The ecology of an African savanna fruit bat community: resource partitioning and role in seed dispersal.

THE ECOLOGY OF AN AFRICAN SAVANNA FRUIT BAT COMMUNITY: RESOURCE PARTITIONING AND ROLE IN SEED DISPERSAL by DONALD W. THOMAS Thesis submitted for the degree of Doctor of Philosophy of the University of Aberdeen FEBRUARY, 1982 I declare that the work presented in this thesis was undertaken and completed by myself. It has not been submitted in any previous application for a degree. All verbatim quotations are identified with quotation marks and the sources of information have been fully acknowledged. Donald W. Thomas 11 February, 1982 ACKNOWLEDGEMENTS Where does one begin to gratefully acknowledge all the help and support, both material and moral, that has been so readily offered? First, I wish to express my gratitude to the inanimate; to Africa, which captured my imagination, inspired this study, and spawned a people so warm and hospitable. I wish to thank Dr. Adrian Marshall, who followed the progress of this work with interest, who offered many valuable suggestions, and who showed patience when necessary. I thank Roger Vuattoux, director of LAMTO, for making available all the resources of the field station unstintingly and for making my stay a rewarding experience. I also gratefully acknowledge the help and friendship of Konan N'dri, Konan Germain, Kouadio Mix, and all the rest of the staff at LAMTO. Thanks are due to M. Jean-Luc Tournier, director of the Station de G6ophysique, for always making the resources of his laboratory freely available. Thinks go to Jean-Louis and Francoise Tireford who enriched my stay with warmth and hospitality and many the long evening of Tarot and beer. Dr. Paul Racey deserves special mention. His enthusiasm for biology and stimulating conversations were refreshing to encounter. The smooth production of this thesis was made possible through the kind help of Dr. R. Ralph, who permitted me to have access to the facilities of the Offshore Marine Studies unit and their word processor. Finally, but certainly not of lesser (much!) merit, I wish to make special acknowledgement of Martin Nicoll and Andrew McWilliam, with whom I passed many the hour in stimulating disagreement. This study was supported through a Natural Sciences and Engineering Research Council of Canada Doctoral Fellowship and through a National Geographic Society research grant for 1980/81. ABSTRACT This study was undertaken to examine the structure of a fruit bat community, its seasonal variation in composition, the mechanisms of partitioning of fruit resources, and the role of the bats as seed dispersers in southern Guinea savanna in Africa. The community is structured on two levels. A group of four species remain resident in the savanna-forest mosaic zone year round and three species invade the community from the southern forest zone seasonally. Diets indicate that the resident species (Epomops buettikoferi, Hypsignathus monstrosus, Lissonycteris angolensis, and Mieropteropus pusillus) forage in three mutually exclusive zones and the diet overlaps between these species are low, except in the case of E. buettikoferi and M. pusillus. The latter two species overlap heavily-in both diet selection and in foraging zone, and evidence is presented showing them to be in severe competition when resources are limiting. They co-exist only in high fruit density patches. The migrant species (Eidolon helvum, Myonycteris torquata, and Nanonycteris veldkampi) invade the community at the onsert of the rains when the diversity of fruit resources increases over the dry season low. Each uses one of the three foraging zones and co-existence is mediated through: 1) the use of different fruit sizes and 2) overlap with the resident species only on fruits that are super-abundant. Dry season limitation of the resident species' populations appears to be important in permitting the invasion of this community. -The seed rain generated by bats and birds was measured with ground-based collecting sheets on transects through the savanna. Seed rains were impressively high, with each square metre receiving on the average 3.3 seed loads annually. Bats accounted for 95.7-98.2% of the measured rain. The bat dominance appears due to the higher proportion of time spent in flight compared with birds. • TABLE OF CONTENTS SECTION  PAGE 1. INTRODUCTION  1 2. STUDY AREAS  7 2.1 LABORATOIRE D'ECOLOCIE TROPICALE DE LAMTO  7 2.1.1 INTRODUCTION  7 2.1.2 CLIMATE  8 2.1.3 VEGETATION  9 2.2 WANGO FITINI 11 2.3 OTHER SITES 12 3. MATERIALS AND METHODS 14 3.1 MIST NETTING 14 3.2 REPRODUCTION AND GROWTH 16 3.3 DIET 17 3.4 MOVEMENTS 19 3.5 POPULATION SIZE 20 3.6 FRUIT CONSUMPTION AND SEED DISPERSAL 21 3.7 SEED PREDATION AND GERMINATION 22 3.8 FRUIT TREE ABUNDANCE AND FRUIT PRODUCTION 22 3.9 CONTROLLED FEEDING EXPERIMENTS 23 3.10 STATISTICS 25 3.11 NOMENCLATURE 25 4. RESULTS 3.0 4.1 COMMUNITY COMPOSITION 30 4.1.1 SEASONAL COMMUNITY COMPOSITION: RESIDENTS AND MIGRANTS 31 4..1.2 MIGRATORY SPECIES 32 4.2 REPRODUCTION AND GROWTH OF RESIDENT SPECIES AT LAMTO 37 4.2.1 BIRTH PERIODS 37 4.2.2 OESTRUS 39 4.2.3 MATING PERI: I DS 40 4.2.4 114_LE CALLTIG BEBA7IOUR 43 SECTION  PAGE 4.2.5 RECRUITMENT AND GROWTH: E. BUETTIKOFERI  42 4.2.6 RECRUITMENT AND GROWTH: M. PUSILLUS  46 4.3 REPRODUCTION OF MIGRANT SPECIES  50 4.4 REPRODUCTION AT WANG° FIIINI  52 4.5 ECOLOGY OF THE SPECIES AT LAMTO  53 4.5.1 ROOST SITES  53 4.5.2 ACTIVITY PERIODS  54 4.5.3 HABITAT USE  55 4.5.4 DIET AND RESOURCE USE  57 4.5.5 ABUNDANCE OF BAT EXPLOITED FRUIT SPECIES 77 4.5.6 FRUIT PRODUCTION OF F. CAPENSIS  79 4.5.7 MOVEMENTS AND POPULATIONS:  MICROPTEROPUS PUSILLUS  82 4.5.8 EPOMOPS BUETTIKOFERI:  MOVEMENTS AND POPULATIONS  87 4.5.9 RADIO TRACKING  96 4.6 FRUIT CONSUMPTION AND SEED DISPERSAL  103 4.6.1 F. CAPENSIS IN THE SNB  103 4.6.2 F. CAPENSIS ALONG GALLERY FOREST EDGE  106 4.6.3 ADENIA CISSAMPELOIDES  107 4.6.4 FAECAL AND SEED RAIN  107 4.6.5 SEED PREDATION  110 4.6.6 GERMINATION OF FICUS CAPENSIS  112 4.7 FRUIT COMPOSITION  114 4.8 CONTROLLED FEEDING EXPERIMENTS  114 4.9 ECOLOGY OF THE SPECIES AT WANG° FITINI  120 4.9.1 ROOST SITES AND CALLING BEHAVIOUR  120 4.9.2 HABITAT USE  121 4.9.3 DIET AND FOOD RESOURCE USE  123 5. DISCUSSION  128 5.1 THE WEST AFRICAN PTER0i0DIDAE  128 5.2 THE WEST AFRICAN MIGRANTS  129 5.2.1 EIDOLON HELVUM  130 5.2.2 MYONYCTFRIS TOROUATA  133 5.2.3 NANONYCTERIS VELDKAMPI  136 5.3 COMMUNITY STRUCTURE AND RESOURCE PARTITIONING  137 5.3.1 THE DATA BASE  137 5.3.2 DIET OVERLAP AND SPECIES PACKIEG  138 5.3.3 NORTHWARD MIGRATIONS  141 SECTION PAGE 5.3.4 RESOURCE PARTITIONING 144 5.3.5  E. BUETTIKOFERI AND M. PUSILLUS 148 5.4 THE TIMING OF REPRODUCTION 155 5.5 FRUITS AND FRUGIVORES 161 5.5.1 FRUITS AS A DIET 161. 5.5.2 PROTEIN AND ENERGY BUDGETS 164 5.5.3 PLANT STRATEGIES 178 5.5.4 BAT FRUITS 179 5.5.5 SEED RAIN 183 6. SUMMARY 188 7. LITERATURE CITED 190 -11.  INTRODUCTION Early in their history angiosperms evolved arillated seeds and animal dispersed (zoochoric) fruits. The development of these features, if not causative, was closely associated with their early radiation and subsequent domination of most plant communities (Corner 1949). Now, in Neotropical deciduous and wet forests 53-79%, and in African wet forests up to 80%, of the tree and shrub species produce fleshy fruits consumed principally by birds and mammals (Jones 1956, Daubenmire 1972, Frankie et al. 1974, Hilty 1981). Since the development of the primitive arillated seed, the co-evolution of plants and their animal seed dispersers has given rise to a vast array of fruit forms, sizes, nutrient contents, display positions, and fruiting phenologies. Since all of these features directly affect the availability and attractiveness of fruits to dispersers and hence the reproductive success of the parent plants, they are under strong selective pressure. This has best been demonstrated by Herrera (1981) who showed that variation in the ratio of reward (pulp) to ballast (seed) in the fruits of different Smilax aspera (Liliaceae) populations was related to competition for dispersers. In populations where competition for a limited number of dispersets was severe and where the fruiting season was restricted by climate, selection bad favoured an Increased investment in the pulp and less in the seed (greater reward to ballast ratio) in order to make fruits more attractive to the avian .dispersers. Along a similar line, Snow (1965) argued that competition between sympatric Miconia species for dispersers in a tropical environment had led to the staggering of fruiting seasons. Snow (1971) pointed out that fruit are produced to attract frugivores and so are usually conspicuous and sometimes nutritious. However, some fruits may be taken by a wide range of frugivores while others may be used by only a few of the potential frugivores.. Extreme examples of this are Ficus sumatrana (Moraceae) in Malaysia which is visited by up to 25 species of birds and eight species of mammals (not mentioning nocturnal mammals; McClure 1966) and Loranthus species (Loranthaceae) in the Neotropics which are used primarily by the mistletoe thrush (Dlcaeum -2— sanguinolentum (Howe and Estabrook 1977). Frugivores differ in their treatment of seeds and movements around fruiting trees (Krefting and Roe 1949, Howe and Primack 1975, Howe and Vande Kerckhove 1981) and plants have differing requirements for "safe sites" (those offering the minimum requirements for seed survival, germination, and growth; Janzen 1971, Harper 1977) . . Selection, then, should act to modify fruit characteristics in such a way as to make fruits most attractive and accessible to frugivores offering the highest "quality" of seed dispersal characteristics (sensu Howe and Estabrook 1977). Thus plants often attract some subset of the frugivores available. The breadth of this subset is related to the spatial and temporal predictability of "safe sites". If, for example, "safe sites" are unpredictable in space and time, then plants Maximize the distribution of seeds in the environment by using the greatest range of frugivores possible (a large subset; Ficus strategy). If "safe sites" are predictable in space and/or time, selection favors fruit characters which restrict the subset to species frequenting the appropriate sites (Loranthus strategy). McKey (1975) and Howe and Estabrook (1977) have dealt in detail with the potential effects of varying nutrient value (sugary, low protein fruits vs lipid— and protein—rich fruits) and fruiting schedules (mass fruiters vs extended fruiters) on the types and numbers of frugivores attracted. The net result of the varying fruiting characters (size, shape, nutrient content, and fruiting schedule among others) is to make different fruits available to different frugivores. Superimposed on this, competition between frugivores acts to reduce diet overlap even more. Fleming (1979), reviewing the literature on frugivory, concluded that there was low dietary overlap between even closely related frugivores and that competition among frugivores was uncommon. Lack (1976), rationalizing the apparently high overlap in.the diets of frugivorous birds in Trinidad, also stressed the underlying differences in diets. The differences in diets of frugivores mean that knowledge of the partitioning of fruit resources within frugivore guilds is not only of interest to zoologists, but is also critical to the understanding of plant community composition, seed movements, and the ability of plants to recolonize disturbed sites. -82.1.2. CLIMATE The climate of West Africa is governed by the seasonal northward and southward movements of the inter-tropical convergence zone, separating the moist equatorial maritime air mass from the dry continental air mass, between approximately 5°N in January and 17-21°N in July (Ireland 1962; Figure 2). A rain belt follows approximately 320-480 km behind the front, so at a particular site the rains begin when the front is three to five degrees latitude to the north. Since the rain belt is relatively narrow, at southern sites the rains may pass to the north in July/August creating a short dry season. This mid-rainy season hiatus is less marked or absent at more northern sites. As a result of the inter-tropical convergence zone movements, the climate at LAMTO shows four more or less defined seasons: a major dry season (November to February), a major rainy season (March to July), a minor dry season (July to September), and a minor rainy season (September to November; Figure 3). While rainfall graphs and hythergraphs show four measureable seasons (Figures 3 and 4), it is unclear whether plant communities perceive the July/August minor dry period. Neither forest nor savanna species show any noticeable increase in leaf fall during this period (Devineau 1976a, Menault and Cesar 1978) and climatographs (Figure 5) show either no potential water deficit (by the method of Walter and Mueller-Dambois 1975) or only a brief deficit In August (from the data of Eldin in Devineau 1976a).- For plants, this dry period may well be buffered by soil retention. Whatever the situation, the minor dry season is brief. For this reason, unless otherwise specified, I will for convenience refer to only two important seasons: the wet season lasting from March until October and the dry season lasting from November until February. At LAMTO, average daily maximum temperatures taken over 18 years range from 30.1°C in August to 35.3°C in February. Daily relative humidity, averaged from readings taken at 0600h, 1200h, and 1800h, ranges from 71% in January to 84% in June. The mean annual, precipitation over the past 15 years was 1246  251mm (data from J-L. Tournier, pers. comm. and Lecordier 1974). 20- - _ ,  ,  .1  ,  1 J FMAM . JJ AS OND MONTH FIGURE 2: The approximate position of the inter—tropical convergence zone throughout the year in West Africa. The stippled areas indicate the months with rain at LAMTO (6°13'N) and Wango Fitini (9°50'N). (Data from Hopkins 1965). FIGURE 3: The monthly rainfall at LAMTO through 1979 and 1980 (data courtesy of J—L. Tournier, Station de Geophysique de LAMT0). The inset shows the 15 year average rainfall for each month (mean ± standard error). I  I  I cS  cl, c)  c) (NI  ,-- P A (- 1 -9 11VdNIV2:1 30V2EAV 1 II  - 0 0 Z 0 (f) < 2 < 1  H  2 (I) < --) 2 < 0 0 U-) 2 U- --) i•I  I  I  I  I  1  T  I  I o  Q D  o N%-- o c3 .....t ) ( 11 - 1u ) 11\iNIVd — cD _  taLl Co 5 pr4 X—) X —) X 0.11.01. 1  I  I.  I U " )  c•4 • co  co  (Y) (J0)  XVN 1NV1A1 FIGURE 5: Two types of climate diagrams for IAMTO. (A) shows the mean monthly rainfall  ) and the evapotranspiration potential (+---t) [data from Devineau 1976a]. (B) shows the mean monthly rainfall  and the mean maximum daily temperature  where 10°C is equated with 20mm of rainfall on the Y axis. See Walter and MuellerDambois (1975) for a justification. The stip n led areas indicate months where evapotranspiration exceeds rainfall, leading to a potential soil water deficit. A 20016012080_ 40_ E E 200- -J < 160LL Z < cc 120_ 4080- — 30o 0 — 2040o _ m 10Lu 1--- _ -  j F M AM J r i A  b ti D MONTH -92.1.3. VEGETATION The vegetation at LAMTO is a mosaic of open to Wooded savanna burned once yearly (73.6% of the surface area), an 80 ha plot of experimentally fire-protected and regenerating savanna (3.2%), and closed or broken canopy forest (23.2%). Savanna is essentially a fire-maintained vegetation association (Lamotte 1978) and as a result is notably poor in both density of woody plants and in species diversity. Three height strata are recognizable: a herb stratum with few woody species (0-2 m), a shrub stratum (2-8m), and a tree stratum (over 8m). The 0-2 metre herb stratum dominates in biomass, accounting for from 100% to a minimum of 79.9% of the total plant biomass at all savanna sites (4enault and Cesar 1979). The over 8 metre stratum is generally represented only by Borrasus aethiopum. In the 2-8 metre stratum some 30 species of woody plants are represented. Of these, 16 are uncommon, being represented in only 33% of censused plots (n=6), 10 species are found in 66% of censused plots, and only four species (Bridelia ferruginea, Cussonia barter!, Crossoptery febrifugum, and Piliostigma thonningi) are found in over 70% of plots (calculated from Menault 1971). These latter four species account for over 90% of both the biomass and stems per hectare of woody plants at LAMTO. The density of shrubs and trees over two metres in height varies from 0 to 1500 stems/ha (x= 250 stems/ha; Menault and Cesar 1979). Forests cover nearly 25% of the surface area at LAMTO and contain approximately 400 species of trees and shrubs (J-L. Devineau, pers. comm.). These form discrete blocks varying from 20 to 35 metres in height and having sharply defined boundaries withsavanna. For convenience forests are usually, divided into two types, gallery and riverine, but in reality they represent a continuum of plant communities from drier inland (gallery forest) to moister riparian (riverine forest) sites (Devineau 1976b). For excellent reviews of the architecture, densities, and species composition of forests at LAMTO see Devineau (1976a, 1976b). Gallery forests precisely follow hydrographic features of the landscape (depressions around seasonal streams leading into the Bandama River; see Figure 6) and as a result are normally not more than 15 to 20 metres wide. This association with depressions is due in part to the greater soil moisture and in part to the fire-protection offered by the broad-leafed pereriials that establish there (eg. Thaumatococcus danielli). Gallery forests are dominated by Cola gigantea, Mallotus oppositifolius, Diallium guineense, Malacantha alnifolia, and Teclea verdoorniana (among other species), but also include such important species as Elaeis guineese, Antiaris africana, Chlorophora excelsa, and Diospyros mespiliformis. These latter two species and Cola gigantea proved to be important in this study. The drier fringes of the riverine forests included many of the species common to the gallery forests, but C. gigantea, C. eXcelsa, and D. mespiliformis typically are less common or absent. A variety of species including Baphia pubescens, Nesogordonia papaverifera, Lasiodiscus milbraedii, and Pancovia bijuga reach their highest densities at these sites. On the moister river banks and islands these species are replaced by Croton scarciessi, Pterocarpus santalinoides, Cola laurifolia, Cynometra magalophylla, Manilkara obovata, and Parinari congensis; all species that are rarely found on the drier sites. The edge of gallery and riverine forests have their own specific plant associations dominated by many of the more or less fire-tolerant, fast growing, and shade intolerant woody species and climbers. In this two to four metre wide zone, Ficus capensis, F. vallis-choudae, Smeathmania pubescens, Anthocleista nobilis, Nauclea latifolia, Mimusops kummel, .01ax subscorpiodea, Napoleanea vogelii, Adenia cissampeloides, A. meigei, and A. lobata are abundant. Forests are continually ep-nding into savanna as is indicated by the presence of Borassus aethionam inside the gallery forests and by the observed advance of the edge into mapped sites (R. Vuattoux pers. comm.). It is through the advance of these fire-tolerant species that the ground is prepared (i.e. tbe grass cover reduced and the soils maintained moister by leaf co,.er ani shading) and the microclimate made favorable for the germinatioa -nd growth of the less fire-tolerant species (Merault 1977). ‘t4rq lafiterl! aeadr %t. a 0,  e o ono cssz' Oo 4 5441'''4 'eye g° ° 0 0-4301. 0  6 p. N4k 1 471 \ SCALE  4 1 km 0 b  CZ5 0  q3 (4); % at. 00 0 oo 0 :.•. SAVANNA GALLERY FOREST RIVERINE FOREST ••  • FIGURE 6: A map of LAMTO showing the major vegetation blocks within the 2500 ha. reserve and the sites where most of the work was carried out. The main netting sites were the SNB, the riverine forest (A), and nine gallery forest edge sites (B-J). The faecal collecting-sheet transects were set out in the SNB and along three gallery forest edges (1-3). a -163.2.  REPRODUCTION AND CROWTH Females were recorded in six reproductive classes: 1) immature: smaller than adult size, not having previously given birth as evidenced by small and unsuckled nipples (nulliparous). 2) nulliparous, not-pregnant: adult in proportions but not pregnant. 3) nulliparous, pregnant: pregnant for the first time. 4) parous, non-reproducing: having previously given birth as evidenced by large and suckled nipples, but neither pregnant nor lactating. 5) parous, pregnant: having previously given birth and palpably pregnant. 6) parous, lactating: lactating (determined by the expression of milk from the nipples) at the time of handling. In cases where females were simultaneously pregnant and lactating, they were classed as 'parous, lactating' unless otherwise specified. I often refer to females as immature or adult. Adults  comprised all parous females and nulliparous, pregnant females. • Since pregnancy was determined by palpation only, early stages of embryonic development were undetectable. Due to the narrow abdomen and the fluid intestinal contents of fruit bats, chorionic sacs approximately 3mm in diameter or larger were detectable and pregnancies were probably missed only for a brief period early in gestation. Males were classified in two categories: 1) immature: having abdominal testes or small scrotal testes and not having yet developed secondary sexual characters (epaulettes for E. bucttikoferi, E. gambianus, M. pusillus, and N. veldkampi; nasal Inflations for H. monstrosus; throat ruffs of longer coarser hairs for L. angolensis and M. torquata). This category also included pubertal males showing signs of increasing testes size and developing secondary sexual characters. 2) adult: having fully developed testes and secondary sexual characters. To analyse the growth of cohorts of young as graphs of weight against Lime it was necessary to have a common time alds for all individuals of -17each cohort. Since parturition dates cannot be determined precisely,I arbitrarily measured time as days past 15 March for cohort 1, or 15 September for cohort 2 of each year. These dates approximated the middle of the peak parturition months. I stress that this measure is arbitrary and is not intended to represent the days elapsed since birth. Since cohorts could have had modal parturition dates before or after these dates, the elevations of different curves cannot be compared. Slopes, representing the growth rates, can be compared since they depend only on changes in weight with time. Equations for all graphed curves were machine calculated using the least squares method (Sokal and Rohlf, 1969). The curves best describing data were those having the highest regression coefficient (r 2 ). Weight data for cohorts were also fitted to the logistic growth equation following Rickleffs' (1967) method, involving the conversion of individual weights to percents of asymptotic (=adult) weights. The asymptotic weights were taken as 190g for male E. bnettikoferi, 120g for females, and 31g for both sexes of M. pusillus. The slope of the straight line plot obtained from this method was used to calculate the logistic growth constant (IC) for comparison with previously published data. 3.3.  DIET Any faeces voided in the net or during handling were collected for later identification of the seeds (if present) or pulp, as well as for microscopic examination to determine pollen and/or plant vegetative tissue content. Identification of the fruit species from the seeds was done by qualitative comparison with a personal reference seed collection from known fruits from the LAMTO or Wango Fitini region. Any unidentified seeds were planted at LAMTO or at the Department of Botany, University of Aberdeen, for lateT recognition of the seedlings. I also planted samples of identified seeds to verify my ability to recognize seeds. Pulp faeces containing no seeds usually had characteristic colours and textures (and in the case of Vitex doniana a characteristic smell). These were identified by comparison with faeces produced by captive bats fed suspected fruits. When coupled with regular tours made to assess the fruiting species available and observations of bats feeding, faeces identification was in most cases remarkably simple. -isThe analyses of the diet of the resident and migrant bat species at LAMTO are based on 589 faeces (net faeces) collected from individuals netted at all sites over the two years. I also collected 1236 faeces from below the E. helvum roost (roost faeces). These have been treated separately since they swamp the smaller E. helvum samples from other times of the year. I have also included scattered observations of bats feeding on particular fruit species to supplement the faeces data. The use of diet items by the bat species were analysed from the te-cal data pooled for the entire year in order to maximize sample sizes. While this may under-represent important but highly seasonal fruit species, this was unavoidable. In this way seasonal changes in species use was masked, but I have drawn attention to seasonal variations wherever this was informative. The relative specialization of the diets of the bats was analysed by calculating: EVENNESS= [ p ij log i op ij ]/ [logioN] where p ij is the proportion of diet item j in the diet of bat species i and N is the total number of fruits used by the species. Diet overlap between species pairs was calculated by: OVERLAP = 1 where Pi j and P hi are the proportion of item j in in the diets of bat species i and h. As an independent measure of diet species selection by E. buettikoferi, I located 32 feeding roosts and recorded the fruit remains below them over 366 roost nights. The species were counted as used if present under a roost regardless of their relative abundance. .I also collected hair samples from the head and shoulder regions of bats to determine whether they had recently visited flowers based on traces of pollen on the fur. This presupposes that the bats acted as pollinators and not nectar theives, gaining access to nectar by piercing the carolla and thus avoiding Contact with the stamens as has been noted with Nanonycteris veldkampi feeding on Spathodea campanulata (Ayensu 1974). This latter feeding strategy would be undetectable by my techniques. Hair samples were immediately placed in clean paper pouches and kept sealed until later mounted on slides and examined under a compound microscope. Pollen was recorded as present or absent in a -19sample and identified to species if possible by . comparison with pollen samples from known flowers. I collected hair samples from 336 bats at LAMTO during March/April, June/July, September/October, and November to January periods and from 113 bats at Wango Fitini in July/August and November. 3.4  MOVEMENTS To assess the movements and site fidelity of bats at LAMTO I relied heavily on recapturing marked individuals with mist nets. I have treated each site along gallery forest edges, in the SNB, and in the rivSrine forest as a point and I did not consider bats recaptured in different nets at the same site as having moved. In this way movements • were only among gallery forest edge sites or between these and the two other sites (the SNB and riverine forest). Treating these movements in section 4.5.7 necessitated making at least one implicit assumption; that, in the absence of any habitat preferences, individuals would be as likely to move among gallery forest sites as between these sites and the SUB. Structural differences between the gallery forest edge (a narrow "corridor" of habitat) and the SNB (a planar habitat) may have affected the validity of this assumption, but since there was no way to test this problem I have treated movements directed towards one habitat as indicating a real preference for that habitat. In addition, I attached radio transmitters and tracked E. buettikoferi to provide more detail on the activity patterns and range of individuals. Initially I had hoped that tracking would be a major 'component of this study, but due to time conflicts and the difficulties of following mobile bats in long grass savanna at night this proved to be of only secondary importance. I located individuals 'carrying transmitters using a hand held four element Yagi antenna and AVM LA-12 receiver. I attempted to stay within . 100m of a bat's position and maintained a log of its movements and the amountiof time spent flying or hanging in. trees. Flying was easily determined due to variations in signal strength caused by the waving of the antenna in flight. This has been used successfully in previous studies on NeOtropical bats (Heithaus and Flem ing, 1978, Morrison 1978). -20-- The transmitters were built following the circuitry of Thomas (1979) modified to incorporate a circuit board to speed construction, a more efficient - 1 4wavelength antenna, and a more powerful 2.8V lithium battery. The final circuitry was encapsulated in silicon rubber (Dow Corning ATV 735 SILASTIC) and had final weights between four and five grams (or between 2.5% and 4% of the bats' body weights). The transmitters had a theoretical life of 100 days; however in most cases that acheived was much less. The most common cause of failure was due to the shearing of the antenna at the transmitter/potting junction. The transmitter, Yagi antenna, and AVM receiver provided a working range of one to five kilometres depending on the height of the receiver or transmitter above the surrounding terrain. Initially the transmitters were glued to the dorsal fur with Silastic (see Bradbury 1976, Heithaus and Flemming 1978), but recaptures of two individuals with large abcesses forced me to abandon this method with E. buettikoferi. Subsequently transmitters were attached with a collar assembly which maintained the transmitter in a mid-dorsal position. In no case did I find that collars caused any damage to the bats. 3.5.  POPULATION SIZE Population sizes for E. buettikoferi and M. pusillus were estimated using a modified Jolly Seber mark-release-recapture model available on computer at the University of Aberdeen (Dr. S. Buckland, pers. comm.). This model was chosen since it accepts recruitment and losses to populations through immigration/birth and emigration/death and also provides confidence limits for estimated parameters. Since the Jolly Saber model has the weakness of occasionally giving nonsensical estimates of survival (greater than unity) or negative recruitment (as opposed to death) the model was 'modified to restrict estimates for parameters within the zero-to-one range. Buckland (1980) preserts the mathematical rationale and procedure for this modification. To create input data for the model, captures were pooled for one month intervals. Individuals captured once during any one month sample were ignored for any subsequent captures. -21-- 3.6.  FRUIT CONSUMPTION AND SEED DISPERSAL - To assess the importance of Megachiropteran bats as fruit consumers and seed dispersers, I concentrated on the gallery edge zone and the SNB, primarily because these were dynamic zones in a constant state of invasion by forest or savanna species. I selected Ficus capensis and Adenia cissampeloides for quantitative measures of fruit consumption. Using non—toxic latex paint I individually numbered fruits of the two species. I then noted the time of disappearance (=consumption and dispersal) or fall (=failure to disperse) by means of dawn and dusk tours. Fruits consumcd during the night were assumed to be taken by bats and those taken during the day primarily by birds and squirrels.  Other possible diurnal dispersers were monkeys (Cercopithecus aethiops and C. petaurista) and nocturnal dispersers were pottos (Perodicticus potto). However, these species were uncommon at LAMTO due to hunting pressure and I never observed them taking F. capensis or A. cissampeloides fruits in the areas where I worked. Through this marking program I followed the fate of 1049 F. capensis fruits on 38 trees in the SNB and 243 fruits on seven trees along one gallery edge. A sample of 100 A. cissampeloides fruits were marked at this latter site. I measured the rate of dispersal of seeds in f , e - ces within the SNB and into the five metre wide belt of savanna bordering three gallery forest edges (see Figure 6) using plastic collecting sheets. These sheets were set out in the open (not under trees) so the faeces found were only of •flying bird or bat origin. Initially I set out three transects (one 20m 2 in the SNB between 5 January and 9 June 1980; one 20m 2 along a gallery forest edge between 19 May and 14 October; one 40m 2 along a gallery forest edge between 5 September and 14 October) that I visited at three day intervals to count the fall of faeces. These transects did not enable me to identify the dispersers (birds or,bats), but did prove to me that seed rain was measureable.  From 20 October to 5 December I monitored one 24m 2 transect (SNB) and two 110m 2 transects (gallery forest edges) with dawn and dusk tours. By noting the time of appearance of faeces (day or night) the dispersers were identified as birds or bats. -22.— 3.7.  SEED PREDATION AND GERMINATION I measured the predation rate on seeds of F. capensis associated with fruit parts (simulating fallen fruits or unconsumed fruit parts) and on seeds collected from bat feces. In each of 20 plastic petri dishes having four 3mm access holes, I placed 20 faecal seeds and to 10 of these dishes I also added pieces of fruit. The dishes were set out, alternating dishes of seeds with fruit with dishes of seeds alone, at four metre intervals along transects in savanna. After 24 hours the number of seeds remaining in each dish were counted. Before subsequent runs the dishes were washed in concentrated acid (H2SO4) to remove any possible scent trails left during previous visitations. No two experiments were run along the same transect. The germination rate and final percent germination of seeds collected from F. capensis fruits (with and without the surrounding ovary wall), from rejecta pellets, and from feces were measured. Fifty seeds were placed on damp cloth strips in each of 10 covered petri dishes for each experimental group. The dishes were examined and dampened daily and any germinated seeds were counted and removed. Early in the experiment it became evident that seeds still covered with the ovary wall were poor germinators. To test for the presence of inhibitory chemicals, I placed faecal seeds under two experimental condititions. In one group (10 dishes of 10 seeds) seeds were placed on damp cloth strips. In another group (10 dishes of 10 seeds) seeds were set on cloth pouches containing pulverized ovary walls and flower Tarts collected from fig syconia. In this way the latter group were in chemical but not physical contact with the flower parts. 3.8.  FRUIT TREE ABUNDANCE AND FRUIT PRODUCTION To assess the abundance of the s , hrub and tree species that I knew to be exploited by bats in the gallery edge zone, I made complete counts of all the individuals of "bat plants" along transects totaling 3600 metres of edge at five sites. In addition I counted the number of F. capensis -23trees over 5 cm in diameter at ground level along three 20 metre wide transects in the SNB, totaling a length of 800 metres and an area of 1.6 ha. From this sample I later estimated the total F. capensis population in the 80 ha SNB. Ficus capensis proved to be one of the species of fruits most commonly exploited by bats. To gain some insight into the production and availability of F. capensis fruits in the SNB and gallery edge I made counts of ripe (red or orange) fruits twice weekly in the late morning or early afternoon. In this way I followed the fruit production on 54 trees in the SNB between 8 November 1979 and 5 November 1980 and on 46 trees along 1.5 km of gallery edge between 21 March and 5 November 1980. 3.9.  CONTROLLED FEEDING EXPERIMENTS I kept three E. buettikoferi and eight M. pusillus in a large flight cage (8x4x4m) for 17 and 65 days respectively during Which I measured their intake of F. capensis fruits. Each individual was weighed before and after confinement to monitor weight gain or loss during the experimental period, but was not handled between these times. Each night I provided weighed quantities of fruits on an ad libitum basis (usually 150 g for M. pusillus and 500 g for E. buettikoferi). At various times through the night and/or the following morning the remaining uneaten fruit and masticated pellets (rejecta) were removed and weighed. This permitted the calculation of intake by the difterence . [INTAKE = weight given - (uneaten 4rejecta)]. Samples of rejecta pellets were oven dried to constant weight at 80°C to determine the moisture content and a dry to wet weight conversion factor. The wet weights of rejecta pellets colle.cted before 0800h were used directly in calculations since evaporation was shown to be negligible in the ca. 100% night-time humidity. Rejecta pelleLs collected after 08001i were oven dried and the wet weight was calculated using the conversion factor. Protein and energy values were determined for the fruits of F. capensis as well as for F. vallis-choudae, F. ovata, Chlorophora excelsa, Bridelia ferruginea, Nauclea latifolia, and Vitex deniana. Only those parts of the fruits normally ingested wele analysed. For the Ficus -24B. ferruginea, and V. doniana the syconium or pericarp was separated from the seeds, but for N. latifolia the seeds and pulp proved inseparable and were used together in the analyses. For C. excelsa, the sponge-like fruits were pressed in a flexible fine meshed screen to separate the seeds and fibre from the fluid. Only the fluids were used in analyses. For nitrogen determination, pulp or fluids were oven-dried to constant temperature at 80°C. Five gram samples of the dried material were analysed at the Office de Recherches Scientifique et Technologique Outre-Mer (ORSTOM, Adiopoudoum6, Ivory Coast) using a standard Kjeldahl digestion and colorimetric technique. The nitrogen value so obtained was multiplied by 6.25 to estimate the total protein content (Lloyd et al. 1978). Soluble carbohydrates in the pulp or fluids were measured using a temperature-compensated refractometer (Bellingham and Stanley, Tunbridge Wells, England). Carbohydrate measured in this way is expressed as grams of sucrose equivalents per 100g of solution. Since monosaccaride sugars with half the caloric value of sucrose per gram have half the refractive index the refractometer readings can be translated to energetic values (Kcal) regardless of the actual sugar composition of the fruit. I will term all readings as sucrose equivalents. The energy content of fruit pulp was calculated by: Kcal= [4.20][%1120 in pulp][measured % sucrose] 1- [measured % sucrose] where 4.20 is the caloric value (Kcal) of one gram of sucrose. Energy ingested is only available for metabolic use after assimilation. To determine the efficiency of assimilation of soluble carbohydrates by bats I trained two E. buettikoferi to accept 1cm 3 pieces of sponge soaked in 5, 10, and 15% sucrose solutions. Each bat, after reaching water equilibrium (defined as not having urinated during the previous 30 minutes), was fed a particular sucrose solution until satiation. The sucrose concentration of the feces/urine produced during or within 10 minutes of the experiment was measured with the refractometer. -25-- 3.10.  STATISTICS Parametric statistics follow Sokal and Rohlf (1969) and non—parametric statistics follow Seigal (1956). The rejection region in all tests was p‘ 0.05. Wherever means are treated statistically they are presented as mean ± standard deviation unless otherwise stated. 3.11.  NOMENCLATURE Nomenclature of the bat species follows Rosevear (1965) and Bergmans et al: (1974) and that of plants follows Hutchinson and Dalzeil (1957). The bat and plant species along with authorities are listed in Tables 1 and 12. Hm n=92 I  I  I  I  I Mp n=414 Eb n=996 0.150.05- = uJ :". ); 0 15iIfl 0.05LIJ I- < Li •=t LJ 0.02JFMAMJ JASONDJFMAMJJ ASOND 1979  1980 MONTH FIGURE 9: The catch rates and sample sizes of E. buettikoferi (Eb), M. pusillus (Mp), and H. monstrosus (Hm) at LAMTO through 1979 and 1980. 0.005 J FM AMJ J ASONDJ FMAMJ J ASOND 1979  1980 - MONTH FIGURE 10: Catch rates and sample sizes of L. angolensis (La), M. torquata (Mt), E. helvum (Eh), and N. veldkampi (Nv) at LAMTO through 1979 and 1980. Eg n=223 • Mp • n=230 • • 0.15 0.05 Nv n=13 Mt n=12 0.06 0,02  •41--f—•—r-1 -1--,--1 ,  J FMA MJ JASON D • MONTH FIGURE 11: Catch rates and sample sizes of E. gambianus (Eg), M. pusillus (Mp), M. torquata (Mt), and N. veldkampi (Nv) at Wango Fitini. -31On this basis, the LAMTO community was comprised of seven core species; E. buettikoferi, E. helvum s H. monstrosus, L. angolensis, M. pusillus, M. torquata, and N. veldkampi. It is primarily with these species that I concern myself in this study. The Wango Fitini core community consisted of four species; E. gambianus, M. pusillus, N. veldkampi, and M. torquata. E. helvum was present in the region seasonally (see section 4.1.2), but was never captured at Wango Fitini. 4.1.1 SEASONAL COMMUNITY COMPOSITION: RESIDENTS tND MIGRANTS Figures 9, 10, and 11 present the monthly catch rates for the core species at both LAMTO and Wango Fitini. At both sites all species showed considerable fluctuations, although M. torquata, E. helvum, N. veldkampi, and L. angolensis varied most, cycling from absence to relative abundance seasonally. The fluctuations arose from two principal factors: 1) seasonal birth periods and the subsequent recruitment of immature individuals into the flying population (see section 4.2), and 2) clearly defined seasonal migratory movements. The coincidence of the recruitment periods for the species present throughout the year with the movements of the migratory species into and through each community, however, tends to mask the inter-specific differences. Figure 12 presents the species at LAMTO and Wango Fitini in terms of their representation in net samples and their recapture rates. The species in each community cluster into two groups: 1) those present in over 80% of sampling periods and having high recapture rates, and 2) those present in fewer sampling periods and having low recapture rates. Assuming that individuals of species that remain in the area throughout the year are available for recapture, while individuals of species that move seasonally through the communities are not available for recapture, these two groups correspond with the resident and miglant species respectively. '100 80 60 cj  40 20 UJ CC (.4 FZ 100 LL 80 60 4 20 10 12 14 16 18  50 52 10  12 A % RECAPTURED FIGURE 12: The proportion of marked individuals that were recaptured at some time plotted against the percent of sampling months when each species was caught at (A) LAMTO and (B) Wango Fitini. Resident species had high recapture rates and were caught in over 80% of sampling periods, while migrant species were never recaptured and were caught in fewer than 80% of sampling periods.Bat species designations are as in Table 1. • 0.08 - 0.04 A. • J FMAMJ J AS OND B. . • J FMAMJ J ASO ND J FMAMJ J A SON D 1979  MONTH 1980 C. 0/0 100] FIGURE 13: The catch rate of M. torquata at (A) Wango Fitini and (B) LAMTO during 1979 and 1980. For LAMTO, the upper line represents the total catch rate and the lower stippled line represents the catch rate of females only. (C) shows the proportion of females (stippled) in the catch at the Tai Forest site in November. f•m, T77' Thus at LAMTO E. buettfkoferi, H. monstrosus, M. pusilus, and (possibly) L. angolensis, and at Wango Fitini E. gambianus and M. pusilus are resident while both communities share M. torquata, E. helvum (at least . regionally),and N. veldkampi as migratory species. The seasonal patterns of these latter three species warrant further elaboration. 4.1.2. MIGRATORY SPECIES Myonycteris torquata MOVEMENTS: Figure 13 and Table 2 show a clearly bimodal distribution of M. torquata annually at LAMTO with corresponding absences and peaks in the catch rate at Wango Fitini. M. torquata was absent or greatly reduced in numbers at both sites between December and April, corresponding with the major dry season. Associated with the northward movement of the inter-tropical front and the onset of the rains at LAMTO, M. torquata first appeared in netting samples at this site in April (1979) and May (1980) and became the most numerically abundant species in June (1979) and May (1980). Neither during this annual peak nor between years were any individuals ever recaptured. At Wango Fitini, M. torquata was first taken in May (1980), only 12 days after the first individual was taken to the south. Associated with the passage of the inter-tropical front to its most northerly position in August, catch rates declined at both sites and subsequently increased in September to November. It is impossible to determine when the second peak occurred' In the north since no netting was carried out between August and November. During November, when catch rates were declining at LAMTO, M. torquata was the most abundant species at the Tai Forest site, comprising 53.2% of the captures (n=111). SEX RATIO: At both LAMTO and Wango Fitini males dominated in all monthly samples. The variation in the proportion of males in samples from the two peaks and between the sites indicated that there were sexual. differences in the movements. Adult and immature fenales together comprised an average of 24.6% (n=134) and 23.7% (re . --59) of the first peaks in 1979 and 1980 respectively. The maximum proportion of females -33occurred in May 1979 (33.3%, n=24) and in June 1980 (27.8%, n = 18). In all months the sex ratio departed significantly from unity (p<0.05). During the first peak, the catch at Wango Fitini was entirely of males (n=10) and at both sites only males were taken during the second annual peak (LAMT0=107; Wango Titini=2). While between September and December only males were captured at savanna sites, at the Tai Forest site in November the situation was reversed. Females constituted 64.4% of the sample (n=59), representing a significantly unequal sex ratio (X 2 =4.9, df=1, p<0.05). Eidolon helvum MOVEMENTS: Observations of the sequential build-up and decline of E. helvum colonies at Abidjan (5'12' ;4°00'W), LAMTO (6°13'N;5°02'W), Ferkessedougou (9°30'N;4°30'W), and San, Mali (13°15'N;5°00'W) considered together with the netting samples from LAMTO showed that this species exhibited a pattern of movements relative to the rains, similar to that of M. torquata (Figure 14). Through the major dry season (October to March) E. helvum roosted in at least one large colony in the forest zone, located in the central park of Abidjan (Parc du Plateau). Beginning in October, the Abidjan colony increased from several hundred individuals to an estimated peak of 200,000-500,000 individuals in January/February when females gave birth. In March, this colony declined rapidly until it numbered only a few thousands and thereafter declined slowly to a low of several hundreds in July/August. Beginning in late February (19. February 1979, 29 February 1980), I first observed E. helvum roosting in small widely scattered groups of 8-15 Individuals in the open savanna Borassus palms (B. aethiopum) at LIMO. Within two weeks of these first sightings, an estimated 100,000 individuals arrived overnight (5 March 1979, 12 March 1980) and formed a consolidated colony in the Borassus palms, covering an area of approximately 50 ha. In 1979 the colony arrived one month prior to the first significant rains of the year and in 1980 during the first rainy 13°15' 0.3 ztz ..111 n 1.0 0  9°30' 3 z,11:211 100 5°12' J FMAMJJ ASOND FIGURE 14: A schematic diagram of showing the approximate timing of arrival and departure of E. helvum colonies at Abidjan (5°12'), LAMTO (6°13'), Ferkessedougou (9°50') and San (13°15'). Estimated colony sizes are given in thousands. XXXX indicates the months with the highest catch rates for E. helvum at LAMTO (see also Figure 10). -34month. The presence of E. helvum in the area was not reflected in the netting samples. While individuals could be seen foraging in the canopy of the gallery forests, these did not descend to net heights. The colony remained at LAMTO for 45 days (1979) and 37 days (1980) and departed as a group in a single night; 20 April 1979 and 19 April 1980. Three days and eight days prior to the definitive movement of E. helvum from the region, the colony was extremely agitated during the day and over a period of several hours shifted the roost to a new site 1.5km distant. Based on the numbers of faeces found under the new roost daily, it appeared that foraging success was reduced during these last days, although this was impossible to quantify. It is noteworthy that no E. helvum were ever seen roosting at LAMTO between 1960 and 1978, nor in 1981 (R. Vuattoux, pers. comm.). After the departure of E. helvum from LAMTO, I located another colony at Ferkessedougou (latitude 9°30'N), nearly 380km to the north. I estimated this colony to number between 2,000 and 5,000 individuals on 25 April 1980. I could not determine when this colony arrived, but the villagers indicated that it had appeared "recently". The colony remained until at least 31 May 1980, but was not present between 26 July and 21 November 1980. On 12 July I also located a colony numbering no more than several hundreds at San, Mali (latitude 13°15'). I have no indication when this colony arrived or departed. Following the progression of colonies to the north with some reaching at least latitude 13°15'N by July, there was an increase in E. helvum in the netting samples at LAMTO, peaking in September 1979 and October 1980 (see Figure 10). During -this period I could locate no colonies either in the vicinity of Wango Fitini and Ferkessedougou or near LAMTO [nor were any E. helvum for sale in the local markets, which is a good Indication of the absence of colonies from the region]. The peak at LAMTO and subsequent decline to a minimum in December coincided with the build-up of the Abidjan colony. -39triggered by rainfall. In 1979 the first birth period began and the peak lactation month occurred prior to any measur able increase In rainfall over the previous two months, but in 1980 this coincided with an increment in rainfall over the January low. In both years the second birth period began during the month with the lowest rainfall for the minor rainy season (August). Birth and lactation periods were highly synchronized within each species. No females of either E. buettikoferi (n=71) or M. pusillus (n=49) caught between June and July or December and February were ever found to be lactating. The exact duration of the lactation periods for individuals is impossible to determine from the field data, but the maximum possible length can be estimated by the spread of lactation (earliest to latest dates for lactating females) for the entire population. No single female could lactate longer. For the first lactation period E. buettikoferi and M. pusillus had spreads of 59 and 58 days respectively and for the second lactation period spreads of 93 and 79 days respectively (Table 4). In both cases the second lactation period had a greater spread than the first. This could be due either to a real difference in the length of individual lactation periods between the seasons or merely to less synchronous births. During the first lactation period one E. buettikoferi and two M. pusillus were captured carrying newly born young (weights: E. buettikoferi =20g; M. pusillus =4g). These individuals were recaptured still lactating 42 days later (E. buettikoferi) and 39 and 43 days later (M. pusillus), giving a minimum duration for lactation between 39 and 43 days for both species. These data indicate that first lactation periodslast between.seven and eight weeks and that the second lactation periods may last up to 13 weeks. 4.2.2. OESTRUS Early in the lactation periods females were rarely palpably preguant, but after March and September for E. buettikoferi and April and -40mid-October for M. pusillus, all lactating females were also palpably pregnant (E. buettikoferi =21; IC pusillus =7). One lactating female E. buettikoferi killed on 15 April 1980 had an implanted embryo with a chorion measuring 4mm in diameter. Most females of proven reproductive capacity (i.e. parous) were either pregnant or lactating when handled, indicating that few failed to reproduce during a given period. Overall 11.4% of parous E. buettikoferi (n=166) amd 8.8% of parous M. pusillus (n=114) were nonre p roductive. While early pregnancies may have been overlooked (over-estimating the frequency of reproductive failures), all pregnant females should heve been detectable by palpation between June and August and December and February (the second half of the gestation periods). This enables a comparison of the failure rate for females carrying pregnancies through the wet and dry seasons (June to August and December to February respectively). For neither species was there a significant difference in the failure rate between seasons (E. buettikoferi: n=88, X 2 =1.25, df=1, p 0.05; M. pusillus: n=51, X 2 =1.70, df=1,p 0.05), nor was there a significant difference between the species in the proportion of females failing to reproduce (X 2 =0.21, df=1, p 0.05). The major dry season apparently had no effect on the probability of females initiating or maintaining pregnancies. Six parous E. buettikoferi and 10 parous M. pusillus were recaptured during two or three successive breeding periods. All females were either lactating or pregnant during each period. 4.2.3. MATING PERIODS The seasonal parturition coupled with the evident post-partum oestrus indicate that mating must be seasonal as well, occurring early in the observed lactation periods (February to April and August to October). However, adult male E. buettikoferi and M. pusillus recaptured between successive mating and non-mating seasons showed no significant variation in testes length between the seasons (E. buettikoferi: xmating=7.6mm, x non-mating = 7.6mm, t=0.04, df=66, p 0.05; M. pusillus: xmating=6.0mm, xnon-mating = 5 .6mm, t=1.56, df=34, p 0.05). Only recaptured males known • •  6'6  6'ö  6 co °  CNI  0  CO  CO•d• I  I co '6' Cd • 0 = - z 0 • cr) • • -2 4.) C 1 / 4 1 0 Cd 4-1  C.) 44  1-1 O 1 4-1  • .O  •r-1 O 0 CN1 C.)  b0  • n-1 e-1/4 CO 1/40 0.) 14 W X u N  c•i Cr) ON ▪  •--I  • - u Cd  Cd  • • o • b8 4 0 0 •0 C.)  • -4  %.0 • .01  r••• •: t • %0 44 O CO X CO O a) CN - • C ▪ d b0 130  0  II 4•1  • •-4  >4 4 4 b0 bo • • • r4  PC) a)  a) :3:3 W co %.0 CU .1-1  1-4  Cr) 1/40 CO 0.) bo b0 0 >4  • •-I  c 1.4  CO  C•1 In• W • •  a)  II 4-1  >4 bo O 44 ••4 . g  co  •• O I-1  a)  4 a)  44 co  cd ( 6 ) 11-1013M A008 • • gee 0 4 ' 01' JD • 1-1-1 . • .. • •° • 111 •tb .• • a  I  * . 0 (1)  ..-1  4-1  •r4 $.4  co  co • CO 0 0 a) O 4  0 •:t  • 4-I  0  '0  •,-i $4  C.)  Cl)  •--I O 4-)  • cj /1 4-1  Cl) • 0 4-1 • 0 n -• n 4-1 4J —)  Cl)  I-1 0  /-1  0 I-1  ca 0)4-4 34-1  ..  cn --)  CO oN  co  0 1--•  Cl)  0 ▪ m 0 -H Cl)  —i  4-1  4-1 4-1  C'1-1  C1 2  .1.)  C 0 a) CN1  •  cr O En Gil 4-1  a) • H  ;-1  Hi 4  $-1  o  cO  • oi  •  El  1-f) 1 4-1  0  Cl) 0 O c) 4-1 ..-  0 2  .r.1 e•- n  a)  V %.0 4-)  fr4  $4  P. .t..'  N../  0  • a)  4-1  4-1  ON O -  ca  al -- I - LL  ,0 cn 0 N-  04-  -F • o • N  -r-1 ril —I  al >4 CO  C) CA U)  .-  $4  4-1  CNI —.)  r-f  .--I  CO  4-1  • Fl 4-1  4 - 1 71 C • CO  0 =  al I—  a) $-1  b0 H 9-1 0 "0 •-1 O 4 0 Cl) ='4-40 CO  • • O c)  Cll Ci •P CO  . ** -1  1:4  al  •• z  a) <4 0 3 r-- r-i  9-1  .--1 fat  4-I  r-I  • FI  4-1  al  ,--1  VD 0  0  4.)  al Lr) O CO  0 0 CO ^0 + 0 • H 1--I  0 • H  al  ca ›C b0 co  cn co  a)  bi) C n 1 (/)  4-) 4-1  $-1 0 8 4 4 ozl  .--1 CO b1:1 0 -4 • T-1  4-1  0"  H U)  a)  co  0 >4 •<  3 3  al 0 .0  • • >-,  a)  a)  4-)  PO '0 >,  P. o • H  0  co  •• .1: 21 co  CO  ,--I —)  CO  . cu cf) 4-1a)  41 Hi O 1-4 NI al  co Ln O 0 0 a) --)  o  1.4  bo ,---1  + 4-1  la.  a)  al .i.)  3-1  El  X • a)  ca.  a) ..o 4-1 4  4-1 CN1 W  4-1  4-1  • 4 - i  0 a) 0 Cl) boa) 14 • H 0 co 0 H -al  • -1  CU  4-)  >-1 4  3 $ 4 0 a)  0  (  e  .. 4 4 Cl) 64 E1 in  14  .1-1 2 O ' 6  O  6  06  O r 7 1 pc sj  2  co ( D ) 11-1913M 1000 701. Z01. 001. 96 96 76 Z6 06 = 99 LD 99 w 06 99 99 79 Ze 09 2Z. 24 6 1:=3, co sivannioNi •()N FIGURE 19: The distribution of forearm lengths in samples of male and female E. buettikoferi captured at LAMTO during 1979 and 1980. The dip in the histogram at 96-98mm forearm length for males corresponds with the size at puberty. • /Mb NNO CD - CD - -C) • • (ID CD .111M 11. • • • •  • • •  •• 0 X• •,..:•• •ou  • ft..... •  e ~ o ,, o •1:66.:. e.%*: •• —1 .14 Iwo • • • n ••• NI  i I  I  I  1 O CD  CD  CD  CD  0 0  CT)  00  CD  CT)  00 ( ww ) HiON31 W8V380d FIGURE 20: The regression of forearm length on weight for male and female E. buettikoferi. Regression lines were calculated using individuals weighing less than 160g for males and 110g for females. -41to be adult were used in order to exclude pubertal males-of near adult size from the analyses. 4.2.4. MALE CALLING BEHAVIOUR Male epomophorine bats of the genera Epomophorus, Epomops, Hypsignathus, Micropteropus, and possibly Nanonycteris signal to females by means of conspicuous vocal, visual, and olfactory displays. Excepting H. monstrosus, males of all species have glandular invaginations on the shoulders (epaulettes) which are endowed with tufts of white hair. Males display by emitting short loud calls at approximately one second intervals, simultaneously flapping the half-opened wings and everting the epaulettes. For descriptions of the displays and their functional significance see Wickler and Seibt (1976) and Bradbury (1976). A detailed study of male calling behaviours was beyond the scope of this study, but I will present some data relevant to later sections. E. buettikoferi: Adult males of this species maintained adjacent but loosely packed territories (estimated inter-individual distances were 30-40m) scattered through the riverine forest along the Bandama River and rarely in the inland gallery forests. Males displayed at these sites for up to eight hours on most nights and they maintained these territories throughout the year. Six territories, on which 1 monitcred occupancy (a male present and calling) in each month of the study, were occupied at some time during all months. Over 101 nights between January and November 1980,1 recorded whether or • not males called, how many were calling (full, half, none), and what time the calling began. I also noted the moon condition as "bright" when the moon was above the horizon between 1900h and 2400h and half phase or greater, and as "dark" when the moon was below the horizon at these times, less than half phase, or covered by heavy cloud. Males showed clear differences in their response to moon condition between the mating and non-mating seasons. During February/March and August, all males called on all nights regardless of the moon condition. There was no significant difference between the starting times on "bright" and "dark" nights (t=0.79, df=53, p>0.05). The mean starting time was 1942h. -42In contrast, during the non-mating season calling was strongly affected by the moon condition. On only 7% of "bright" nights (n=30) were all males calling and on 33% of the nights no males called at all. On 75% of "dark" nights (n=16) all males called, and on the remaining 25% half the males were judged to be calling. Starting times on "bright" nights when males did call (X=2204h, n=20) were significantly later than on "dark" nights (t=6.21, df=34, p<0.05) when males began calling at 2011h. M. pusillus: I heard males of M. pusillus call only briefly during April 1979 and never thereafter at LAMTO, although I did hear males call at sites.to the north and I could recognize this species' call. 4.2.5. RECRUITMENT AND GROWTH: E. EUETTIKOFERI Immature bats first appeared in the flying population in May/June and November/December for cohorts 1 and 2 each year. Male and female cohorts could be followed as they grew to adult proportions (Figures 17 and 18), but became difficult to distinguish from older adults above the weights of 160g for males and 110g for females. In the following analyses these weights are used as the upper limits of discernable cohorts (see section 3.2.). The distribution of forearm lengths in the total captures and the regression of forearm length on weight are presented in Figures 19 and 20. Both males and females first flew and were captured at forearm lengths of 74-76mm and weights of 45-65g. It is impossible to determine .the exact age of recently weaned individuals caught in May/June and November/December; however, assuming that average individuals were born at the start of the peak lactatioa month (1 March, 1 September) with a mean weight equal to that of newly-born juveniles carried by capturea females (R=23.0g, n=4), a rough estimate of the average weight incremant per day through the lactation period can be made. Five individuals, caught weighing under 65g (range =47-64g) and estimated to range between 42 and 83 days of age, were calculated to have gained an average of 0.64g/day (range=0.32-0.88g/day). • -43ANALYSIS OF COVARIANCE COMPARISON COHORT df  F ratio Probability Imm orvs Imm ? 1,  1979 IA112  0.48 ns Imm crvs Imm 2,  1979 1&121  0.42 ns. Lmn d'vs imm 1,  1980 1&129  0.53 ns Lmn or(cohort 1, vs 1979) 1&179  2.12 ns Imm or(cohort 2, Imm e(cohort 1, vs 1980) 1979) 1&206  1.39 ns Imm e(cohort 1, Imm 7. (cohort 1, vs 1980) 1979) 104  0.35 ns Imm  (cohort 2, Imm g (cohort 1, vs 1979) 1979) 1&52  0.02 ns Imm  (cohort 1, 1980) TABLE 5: Comparisons of the growth rates (slopes of the regression lines) of male and female cohorts of E. buettikoferi at LAMTO. ns=not significant at p.10.05. --44— A. COHORT SEX‘' GROWTH RATE (g/day) 95% CL upper  lower 1,  1979 0.292 0.314 0.270 1,  1979 0.260 0.319 0.201 2,  1979 0.266 0.294 0.237 2,  1979 0.232 0.313 0.151 1,  1980 er 0.318 0.356 0.279 1,  1980 0.288 0.386 0.190 All cohorts x=0.276 B. 95% CL COHORT SEX ASYMPTOTE (g) K upper lower 1, 1979 190 0.0072 0.0088 0.0056 1, 1979 120 0.0128 0.0200 0.0056 2, 1979 190 0.0068 0.0084 0.0052 .2, 1979 120 0.0108 0.0172 0.0036 1, 1980 190 0.0076 0.0094 0.0060 1, 1980 .  120 0.0116 0.0156 0.0675 Table 6:  (A) Growth rates for male and female cohorts of . E. buettikoferi calculated from fitted regression lines (see text) and (B) logistic growth constants calculated according to Rickleffs (1967). Both growth rates and K are presented with their upper and lower 95% confidence limits. -48-- A. • GROWTH COHORT  EQUATION  N  R2  GROWTH RATE (g/day) dl  1,  1979 log10Y=0.3161ogi0X+0.701 14 0.64 0.123 g:  1,  1979 logl0Y=0.238logioX+0.872 8 0.39 0.152 or:  2,  1979 log ia=0.30510 g 10X+0.741 18 0.49 0.069 g:  2,1979 log10Y=0.420logioX+0.510 15 0.43 0.097 e:  1,  1980 log 10 Y=0.487log10X+0.428 17 0.78 0.139 g:  1,  1980 INSUFFICIENT DATA B. 95% CL COHORT  ASYMPTOTE (g)  upper  lower 1, 1979 32 0.010 0.028 0.000 2, 1979 32 0.014 0.021 0.007 TABLE 7: (A) Growth equations and growth rates for male and female cohorts of M. pusillus at LAMTO and (B) logistic growth. curve constants (K) calculated according to Ricklef_s (1967). Growth rates were calculated as average rates over the 20-25g weight interval. -49-- ANALYSIS OF COVARIANCE COMPARISON  COHORT  df  F ratio  Probability Imm orvs Imm  1, 1979  1&18  0.37  nsImm ?'vs Imm 7  2, 1979  1&29  0.46  ns Imm ?(cohort 1, 1979) vs  1&27  3.22  ns Imm e(cohort 2, 1980) Imm e(cohort 1, 1979) vs  1&28  0.01  ns Imm e(cohort 1, 1980) Imm  (cohort 1, 1979) vs  1&19  0.78  ns Imm ? (cohort 2, 1979) TABLE 8: Comparisons of the growth rates (slopes of the regression lines) of male and female cohorts of M. pusillus at LAMTO. ns= not significant at p,5 0.05. -50Male cohorts, as with females, reached approximately adult weights by the age of six months (Figure 21). Males under six months had testes under 3mm in length and undeveloped epaulettes. Of 13 males recaptured at ages of six to seven months, 31% were beginning testicular development (testes=3-4mm) and invagination of the epaulettes. All of four males caught at eight months had partially developed testes (31=4.5mm) and 100% of 12 males recaptured at nine to 12 months of age were indistinguishable from older adults. For the purposes of population analyses I assume that males became adult at nine months of age. 4.3.  REPRODUCTION OF MIGRANT SPECIES Neither E. helvum nor M. torquata reproduced at LAMTO; however, captures and observations here as well as at Abidjan and the Tai Forest site provided some insight into their reproduction. The data for N. veldkampi were inconclusive. E. helvum: Dead new-born juveniles were found under the Abidjan roost in February 1979 (the first observation attempted) and in January and February 1980. When the colony arrived at LAMTO in March of both years, a large proportion of the females carried juveniles. During the first two weeks, the juveniles remained at the roost site throughout the night and -could be heard calling at first, then later seen flying between roost trees. Towards the end of March and in early April the numbers of juveniles seen at the roost at night declined until by the last week none remained. At this time the juveniles were apparently sufficienily old to forage with the adults. When the colony moved to the new site 1.5km distant, I could see no females carrying young. During the return migration (November to January) I caught. nine parous females of which 55.6% were pregnant with large foetuses and one (caught In January) was lactating. -51H. torquata: At LAMTO female M. torquata were only caught during the northward migration (May to July) movement and never during the return. Most of the parous females (86.5%, n=37) were pregnant, carrying small palpable foetuses that, on the basis of equivalent sizes with E. buettikoferi and M. pusillus, indicated an August/September birth period. In November in the Tai Forest, 100% (n=26) of the adult females had small palpable foetuses, suggesting another birth period in February/March. During the first annual peaks at LAMTO, 24.8% of males (n=137) did not have the throat ruff of long dense hairs characteristic of adult males, but during the second annual peaks only 2.9% of males (n=107) lacked the ruff. Body weights of males with ruffs caught during the first peaks did not differ significantly from those of males caught during the second peaks (t=0.07, df=62, p>0.05); however, testes lengths and throat gland secretions of these males did differ between the two times of year. Males with developed throat ruffs had significantly smaller testes (5=5.6mm, n=36) than those caught in September to December samples (R=6.9mm, n=24; t=3.57, df=58, 1)40.05). The ruffs of males caught between April and July were clean, dry, and grey, while those of males caught in October to December were covered with a yellow oily exudate. N. veldkampi: At LAMTO there were insufficient captures of N. veldkampi to define any clear reproductive pattern. During the first catch rate peak two parous, non—reproductive females were caught. During the second catch rate peak, three females were pregnant, two were lactating (October), and one was parous, non—reproductive. In the Tai Forest in November, four adult females were captured. Of these, three carried small palpable embryos and had recently been lactating, and one was both pregnant and lactating. The combined data for LAMTO and the Tai Forest suggest a birth period occurring during October and November. 5 .•  •••• J'F ' M'A  M1J 12 77; Eg n=20 11 0 N ID Mp n=42 20 100A'S m Nv n=5 100PREGNANT  LACTATING FIGURE 24: The timing of pregnancies and lactation for E. gambianus (Eg), M. pusillus (Mp), and N. veldkampi (Nv) at Wango Fitini. Figures above each histogram are the sample sizes for each sampling period. -52-- 4.4.  REPRODUCTION AT WANGO FITINI Samples of parous female E. gambianus amd M. pusillus from April, May, July, August, and November were small (20 E. gambianus, 42 M. pusillus), but provided an indication of the timing of reproduction. There were two lactation periods for both species, one occurring in April and May and the other around November (Figure 24). None of the lactating females caught in April (4 E. gambianus, 8 M. pusillus) were detectably pregnant, indicating that this month fell near the beginning of the lactation period, before foetuses were large. All of the parous pregnant females caught between 27 and 30 May (1 E. gambianus, 2 M. pusillus) had been lactating recently and carried small palpable embryos. In November all parous females (11 E. gambianus, 20 M. pusillus) were either lactating or had recently finished and were palpably pregnant. These data indicate that March or April and October or November were the peak lactation months at Wango Fitini. Two lactating N. veldkampi were caught in April and May, coinciding with the lactation periods of the above species. -534.5.  ECOLOGY OF THE SPECIES AT LAMTO 4.5.1. ROOST SITES E. buettikoferi: I located 14 different roost sites, five used by radio-tagged individuals which I monitored for a total of 82 roost days (2 immature males for 1 day each, 1 adult male for 29 days. l'adult male for 6 days, 1 lactating female for 45 days) and nine which I located on walks. With the exception of the lactating female, all individuals roosted in the riparian forest along the Bandama River, either on islands (3 of 4 radio-tagged males) or on the main bank (1 radio-tagged male, nine unidentified individuals). All individuals roosted alone in dense foliage at heights of 2-5m and 11 of the 13 were in trees overhanging and virtually enclosing side channels of the river. The two individuals not roosting directly over the water were within one to four metres of the edge. Each radio-tagged adult male used the same roost each contact day, one for 29 consecutive days. E. buettikoferi appeared to be extremely sensitive to disturbance. None of the nine individuals that I frightened from their roosts returned on subsequent days. The single lactating female occupied the same roost with her current offspring over 45 consecutive days. This site was located 15m above the ground in a broad-leafed, dense canopied tree (Cola gigantea), along the edge of a gallery forest 2.5km inland from the Bandama River. The local Baoul6 people indicated that roosts of E. buettikoferi were most commonly located along the Eandama River and that this species usually roosted alone. H. monstrosus: I located five different roosts, four directly on the edge of gallery forests and one in the centre of an isolated forest patch. All had a common form; that of an open-Qfded enclo q ure, 4-5m in La I n=108 xx Mt n=235 )* CD  0 0 0 0  0 0 0 L()  0) , Cr) 0  .--- CNI (N TIME PERIOD 0 0 CD CD 15 100 EID  E b " ft  n=515  n=219 5 oMp n=284 Hm n=58 150 E3 1 0 0 LU Li-I,_ E-- L_) <C  so LU Li cc 0_ x I— Li200 Li CD — 150 100 50 c) 0 0 0 04 0 0 0 al CN CV 0 FIGURE 25: Activity periods of five pteropodid species at LAMTO. Activity was measured by net captures during two hour periods through the night and is expressed as a percent of the captures that would be expected, given an even distribution of activity. ** indicates a significant departure from an even distribution (1)('0.05). -54height created by a dense overhang of climbers. One roost was occupied by a single adult male. The others held from two to 15 females and/or immature males and at least one had one adult male. Only one roost was occupied for more than one day; however, all those abandoned were roosts where I had disturbed the bats. The single successively used roost held a group fluctuating from eight to 15 individuals on the six occasions when I checked over a three week period. Once abandoned, the roost was never occupied again over the following 15 months. E. helvum: The roost of E. helvum at LAMTO does not warrant any further description. For other detailed descriptions of roosts in West Africa see Huggel-Wolf 1965, Okon 1974, and Funmilayo 1976. OTHER SPECIES: No roosts were ever located for M. pusillus, N. veldkampi, L. angolensis, or M. torquata at LAMTO or elsewhere. 4.5.2. ACTIVITY PERIODS The first captures in nets placed near fruiting trees (where bats should arrive first) were usually between 1845h and 1900h. From observations it was clear that bats began foraging at this time. The last bat • captured was at 0530h, and in 280 net hours between 0500h and 0600h only two individuals were taken. I thus define nightly activity as beginning at 1900h and teroainating at 0500h. Not all species distributed their flying activity evenly througn the 10 potential foraging hours of night (Figure 25). Adult male E. buettikoferi, L. angolensis., and M. torquata all had significantly higher captures than expected in the middle of the night (2300-0300h). In contrast E. buettikoferi females and immature males (grouped), H. monstrosus, and M. pusillus showed no significant peaks in catch rate at any time of night. The mid-night activity peak for adult male E. buattikoferi was clearly linked with their calling periods in thc riverine fore. Males, when -55they called, began between 1915h and 1955h (x=1942h) on "bright" and "dark" nights during the mating season and between 1945h and 2230h during non-mating seasons ("dark" nights R=2011h; "bright" nights R=2204h; see section 4.2.4.). This effectively removed adult males from the population of flying bats susceptible to netting before 2400h on most nights. Males stopped calling for 1.5 to 2.5 hours after 2400h, during which time they foraged. This resulted in the observed peak in captures at this time. For more details of the activity of radio-tagged adult males see section 4.5.9. 4.5.3. HABITAT USE The differential use of the six netted habitats or zones at LAMTO (savanna, SNB, riverine forest, gallery forest edge, gallery forest at ground and under-canopy levels) could only be tested for the resident species. The rapid, high amplitude fluctuations in the catch rate of the migratory species meant that to assess habitat use all sites would have had to have been sampled simultaneously. Sampling one site on a high catch rate night would have artificially biassed it relative to other sites sampled on low catch rate nights. With the exception of L. angolensis, all species used savanna and the two levels inside gallery forests significantly less than expected. When I noted the low net yields early in the study I abandon , ed sampling at these sites. Thus, the main sampling effort focused on the SNB, gallery forest edge, and the riverine forest. Among the latter three .sites, immature female E. buettikoferi, H. monstrosus, and L. angolensis showed no significant preference (Table 9). M. pusillus used the SNB more than expected and all other sites less than expected. E. buettikoferi males and adult: females were captured significantly more often than expected in the riverine forest. The high activity of E. buettikoferi in the riverine forest was created by the calling aggregations of males and (apparently) the visitations of females, not by foraging activity. None of the species of fruits present in the faeces of E. buettikoferi were ever found in the riverive forest (eee section 4.5.4.) and the proportion of individuals voidi-Ig faeces when caught was significantly lower at this site than at others -62-- SPECIES  MEANS OF RECOGNITION ANACARDIACEAE Mangifera indica Linn. CUCURBITACEAE Momordica spp? EBENACEAE Diospyros mespiliformis Hochst. EUPHORBIACEAE Bridelia ferruginea Benth. LOGANIACEAE Anthocleista nobilis G. Don MORACEAE Chlorophora excelsa (Welw.) Benth. Ficus capensis Thumb. F. leprieuri Miq. F. lyrata Warb. F. ovata Vahl. F. polita Vahl. F. scott-ellioti Mildbr. & Burret F. thonningii Blume F. vallis-choudae Del. F. vogelii (Miq.) Miq. OBSERVATIONS, PULP, ROOST SEEDS ROOST, SEEDS OBSERVATIONS, PULP ROOST, SEEDS OBSERVATIONS, SEEDS GERMINATED, OBSERVATIONS, ROOST OBSERVATIONS GERMINATED, OBSERVATIONS GERMINATED, OBSERVATIONS, ROOST OBSERVATIONS OBSERVATIONS OBSERVATIONS GERMINATED, OBSERVATIONS, ROOST GERMINATED, OBSERVATIONS PAS SIFLORACEAE Adenia cissampeloides Harms.  GERMINATED, OBSERVATIONS A. meigei Ake Assi  GERMINATED, OBSERVATIONS, ROOST Smeathmannia pubescens Soland.  GERMINATED, OBSERVATIONS RUBIACEAE Nauclea latifolia SAPINDACEAE Pancovia bijuga Willd. SOLANACEAE Solanum torvum Swartz S. verbascifolium Linn. STERCULIACEAE Cola gigantea A. Chev. VERBENACEAE Vitex doniana Baker MYRTACEAE Psidium guajava Linn. ROOST, SEEDS OBSERVATIONS GERMINATED GERMINATED OBSERVATIONS, PULP, ROOST OBSERVATIONS, PULP, ROOST OBSERVATIONS, ROOST, SEEDS TABLE 13: (A) Characteristics of some of the plants commonly eaten by the Pteropodid bats at LAMTO. Fruit positions were judged subjectively and (F) and (C) stand for free or cluttered fruit positions respectively. All measurements arein mm. (B) The numbers and percents (in parentheses) of species with given characteristics. Under fruit colours, brown and blue were classed as dark. CTI 3 0 ,--1 , - 4 W a) w P H CD A  o 0  H ?-1  H 0  a) U  ›-1 0 )-4 .*.1  w w 0  P rr.  H 0 44 0 •-1 b0 "  N4 I •rf '1J 1.11 <C 4  cc, H Cl) c C 1 3 Tzi H cU 0 cj 4-1 <  co •r4 00) X 0 ct 4., 01 0 4-1 H r-I cv ca r  ! 4J 1 4 J 4-1 r-I 1 o H bC u o I -r-i . .ri H p 4 W D 0 [tI P ›, <4  C2, Z P1 4H 2 4 74 .) zo Ps]  0 .0  S . 1 <g  0 Cl) 7a —: P r t 4 - 1 o ' H  4-1 .. -I  () 0  C.) 1 rq 0 .64 0 CD E r) 0 c) CIO  0 H Cm)  rza ',4"4 tza c ..D 0 0 C) 44 • 44 e 44 a l co b  • 4414 • rT. C ; U•r' Q) 4-1 —64— c s a0  VD  r-t  .—I •—I  1-1  t--I tr) Cl) PI un CD C) CD cD 0 0 00 CD c) CD c) In CD u l CD CD c) CD v) CD CD UI ,4  ,4 ,4 4.4 4-1 1-I 4-.1 4-.)  4-1  4I 4I 4-1  44  4..)  4 -)  4-)  4 - I 4-3 a) cu W ww  a)  w  a)  a)  0)  a l  a)  W  a) w w n ••• w 3 w 3 w 3 CU  (11  a.)  0. 1 a)  CD  1)  cl)  a)  0) 3 3 3 3 3 3 3 3  3 _3_ a) 3 U) V) un cil (1) CD En C/) co V) V) cr) c/) Cl) cr) Cr) un  cD CD CD u 1 N. CD 01 Cl) 0C.1 CV n 4 0 /  ,t -4.1 n1  vD ol n - n n - ▪ •  e-, ▪ rx4  Pr - i C x 4 r x4  Cif • C14 $ - 1 •  P PPPPPP mummom aimmo H H H r-1 ,4 4 . 1 r1 i • -1 4 — i *1-4 °T4 (144-U44. 3  3  3 c•-• c•-• g 0 a) g 0 0  o00  0o  0 w  r-I  W  0  0 r-I  1--1 Q)1:1 1-1  CL)  Q)'  CU  r-4 1:1  1-1 P PP ,4 PI P pa P0)  ()  1-i  P  CO  P  a)  0)  (1) 0  ›-i 0 0  0 P1 ';,-, W R  PI Z ,..1 0 4-)  cr)  4 - 1  41  4 - 1  41  4I  41  En  4-1 Cl) 0 m .--,MMWMWM"---. M WW M W 0.)  a)  a)  W  0.)  Q)  w  (1.1  co  W ta0 Ti P o P. M b0 ,li P tIO P P 3-1 P P P ttO P 0 lia. o o o 0 0 o-o o P4 44 cn 4.1 rr. LT4  Fx4  IL'  C x 4 44  Fr-i  Lt.'  41  rz4 Q.)  al  a )  0)0)  W P a) 4-1  4-)  41 4-1  4-1  Li !•,-,  >-. -.0 0 4-4 o ww w cu CI) 41  c1-  ,s  4  ..0 43)  a)  wP. cc . W  a w  W 4-4 4-1 4-I  W4 4-  4-1  W  4-1 H PP P P P (P, c, (P  C, G4 P  fa. CD H H H H [-I rx1 41 1'4 H 4.1 CL1 H PL1 •••••1 r•-1 in  C> c s 4  C•1 0 Lf) Pzi 0 A p r; 1 n •••• n CO Cr; c's1 lf) Cssi n C, CO c'n csj C r ) Cs) 01 a) LI C.7 0 Pe. LI a) rI C., a) 4.) 9-1 fa, LI -65- •4" i n In Cl) Cl) rcl N P:1 in 0 CN.1 9-4 Les i (NJ 0 w  4 - ) 4-1 4J  )4 /-I  4-1 H a) Q) Q)  4-4  4..)  CV ( 1 )  a) Cl)  a) cl) a)  a)  a)  4..1 4-)  a) •t 4  3 3 3  aJ  a)  1.1 4-1 E-1  VD Cl) Ci)  Cl) Cl)  cn  Pa PC) r/3  0 0 0 N  ,--I LI) 04  0  0  0 0  0 H  lf)  H  —4-4  CY) CID •-• ..",.. >4 ria ,•- n ......, ,..... p 4  ••-•  /-•  ••••. e-s "....,  1.4  0  rx4 I:14 U) U)  ...-.  ..."  ..... ...... Z  0 0 ›.. CD  0 0 I-4  >1  1-4 r-I H  H H il3  P  $.4  CI 0 H 0 0 I-I  Cd  23  CI 0 H  -0 "i H  H  H  4-4 •r4 Cl)  0 0 • r-S  r-i  1--1  151 4 a O a) a) N  4-I  •,-- n 44  Pn P-I •d•Q 4  4  a1 a) H Ei g  3 0 o g OW,--( .-I  a) ,--c 0 O f-i a) P 0  0 ›- 1 pa 3  3  --) 0  0 0  ij H  a) )-I H  a) 1-1  a)  H 0  1-4 a)  r4  m 0 -0 -0 O 0 O 0 O 0 1-4 1.4 Z  CU  1:)0 bt)  Cu o 4 - )  5 H  ^C7/ -0  co ,  .-, H 0) 0) CUCU  a)  0)0)  a)  al < 4  la0 t).0 b.°  b0  P 1-I  $4  1> C.,  10 71 MI  C -ii  CV  0  al o w 1:4 [4  PO  Pa Pa  44  Cl) 14 $4 S.4 0)0) .0 X) , r 3  ..0  .A3  X) Xi '4  0 5 0 •rt • H  I-I H  0 0 Li  0 0 1 ... 1  )-4  a) W  a) I) O r-1 r-I -0  ,.0  -0  4 4  14  s..1 c.4  0 0 cn  co  Cl)  Cl') co  H  E-1 co Ca'0  a)  a)  -0  c.)  •,-)  co  0  (1)  as  fl  H  XI  4-1 a)  I - I  (1)  •-4  Vi . 04  04  0  bD  E t--1 '',  1E1  4-I  0  7-) 0 1  1-)Cu  2/  •r-i  '--..  > L I-4  1  Cd Q 4.4  C.) CO  •,-) ca  0,  4-1  Po v-i  W  •= ..0  rm. ') 0c)  0 ,_, -, .,  7-4 • ,-4  tx1 4-! ▪ .  'c.' •,-1  • ,-1 • ,-1 0  10 i-'.  rx-2  -,z  4) Cl)  <4 cO  14 0 E.=.1  cO  -4b  c....) 0 I.-)  tio 5  4.1 cc i -< -4  u 5 ..,- ...  . 1 .,--,  ,. ..,j 44 co r-c .=  00)  0x  ---4 0 s-c  to  H . .-1 ci) 4J  . s: .-+  ..-_ 7 _,* o  ' ,2- . 0 0  0  1-7-1 ::<  cf") 0 Ei 03  HC)  I--1 ci  -ct 03 ::,  0  a'  WO)  al  ill Z..-3H  [= - 1 ,--t  CJ  - , --r:i -d • Fi  CLI  !Id 29  00 •  HO  PA .1-I 04  cn  ç:2',  in P6(  co in W  cn c...") -66-- variety of characteristics, such as: smaller or larger seeds, often pronounced roughening or pitting on the seed coat, often'a mild to pronounced longitudinal ridge, often,an asymetrical tapering of one end, and one species had a gelatinous seed coating. Since part of the following analyses depended on my ability to correctly classify the seeds that I encountered and also on a knowledge of the characteristics of the parent plants, I planted and grew 50 Ficus A and 35 Ficus B seed groups. The 50 germinated Ficus A were represented by 40 F. capensis and 10 F. vallis-choudae. This confirmed my previous Impression gained from visual comparisons of the seeds. F. capensis and F. vallis-choudae are both free-standing tree species commonly found along gallery forest edges and both are invasive species in the SNB. Of the 35 Ficus B seed groups planted, 20 had a gelatinous seed coating and 15 did not. The former group consisted of a single yet unidentified species that was clearly epiphytic in form. The latter group of 15 Ficus B consisted of three species: 10 F. ovata, four F. vogellii, and _ one F. lyrata. All these species are common epiphytes in the gallery forests at LAMTO. F. lyrata and F. ovata produce large fruits (ovet 2cm diameter) and were collected only from H. monstrosus faeces. F. vogellii produces small fruits (less than lcm) and was found in E. helvum faeces. It is impossible to determine the entire complememt of Ficus B species, but from observations of bath feeding, it certainly also included F. leprieuri, F. polite, F. scott-ellioti, and F. thonningi; all of which are also epiphytes. Based on the data, I consider the groups of species comprising Ficus A and Ficus B to represent two contrasting life-forms; the former being free-standing trees of the gallery forest edge and SNB and the latter being epiphytes found in the forest canopy. For simplicity, I will refer to these groups as one species each although they clearly represent more. Curves representing the increment of fruit species found in the diet with increasing sample size (Figure 26) for E. bvettikoferi, 30-  • 20Li LU 30 60 ' 90 120 10 180 210 240 270 300 Mp Mt Eb 8-  ,/ • .* .  .' .. ../  .' . r .  .  .  ./ 10  '  1  1 15 20 25 30 ' 35 40  '  ' 45 50 55 60 65 70 715 No. OF FAECAL SAMPLES FIGURE 26: The cumulative number of fruit species found in the diets of E. buettikoferi (Eb), M. pusillus (Mp), and M. torquata (Mt) at LAMTO, with increasing sample size. 4- -67-- M. pusillus, and M. torquata all plateau before the end of sampling. This indicates that few new species are likely to be found in their diets at LAMTO. For the remaining bat species, continued work would likely add new fruit species to their diets; however, the dominance of a small number of fruits means that the addition of rare species would not substantially change the conclusions. Seasonal Fruit Availability Only 18 of the 34 fruit species were found in more than three faecal samples (Table 11), and these gave some indication of the seasonal abundance of fruits. Thirteen of these species produced fruits during six or fewer months of the year (seasonal fruiters), two species produced fruits during seven to eight months (extended fruiters), and three species produced fruits year round (continous fruiters). Among the 13 seasonal fruiters, seven species produced their fruit crops only during the wet season, four species produced fruits late in the wet season and into the dry season, and only two species ripened fruit during the dry season (Table 14). The net result of the various seasonal, extended, and continuous fruiting strategies was to create a year-round fruit supply. Only a single two month period had as few as six species of fruits available (January/February) and the rest had eight to 12 species. This, however, over-estimates the abundance of fruits during the height of the dry season (late January to mid-February). Bridelia ferruginea fruits were depleted by early January of both years, Chloronhora excelsa never ripened fruits before the last week of February. and D. mespiliformis fruits were only found in early January. During the height of the dry season only three species continued to produce fruit; the free-standing and epiphytic Ficus species and Solanum verhascifolium. The latter species only produced a few scattered fruits during this period (although I could always find some plants with fruits) and had its peak fruiting season doring the wet season. + + + +  +  +4  ++ 6 8- >-; P4 >4 P4 >4 i. >-1 P4 A t--4  1--1  -..... A •.--.. A ...., A ....... (.7 Z H Z cr3 <4  EI o  --- cta  CO <-4  - . 4  43 H rzl A rza A Z Txl H r4 H A 44  f xl C;) Z 4.1  W H (/) PEI  H cia  Cl)  cn < 4 P P P VI 1 .3 >-I  H r:4 A Ej H H >4 w >-I p4 A E - 4  E-I  E - 4 V.-I  1  fxj ,S El H  E - 4 rxj 4.1 cz) ++++ +4 0 .-I CV  n n /4 U .--1  Qt-1  4-1 1-1  CL)  c.J CJ 1.4 .0 4-3  (I)  4.3 to P4 ,0  .c4) 0 "r 4 Co) "r-1 0 0 44 • 71 4-1 '0 0 144 w o "0 04J 0 0 11 44 CO  CO  (I) 4.) H CO 0. C.) r - i  0) r4 CU  0  C.)  ..; t'S  C.)  4) 9-4  rd 4J Cr)  4-1 4-1 F:).  0 '4-. ..1  C.)  CO G) 3)  0 C.)  Ca  .1 - - )  C-I ' 0 1 4-4  co •r4 "ri  r.) (A  1-1  "1-1 CD a) '0 14 4-3,11  0 (1) 4-4 41 s r( 4 - ) O .-4  4-4 t-I  tl)  ri t4-I  F_E1 C., 0t44 4-1 4-1 O '0 r.. . '0 0 0 co  0  (x) co a) 0  Ckt 4-I  0  4- .1 a) 0 4 . 4  Cl) Eil CU  o a. co o 4.a co  co  ;-.1  LI)  L'.4  '.. CO  El .--1 Cr)  0 cd 4.) 4-1 . H r- . 1 '0 O 0 0 (J) 4) .0  4 CU  0,  (1) C.) co  • -0 O 43 ..-1 Cl.)  cd  CO 0)  00 0 4-1  .., 0 p  C!,  .)-) 0 P4 rZ r..1  4..,  l'..3 0 .ri 4)  co ,--1 O 41 0 rt1 Q 4-4 ,0 Ul Cl)  Cd  r--1 tt 0 EC)) 0 "1-1 0 U CO 'd 7.4 Pc, CI ' 4 - I Q) W -4 .n.  $.4 H + 00 <4 +4+  ++  4+ a. + + +  +4 0 r-4 1-4 P-1 ++ +  +4 CO C4 PI + + +  + n C) <4 04 N 1-.1 ii (15 Cl) I •r-t 0 4-4 a) rta 4-4 0 ca a) ci 1 t cii cr) C.) '0 r-1; 0 Z 00 {.-4 •,-( H • r -1 ct1 0 CZ/ Cii Cr) Cl! C) (1) n , ,c. - H -11 ›• <4 VD 14 H cia F.4 ra-1 Cl) Cl Ci rx4 CC Opt a) n -• E , ci CO 1  0 r-1 Cli .H  0 P-r-t 5 ci t-1 0 • Cl) (xi r1 C.) r1 4 13 • n -+ CL) 03 4n ri) ---1 0 ci ci ;-4 0 4 Of, 71 CU 0 _0 0 C., ciiJ 0) cii ci 0 be to rd r-1 0 (0 cD J./ cc,  Q.) 0 5 CC  'A 4-1  0 -1-1  1-1 Ca  0 .r4 0 .H cii 1-4 1L4 11: ra :=V1 0 14.4 Cii ci cii › ca •f--) e-,' 5 17) C :1 4 ci rri 5 cr. ca T.) 4 J ot-{ 1. en C.-) N P4 C': cx., 0 E9, crN 0 co -69- .0 ti-)  co  1"-- ).0  Cr) Cr)  csi 0 0 0 4: 0 0 cN) c q -X  (NI cNI  in 0 co *  tn Lt)  4:  CV CV O. CV 01 * CV CV C()  CV  C r )  *  C .. 1  C'.1 0 *0 0 0 0 * C r ) -4' CV CV NCY. )0 CV .0 CO U . ) CN 0 IC C r )  Cr) IC Ls-)  * 0 0 a) 1-1 CD a) 0 g H r: 0. co 0 c.) cr.) a) 4: co Cl) •1-f (/) Ca E-4 4-t 0 $-1 cc •  • H 1-1 Cl) 1-1 0 co 0 0 4./ • A. Eb  Mp  Hm-  La  Mt  Eh(N)  Eh(S) EVENNESS  0.67  0.75  0.79  0.77  0.29  0.25  0.65 B. Eb  Mp  Tim  La  Mt  Eh(N)  Eh(S) Eb  0.71  0.38  0.42  0.12  0.16  0.26 Mp  0.28  0.38  0.09  0.09  0.18 Tim  0.33  0.08  0.01  0.13 La  . 0.48  0.14  0.09 Mt  0.05  0.11 Table 16: (A) Diet evenness and (B) overlap values for the pteropodid bat species at LAMTO. Bat species designations are as in Table 1. Eh(N) and Eh(S) represent E. helvum on the northward and . southward migrations respectively. -75-- The E. helvum/H. monstrosus Pair Although E. helvum and H. monstrosus foraged in the same forest canopy zone early in the year (northward migration period), they selected different fruits and had the lowest diet overlap for any' speciespair (1.1%). Partitioning within the zone appeared to occur on the basis of fruit size and was mediated by distinct morphological differences. The conspicuous feature of the diet of E. helvum was the small size of the fruits that it exploited. On the northward migration E. helvum relied primarily on C. excelsa, a tree producing fruits approximately 25x8mm in dimensions. On the return migration this species used B. ferruginea, a tree producing fruits only 6mm in diameter, for 50.0% of its diet. The limited germinations of Ficus B seeds from the faeces (4) were of all of F. vogellii. On 12 nights I observed E. heJvum feeding on F. leprieuri (5 nights), F. thonningii (5 nights), and F. vogellii (2 nights). All these Ficus species produce small fruits under 15mm in diameter. When feeding in these trees E. helvum did not remove the fruits from the tree, but merely fed by crawling along the branches and consuming the fruits in place. Their remarkably long legs and thumb are well adapted for climbing and make E. helvum extremely agile in trees (see also Kulzer 1971). It appeared to be this morphological adaptation to climbing that made this species capable of efficiently exploiting small, dispersed fruit packages. Eleven Ficus B seeds from H. monstrosus faeces, when germinated, proved to be F. ovata (10) and F. lyrata (1), both of which are species with large 30-40mm diameter fruits. In addition, on over 20 occasions I observed this large (and readily identifiable) species fly into and remove fruits from F. ovata, F. scott-ellioti, and F. polita trees emergent above the gallery forest and riverine forest canopy. All these species produce large fruits 30-40mm in diameter. During observations at fruiting F. leprieuri and F. thonningi trees which were being visited by E. helvum, I never observed H. monstrosus. All evidence indicated that H. monstrosus used only the large fruited species and I suspect that this relatively clumsy climber could not have effectively handled the smaller fruits. -76The L. angolensis/M. torquata Pair L. angolensis and M. torquata had the second highest overlap (47.7%) among the species pairs. S. verbascifolium constituted the single most important fruit species for both bats, with L. angolensis using it for 43.8% and M. torquata for 89.5% of the diet. Daring the dry season et LAMTO M. torquata retreated south to the forest zone and the two species were geographically segregated. During the January to April dry season period L. angolensis continued to use S. verbascifolium, but spread its diet among other non-Ficus species (S. verbascifolium=20%, others=80%, n=10). When M. torquata moved through the community with the onset of the rains and the peak fruiting period of S. verbascifolium, it is unlikely that food was a limiting resource and that the two species were in serious competition. ALTERNATE FOOD SOURCES None of the 16 unrecognizable net faeces and 39 unrecqnizeable roost faeces had any fragments of plant walls or insect exoskeletons to indicate that bats had been feeding on buds, leaves, or insects to supplement their nectar, pollen, or fruit diets. In 90 faeces containing seeds I did find numbers of lepidopteran scales. Surface swabs from ripe fruits, however, produced large numbers of scales. Controlled feeding of such fruits to captive bats demonstrated that these scales turned up in faeces in approximately the same proportions as was observed in the net faeces. It is unlikely that the bats actively hunted and caught moths. Among 85 Ficus faeces that I examined, 35.3% contained some remains of fig wasps (Agaonidae). These were potentially an alternate source of protein. I examined 50 F. capensis fruits at the stage of ripeness when they were removed by bats (first or second day after softening and turning orange/red, easily detached) to determine the incidence of infestation. Sixteen percent of the fig syconia were still infested, having a mean of 6.4mg fresh weight of wasps. At a level of 18% protein by fresh weight (Morton 1973), this represents 1.2mg of insect protein In an infested fig. -774.5.5 ABUNDANCE OF BAT-EXPLOITED FRUIT SPECIES GALLERY FOREST EDGE Table 17 shows the numbers of F. capensis and other fruit species known to be used by bats along five transects totalling 3.6km of gallery forest edge. Along these transects I found nine species: Adenia cissampeloides, A. meigei, Anthocleista nobilis, F. capensis, F. ovata, F. vallis-choudae, Nauclea latifolia, Smeathmania pubescens, and Vitex doniana.  The numbers of the latter two species were under-estimated since they often occurred in the savanna near the edge, but not forming part of this community and so were not counted. The census shows that bat-exploited fruit species are common. A bat foraging along the gallery forest edge would have to move a mean distance of only 18.4m to encounter another potential fruit source. However, all the species were seasonal fruiters with the exception of the Ficus species and this under-estimates the real inter-fruiting tree distance. Only F. capensis produced abundant fruits on a continuous basis and this was the most reliable fruiting species. The mean inter-tree distance for F. capensis was 54.9m and there were a mean of 21.5 trees/km of edge. From large scale maps of the area I estimate . there to be 140±6.7km of edge within the 2500ha reserve. On this basis the total number of F. capensis can be estimated at 3010 trees along the gallery forest edge. SNB Table 17 shows the results of three 20m wide census transects taken in the SNB at distances of 50, 100, and 150m from the adjoining gallery forest edge. There was no significant difference in the number of F. capensis trees in the three transects (X 2 =1.7, df=2, 00.05), giving a mean of 51.3 trees/ha. On this basis in the 80ha SNB there were approximately 4104 F. capensis trees or 1.36 times the number in the entire gallery forest edge zone. The structural differences between the gallery forest edge (linear) and the SNB (planar) tcgether with the different densities results in Fz.4 1-4 000 In c.) 0-1  0-1 .-4 c ‘ i  er) Pc1 Fa pa Z • Cl)  ri) In tn • •  •  •  . — 1 0000 • CV C r ) * 4" In 4-)  Cl) a) rx1 O 0 ›N (-1 4-1 Cl) H 1-1 110 QS 4 4-) Cl) 4.) C) to 0 0 co 0 .H 0 4-1 . • c 4s.' n-f cd U)  (I) a)  1.) 4-1  a) "0 4-i  Cl)  )-1 O 0 • CL) O 4-1  (ts co a)  CD  Cl) a I 1-1  CI) cu P. 4.1 cri  ti o 4-1 1-1 •  4-f  di U)  c.) O 0 a)  cd CU  P4 4-I . H 4i .r4 co O -7800 cn 0 CD • • • • • 0 c-, cn . 4 . o  ....1-  .-.1  —+  .--1  cn ••- n )-4 CO CD 0\ • •  • • % .0 0 CO ('4 1`. Cr) CV in 0 N. 0 1`, • •  •  • .0-  '00 cn CV 0 1/40 Ifl cf) 0 in CO N. %.0 N. 0000 0  •  •  • • 4" Cn 0 'lir ) ‘.0 CO CI -- I-. '  e-% ••- n CV C r ) In • •  • C r ) VD ON 1/40 0.1  CV CV In C 4 1 Csl " 4 r•-• • •  •  • 00 C4 COO CNI - -4  ce) N4 (NJ 0 In CV CO C r ) In )-4 ce) • O CV CV  e-1 ‘C)  in n 0 1/40 r-) C r ) C V CV cr) • •  •  • 00 0 000 CV 04 cv 000 0 Ln cr) 04 04 • • • —1 • • 6 CN lindd HUM S33d1 dO FIGURE 29: The percent of F. capensis trees in the SNB (n=54) and along one gallery forest edge (n=46) that had some ripe fruit on a given census tour. • -79-- greatly different inter-tree distances. If trees were distributed evenly through the SNB, then each tree would have had six equidistant neighbours, each 15.8m away. The trees approached a random dispersion (not quantified) in the SNB, but the mean distance to the six nearest neighbours remains 15.8m (although the variance on this mean would be high). In contrast, the mean distance between a tree and its six nearest neighbours in the gallery forest edge was 109.8m or 6.9 times greater. 4.5.6 FRUIT PRODUCTION OF F. CAPENSIS Although individual trees did not produce fruits constantly, the population of trees monitored regularly along the gallery forest edge (n=46 trees) and in the SNB (n=54 trees) produced fruits year-round (Figure 29). On 74.5% of the twice weekly tours along gallery forest edge (n=47) and on 98.8% of tours in the SNB (n=8I) there were some ripe fruit. There were significant differences in the rate of fruit production between the two habitats. Significantly fewer edge trees had ripe fruit on any tour (x=6.6%) than SNB trees (x=23.4%; sign test, p<0.01). This was due to the fact that individual edge trees had fewer cycles of fruit production than did SMB trees (Table 18). For analyses I have defined a production cycle as the presence of one or more fruit on a tree on two consecutive tours (or three or more ripe fruit on only one tour) bounded by four tours (two weeks) without any ripe fruits. In the SNB only 16.7% of the trees (n=54) never cycled and 53.7% of the trees cycled two or fewer times. Along the gallery forest edge 47.8% of the trees never cycled and 91.3% cycled two or fewer times during the same period. The form of trees also differed between the two sites. Along gallery forest edge only 4.3% of the trees had more than one stem, while in the SNB 68.5% of trees consisted of clusters of two to eight stems. I never determined whether these stems shared a common root systems or consisted of separate individuals. However, there was no significant correlation between the number of stems and fruit cycles (F = 0.15, n-51, p>0.05), indicating that stems did not cycle independently. % OF TREES CYCLING X TIMES # OF CYCLES  GALLERY FOREST EDGE  SNB 0 47.8 16.7 1 21.7 14.8 2 21.7 22.2 3 8.7 14.8 4 0 25.9 5 0 3.7 6 CONTINUOUS 0 1.9 Table 18: The frequency of fruit production cycles among 46 F. capensis trees along one gallery forest edge and 54 trees in the SNB at LAMTO between March and November 1980. -81— There was no clear relationship between fruit'cycling and. stem diameter. Along gallery , forest edge the number of cycles was significantly correlated with stem diameter (F=6.0, n=39, p40.05), but diameter explained only 14% of the total variation in fruit production. In the SNB there was no significant correlation between stem diameter and fruit cycling (F=0,07, n=17, p>0.05) for those trees with only one stem. FRUIT BIOMASS While the total biomass of fruits produced monthly or yearly cannot be estimated from the census tours (some fruits were ripened and removed between tours), the number and biomass of fruits available to frugivores on each census day and night can be estimated. GALLERY FOREST: Along the gallery forest a mean of 6.6% of trees bore fruit at any given time. Trees with some fruit had a mean crop of 10.4 fruits (n=146 tree days with fruit). Thus of the 3010 trees in this zone, 199 bore a mean of 10.4 fruits on a given day, for a daily crop of 2070 fruits. At a mean weight of 21.1g/fruit (n=58) this represents 43.7kg fruit/day in the entire zone. The production, however, often dropped well below this level. SNB: The proportion of trees bearing'fruits on a given census tour varied greatly, since the trees appeared to be roughly synchronized in their fruit production (Figure 29). Over the year a mean of 23.4% of trees produced some fruits on a given day. Each fruiting tree bore a mean of 8.3 fruits (n=1024 tree days with fruit). Thus, in the SNB an average of 960 trees bore 8.3 fruits for a mean daily crop of 7968 fruits. This represents 168.1kg fruit/day. During low periods in fruit production (January, August/September on Figure 29) a mean of 8% of trees had fruits. At these times the daily crop was 2725 fruits and the daily biomass was 54.8kg. This decreased availability, especially during the January period when few alternate fruits were available, appeared to be of key importance in governing M. pusillus populations in the SNB. -824.5.7 MOVEMENTS AND POPULATIONS: MICROPTEROPUS PUSILLUS MOVEMENTS During the study I made 411 captures of M. pusillus (Table 19). Of these, 178 were of individuals that I marked, 41 were of individuals that I released unmarked, and 192 were recaptures of previously marked individuals. Ninety-seven (54.5%) of the marked individuals were recaptured one or more times (Table 19). Bats originally marked in the SNB were no more or less likely to be recaptured than those marked elsewhere (X2=0.0, df=1, p>0.05). All sex-age classes had the same recapture rate (X2=0.1, df=3, p>0.05), indicating that immature and adult bats died or emigrated from the study area at the same rate. The movements of M. pusillus were strongly affected by the SNB. The site fidelity (measured as the number of recaptures at the original banding site vs at a new site) of individuals originally marked in the SNB was significantly higher than for individuals marked along the gallery forest edge (X 2 =87.5, df=1, p<0.001; Table 19). Of 151 recaptures of individuals originally marked in the SNB, 91.4% were at the same site and only 8.6% were outside this habitat. In contrast only 19.5% of recaptures of individuals marked at gallery forest edge sites (n=41) were at the same site. Thirty-three of these recaptures were at new locations and 87.9% of the movements were into the SNB. These movements suggest that the population of M. pusillus was based in the SNB with only a few individuals making temporary excursions into the surrounding area. This approximates a closed population occupying what could be considered essentially an island habitat - the SNB. This justifies a more complete mark-release-recapture analysis of recruitment and population size. POPULATION STRUCTURE When bands were in short supply, I preferentially marked immature males and females, so the figures in Table 1913 under-represent the true A. MARKING SITE INSIDE SNB  OUTSIDE SNB  COMPARISON # Marked  124  54  X270.0 # Recaptured  71  26  '  p>0.05 % Recaptured  .57.3  48.1 B. SEX—AGE GROUP ADULT di s IMMATURE Or ADULT  IMMATURE Y COMPARISON # Marked  44  47  45  42  X2=0.1 #Recaptured  25  25  25  22  p>0.05 % . Recaptured  56.8  53.2  55.6  52.4 C. LOCATION OF RECAPTURES ORIGINAL  AT SAME SITE  AT NEW SITE BANDING SITE # RECAPTURES  N  N  %  COMPARISON SNB  151  138  91.4  13  8.6  X2=87.5 OUTSIDE SNB  41  8  19.5  33  80.5  p<0.001 D. MOVEMENT 10:  SNB A OTHER SITE N  %  N  % 29  87.9  4  12.1 TABLE 23: The recapture and movement characteristics of M. pusillus at LAMTO, showing: (A) The recapture rates for bats originally marked in the SNB or elsewhere, (B) the recapture rates for adult and immature males and females, (C) and (D) the locations of recaptures relative to the original banding site. -89TABLE 21: The recapture and movement characteristics of E. buettikoferi at LAMTO, showing: (A) the recapture rates for adult and immature males and females, (B) the recapture rates for males and females originally banded in the SNB or elsewhere, (C) and (D) the locations of recaptures relative to the original banding sites for males and females. (no) indicates a non-significant X 2 comparison. -90-- A. MALES  FEMALES ADULT TMM. COMPARISON  ADULT INN. COMPARISON # MARKED  155  290  104  110 # RECAPTURED  41  70  X2-0.3  12  10  X2=0.4 % RECAPTURED  26.5  24.1  p>0.05 (ns)  11.5  9.1  p>0.05 (ns) B. BANDING SITE (4)  BANDING SITE () SNB OTHER COMPARISON  SNB  OTHER COMPARISON # MARKED  182  263  X2=15.4  70  144  X2=7.8 # RECAPTURED  63  48  p<0.01  13  9  p<0.05 % RECAPTURED  34.6  18.3  18.6  6.3 Si!. MALES: LOCATION OF RECAPTURES ORIGINAL  AT SAME SITE AT NEW SITE BANDING SITE  # RECAPTURES  COMPARISON SNB  121  99  81.8  22  18.2  X2=46.6 OUTSIDE SNB  68  21  30.9  47  69.1  p(0.001 MOVEMENT TO:  SNB ..][..NEW SITE N  %  N  % 20 42.6  27  57.5 D. FEMALES: LOCATION OF RECAPTURES ORIGINAL  AT SAME SITE AT NEW SITE BANDING SITE  # RECAPTURES  COMPARISON SNB  20  13  65.0  7  35.0  X2=2.2 OUTSIDE SNB  13  5  38.5  8  61.5  p>0.5 (ns) MOVEMENT TO:  SNB  NEW SITE N  %  N  % 4 50.0  4  50.0 -91 - Males appeared to be less mobile than females. The high proportion of recaptures that were in the SN13 (81.8%) indicates that males spent most of their time in this habitat (see also section 4.5.9). This justifies the application of a mark-release-recapture analysis to male . populations. ' POPULATION STRUCTURE When bands were in short supply I preferentially marked immature Individuals. For this reason the figures in Table 21A uncler-revres.eat the true numbers of adults of both sexes. I have used the total captures of 285 adult males, 402 immature males, 166 adult females, and 140 immature females as the data base for the following analyses. SEX RATIO: The sex ratio among adults deviated significantly from unity (X 2 =31.4 df=1, 1)(0.001; Table 22), having an under-representation of adult females relative to adult males. Among immatures, the original data cannot be compared directly since the sexes matured at different ages. Immature females became pregnant at the age of six months and were classed as adults thereafter. Immature males achieved puberty and were classed as adult at 15 months of age. The ratio of males to females among immatures should 2.5:1 (15/6 months) to show an equal sex ratio. The ratio of males to females among immatures did not deviate significantly from that expected (Table 22). Immature W : Adult v7 Ratio: Each female produced 0.5 female offspring • every six months. These offspring matured and were classed as adults by the time of recruitment of the following cohort. Thus the ratio of immature females to adult females should be 0.5:1. The proportions in the ' captures deviated significantly from those expected (Table 22). Adult females were under-represented relative to immature females. Immature ore:Adult ?? Ratio: With an average production of 0.5 male offspring per adult female every six months and with males . maturing at 15 months, there should be 1.25 immature males per adult female in the captures. The observed proportions deviated significantly from those expected (Table 22). Adult females were under-represented relative to Immature males. -92A. ADULT SEX RATIO -  TOTAL IMMATURE SEX RATIO TOTAL OBSERVED EXPECTED X 2 B. 285 225.5 15.7 166 225.5 15.7 451 31.4* 402 387.1 0.6 140 154.9 1.4 542 2.0 (no) ADULT FEMALE ADULT 4  imm. A s : OFFSPRING RATIO TOTAL  ADULT . pp  imm. TOTAL OBSERVED EXPECTED X 2 C. 166 204 7.1 140 102 14.2 306 21.3* 166 252.4 29.6 402 315.6 23.7 568 53•3* ADULT MALE : OFFSPRING RATIO ADULT R IMM. a TOTAL  ADULT ee IMM. gp TOTAL OBSERVED 285 140 425 285 402 687 EXPECTED 283.5 141.7 305.3 381.7 X 2 0.0 0.0 0.0 (no) 1.4 1.1 2.5  (ns) TABLE 22: (A) Sex ratios among adult and immature E. buettikoferi and (B) and (C) the ratio of immatures of both sexes to adult females and adult males. * designates a X 2 comparison significant at p< 0.01 and (no) designates a non-significant X2. -93The sex ratio among adults and the ratio of adult females to immatures of both sexes all show adult females to be under-represented. Adult females appear to be less frequently caught after parturition than expected. This could arise from three causes: 1) post-parturition emigration, 2) post-parturition mortality, or 3) a low tra4bility of adult females relative to other sex-age groups. The first two possible causes would have reduced the recapture rate of adult females relative to immature females. This was not observed. This suggests that for some reason adult females were present, but merely not captured. If there was in fact an equal sex ratio among adults, then the numbers of adult males would be good predictors of the immature population size. Table 24 shows this to be the case. Neither immature female to adult male nor immature male to adult male ratios deviated significantly from that expected. This lends support to the conclusion that adult females were present, but merely less trappable. If this is true, then the real population can be estimated by equating adult female numbers to those of adult males, giving the sex-age classes In the proportions of: 285 adult males: 402 immature males: 285 adult females: 140 immature females. All females represent 38% of this population and adult females represent 26%. Females do not comprise 50% of this hypothetical population (as an equal sex ratio at birth would generate) because by equating the numbers of adult males and females I have assumed that they have equal life expectancies after reaching puberty. Since males remain immature nine months longer than females the surplus of males over females in this immature category (1.5 recruitments) would boost the overall male numbers. While I cannot support this assumption (equal post-puberty longevity) for E. buettikoferi, this has been indicated for a number of insectivorous bats (Bradbury and Emmons 1974, Tannenbaum 1975, Bradbury and Vehrencamp 1976a). That this holds true for E. buettikoferi is suggested by the fact that the hypothetical populations so generated "work" empirically, explaining the observed proportions of immatures in the population and (as we shall see) the recruitment. The, high recapture rate and low movements from the SNB indicate that males which were original1y caught there spent most of their time in the confines of this 80 hectare habitat. Since Tulles and females differ in their movements I have calculated Jolly-Seber population estimates for MONTH (1979/1980) MALE POPULATION SIZE th SE ESTIMATED TOTAL POPULATION SIZE JUNE 160 ± 200 221  . JULY 177 * 107 244 AUGUST 178 *  79 246 SEPTEMBER 167 *  64 231 OCTOBER NOVEMBER 223 *  66 308 DECEMBER 203 ±  44 280 JANUARY 168 *  40 232 FEBRUARY 153 *  44 211 MARCH 148 *  56 204 APRIL 160 th  79 221 MAY 167 *  76 231 JUNE 266 ±  82 267 JULY AUGUST 275 *  98 380 TABLE 23: Monthly Jolly Seber population estimates for male E. buettikoferi in the SNB at LAMTO, as well as the estimated total population including females. SE = standard error. ^ / 'N v... - - • - -- \ / - Ni 400- - 300- - .........41 1.1.1 NI 200i" .7) a: CD a. 100IIIIIIIIFIIIIIIIIii, MJ J ASONDJ FMAMJ J ASOND MONTH TABLE 31: Monthly Jolly Seber population estimates for male E. buettikoferi in the SNB at LAMTO. Estimates are presented ± one standard error. -95The high recapture rate and low movements from the SNB indicate that males which were originally caught there spent most of their time in the confines of this 80 hectare habitat. Since males and females differ in their movements I have calculated Jolly-Seber population estimates for males only. I then calculated the total population based on the numbers of females that must have been present at the sampling times to create the sex-age ratios outlined above. Table 23 and Figure, 31 show the monthly population estimates for E. buettikoferi males and for the calculated numbers of both sexes in the SNB. Between June and September 1979 (Figure 31) the population in the SNB remained relatively stable numbering between 160 and 178 males or 221 and 246 individuals of both sexes. The population peaked between September and November, corresponding with the recruitment of cohort 2, 1979 into the flying population. The population declined through the dry season to a low of 148 to 160 males (204-221 individuals) in March and April. The following increase between May and June coincided with the recruitment of cohort 1, 1980. Due to low sample sizes no population estimates could be calculated after August 1980. No mention has been made of the standard errors of the population estimates. As with M. pusillus, the accuracy of the monthly population estimates can be verified by comparing each observed population increases after recruitment with the expected increase based on the number of adult females estimated to be present during the months prior post-recruitment population sizes. In both possible comparisons, there was no significant difference between the two. The number of adult females estimated to be present predicted accurately the observed .increase in the populations with each recruitment, indicating that the original population estimates were accurate. PRE-RECRUITMENT MONTH  POP. SIZE • PREDICTED POSTRECRUITMENT SIZE OBSERVED POSTRECRUITMENT SIZE  X2 JULY 1979 FEBRUARY 1980 244 211 308 266 308 267 0.0 0.0 X 2 = 0.0 -.96— Assuming a mean weight of 150g for E. buettikoferi individuals (ca. 120g for adult females, 100g for immature females, 140g for immature males, 200g for adult males), then the biomass of this species in the SNB is as follows: POP. SIZE INDIVS./Ha BIOMASS/Ha (Kg) mean 256..9 3.21 0.48 minimum 204 2.55 0.38 maximum 380 4.75 0.71. The mean of 3.2 individuals/Ha in the SNB gave rise to a catch rate of 0.10 bats/net hour. If individuals behave in a similar fashion while foraging along the gallery forest edge, then the catch rate in this latter zone (0.05 bats/net hour) should estimate the density of E. buettikoferi there. On this basis, there were an estimated 1.6 Individuals/Ha of gallery forest edge or a biomass of 0.24Kg/Ha. 4.5.9 RADIO TRACKING Through the study I attached radio transmitters to 27 E. buettikoferi (9 adult males, 12 immature males, 1 lactating female, 5 immature females). For six individuals the tags were fastened by gluing to the dorsal fur. The subsequent recapture of one of these individuals with a sizeable ab A ess under the tag position suggested that this means of attachment caused at best discomfort and almost certainly severe behavioural disturbance. As a result I was obliged to discard all data for these bats. For the remaining 21 bats I attached the transmitters with nylon collars. Four recaptures showed no evidence of abrasions or damage to the bats, but all four transmitters had lost their antennae. This meant that the permanent loss of contact with an individual could not be construed as a permanent movement from the area. The following observations were made on the only six individuals (2 adult males, 3 immature males, 1 lactating female) that were located on more than one night. —91-- ADULT MALES The two adult males were radio—tagged following their capture in the SNB during February (male A) and October (male B) 1980. Both periods were during mating seasons. Males A and B were followed for 10 and 15 days . respectively, the former during a waning moon (days 19 to 1 in the lunar cycle) and the latter during a waxing moon (days 1 to 15). Both males called at sites in the rivetine forest during the greater part of most nights, but male B abandoned calling during the last three days approaching a full moon when most other males had also stopped calling. The roosts for both males were some distance up the Bandama Rivet out of tracking range and I never successfully located either. On all calling nights for male B and five calling nights for male A, they moved directly to the same previously used calling sites where they remained quiet until other males began to call between 1933h and 1945h. On five nights male A first moved into the SNB and spent a mean time of 22.1 minutes feeding before going to the calling site. After arriving in the SNB, male A located an F. capensis fruit and removed it to one of three feeding roosts that it used through the tracking period. It processed the fruit, taking a mean of 15.2 minutes, then either moved directly to the calling site or found another fruit, processed it, and then moved to the calling site. On two nights male A fed on two fruits and on three nights on one fruit during this early feeding period. Males called constantly until after 2400h, interupting calling at mean times of 0054h (male A) and 0159h (male B). At this time each male moved to the SNB and fed for a mean of 114 minutes (male A) and 82 minutes (male B). Feeding activity was restricted to short (ca. 30 second) search flights and longer handling periods (male A: R=22.1 minutes; male B: 7:=19.5 minutes) which I term feeding bouts. The longest search flight was only three minutes and both males together accumulated 123 minutes of flying time during the total contact time of 2124 minutes away from the calling sites. This represents 5.8% of the foraging time spent in flight. Male A used three different feeding roosts over the 10 contact nights, always using more than one on a given night. Male B used two feeding roosts, both on all nights. I only located one feeding roost for each male and so cannot be certain of all the diet items taken; however, all the fruit remains at the two roosts A. ANT CATEGORY FICUS CAPENSIS % FALLEN # TREES NIGHT DAY SNB NO ANTS 77.5 (32.1-100) 14.5 (0-42.9) 8.6 (0-46.4) 19 LIGHT ANTS 74.4 (46.4-100) 13.6 (0-53.6) 7.5 (0-17.2) 9 HEAVY ANTS 62.0 (44.0-75.4) 7.9 (0-17.6) 33.8 (19.3-50) 10 GALLERY EDGE NO ANTS 30.7 (0-57.7) 52.1 (30.7-100) 17.2 (0-27.5) 7 ADENIA CISSAMPELOIDES LOCATION NIGHT DAY % FALLEN # STEMS GALLERY EDGE 81 9 11 1 B. NIGHT 1 FICUS CAPENSIS NIGHT 4 # FRUITS NIGHT 2  NIGHT 3 % CONSUMED 54•5 83.1 91.8 92.6 214 % FALLEN 1.7 5.6 7.4 7.4 17 % REMAINING 43.8 11.3 0.8 T0TAL---231 TABLE 25: (A): The proportion of F. capensis and A. cissampeloides iruits removed by diurnal and nocturnal frugivores or falling unconsumed in the SNB or gallery forest edge. Ranges are in parentheses (B): The rate of removc.1 or fall of F. capensis fruits in the SNB. -105n 1 =9, n2=21, p>0.05). Thus there was no evidence of a partitioning of the two height strata between bats and diurnal frugivores. To determine the rate of removal of ripe fruits, a sample of 231 fruits were marked on the first day of ripening (indicated by the change in colour from green to deep orange/red and the softening of the pulp). Of these, 56.2% had been removed or had fallen following the first night and only 0.8% remained on the trees after three nights (Table 25). "HEAVY ANT" TREES: On trees bearing large ant populations, a significantly greater proportion of fruits fell than on trees with few or no ants. Diurnal frugivores removed the same proportion of fruits on "heavy ant" trees as on "light" or "no ant" trees (U test, z=0.60, n1 = 10, n 2 =28, p>0.05), but bats removed a significantly snlaller proportion of fruits on "heavy ant" trees (U test, z=3.33, n1=10, n9=28, p<0.001). The increased number of fallen fruits were those not taken by bats. On "heavy ant" trees bats removed only 62.0% of the marked fruits compared with 78.5% on "no ant" and "light ant" trees. It is interesting to note that the diurnal frugivores did not respond to the increased availability of fruits by increasing their consumption. Despite the reduced impact of bats on the fruit crops of "heavy ant" trees they still removed significantly more fruits than did diurnal frugivores (U test, U = 0.0,.n1 = n2=10, p<0.001). If ants themselves were responsible for the change in dispersal characteristics (and were not merely correlated with some other factor commonly affecting both, eg. location) then manipulating ant abundance .should produce predictable shifts in the proportions of the fruits removed by bats. To test this I el,,iminated ants on two "heavy ant" trees by removing the colonies and I also located one previously measured "light ant" tree on which ant populations had increased. Trees with decreased ant numbers increased the proportion of fruits removed by bats and the tree with increased ant populations decreased the proportion of fruits taken by bats. Both shifts were in the expected direction. -106Observations of these diurnally active ants suggested that a particular nocturnal behaviour was responsible for the changes in fruit consumption. During the day the ants were distributed throughout the trees and on the ground below. Some proportion of individuals foraged for insects while others collected "honeydew" from coccoideans (Insecta: Coccoidea). I did not quantify their dispersion in the trees, but it was my impression that the coccoideans were most abundant on the fruiting branches and fruits than on other parts of the trees. Ants moved onto these branches, collected the "honeydew", and moved off towards the nest distended with fluid. This continuous traffic resulted in there being relatively few ants on the fruiting branches at any one time.. At night the ants did not forage and large numbers remained on and.around the fruits. On "heavy ant" trees many of the fruits were completely covered and any nocturnal frugivore (bat) would have been forced into contact with large numbers of these aggressive ants. That the bats had no effective means of avoiding the ants if they attempted to feed on ant-infested trees was evidenced by occasional captures of bats with O. longinoda attached to the muzzle and eyes. 4.6.2 F. CAPENSIS ALONG GALLERY FOREST EDGE During September 1980 I marked 243 fruits on seven "no ant" trees along one gallery forest edge. The dispersal characteristics of the fruit crops on these trees was markedly different from those of equivalent trees in the SNB (Table 25). A significantly greater proportion of fruits fell unconsumed along the • gallery forest edge (k=17.2%) than in the SNB (R=8.6%; U test, U=31.4, n1=7, n2=19, p<0.05). Along the gallery forest edge bats removed a significantly smaller proportion of fruits (R=30.7%) than in the SNB (R=77.5%; U test, U=5, n 1 =7, n 2 =19, p<0.01) and diurnal frugivores removed a significantly greater proportion (R=52.1%, U=13, n1 = 7, n2=19, p<0.01). Bats and diurnal frugivores did not differ significantly in their consumption at this site (U test, U=I3, n 1 =n 2 =7, p)0.05). The lower impact of bats on F. capensis fruit crops along the gallery forest edge in September relative to that measured in the SNB during . -107other months may in part be due to a reduced emphasis on F. capensis in the bats diet during this period. Although the total faecal seed rain measured in this zone did not decline over that of previous months (see Figure 32), the proportion of F. capensis in the faeces declined from 100% in May/June and July/August samples to only 23.0% in September. This was the lowest level for any sampling period (see Table 26). September corresponded with the peak fruiting month of A. meigei along ' the gallery forest edge. Faecal samples showed that the bats fed heavily on this species in this month, suggesting that decline in bat impact on the fruit crop of F. capensis was due to a shift in the diet and probably was only transient. 4.6.3 ADENIA CISSAMPELOIDES ALONG GULERY FOREST EDGE marked 100 fruits on a single A. cissampeloides plant in September, when the low bat impact on F. capensis was measured. On this species bats removed 81% of the fruits, diurnal frugivores only 9%, and 11% of the fruits fell unconsumed (Table 25). 4.6.4 FAECAL AND SEED RAIN On transects along three gallery forest edges, collecting sheets set out for a total sampling effort of 9828 m 2 days yielded 121 faeces. In the SNB an effort of 2220 m 2 days yielded 307 faeces (Table 26). In this faecal rain, nine fruit species were identified; namely A meigei, B. ferruginea, C. excelsa, F. capensis (which probably included some F. vallis-choudae), N. latifolia, S. verbascifolium, S. torvum, and V. doniana. All were species known to be used by bats. Only F. capensis was found at all sites during all the sampling periods. The remaining species were more seasonal. The fruit species most commonly used by the migrant bats, E. helvum and M. torquata, were only found during the months that these bats were present. B. ferruginea (listed under "seedless" in Table 26) and C. exceisa were only found in March and November respectively, when E. helvum moved north and south through the LAMTO community. S. verbascifolium was only found in May/June and September to November samples when M. torquata was most abundant at cel  co cn c p •-1 I I I I I I I I 0000 0000CDC0-t • —1 .--1 0 0  c • „  ^cs Q) o  • co 0  CO  u) .•- s s ..--. co cn al ON ,..../ ....., CO CO If  I -0 0 G N I G N I .  • CC ..--. ...-... C') cn r - .. . r--. 5./ 41.„,./ IlliiIIII I  I  I  I  I  I  I  I  I  I I  III  11111  III CO U)  -"*".-. '17  (1.1  Cl) G)  '' '-  n-4  QJ (1) n ,. ..,  0-1 e ..,  c.)  c) cn  ......• .-..  co  a)  a)  ›..,  ca, Cl) 4-4 r e)  En qi a)  cr) a-1 rc) cd . r-i  a) C..) CV O t 4-1  rz.1 co  0 Cu s  •,-4  ca  ril -  ..... 9-i L.) Cl) Cl) 44 O 0 a)  H O 2) r-1 n -1 . -4 04 '0 'Ti 'V -- -4 .1 : 4 . 0 .1-1  Cl)  F-4  E-I  r....., • r_-: 1.1 0 0 GA CO E A E4 iz-1 cs.) c I L( Lt)OD .--, (11 •.../ ..--. U1 ,...../ . 0 N. r•-4 1 , .. CD %0 0 CD CD c y r) CDOC.0 000000 .--. r-- •• • CD 0 0 0 •  • CC p.so"'". 00 00 5. CP%  Cr% u") CD 0 H 1-1 •  • CC z z n -• CO 0,3 CN rf) Cl) 4.1 c1.1 Ci E-4 !LI <1 P=, 0 0 (I) o) E-, 0 0 Cr, P4 (.0 Ci 0 Cl:r ! H cd N cd 0 L 4 - I  21 p . '' -- ' H  r-I ....."  S..../ '-''--S co  crN s.../  %..., ...1. Ci 4-1 0 a) P. 4-1 t4) %.0  0  .-1  r-I  CV  1 r 1  0  Ln  0 03 00 111  CO  CV  1/40 0 0  0  0  C 0  *--i '-' 4 C s i •1.-4 CV 0  CN CO 1 , 4 0 0 0 0 0 CD 0 0 0 0 0 %.0 00 0 0 0 0 0 In •  •  •  •  •  •  •  .../- 0 • 0 •  •  • 0 0 0 0 0  0  0  0 0 0 0 .1-1 cG . 0 e n e • - n CD 01  C r ) ...-.,  •ft..I. .... t  ... . t  m  m  0  ....t e.'s e-s.  C r ) Lo,., • nnn •  •• nn • . m  -•  ..4.  m co  co I-I CJ 0.) cO 1:14 • C o  o  C s 4  r-I  Cn  1/40 0  0  0 0 0 n C) 0.) r--I 0 0  C. ) 0  , - .4  ,-1  GNI 000000(000 OCD n O td) • .  •  •  •  ••  •  •  •  •  ,--4 7:7 0 0 0 0 0 0 0  0  0 0 0 e•- n CI) E-I VD 0 C n 1 $41 0 co  r-I  —I  csi  cn 0 4-i 0 cp o  0 cp 0  0 v-1 s--1 P.1 III I  I  II  I 0 CD 0 0 0 0 CD 0 CD 0 0 ce) II II  I  III •  •  •  •  •  c n ! Q) 0 0 0  0 0 0  C 0 0 bl) /1 11 %.10 bO ata c." <4 I 1 I I I I I I Cn  V) c f )  cN1 es) 0 0 0 rG 0  0 114 CO  in <4 rx-4 rag H 0 C) C) 1• 4 -1  0  0  0 • Ci -4 0 CO 14-4 •rn •rt H Cl) Ci 0.1 a) 14-1 crj rz P. $-1 Cu 0 C.) • • Cl) C=4 FIGURE 32:" (A) The total faecal rain, (B) the rain of F. capensis faeces, and (C) the percent of the total faecal rain that was F. capensis in the SNB and along gallery forest edge through the sampling months. - 6 L.() ''.. . ... - El) (N 0 LI)  (. 1  c E! , D n G 6 i G  m G d ,9 P 8 - do Ava zitm-oN , IC)  ifl) ,---. G  G d AVCI zIAVoN >-- Ct uj UJ Lo < W  CO LO V) . 6 c p o 6  6 ,- -  c, • 1-r)  6 c) N ,--- 0 ..._, -. , ‘--• .a - .... -," - ; -  n Z 0 (f) _ < --) --) 2 < 2 Li. - --) = 1.— Z 0 1 IC) 5 d , a) G G 0 (4) 0 0 0 If) NI 0 I In ,---; G Li) q CD 7- , ..... . ... - .- s., -109— LAMTO. Although S. verbascifolium was found to be fruiting in all months of the year, seeds of this species were not found on the collecting sheets during months when M. torquata was not abundant. On the gallery forest edge transects the faecal rain fluctuated from 0.007 faeces/m 2 day in May/June to 0.015 faeces/m 2 day in November (Figure 30), with a mean rain of 0.011*0.004 faeces/m 2 day. The faeces fell at random on the collecting sheets over the period which they were monitored (X 2 =2.1, df=3, p> 0.05) and there was no evidence of areas with consi:tently high faecal rain. In the SNB the faecal rain fluctuated from 0.032 faeces /m 2 day in November to 0.258 faeces/m 2 day in May/June samples (Figure 30), with a mean rain of 0.148*0.080 faeces/m 2 day. I did not record the distribution of faeces among collecting sheets in the SNB to facilitate a dispersion analysis. Not all the faeces contained seeds. On gallery forest edge transects only 74.4% and in the SNB 97.1% of the faeces had some seed content. The seed rain at these two sites was 0.009±0.004 faecal seed groups/m 2 day and 0.142±0.083 faecal seed groups/m 2 day respectively, it is likely that these values slightly underestimate the seed rain since some seeds were removed by ants. however, the collecting sheets, raised above the ground surface on the supporting grasses, did not appear to be often vised by ants and it was my impression that seed predation on the sheets was minimal. Of the nine frnit species represented in the faeces only F. capensis was present in all sampling months and at all sites. Overall,on the gallery forest edge transects 53.3% (n=90) and in the SNB 94.0% (n=298) of the seed—bearing faeces wer2 from this single species. The propertion of . F. capensis varied seasonally from 100% at both sites to a low of 23.1% along the gallery forest edge in September and 77.8% in the SNB in October. The mean. F. capensis seed rain on gallery forest edge transects was 0.005+0.003 taecal seed groups/m 2 day and in the SNB was 0.132*0.075 faecal seed groups/m 2 day. Given that seed rain was random, at the measured seed rain every square metre of savanna in the zone bordering gallery forests would receive 1.8 faecal seed groups each year. If the rain was approximately rat-Qom in the SNB, each square metre received 48.1 F. caceasis scea groul.s 2ach year. -.110BAT vs BIRD INPUTS During October and November I visited the collecting sheets at dawn and , dusk in order to quantify the rate of diurnal and nocturnal inputs. Since the sheets were set out only under open sky, inputs were from bird and bat origins only. Table 27 shows the results of these samples. At all sites bats accounted for the majority of the total faecal rain (with and without seeds). In the SNB 92.7% (n=55) and on the two gallery forest edge transects 95.5% (n=44) and 92.1% (n=38) of the faeces were deposited by bats. Considering the seed-bearing faeces only, the importance of bats was even greater. In the SNB 95.7% (n=46) and along the gallery forest edge transects 100% (n=31) and 95.3% (n=24) of the seed-bearing faeces were of bat origin. A significantly smaller proportion of bird faeces (33.3%, n=9) than bat faeces (76.6%, n=128) carried seeds (X 2 =8.1, df = 1, p<0.01). 4.6.5 SEED PREDATION Faecal seeds set out alone (simulating seeds dispersed in faeces) or associated with fruit parts (simulating faecal seeds and seeds in rejecta pellets left below feeding roosts) suffered heavy predation during 24 hour presentation experiments. The main seed predators that I noted were ants, but I did not identify the taxonomic groups involved. I often interrupted ants in the process of removing seeds from the presentation dishes and it seemed likely that given an additional few minutes all these seeds would have been removed. For this reason I used the removal of more than one seed from a dish to indicate discovery and assumed that once discovered all the seeds would he' preyed upon. . In 12 trials a significantly greater proportion of "seeds plus fruit" dishes were discovered than "seeds alone" dishes (Sign test; p<0.01). A mean of 91.7% of the "seeds plus fruit" dishes were discovered in 24 hours, whereas only 71.8% of "seeds alone" dishes were discovered in the same period. NOCTURNAL AND DIURNAL FAECAL RAIN OCTOBER NIGHT DAY NOVEMBER NIGHT  DAY OVERALL NIGHT  DAY SNB ALL FAECES 35 1 16 3 51  (92.7) 4  (7.3) FAECES WITH SEEDS 29 1 15 1 44 (95.7) 2 (4.3) F. capensis FAECES 28 0 15 1 43 (97.8) 1  (2.2) GALLERY FOREST EDGE (SITE 1) ALL FAECES 4 1 38 1 42 (95.5) 2 (4.5) FAECES WITH SEEDS 3 028 0 31 (100) 0 F. capensis FAECES 2 0 13 0 15.(100) 0 GALLERY FOREST EDGE (SITE 2) ALL FAECES 7 1 28 2 35 (92.1) 3  (7.9) FAECES WITH SEEDS 3 1 20 0 23 (95.8) 1  (4.2) F. capensis FAECES 2 1 14 0 16  (94.1) 1  (5.9) TABLE 27: The number of faeces falling on ground-based collecting-sheets during the night and day in the SNB and along two gallery. forest edges during October and November. Percents of the total rain arein parentheses. Wolton, R.J., P.A. Arak,.H.C.J. Godfray, & R.P. Wilson. 1982. Ecological and behavioural studies of the Megachiroptera at Mount Nimba, Liberia, with notes on the Microchiroptera. Mammalia, in press. Wunder, B.A. 1975. A model for estimating metabolic rates of active or resting mammals. J. Theor. Biol. 49:345-354. Leck, C.F. 1969. Observations of birds exploiting a Central •1 American fruiting tree. Wilson Bull. 81:264-269.