafter referred to as the omnivore guild) contains three species, all of the family Phyllostomatidae -- Glossophaga soricina (Glossophaginae), Phyllostomus discolor (Phyllostominae), and Phylloderma stenops (Phyllostominae). The mean body weight for P. discolor on Barro Colorado is 4.36 times larger than that of G. soricina (Table 18), a larger difference than is found between species adjacent in size in any other guild. The large


Table 18. Weights in grams of omnivore bats on Barro Colorado.

Bat species Mean Standard Sample Wgt f Wgt *
(x) Deviation size ig sm
(S. D.) (N)

G. soricina 9.8 1.0 9 -P. discolor 42.8 3.9 27 4.36

male P. discolor 44.6 3.6 17

female P. discolor 39.7 2.1 10 -P. stenops 61.0 -- 1 1.42

'Weight of larger species divided by weight of smaller species in the pair compared.


gap in body weight between G. soricina and P. discolor exists because of the recent extirpation of a bat species belonging to this guild. As recently as the early 1950's, Lonchophylla robusta (Glossophaginae), was alive on Barro Colorado (Hall and Jackson, 1953). This species eats nectar, pollen, fruit, and insects (Howell and Burch, 1974). L. robusta from Costa Rica weigh about 17 g~and if this species were still present on Barro Colorado the ratios between body weights of the four omnivore guild members would be 1.4, 2.6, 1.4. The large ratio between L. robusta and P. discolor actually would have been somewhat less than 2.6 because of the sexual dimorphism in body weights of P. discolor. The dimorphism in body weights between male and female P. discolor is very slight (Table 18) but significant (P <.05, Student's t-test). Food selection

Nectar and pollen are consumed by guild members almost exclusively in the dry season, as large flowers suitable for bat use are in bloom only then (see Phenology). The few data available suggest that during the wet season fruit and insects become dietary staples (Table 19) That insects were not present in the food samples from G. soricina on Barro Colorado is probably because of poor sample size and the fact that this species moved out of the study area during the wet season. Nothing beyond the observations of Jeanne (1970) of P. stenops eating social wasp larvae and my two observations of. fruit eating is known about the diet of this bat.

Phyllostomus discolor is neither an extreme specialist nor generalist in terms of food species (niche breadth = 1.65). Several types of flowers are visited for pollen and nectar in the dry season. And in addition to insects, several types of fruits are eaten in the w~t season. The


Table 19. Seasonal use of pollen and fruit by the omnivore guild on Barro Colorado.

Food species No. of dry season No. of wet season
samples samples

P. discolor, N = 23 Pollen:

Ochroma lagopus 6

Pseudobombax septenatum 6

Unknown 202 Fruit'

Cecropia exima 2

Unknown 124 3

Insects: 3

G. soricina, N = 6


Ochroma lagopus 3

Unknown 201 I


Cecropia exima 1

Piper 109 I

P. stenops, N = 2


Unknown 110

Unknown 151 1


available data are too limited to consider niche breadth values for P. stenops and G. soricina, or to calculate niche overlaps between guild members.

Vertical stratification

All of the flowers and fruits eaten by P. discolor and 83% of those eaten by G. soricina in this study area grow in the subcanopy and canopy of the forest. Both species were captured most frequently in the upper levels of the forest, 3 of 4 for Glossophaga and 40 of 54 for Phyllostomus. For P. discolor preference for flying above groundstory shrubs is highly significant (P <.Ol, Chi Square Test). Habitat selection

Phyllostomus discolor was common in the mature forest and second growth but uncommon over creeks. Some of the important tree species producing flowers and fruits eaten by Phyllostomus are common only in second growth (e.g. Ochroma); others are common only in mature forest (e.g. Pseudobombax); and still others are common in both habitats (e.g., Cecropia).

Feeding behavior

During the dry season bats are frequently captured with pollen

heavily dusted over the anterior parts of the body. It is likely that these animals visit a number of flowers in succession, consuming nectar and performing pollination services at each flower, and then later perch to ingest pollen by grooming it from the fur and skin.

None of the bats in this guild were captured carrying fruit in the mouth,and it is not known whether they use night feeding roosts.

Sixty-nine percent of all P. discolor captured in all-night samples were taken within two hours of sunset. Such a strong unimodal pattern


of flight activity (Fig. 9) also is reported by LaVal (1970) and suggested by Heithaus et al. (1974) for this species in Costa Rica. My data are insufficient for discussing the flight activity cycle of Glossophaga; however, LaVal (1970) reports a strong peak in activity at dusk and in the first hour of darkness just before the peak in P. discolor activity. Sanguivore Guild

ky size

Of the three extant vampire species, only Desmodus rotundus, the common vampire, occurs on Barro Colorado Island and in the surrounding vicinity. The pre-meal mean body weight of D. rotundus is 33.5 g. Food selection

Wild vampires feed only on the blood of homoiothermic vertebrates (McNab, 1973). While vampire feeding behavior and prey selection is well documented in agricultural areas where domestic livestock are the chief food source (Turner, 1975), nothing is known of the prey species of vampires in remote areas where only wild animals are potential hosts. Vertical, stratification

Where domestic animals are the source of food, vampires fly almost exclusively within 3 m of ground level (Bonaccorso, unpublished data). It is possible that vampires more commonly fly in the canopy level in isolated forests where arboreal species (e.g., monkeys and birds) may be important sources of blood meals for these bats. On Barro Colorado Island two vampires were captured in subcanopy nets and one in a ground net.

Habitat selection

Vampires were clearly more abundant on Buena Vista Peninsula than on Barro Colorado Island. Desmodus was the fourth most abundant species in


u (a
4-) C







0 I



u 0

4) 0 :3 U
tr (n
Q) L

E m 0
Ln 0






z 00

z 0 D
0 u

E 0
-&- CV


LO 0 tn 0 C)
cle) CII) CN CN



the Buena Vista samples (7.9% of the total captures)whereas on Barro Colorado Desmodus was one of the least commonly captured species (0.2% of the total captures, and see Fig. 4). Horses, cattle, pigs, and fowl of the scattered farms in the Buena Vista-Frijoles area provide a dependable and abundant food source that "los vampiros" constantly parasitize (Fulo Sanchez, pers. comm.). Gleaning Carnivore Guild

Bod size

The largest feeding guild within the bat fauna of Barro Colorado,

the gleaning carnivore guild, is formed by nine species of phyllostomine bats. This guild also presents the largest range in body size within any of the BCI feeding guilds (Table 20). Micronycteris megalotis, one of the smallest bats on Barro Colorado, has a mean body weight of 6.3 g., while Vampyrum spectrum, the largest species on the island, weighs about 120 g.

The increment in body weight between successively larger species is more irregular within the gleaning carnivore guild than in any other guild on the island (see Fig. 3). Two species have mean body weights of close 'to 15 g., and three species have mean body weights of 31 to 36 g. On the other hand all three species of the genus Micronycteris differ from the next smaller species by a factor of 1.5. Food species selection

Little precise information can be offered at this time concerning the prey species eaten by members of this guild. Excepting Vampyrum spectrum, guild members feed predominately on insects most of the year. I have a large collection of fecal samples from these species but as yet have found no one able to identify the minutely fragmented insect exoskeletons that constitute these samples.


Table 20. Weights in grams of gleaning carnivore bat species on
Barro Colorado.

Bat species Mean Standard Sample Wgt, / Wgt
() Deviation size Ig sm
(S. D.) (N)

M. megalotis 6.3 0.6 6 -M. brachyotis 9.7 1.1 3 1.53

M. crenulatum 14.7 0.7 Y12 1.52

M. hirsuta 15.2 1.2 7 1.03 1.57

T. cirrhosus 31.0 3.8 13 2.09

T. sylvicola 32.6 3.6 10 1.04

T. bidens 35.6 2.3 7 1.09

P. hastatus 91.2 4.0 7 2.56

V. spectrum 120.0 -- I 1.31

Weight of larger species divided by weight of smaller species in the pair compared.


On the basis of characteristics of echolocation, Novick (1971)

hypothesized that large-eared insect- and vertebrate-eating bats, such as are found in the gleaning carnivore guild, are adapted to distinguish and capture prey items resting on foliage. Ross (1967) and Wilson (1971b) have shown in food habit studies that three such large-eared species, Antrozous pallidus, Macrotus waterhousii and Micronycteris hirsuta, do feed primarily on large insects that spend much of their time perching on vegetation or on the ground. Gardner's (1975) review of the scattered information on food habits of the bats in this guild further confirms that food items such as lizards and large insects probably are gleaned from foliage.

Micronycteris. Wilson (1971b) reported that large roaches, Orthoptera, and scarabeid beetles are the most important items in the diet of M. hirsuta on Orchid Island, a small island adjacent to Barro Colorado. During the dry season, fruit became an important component of the diet of this species as indicated by droppings below the study roost. My food samples show that M. megalotis and M. brachyotis also switch in part to fruit diets in the dry season. M. brachyotis also eats nectar and pollen. An individual captured in mid-December was thoroughly dusted with the pollen of a balsa tree (Ochroma lagopus).

Tonatia. A very large male cicada (Fidicina mannifera) weighing 2.5 g was carried into a net in the mouth of a Tonatia bidens in July of 1974. The prothorax of the cicada had been crushed by the bat's teeth and the cicada was dead when removed from the net. Because this event occurred in the mating season of the cicadas, amongst the loud nocturnal chorusing of the males, it posed the question of how Tonatia locates such insect prey. Do bats locate such prey items via echolocation or sounds produced by the insects?


Two T. bidens, one male and one female, were released in a large outdoor flight cage, one at a time, on BCI. Both individuals were immediately attracted to the sounds of calling male cicadas that I held by forceps inside the cage. The cicadas were plucked from the forceps by the flying bats and eaten with gusto at a perch. Female cicadas held so that their wings could not move in the forceps were ignored by these bats; however, when the wings were allowed to flap nosily, the bats again were attracted to the cicadas and ate them. During later experiments large nocturnal grasshoppers, katydids, beetles, and moths (species not identified) placed on the ins ide cage screening were "gleaned'' from the screening and eaten by these bats. It is obvious that Tonatia bidens was able to locate cicadas from sounds produced by the cicadas, but whether other large foliage-clinging insects, many of which produce ultrasound, were echolocated or detected from insectproduced sounds remains an interesting question for future research.

Only insect fragments were found in the fecal samples of T. bidens and T. sylvicola from BC!.

Phyllostomus. Insects and fruits were found in the fecal samples of Phyl'ldstomus hastatus. It also has been reported by several authors to eat birds and rodents (Gardner, 1975).

Vampyrum. Vampyrum spectrum, the false vampire bat, is the largest New World bat. It appears to feed primarily on birds and small mammals, though investigators report some fruit and insects in its diet (Gardner, 1975). A hollow tree roost monitored by J. Bradbury (pers. comm.) in Costa Rica had a steady flow of feathers from parrots, trogons, cuckoos, anis, and many other birds appearing at its base. D. J. Howell and I kept a Vampyrum alive in captivity for three weeks on a diet of small bats


and birds ranging in size to large doves (Howell and Burch, 1974). When released in a large room with small fruit bats (10-20 g) the false vampire would fly up behind its flying victim and slap it into its jaws with a wingtip. One morning at sunrise on Barro Colorado, a false vampire circled about an Artibeus jamaicensis I was untangling from a net. The Vampyrum was apparently attracted by the alarm calls of the fruit bat and circled for over a minute before leaving.

Mimon. Fecal samples of Mimon crenulatum from BCI appear to contain only insect chitin.

Trachops. Fecal samples of Trachops cirrhosus from BCI appear to contain only insect chitin, but this species is reported to eat lizards such as anoles and geckos that are gleaned from vegetation, as well as some fruit (Gardner 1975; Howell and Burch, 1974). Vertical stratification

Mimon crenulatum and Tonatia sylvicola show a significant preference for flight in the groundstory level of the forestand the small samples for Micronycteris brachyotis and Tonatia bidens are just barely below significance levels for showing a preference for flight activity in the subcanopy-canopy level (Table 21:). Thus the two similarly sized species of Tonatia appear to forage in separate vertical strata.

Should future research increase sample sizes on vertical stratification it would not surprise me if none of the three species of the genus Micronycteris show a strong preference for one particular vertical stratum. All three species are very different in body size and probably specialize on mutually exclusive sizes of prey items.

Though the data are very limited, the two very large species in this guild, Phyllostomus hastatus and Vampyrum spectrum were captured or seen


Table 21. Vertical stratification of gleaning carnivore species on Barro Colorado.

Bat species No. of bats captured No. of bats captured
at ground level, at subcanopy levels,
I to 3 mm 3 to 12 mm

M. megalotis 1 3

M. brachyotis 6 12

M. hirsuta 6 3

M. crenulatum 9* I

T. cirrhosus 5 2

T. sylvicola 18** 3

T. bidens 2 8

P. hastatus 0 4

V. spectrum + 0 3

Significant by Chi Square Test (P < .05).

** Highly significant by Chi Square Test (P <.01). + Based on two net captures in Costa Rica and one visual sighting on on Barro Colorado

Yates Correction for Continuity is used on all Chi Square Tests (Sokall
and Rohlf, 1969).


flying only in the subcanopy-canopy levels of the forest. The groundstory (0 to 3 m) on Barro Colorado has the most dense foliage cover of any of the vertical strata of the BCI forest (E. Leigh, unpubl. data). P. hastatus and V. spectrum may be too large to maneuver well through the thick groundstory vegetation.

Habitat selection

Several gleaning carnivore species prefer either creek or forest habitats on Barro Colorado to the exclusion or near exclusion of the other habitat. Micronycteris brachyotis and Tonatia sylvicola are both common species in the forest station samples but were totally absent from creek samples (Fig. 6). Trachops cirrhosus represents 4% of all individual bats sampled at creek stations (8th most abundant species in creek samples) but only 0.5% of the individuals sampled at forest stations (17th most abundant species in forest stations). All other species in the guild are approximately equally abundant in creek and forest samples. Comparisons with Buena Vista second growth samples are not made because most species in this guild were under-represented at Buena Vista from lack of harp-trapping.

So far I show that most bat species within a given feeding guild partition food resources on the basis of food particle size. However, some ofthe gleaning carnivores seem to use additional mechanisms to partition food resources between similarly sized species.

The existing data make it appear that a spatial mechanism, specialization in foraging microhabitat, permits Trachops cirrhosus, Tonatia sylvicola, and Tonatia bidens, all similar sized gleaning carnivores, to partition food resources within the same macrohabitat. I suggest that T. cirrhosus specializes on prey items it can glean from low


foliage along creeks, that T. sylvicola specializes on prey items it can glean from groundstory foliage in the forest, and that T. bidens specializes on prey items it can glean from trees in the forest and along creeks or from the ground. Future research likely will show that all three species eat invertebrates and vertebrates weighing between 2 and 15 g, including lizards, frogs, and large insects. Feeding behavior

The gleaning carnivores eat rather large prey items relative to their body weight. It probably is common for them to carry prey to a feeding roost or day roost for consumption (Wilson, 1971b; Bradbury, pers. comm.).

Data on activity cycles are too scant for meaningful analysis.

M. brachyotis, M. crenulatum, T. sylvicola, and T. cirrhosus appear to have a major peak of activity in the first two hours after sunset. Slow-flying Hawking Insectivore Guild

Eight species belonging to three families constitute the guild of slow-flying hawking insectivores. Four species belong to the Emballonuridae, two to the Vespertilionidae, and two to the Mormoopidae. An additional species, Thyroptera tricolor of the Thyropteridae, is known in recent years only from a single 1973 sighting on BCI. T. tricolor perhaps should be included in this guild if a population still exists on BCI, but the species is probably near extirpation on the island because of plant succession that has resulted in the disappearance of most large-leafed groundstory plants (e.g. Musa and Callithea) used as roosts (Findley and Wilson, 1974).


Body size

Mean body weights of the species in this guild range from 4.2 to 22.6 g (Table 22). Wing morphology and flight behavior (Bonaccorso, unpubl. data) suggest that species within the same family are most similar in foraging behavior. Thus, species are grouped in subguilds by families. The two mormoopids differ in mean body weight by a factor of 1.37, a figure that suggests these two species may divide food resources solely on the basis of particle size. The two vespertilionids, on the other hand, are very similar in body size. Analysis of the relationships among body weights of the four emballonurids is complicated by small sample sizes and sexual dimorphism. The mean body weights of males of the four species differ by a factor of 1.25 to 1.47. Food selection

All species of this guild appear to feed on fairly small flying

insects. Prey items are eaten on the wing rather than carried to feeding roosts. Some emballonurids hover around tree foliage and probably feed to some extent on insects attracted to host trees. One BCI fecal sample from Pteronotus parnellii examined by Terry Erwin contained leg parts of a small beetle of the family Alicuidae. All other samples await analysis.

Vertical stratification

Pteronotus parnellii almost exclusively restricts its flight to

within 3 m of the ground (Table 23). Myotis nigricans and S. bilineata apparently fly with nearly equal frequency in groundstory and subcanopy levels of the forest. Peropteryx kappleri is probably a specialist on insects of the subcanopy, as indicated by the capture of all four BCI individuals in high nets and numerous visual observations made by the author in Belize (unpubl. data).


Table 22. Weights in grams of slow-flying hawking insectivore bat species on Barro Colorado.

Bat species Mean Standard Sample Wgt / Wgt*
() deviation size lIg sm
(s. D.) (N)


S. leptura ** 4.2 -- 1 -C. maximilliani ** 5.2 -- 2 1.25

S. bilineata males 7.7 0.56 11 1.47

S. bilineata females 8.7 0.7 3 -P. kappleri ** 11.2 -- 2 1.46


R. tumida 4.2 -- 2 -M. nigricans 4.4 0.67 II 1.05

Mormoop i dae

P. suapurensis 16.5 -- 1 -P. parnelli 22.6 1.48 30 1.37

Weight of larger species divided by weight of smaller species in the pair compared.

** Males and females are probably dimorphic in body weight.


Table 23. Vertical stratification of slow-flying hawking insectivore species on Barro Colorado

No.of bats captured at No. of bats captured at Bat species ground level, 1 to 3 m subcanopy levels, 3 to 12 m

S. bilineata 7 4

C. maximilliani 2 0

P. kappleri 0 4

R. tumida 1 0

M. nigricans 3 3

P. parnellii 74** 1

** Highly significant by Chi Square Test (P <.01) with Yates.

Correction for Continuity (Sokall and Rohlf, 1969).


Habitat selection

P. parnellii is the second most abundant species in the forest station samples but is very rare in the creeks (Fig. 4). The only specimen of P. suapurensis was captured in the forest.

Myotis nigricans was captured only at forest stations, whereas Rhogeesa tumida was captured only at or near creeks. These two similar-sized species may differ in habitat requirements.

Visual observations of Saccopteryx bilineata were possible because

this species is crepuscular. Individuals repeatedly fly in circles around feeding territories in small clearings of the forest (e.g. treefalls) or over creeks (J. Bradbury, pers. comm.). I have frequently watched territory holders chase intruding conspecifics out of their territories emitting high pitched audible sounds as they fly.

Flih behavior

Pteronotus parnellii is one of the most commonly seen species on Barro Colorado, as it flies low along forest trails. Ultrasonic pulses picked up by a bat detector indicate that P. parnellii feeds as it flys back and forth in long loops along the forest trails and groundstory vegetation.

The flight activity of P. parnellii through the course of the night is bimodal with a major peak of activity occurring one to four hours after sunset and a minor peak occurring eight to ten hours after sunset (Fig. 10). Data on activity cycles of other species are limited, but P. parnellii appears to be the only species in the slow-flying hawking guild that has no strong peak of activity the first hour after sunset. Based on netting, visual observations, and ultrasonic detection, the emballonurids are active from an hour before sunset to an hour after sunset and again at a similar period about sunrise.

4-J 4-J
Cu 4-J '

0 V)




a)0 L.


Cu, S4




co V)


D -3...
0 C:


0 0 V)



Three patterns of reproduction occur in Neotropical bats: seasonal monestry, seasonal polyestry, and year-round polyestry (Fleming, 1973). Present information indicates that a single young is born per litter except in the genus Rhoeq9ssa in which the usual litter size is two (Humphrey and Bonaccorso, 1975).

Canopy Frugivore Guild

Canopy frugivores are seasonally polyestrous, with one birth peak at the end of the dry-to-wet transition and a second about the middle of the wet season (Figs. 11-13 and Table 25; Wilson, 1975). The first birth peak for all species coincides with the beginning of the first predictably steady rains of the year in late April and May, a time of fruit abundance. Large species such as A. jamaicensis and A. lituratus are pregnant (as detectable by palpation) by the first week of January. Small species like A. phaeotus are not in a similar stage of pregnancy until late January. Lactation then proceeds for one or two months during a period of food abundance. There is a postpartum estrousand females are well advanced in the second pregnancy of the year while lactation is still underway (Fleming, 1971).

The second birth and lactation peaks are less synchronized among species because of differences in gestation and lactation periods; the same is true to a lesser degree within species because of individual variation. For example, the second peak of lactation occurs in JulySeptember for A. jamaicensis and A. phaeotis, but not until September87


4-J V)

4-) L










a) 0

0 0 C)
%0 LO N

sgivmA -Linciv Ao IN 3:) N d

Ln Le)




E 4-J


0 S
CL a)



z C4 F


1 1,; I J44 0 I



C) c CN
cu 1 0
> z ol

Z) >

cl 0
-u 0) co
0+ L- u

LF]l F Ell

0 0 0 0 0 0 Co 0 0 0 0 0 C) (D 0
co 10 11 Cq ol 00 r" 10 LO Cl) C14

S31VW3A iinGv JO IN3D i3d






u :3

CL cu








co 10 LO CV> C14

s3ivw.lj iinav JO IN3Dd3d


tn 0
L_ L.
Q) 4-0
E- a) -0
:3 -C c

-C 0-0
-W 4- =1
4J C14 C14 CD C*4 C)
-0 ru U) ru
0 4J n
cu M 1 0 n CD 0 C3 0
M .- >
a) C 0
0- W :z m C) CD _T a
d) 0
-C "a L.W. L_ 4c 0
0 Q) > rn
E L- 0
.- cc (U
-0 -C C14 C) n 0 C)
4.J M
-C 4J


a) CL cli en Q
u -C W
=) Ln
_0_0 C) 0 C14 Q C) C) CD
0 Q)
S- 4.J =1
0 0- c: 00 CD C C) C) C) CD Q
-0 a) Q)
(a L_ EA
L. c 0)
0 0 1- r_ Q0
-0 CN C) C)
c: 0 n
0 ru c __3
0 C) Q m 0
co 0) L. ru
C14 m C) a C) C
C 4-J 0
0 (v (D
4J LLe) U =1
41 M T
C14 CD Lr\ C)
0) -a)I C14 0 C) CD
ro ro 0) (a
E c: U'\ C) CD Ln 0
4- 0 m E
L_ CL a)
C: CL E 0 U
U) U
0) a) u >1 L_ 00 C) CD C) 0 C14 N 0 -zr __t,
C: (U .0 (U
.- (U 1:
E E a cc I C a C) 0 0 0 C) C) a 0
.- W -- E c
4 J 4- ro fu
_0 c rr% C14 C14 C14 m
(D 4- 0) M > 0 41 04J Lu a)
:3 -0 (U
-0 E 1- -0 0 D CL 0) L_ C $CL 0 = Ln
0 4-J E 0 V) E

4-1 u (D Ln 4-J E 0 0 ro 4J _W
Q) u G Q) (D (a :_3 .3 :3 4-J 0 m CD Ul)
4.J a C: 4J V) 2 _0 0 4-J 0
c Ln Ln u ro 0 u x C) c =1 -C
C14 0) 4-J Q) 4.j m a) u :3 u c 0 1Ln c (D 0 0 c V) 41 cn m Q) L0 0 -0 ro =3 m Q) 0 W L_ Ls- E u OL 41 > u tn E -0 u
-0 CL E a) 4J
ru Q) 0 -C (a
U 4J m C-) (i u u 0-1 m I M: I :: I I