Ecological Monographs (1978) 48: pp. FORAGING MOVEMENTS OF A FRUGIVOROUS BAT, CAROLLIA PERSPICILLATA (PHYLLOSTOMATIDAEB)! E, RayMonb HerrHaus Department of Biological Sciences, Northwestern University Evanston, Mlinvis 60201 USA AND ‘THEODORE H. FLEMING Department of Biology, University of Missouri St. Louis, Missouri 63121 USA Abstract. ‘The foraging movements of a Neotropical frugivorous bat (Carollia perspicillara) were studied to better understand its impact on the plants it disperses and the factors influencing its foraging behavior. Radio telemetry was used during 2 wet seasons to monitor the bats: movements in a Costa Rican seasonal tropical forest. We monitored 24 individuals for periods of up to 19 nights. This sample size permitted a statistical analysis of the influence of bats” Weight. bat's sex, moonlight, und the distribution of resources on foraging patterns. Carollia perspicllata is a tefuging species: individuals dispersed from shared day roosts to feed at night, then returned after foraging. At night there were 3 major activity periods. but there was Some activity all night. When foraging, each individual went to 2 to 6 feeding areas that were used consistently through the observation period. Each night it flew back and forth between areas, Search- ing for new feeding areas was uncommon. Average flight distances for individuals were 1.6 kilometres between the day roost and feeding areas, 1-5 km among feeding ares, and 1.6 km to gather fruits within feeding areas, The density at feeding areas decreased with distance from the day roost, as predicted by refusing theory, ‘There was little overlap among different individuals’ feeding areas, but the largest amount of overlap was adjacent to 1 day roost. Bats from the 2 major day roosts shared feeding areas only at locations midway between the roosts. Flight patterns were influenced by bats’ weights, the patchiness of resources, and the amount of ‘moonlight. From these and other observations, we infer that both predation and energetic consider: ations affeet C. perspicillata foraging. In addition. this species appears to respond to long-term factors 28 Well as nigh-to-night conditions so that dynamic models will be necessary to describe its foraging. Key words Bats: Carolia perspicillata: Chiroptera: eoevolution: Costa Rica: energetics: for ‘axing: fragivorys neotropies: predation: radio telemetry: refuging; resource distribution. INTRODUCTION The foraging movements of animals are ecologically important for at least 2 reasons. First, these move- ments may affect the spatial distribution and genetic structure of plant populations through the dispersal of seeds and pollen (Covich 1974, Levin 1974). Second, foraging behavior represents an animal's response to conflicting ecological and evolutionary pressures, so it is of considerable theoretical interest (Hamilton and Watt 1970, Schoener 1971, Charnov 1976, Covich 1976, Wiens 1976, Norberg 1977, Pyke et al. 1977, and others) In thiy paper we report a study of foraging by the frugivorous bat Carollia perspicillata (Phyllostomati- due), This common neotropical species is an important dispersal agent for at least 25 plant species in Trinidad alone (Goodwin and Greenhall 1961), Many of these fruits contain small seeds that are ingested, then later defecated as the bat continues its foraging. The dis- " Manuscript received 20 July 1977: accepted 9 January 1978 tribution of these plants’ seeds is largely determined by the foraging movements of the bats. One coevo- lutionary question in this plant-animal interaction is, can plants respond to selection for different seed dis- tributions by “manipulating” bats” foraging? Hypo- thetically, this manipulation could be accomplished if bat foraging movements responded to modifications in the distribution of their food source. Many theoretical models of foraging are based on an energetic approach (see the previously cited re- views), but other approaches to understanding forag- ing emphasize factors such as predator-escape strat- egies (Buskirk 1976). These other factors would reduce the coevolutionary “control” of bats by plants, (Carotlia perspicillata is a good organism to use to test the predictions of theory based on selection for opti: mal foraging efficiency. Curollia cannot escape long, periods of scarce resources through torpor, nor do they have large stores of fat (McNab 1969). Conse- quently, their foraging behavior should reflect re- sponses to short-term environmental parameters, addition, the movements of individual bats can be128 closely monitored using radio telemetry (Williams and Williams 1970, Morrison 1975, J. W. Bradbury per~ sonal communication). The major objectives of this paper are: (1) to de- E, RAYMOND HEITHAUS AND THEODORE H. FLEMING Scribe the movements of I bat species in detail, (2) to) examine the effect of fruit distributions and predation on foraging patterns, and (3) to examine the utility o! current theory in explaining the foraging of Carollia perspicillata. These points are a first step in under- standing the degree of coevolution between bats and plants. Meriops This study was conducted in Parque Nacional Santa Rosa, Guanacaste Province, Costa Rica. The study area was in Premontane Moist Forest (Holdridge 1967). Here, mean monthly temperatures are similar through the year (near 27°C), but rainfall is highly sea- sonal. An average of 2200 millimetres falls annually, essentially all of it between mid-May and mid-Novem- ber, Our work was done from mid-June to mid-Sep- tember, in 1975 and 1976. During this period the leaf canopy was well developed and many plants were forming fruits (Frankie et al. 1974). Our study area was the area within 4 km of the park administrative area. This area has an elevational range of =270 to 320 metres, and includes both forest and grassland. Radio telemetry was used to locate bats. Eight bats were captured with a hand net inside their day roost. Sixteen bats were captured in mist nets set just outside the day roost exit or near plants producing ripe fruits. ‘After capturing and weighing a bat, we used Silastic® medical adhesive to glue an AVM-SMI transmitter to its back. These transmitters had frequencies between 150.85 and 151.15 MHz, and weighed ~2 g after ad- dition of a battery, antenna, water-repellent wax, and a coating of dental acrylic. Bats were restrained for 1 h before being released. We did not radio tag obvious- ly pregnant females or bats weighing <17 g. Our goal was to locate each radio-tagged individual at 6 or 9-min intervals. Bats were located by trian- gulation from 2 or more receiving stations that were either permanently established or “mobile.” To trian- gulate, each station obtained a directional reading us- ing an AVM Instruments Co. LA-I2 receiver, a direc- tional antenna, and a Brunton transit, Three antenna configurations were used: an 8-element “‘null-peak"* system, a 7-element or 4-element directional antennas. The less bulky antennas were used in dense forest or when the mobile stations were on foot or horseback. ‘Simultaneous directional readings at different stations were coordinated using stopwatches and walkie-talkie radios. The relative intensity of the radio signal and the degree of modulation of the signal was noted at, cach directional reading. When triangulation was not possible, as when bats flew behind hills or out of range of 1 station, at least 1 station attempted to monitor each individual. The accuracy of the system was checked by monitoring hand-carried radios and by Ecological Monographs Vol. a8, No.2 mapping transmitters that were dropped by bats and subsequently located. The accuracy was +2° for the null-peak system and +4° for the 7-clement antenna system To analyze the telemetry data, we located receiving- station positions using aerial photographs of the study ‘area and our maps of roads and forest trails. A bat's, locations during each tracking night were plotted as points on clear plastic overlays. A composite map for all nights was also drawn for each individual. From the composite map we identified feeding areas (FA) in which bats spent at least 30 min on any night or com- bination of nights. We then reconstructed a log of each individual's nightly locations. For each reading we identified the location of the bat and the distance traveled to reach. that point from the preceding location. After major feeding areas were identified, it was usually possible to locate a bat using data from only 1 receiving station. We used the following parameters to characterize each bat's nightly fight behavior: the commuting dis- tances between the day roost and its first and last feed- ing areas, the number of feeding areas used, the dis- tance flown between feeding areas, and the number of moves from | area to another. Analysis of variance was done to test for differences among individuals and to test for possible effects of the variables weight, sex, the day roost used, the capture site, and the brightness, ‘of moonlight. To test for the effect of weight we used 3 weight classes: <20 g, 20 to 20.9 g, and >21 g. We| did not have direct measurements of light levels, so) we used an index of light based on the percentage of) time the moon was above the horizon during a night's tracking period, the percent cloud cover, and the phase of the moon. Foraging nights were then classi fied as having low light intensity (e.g., new moon or, total overcast), medium light intensity (e.g., hal moon), or high light intensity (c.g., full moon with: clear skies). Partial correlation analysis and multiple) regression techniques were used to test for correla- tions among descriptive parameters and the above variables, Supplementary information about food habits and movements was obtained in a concurrent mist-netting and banding program. ‘The distribution of resource plants was determined by mapping an area 100 m wide by 820 m long. Map- ping was facilitated by a 20- x 20-m grid constructed by Steve Hubbell. Additional grids were constructed outside this area for supplementary resource mapping, Biweekly censuses were made of selected resource plants to monitor the availability of fruits to Carollia perspicillata. Phenology results are reported in Flem- ing et al. (1977). ResuLts, Effects of telemetry on bats Did the addition of a 2g transmitter to a 17- to 22g bat affect its foraging behavior? We feel our re-Spring 197% sults largely reflect normal bat behavior for the follow. ing reasons. Eight bats were recaptured or observed in the roost after losing their transmitters. Three bats were in the same feeding areas they hud used while they were tracked. All observed bats were healthy. In addition, all of the bats that were initially captured away from their roosts for radio tagging foraged near the capture site on subsequent nights. From these facts we conclude that adding a transmitter to a bat did not cause it to change significantly its feeding lo- cations. Another test of the effect of transmitter loads is con: sistent changes in distances hats flew through the te- lemetry period. Changes might indicate either signifi- cant weight loss or gradual accomodation, There was no consistent trend toward increasing or decreasing distances. with | exception, Seven bats failed to return to their roost on the first or second night after addition of the transmitter, but did make this commute on sub- sequent nights. Whether this was the result of distur: bance associated with handling or the addition of a 2g weight cannot be determined The only known effect of adding transmitters was that they tended to pull out hair. After transmitters fell off, bats had bald spots on their backs. None of the recaptured hats suffered any serious infections that we could determine ‘There are reasons to expect Carolla perspicillata to bear such a load with a minimum of disturbance First, femaley are frequently observed carrying juve- niles that are one third adult size. even when the adults are far from the roost. Secondly. wing morphology Strongly suggests an ability to carry loads (such as young bats or large fruits), In flight, weight-bearing ability is increased by low aspect ratios, and bats in the subfamily Carolliinae have very low aspect ratios (Findley et al. 1972), Among 24 groups examined by Findley et al, (1972), the Carolliinae were third lowest in this characteristic, A female Lasiurus borealis with higher aspect ratio than Carollia is suspected of hav- ing carried a load of > 180% of her body weight (Stains 1965). Bats sampled for telemetry We were able to radio-track 24 bats for at least 3 nights, 6 bats in 1975, and 18 in 1976, Seventeen bats were males. Their average weight was 20.0 g (range = 17.0 to 21,5 g). Bats carried operating transmitters from 3 to 19 nights (Y = 12.1 nights), but since we did not radio-track every night, the number of tracking nights ranged from 3 to L4 nights per bat = 8.5). Individuals were monitored for an average of 41 hy Carollia perspicillata is a refuging species (Hamilton and Watt 1970): individuals usually roost in barge groups during the day. disperse away from this central place to feed. and then return to the roost after for- aging. For purposes of description, we will divide this Pattern into several components. including use of day FORAGING MOVEMENTS OF ¢ PERSPICHLLATA he roosts. timing of nocturnal activity, location and use of feeding areas, and distances flown per night Use of day roosts ‘Two major roosting areas were located in our study area. The first ("Sendero™) was located in a small cave created by a stream undercutting a low ridge. The cave is =10 mong. | to 1.5 m high, open at both ends, and. is subject to periodic flooding, which forces hats to move temporarily to thick foliage and spaces between large boulders just downstream. In addition to this natural disturbance. there is frequent human distur- bance because the roost is major attraction on the park's nature trail. In spite of these disturbances, the Fost has been used regularly: some banded bats have used it for as long as 4 yr. The roost was used by 100 10 200 bats of 5 species (Curollia perspicillata, Carollia subrufa, Glossophaga soricina, Desmodus rotundus, and Pteronotus parnelii). Most of the bats were C. perspicillata. The second major roost ("Red") was, located 3.5 km NW of the Sendero roost. The entrance is a 0.5-m-wide hole at the top of a dry hill, Since we could not enter the roost, we cannot describe the num- ber of bats, but this roost apparently contained iny individuals as the Sendero roost Some bats spent their days away from the 2 major Foosts, scattered among other sites such as an unused well 250 m west of the Sendero roost, a Bombacopsis, quinata trunk 80 m west of the Sendero roost, and a Cedrella odorata trunk 0.8 km north of the Sendero roost. Two bats split their time between the Sendero and the Cedreila roosts. Subsidiary roosts were not necessarily in cavities: on 3 occasions, we located ra- dio-tagged bats hanging alone among foliage. In sum- mary, most C. perspicillata roosted colonially, but some individuals used subsidiary db " In general, the tagged bats consistently returned to the same day roost. Twelve tagged individuals were always found at the Sendero (or nearby Bombacopsis) roost (V = 87 days). Three individuals failed to return to the Sendero roost just after their transmitters were attached, but started to return consistently after | or 2 days, Five individuals divided their time between the Sendero roost and subsidiary roosts near their for- aging areas: the average proportion of days they used the subsidiary roosts was 0.62 (N = 40 days). We did not monitor the 4 Red roost bats at their day roost. but their morning flights were consistently in the di rection of the roost, and they were not near theit feed= ing areas during the day. To summarize, the probit bility of bats using the main roost area was .86, but the probability of using the same roost on consecutive’ days is even higher because the subsidiary roosts we located were used repeatedly Timing of activity Transmitter signals became highly modulated in both pitch and intensity when the transmitter's anten: