Diet selection and resource use by flying foxes (genus Pteropus).
Flying foxes of the genus Pteropus Erxleben (Pteropodidae: Chiroptera) are important pollinators and seed dispersers in oceanic-island ecosystems (Cox et al. 1991, 1992, Elmqvist et al. 1992, Rainey et al. 1995, Banack 1998; cf. Crome and Irvine 1986, Eby 1991). Their role is crucial to maintaining community diversity by affecting the regeneration and genetic flow of dominant forest trees (Banack 1998). Although substantial dietary information is available for some Pteropus species (see Dobat and Peikert-Holle 1985, Marshall 1985, Mickleburgh et al. 1992; see Wiles and Fujita 1992 for reviews), few studies offer a thorough investigation of the feeding ecology of flying foxes in a single area. Theories concerning diet breadth, diet selection, and the evolution of feeding strategies in frugivorous bats have therefore been based primarily on studies of Neotropical fruit-eating bats (Family Phyllostomidae) along with support from African pteropodid (Family Pteropodidae, non Pteropus) fruit-eating bats (Fleming 1982, 1986). Fleming (1986) concluded that (1) frugivorous bats eat a taxonomically nonrandom subset of fruits; (2) the year-round availability of fruit is a key characteristic of the fruit taxa, which has allowed bats to specialize on them; and (3) the evolution of feeding strategies in frugivorous bats has involved both a specialization on core fruit taxon and the opportunistic consumption of seasonally available fruits. These conclusions would be strengthened by further comparison with other Paleotropical species, particularly from the genus Pteropus, which includes 58 of the 166 species in the family Pteropodidae (Koopman 1993). Because the majority of Pteropus species are found on islands (Rainey and Pierson 1992) and because island species often differ widely from continental species, a comparison between such groups would provide insight into the general foraging theories of frugivorous bats.
This paper examines the three hypotheses outlined above using data from two Paleotropical island Pteropus species, Pteropus samoensis Peale and Pteropus tonganus Quoy and Gaimard, which coexist in the Samoan islands. P. samoensis is endemic to Samoa and Fiji and is divided into two subspecies. P. tonganus is distributed widely from islands near New Guinea through the Cook Islands and is divided into three subspecies (Koopman 1993). Pteropus samoensis and Pteropus tonganus are similar in adult body mass, averaging 379 g and 428 g, respectively (Banack 1996, Richmond and Banack, in press). I examined resource use and seasonal fruit preferences in relation to fruit abundance and fruit availability. Variations in foraging patterns are discussed in relation to periods of fruit scarcity.
MATERIALS AND METHODS
Research was conducted from August 1992 to December 1992 and July 1993 to October 1994 on the island of Tutuila, American Samoa located 14 [degrees] S and 170 [degrees] W in the South Pacific Ocean [ILLUSTRATION FOR FIGURE 1 OMITTED]. Observations were also made on the nearby islands of Ofu, American Samoa and Savaii, Western Samoa. Research consisted of two parts: (1) direct feeding observations made with binoculars, spotting scope, and a night-vision device; and (2) ejecta-pellet collections taken from systematic transects on the northeast side of Tutuila, in Amalau Valley, between the villages of Afono and Vatia [ILLUSTRATION FOR FIGURE 2 OMITTED]. This area was least damaged by the recent hurricanes and contained the highest density of Pteropus samoensis on the island. In addition, there was a nearby colony of P. tonganus that consistently foraged within this area. The study areas consisted of 135 ha of primary forest on steep grades extending from sea level to 366 m and a small amount (15 ha) of agroforest (multi-crop agricultural land) and secondary forest on flat areas adjacent to roads.
I used two non-intrusive methods to evaluate diet - direct observations and ejecta-pellets counts. Daylight observations were made using Zeiss 10 x 40 binoculars and an Optolyth spotting scope (model TBS-80HDF with a 22-60X zoom eyepiece; Optolyth-Optik, Walter Roth GmbH & Co. KG, Pommelsbrunn, Germany) at vantage points allowing a large block of forest to be observed at a single interval. Focal animal and scan sampling techniques were employed. Over 402 observation hours were logged including all hours of the daylight (no. observations/h = 26.0 [+ or -] 11.0 [mean [+ or -] 1 SD]; range: 23.3-44.4 h). Night observations were aided with a night-vision device but were limited to focal trees. All hours of the night were included in observation periods, with not less than 120 min total in any given hour. Observations included trees located in primary forest, secondary forest, and agroforest. Detailed behavioral observations were recorded noting species, behavior, plant interactions by species, plant part, and duration.
Ejecta pellets are created as a flying fox feeds on fruit pulp, compressing the fruit against the palate with the tongue. This action effectively squeezes out the juices, which are swallowed, and creates a pellet of dry fruit pulp. The resulting ejecta pellets represent one mouthful of fruit and are approximately the same size regardless of plant species. In addition, ejecta pellets and fruit abundance were locally correlated, indicating that trees with ripe fruit were found and their fruit consumed by flying foxes. The number of ejecta pellets found therefore provides a quantifiable estimate of resource use by flying foxes.
While fecal pellets are commonly used as an estimate of resource use (Thomas 1988) ejecta pellets provided an even better estimate in this study for three reasons. First, ejecta pellets could be collected without the need to handle animals. Most analyses using fecal samples require capturing the animals and detaining them for short periods. Although some mist-netting was conducted in concurrent studies, animals were difficult to catch and sample sizes were very low (Banack 1996). Ejecta pellets, on the other hand, were abundant and predictable underneath fruiting trees. Second, most fruit consumed by flying foxes had large seeds, which were not ingested. The task was therefore not to identify seeds within ejecta pellets or fecal pellets but to identify the fruit pulp itself. Plant species were easily identified from ejecta pellets as the fruit was relatively unprocessed. Finally, netting introduces a temporal and spatial bias into the data, being biased toward tree species close to the net and recording only food recently consumed by netted animals (as gut passage time for fluids is relatively short and fruit species are not necessarily eaten randomly throughout the night [see Fleming 1988]). Netted animals were caught most frequently as they left the roost to feed (and therefore had empty guts) or around specific fruiting trees. Ejecta pellets, on the other hand, were collected throughout the forest and were unbiased with respect to time and tree species.
While direct observations almost always permitted definitive identification of a flying fox as either P. tonganus or P. samoensis, indirect observations of resource use (ejecta pellets and chewed fruit) permitted only a partial separation between P. samoensis and P. tonganus. Triangular-shaped tooth imprints in fruit indicated flying fox use (as opposed to rat or bird use). Ejecta pellets, on the other hand, consistently contained complete imprints of six teeth (left and right [P.sup.2], [M.sup.1], [M.sup.2]) and occasionally seven or eight teeth (left and/or right [M.sup.3]). These imprints permitted a separation based on a discriminant analysis of tooth patterns, since P. samoensis has a shorter, broader rostrum than does P. tonganus (Banack 1996). The discriminant analysis supported the assumption that both P. samoensis and P. tonganus utilize fruit from forest tree species to a similar degree (Banack 1996). This uniformity supported combining both species of flying fox found on Samoa in a common analysis of fruit preference and permitted the examination of diet breadth, fruit selection, and evolutionary strategies of foraging of the genus Pteropus. Therefore, the presence of ejecta pellets served as an indicator of fruit use by genus Pteropus with partial discriminating capabilities between P. samoensis and P. tonganus.
Tree species were chosen for investigation based upon daily observations of flying fox foraging behavior, indirect observations of use based on pellets found under trees on forest transects, and previous information gathered on resource use from a literature review and from colleagues. Fruit tarps made from plastic- or aluminum-mesh window screening were suspended under fruiting trees 1 m off the ground and tied to adjacent foliage. Tarps with an area of 9 [m.sup.2] were placed under every tree of a selected species along each of eight representative transects. A representative transect traversed 0.5 km and was 25 m wide, with occasional sharp cliffs limiting width to 5 m. One tarp was placed under trees with small canopies and two to four tarps were placed under trees with larger canopies. Each tarp was placed subjectively in an area with fruit above or in an area with fresh ejecta pellets on the ground below. This was done in an attempt to place the tarp in an area that would be most likely to receive fallen ejecta pellets if the tree were visited by a bat. Tarps were placed under every fruiting tree of a focal species that was accessible from the transect (except for one tree species, Inocarpus fagifer, Fabaceae/Leguminosae). Tarps were placed randomly under I. fagifer during months when a large number of trees in a single area contained fruit. Sample size for each species varied according to availability and accessibility of fruiting trees each month (number of trees per species per month = 9.85 [+ or -] 6.15 [mean [+ or -] 1 SD]; range: 1-35 trees). During months when very few of the trees of a given species were fruiting the sample size decreased to as low as one tree. These samples were included in the analysis of preference since the preference index is a measure of resource use relative to abundance in the habitat. If flying foxes sought out fruit that was rare in the environment, then its preference value would be high relative to more abundant fruit. Although the number of tarps sampled in a month was sometimes low, the overall number of trees investigated for fruit and possible tarp placement was much higher (984 trees examined per month; [ILLUSTRATION FOR FIGURE 6 OMITTED]). In any given month, not less than 5% of the fruiting population of a tree species in Amalau Valley included in this study had tarps placed under them. It should be noted that removing the tree species with sample sizes of [less than]3 in a given month from the analysis does not change the relative position of the other species in the computed preference index (Banack 1996).
Ejecta pellets were collected from each tarp after a 72-h sample period. For each 72-h sample I recorded the total number of ejecta pellets, the number of uneaten fruits, and the number of chewed fruits along with percentage of fruit consumed from each chewed fruit and whether or not the seed was damaged. The percentage of ripe canopy fruit overhanging the tarp and the total ripe fruit crop was calculated either by direct counting or by estimating the total based upon the visible portion. Ejecta pellets brought in from another fruit species by flying foxes and dropped into the tarp were easily identified to plant species based on fruit and pulp characteristics. When the sample size for fruiting trees of a given tree species was [less than]5, two 72h samples were gathered in a single month (a minimum of 10 d apart). The same trees were sampled for consecutive months until all fruit was removed from the tree. Tarps were placed under new trees as they began fruiting.
To quantify ejecta removal from tarps by ants, snails, rodents, birds, pigs, or wind, I employed a system of "removal tarps." Removal tarps were tied underneath the collection tarp so that no new pellets could fall in from the above tree. Fresh ejecta were placed in the removal tarp and monitored for loss over a 72-h period.
The number of ripe fruit per tree were counted directly, then averaged across all trees with fruit for that species in that month. Fruit was considered ripe when 50% of an individual fruit had turned color for those plant species that change color as they ripen. For species in which the fruits do not change color during ripening (Palaquium stehlinii [Sapotaceae], Planchonella grayana [Sapotaceae], and Cananga odorata [Annonaceae]), fruit was considered ripe as soon as it reached full size. These definitions allowed for the best measure of monthly biomass because fruits generally ripened quickly during the course of a month after they began to change color. Flying foxes, however, do not generally consume fruits until they are fully ripe. Therefore, fruit ripeness was marginally overestimated for all species at the time the sample was taken.
The proportion of trees fruiting or flowering were recorded monthly for each sampled tree species, using binoculars on transects and using a spotting scope for phenology assessments of individual trees on steep slopes. Phenology sample sizes varied slightly from month to month but represented at least 8% of the total population in Amalau Valley for each species (44 [+ or -] 25% [mean [+ or -] 1 SD)]; range: 8-100% sampled per month).
The density of tree species within the agroforest and secondary forest was surveyed by counting the total number of trees of each species within a 100-ha area. Trees within the primary forest could not be determined in this manner because much of the terrain was too steep to traverse and some plants vary in density on an elevational gradient. Primary-forest trees were surveyed using a spotting scope from three vantage points. Random coordinates were chosen using a random number table, a compass, and a protractor mounted with a plumb bob on the scope. Presence or absence of plant species was recorded within each field of view. A running mean of relative frequency was plotted for each of three species - Syzygium inophylloides (Myrtaceae), relatively common; Palaquium stehlinii, relatively rare; and Planchonella samoensis (Sapotaceae), intermediate density. Sampling continued until the mean leveled off for all three species (n = 127 samples).
In order to calculate a preference index, relative frequency of occurrence was calculated for both the primary-forest species and the agroforest species based on a count or estimate of the trees in Amalau. This method was comparable to the other methods described above, although some subjective error was present. A regression of the relative frequency obtained by the spotting-scope method and the estimate of the total number of trees in Amalau was highly correlated ([R.sup.2] = 0.87, y = 0.8778x + 0.0289).
I grouped secondary-forest trees and agroforest trees in the same habitat category for analysis. Both kinds of trees are found along roadsides and in disturbed areas. These areas have a high density of fruiting trees and observations of flying foxes indicated that the secondary and agroforest areas were searched as a unit by flying foxes.
A Manly-Chesson preference index (Chesson 1978, Vanderploeg and Scavia 1979, Manly et al. 1993) was calculated for each month and tree species using the proportion of a food item in the diet (d,) divided by the proportion of food in the habitat ([N.sub.i]). This ratio was then normalized so that the sum of all ratios in a month equaled one. Formulas applied were:
[P.sub.i] = [d.sub.i]/[N.sub.i]/[summation of] ([d.sub.j]/[N.sub.j]) where j = 1 to k
where i = 1, 2, 3, ..., k, for k tree species; [P.sub.i] = flying fox preference values for species i when foraging in an environment with k species available; [d.sub.i] = the proportion of fruit of species i in the diet; [N.sub.i] = the proportion of fruit of species i in the habitat; and
[d.sub.i] = [summation of] [[Gamma].sub.m]/[c.sub.m]/n [[Beta].sub.i][[Epsilon].sub.i] where m = 1 to n
[N.sub.i] = [[Beta].sub.i][[Epsilon].sub.i][O.sub.i][[Omega].sub.i]
where [[Gamma].sub.m] = no. of ejecta pellets of tree m of species i; [c.sub.m] = percentage tarp cover of tree m of species i; [[Beta].sub.i] = relative frequency of trees of species i; [[Epsilon].sub.i] = percentage of trees in fruit of species i; n = the number of trees examined of species i; [o.sub.i] = mean no. of ripe fruit per tree of species i; and [[Omega].sub.i] = mean mass of one fruit in grams of species i.
A yearly mean preference index was also calculated using the formula listed above but pooling all trees sampled for the year. It should be noted that it is not possible to calculate a confidence interval for the preference index because individual flying foxes were not sampled. Instead an estimate of the preferences of the flying fox population was obtained by sampling trees visited by many individuals of the population.
Pteropus tonganus and P. samoensis both foraged on fruits, flower resources, and leaves in varying amounts throughout the year [ILLUSTRATION FOR FIGURE 3 OMITTED]. P. samoensis and P. tonganus fed upon 36 and 42 plant species, respectively, in Samoa, with 22 plant species overlapping (Appendix). I observed 32 different plant species from 25 genera and 19 families being used as a food source by P. samoensis or P. tonganus in Samoa (Appendix). Forty-six additional plant species have been identified as being used by P. samoensis or P. tonganus either in Samoa or in other archipelagos in the South Pacific (Wodzicki and Felten 1975, Cox et al. 1992, Mickleburgh et al. 1992, Wilson and Engbring 1992, Trail 1994, Banack 1996). Combined, these records indicate that 78 plant species from 58 genera in 39 families are used by just two species of flying foxes throughout their range. based upon records of the same plant species being used by other Pteropus spp. (on other islands) and use of closely related plant species by P. samoensis or P. tonganus, another 10 plant species found in Samoa are probably occasionally used by P. samoensis or P. tonganus but have not yet been detected (Marshall 1985, Fujita and Tuttle 1991, Wiles and Fujita 1992, Banack 1996). I suspect [greater than] 69 plant species (from 51 genera in 36 families) are used by P. samoensis and P. tonganus in the Samoan islands.
I generated a list of 78 species of forest trees growing in Amalau Valley, Tutuila, American Samoa [ILLUSTRATION FOR FIGURE 2 OMITTED]. I included all tree species known to occur within Amalau Valley (Banack 1996, 1998). No shrubs, herbs, vines, or cultivated species were included. The list was checked by two colleagues familiar with both the Samoan flora and Amalau Valley. From this list, I found 47 species (60%) used by either P. tonganus or P. samoensis ([ILLUSTRATION FOR FIGURE 4 OMITTED], Appendix). Use was determined based upon personal observations during visitations, ejecta pellets, reliable records of foraging by P. tonganus or P. samoensis on a given species, or the likely use of six plant species based upon use of closely related plant species. Eighteen of these species (23%) were commonly used during my study and 22 (28%) were used with sufficient frequency for me or my research assistants to observe their use during the 18 mo we conducted research in this valley.
From the list of forest tree species in Amalau Valley, 13 species (17%) were used as both a fruit and flower resource and 7 of those were commonly used by flying foxes [ILLUSTRATION FOR FIGURE 4 OMITTED]. Forty-seven of the 78 tree species form the canopy of the forest and 37 of these (79% of the canopy-forming species) were used by flying foxes for either fruit or flower resources. Seventy-six of the canopy species are indigenous to Samoa and 16 are endemic; flying foxes used 9 (56%) of the endemic plant species.
Ten of the plant species most commonly used for fruit by P. samoensis and P. tonganus were included in a more intensive survey of comparative use (Artocarpus altilus [Moraceae], Cananga odorata, Carica papaya [Caricaceae], Inocarpus fagifer, Palaquium stehlinii, Planchonella garberi [Sapotaceae], Planchonella grayana, Planchonella samoensis, Syzygium inophylloides, and Terminalia catappa [Combretaceae]). Although some species of figs (particularly Ficus prolixa [Moraceae]) were also heavily used when available, they were not included in this study due to small sample sizes and random fruiting patterns. Of the 10 species examined, 5 species (Artocarpus altilus, Planchonella samoensis, Syzygium inophylloides, Palaquium stehlinii, and Inocarpus fagifer) comprised 79% (21, 17, 17, 13, and 11%, respectively) of the total biomass of fruit detected in the diet from November 1993 to October 1994 [ILLUSTRATION FOR FIGURE 5 OMITTED]. The first three species listed represented over half of the total yearly fruit biomass detected in the diet. These plant taxa were dominant during different times of the year - Artocarpus altilus from October through January, Planchonella samoensis in April, May, and June, and Syzygium inophylloides in July, August and September. Inocarpus fagifer contributed over half (58%) of the total fruit biomass eaten during the month of March and was the dominant contributor (39%) in February among the ten commonly used species. Palaquium stehlinii contributed the dominant portion (39%) of the total fruit used during January and Planchonella grayana contributed the largest portion (28%) to the total fruit diet in July. Thus, there was a temporal shift in the dominant plant taxa used through the year.
Preference index values were calculated on both a mean yearly and monthly basis. Food preference values reflect any deviation from random food sampling and provide an indication as to how well the consumer likes the food item relative to other foods (all else being equal). Although food preferences should be repeatable given the same conditions, food preferences have been noted to change with variations in the food requirements of the consumer (physiological conditions), food quality, and environmental features (Ellis et al. 1976). I calculated both an overall preference value based on the pooled sample of all trees included in this study (n = 984; [ILLUSTRATION FOR FIGURE 6 OMITTED]) and a monthly preference value based on those trees that were fruiting in any given month [ILLUSTRATION FOR FIGURES 7 AND 8 OMITTED]. Variations in the monthly preference values reflect both changing preferences and variation existing between individuals within the population of flying foxes sampled.
Preference indices [greater than] 1/k show "preference," and values [less than] 1/k indicate "avoidance" (Manly et al. 1993). Values close to 1/k indicate a food resource used in the proportion in which it occurred in the environment ("no preference or avoidance"). Inclusion or exclusion of plant species in the set of available foods altered the index number but rarely the categorical placement as preferred or avoided. The ranking of species was also unaffected by inclusions or exclusions. The results presented here show preferences in relation to the species assemblage chosen. Because sample sizes were small for some species in some months, preference values were calculated both including and excluding species with sample sizes [less than]3 during a single month. Results showed no change in categorical placement of species as preferred or avoided in any month except November and December (Banack 1996). During November and December Syzygium inophylloides was the only preferred species when it was included in the analysis. I was able to locate only one fruiting tree in 50 trees examined in November and two in 111 trees in December. Only one tree with fruit was accessible to place a tarp. The flying foxes found this tree and ate the fruit, indicating a high preference for S. inophylloides. When this single species was removed from the analysis for those months, a preference for other species was indicated (Banack 1996).
Removal tarps showed no loss of ejecta pellets from 41 of 52 tarps (79%). Of the remaining 11 tarps, pellets were removed at a rate of 7.4% per 24-h period. No differences were noted between fibrous and fleshy fruits. The reasons for removal, however, may be different. Removal was noted by ants, rats, and snails for fleshy fruits such as papayas and breadfruit. Removal of more fibrous ejecta pellets such as Inocarpus fagifer and Palaquium stehlinii was by rats or large animals (pigs or dogs) knocking the pellets out of the tarp. Pellet counts were not adjusted for removal because the removal was infrequent and not correlated to any one factor. Counts of ejecta pellets therefore represent a minimum number of pellets dropped into the tarps.
Fruits of three plant species - Planchonella samoensis, Planchonella garberi, and Terminalia catappa - were highly preferred by flying foxes [ILLUSTRATION FOR FIGURE 6 OMITTED]. In general, the rank-order preference calculated for the entire year was consistent with the rank order of species calculated within a single month, but some variations exist.
Planchonella samoensis had the highest preference in the yearly index and was preferred during six of the seven months it was available (March through August, and October) when analyzed on a monthly basis [ILLUSTRATION FOR FIGURE 8 OMITTED]. During March the fruits were just beginning to ripen and use was not detected. Another indication that this fruit is highly preferred is that two Pteropus samoensis defended fruits throughout the night in May 1994 (Banack 1996).
Planchonella garberi fruit showed a similar pattern of being present but not used during the early stages of ripening, followed by a period of high preference for four consecutive months and the second highest preference value in the yearly calculation. Terminalia catappa had a low population density in Amalau Valley but was eaten in large quantities when fruit density within a single tree was high. Thus, T. catappa was identified as a preferred species because it was sought out by flying foxes and eaten more often than would be expected by its abundance.
Two species, Syzygium inophylloides and Planchonella grayana, showed a preference slightly below random use on the preference index calculated for the entire year [ILLUSTRATION FOR FIGURE 6 OMITTED]. These species, however, showed high preference during individual months [ILLUSTRATION FOR FIGURE 8 OMITTED].
Carica papaya was available year-round and was identified as an avoided species (used in quantities less than were present in the environment) in the analysis of overall yearly preference. When examined on a monthly basis, C. papaya was preferred in February, March, and April when fruits in the primary forest were scarce or in the early stages of ripening. Inocarpus fagifer showed a similar pattern in a monthly analysis of preference values. Although available year-round, it was preferred only during February and March when other fruit was in low abundance [ILLUSTRATION FOR FIGURE 7, 9 OMITTED].
During one half of the year (October through March) [greater than or equal to] 70% of the available fruit resources were found in the agroforest and secondary forest [ILLUSTRATION FOR FIGURE 9 OMITTED]. During the other half of the year (April through September) fruit was most abundant in the primary forest or roughly equal to that in the agroforest. Total biomass of available fruit was relatively constant from November through May but increased dramatically in June, July, August, and October. Although fruit was more abundant in the agroforest for more than half of the year, all primary forest tree species were preferred to those tree species found in the agroforest and secondary forest in the index calculated by year [ILLUSTRATION FOR FIGURE 6 OMITTED]. In monthly preference-index values, at least one primary forest trees was preferred during every month of the year except February [ILLUSTRATION FOR FIGURE 8 OMITTED], while tree species located in the agroforested area were preferred in only four months of the year [ILLUSTRATION FOR FIGURE 7 OMITTED].
In order to determine which environmental factors influenced diet selection, I regressed (SAS Institute 1989: GLM procedure) the number of ejecta pellets observed on my independent variables (number of ripe fruit per tree [RIPE], mean mass of fruit [MASS], relative abundance of species [DEN], number of trees in the population with fruit [PHEN]), and three interaction factors (biomass of fruit in a tree [TREE = RIPE x MASS], density of fruiting trees in the environment [ENVI = DEN x PHEN], and monthly preference index value [PIV]). Observations from the same tree were not significantly correlated with each other between months (run on MIXED procedure [SAS Institute 1989], using individual tree numbers as a random factor), therefore ordinary regression methods were used in this analysis.
In order to determine if the effect of the predictor variables on the number of pellets changed over time, the model was run with month as a class, with month as a variable, and with month as an interaction factor. When all 12 mo were included, the effect of the predictor variables on the number of pellets was significant, indicating that there were more pellets in some months than others and that the interactions between the predictor variables change (with respect to number of ejecta pellets) over time. This indicates that some months were different than others. When the months of July, October, and December were excluded from the analysis, then the interactions were no longer significant. This suggests that during 9 mo of the year, biomass and preference were consistently influencing the number of ejecta pellets, but during July, October, and December something else was influencing diet selection. During July, flying foxes were heavily feeding on fruit from the Sapotaceae family [ILLUSTRATION FOR FIGURE 5 AND 8 OMITTED]. At this time, they were focusing on those trees with the highest number of ripe fruits per tree (overall display, P = 0.0001 for RIPE). During October, MASS, DEN, PHEN, and ENVI were statistically significant (P [less than or equal to] 0.01) indicating diet selection was correlated with a high density of fruit in a single area (i.e., a breadfruit crop). During December, TREE, PIV, and DEN were significant (P [less than or equal to] 0.04). In this month, the fruit consumption of flying foxes was spread between a diverse array of plants [ILLUSTRATION FOR FIGURE 5 OMITTED], and several cues were being used for diet selection. Because nine of the months had the same relationship, I considered them together in a regression analysis. The estimated equation was statistically highly significant (P = 0.0001) with [R.sup.2] = 0.145785 (Table 1). The biomass of fruit in a single tree (TREE) and the preference index value (PIV) were highly significant predictor variables (P = 0.0001). As the mass of the fruit increased, the number of ripe fruit had more of an effect on use (as measured by the number of ejecta pellets produced). These statistics point out not only that preference and biomass were important in the diet selection of flying foxes, but also that the animals were not switching diet based primarily on shifting search images in response to fruit or "prey" density.
In order to understand which characters influenced the selection of a particular plant species, I performed the same regression using data grouped by species and included only those independent variables that had intraspecific [TABULAR DATA FOR TABLE 1 OMITTED] variability (RIPE, PHEN, PIV). The estimated equation was statistically significant (P [less than] 0.05) for only half (5/10) of the species examined (Carica papaya, Inocarpus fagifer, Palaquium stehlinii, Planchonella grayana, and Syzygium inophylloides, Banack 1996). The preference value (significant in four species) and the number of ripe fruit per tree (significant in five species) were the only significant factors in the species models. This indicates that diet selection was a function of preference and fruit display (i.e., plants with a high preference and with a high number of fruit per tree were sought out and consumed in greater amounts) for only half of the plant species examined. For half the species, including the three species with the highest yearly preference values, none of the predictors were significant (P [greater than] 0.05). This would suggest that other factors influence the selection of these plant species in the diet of flying foxes.
The use of flower resources by flying foxes is not represented in the estimation of preference values. Floral resources include nectar (Planchonella samoensis, Palaquium stehlinii), pollen (Freycinetia reineckei), and entire flowers (Eria robusta, Orchidaceae). For many plant species it is likely that both nectar and pollen are being used by visiting flying foxes. I am aware of only one plant species (Eria robusta) where the flying foxes consistently acted as a flower predator. The number of plant species used as a flower resource by flying foxes was highest during the months of February, October, and November [ILLUSTRATION FOR FIGURE 3 OMITTED]. The proportion of the diet represented by floral resources is not known.
Flowers from three tree species (Syzygium inophylloides, Planchonella samoensis, and Freycinetia reinecki [Pandanaceae]) were frequently visited by flying foxes from October 1993 to January 1994. In Amalau, one species of flying fox (Pteropus tonganus) regularly fed upon nectar from banana flowers (Musa spp.). Flying foxes of both species consistently visited and defended Erythrina variegata (Fabaceae/Leguminosae) flowers in July. In addition, I documented a roost shift of P. tonganus to an area with a high density of Palaquium stehlinii just as the flowers opened and I watched eight P. samoensis carefully lick nectar from the flowers of a single Palaquium stehlinii tree at a rate of 20.4 flowers/min.
Other flowers that were used by one or the other species of flying fox during this study include: Cocos nucifera (Arecaceae), Ceiba pentandra (Bombacaceae), Elaeocarpus ulianus (Elaeocarpaceae), Samanea saman (Fabaceae/Leguminosae), Barringtonia asiatica (Lecythidaceae), Syzygium dealatum (Myrtaceae), Eria robusta, Guettarda speciosa (Rubiaceae), and Neonauclea forsteri (Rubiaceae) (Appendix). Additional species have been identified as used by flying foxes as a flower resource in literature records but were not observed during this study: Rhus taitensis (Anacardiaceae), Cananga odorata, Syzygium clusiaefolium (Myrtaceae), Pandanus tectorius (Pandanaceae), and Alphitonia zizyphoides (Rhamnaceae) (Wodzicki and Felten 1975, Cox et al. 1992, Wilson and Engbring 1992, Mickleburgh et al. 1992, Trail 1994).
Flying foxes are generalists and likely feed upon [less than] 69 plant species from 51 genera in 36 families in the Samoan Islands alone (Cox et al. 1992, Mickleburgh et al. 1992, Wilson and Engbring 1992, Trail 1994, and Banack 1996). This represents 60% of forest tree species being recorded as used by flying foxes and 28% of forest tree species frequently being used in a single location (Amalau Valley). In addition, 79% of the canopy tree species were used by flying foxes for either fruit or flower resources.
Differentiation between those resources used by Pteropus tonganus and by P. samoensis is not easily accomplished. Most of the foraging records of flying foxes in Samoa (and elsewhere) have been made opportunistically. Few studies have devoted sufficient time to a single plant species to rule out their use by one species of flying fox or the other (see Cox  for a possible exception). Even with extended observations of a single tree species, rare use by one species of flying fox could easily be missed (compare Elmqvist et al.  with Mickleburgh et al. , Wilson and Engbring , and Trail ). Observations of P. samoensis avoiding available plant resources are inherently more reliable as observations can be made in daylight hours, and distant observations can be made during the day that are not possible at night, even with night-vision devices. In addition, it is difficult to make foraging observations under the primary forest canopy, especially at night. Thus, it is possible to differentiate resource utilization between P. samoensis and P. tonganus to some degree. A complete understanding of how resource use varies for each Pteropus species under different ecological conditions, however, is not yet possible. I have, therefore, included as much information as possible concerning differences between P. samoensis and P. tonganus (Appendix) but group them together as a genus when differentiation was not possible (such as when utilizing ejecta pellets and tooth imprints in fruit as an indicator of resource use).
It should also be noted that my observations represent a period of relatively high resource availability and low flying fox densities during a period following three major hurricanes in Samoa (Tusi in January 1987, Of a in February 1990, and Val in December 1991). These hurricanes caused major structural changes in plant communities (Elmqvist et al. 1994) as well as changes in flying fox populations (Craig et al. 1994). These changes also caused dietary shifts in flying foxes (Pierson et al. 1996; P. A. Cox, personal communication). I expect the number of plant species eaten to be lower under the conditions present during my observations in comparison to pre-hurricane conditions when resource competition would have been more intense (higher population numbers) and individuals may have resorted to eating plants lower on their list of preferences. Note that, from the perspective of an individual animal, diet breadth could decrease with increased competition as weaker animals were forced to feed on a smaller subset of resources. Nevertheless, under these same conditions more plant species would be detected as being used within the entire population when all of the plant species used were totalled together. Thus the food list I obtained did not include all plant species previously noted as being used by these Pteropus species, as foraging patterns reflect the conditions under which the observations were made. Other discrepancies between my observations and food lists recorded by previous observers may reflect some observations of immediate post-hurricane foraging that may not be representative of the diet of P. samoensis or P. tonganus. Still other records may represent evidence of juvenile flying foxes sampling food items for palatability but may not represent food items regularly found in their diet. I found two cases of fruits with flying fox tooth imprints but no evidence that the fruit, or other fruits from that tree species, were actually ingested. Lastly, some records likely represent foods used rarely or occasionally. These plant species may be highly important to the health of the animals by providing low levels of specific dietary nutrients or may be relatively unimportant and eaten only opportunistically. The value of these plant species may also change relative to flying fox densities and increased resource competition.
In addition to fruits, numerous flower and leaf resources were consumed from the primary forest throughout the year. Flying foxes used both fruits and flowers of 17% of the total forest tree species and over half of those plant species were frequently visited. Flowers of [greater than] 16 species of trees are used by flying foxes in Samoa. Twenty-six percent of the total number of tree species in the Amalau forest were used as a floral resource. Additionally, flowers from non-tree plant species are also used by flying foxes. Evidence of the utilization of leaves was obtained in 10 of the 12 mo of the year. Leaf resources thus represented a regular part of the diet but may be particularly important following major disturbances such as hurricanes (Pierson et al. 1996).
Diet breadth of flying foxes in the Samoan islands is thus clearly broad and is comparable to that of Neotropical frugivorous bats (Fleming 1986, 1988).
Consistent with the predictions of Fleming (1986) for frugivorous bats, Pteropus spp. eat a taxonomically nonrandom subset of fruits. Flying foxes preferred fruits found in the primary forest to those found in the agroforest. Although there may be a high degree of individual variation in this regard, this population census found fruits in the agroforest to have a lower preference ranking than all fruit species included from the primary forest. Plant species from the agroforest, however, formed a large portion of the diet when resources in the primary forest were low. Therefore, even though they were not highly preferred, they were important in the diet. There were differences between how P. samoensis and P. tonganus used resources in the secondary forest. In general, the overlap in diet between P. tonganus and P. samoensis was high. However, P. tonganus was found foraging in agroforest areas more often than P. samoensis (Banack 1996). P. tonganus was also found to feed on more species of cultivated plant species than P. samoensis (Banack 1998). Note, however, that plant species in the agroforest were avoided (used in quantities less than random). Therefore, even though the plant species in the agroforest were used often by P. tonganus and in greater variety, they still appear to be lower on a scale of preference than the species examined from the primary forest.
The rank-order preference calculated for the entire year was generally consistent with the rank order of species calculated within a single month ([ILLUSTRATION FOR FIGURE 6 OMITTED], Banack 1996). Discrepancies are due to one of two things. First, the monthly sample reflects an actual shift in flying fox preference for that month. This could occur due to variations in the physiological or morphological state of the animal, including pregnancy or ontogeny, which could alter nutrient requirements or vary handling-time efficiencies; variations in food quality, such as changes in carbohydrates, or secondary compounds; and changes in the environment, including temperature and humidity (Rosenzweig and Sterner 1970, Willson and Harmeson 1973, Ellis et al. 1976). Flying foxes seem particularly sensitive to subtle changes in food quality. They were extremely selective in choosing fruit within a tree, and would smell and occasionally bite 10-15 fruits before either eating one in situ, removing one to eat in another location, or flying to another tree to continue the search. Second, the monthly sample reflects sampling variation due to detecting the food preferences of different individuals in the population or detecting differences between the two species of flying fox. Although there was no evidence of one species of flying fox shifting completely to flower resources during a particular month, subtle shifts in balance may represent a preference bias due to a single species of flying fox in comparison to other months. Thus, one of the limitations of this study is its inability to reliably differentiate between the two species of flying foxes with ejecta pellets. Nevertheless, a discriminant analysis of ejecta pellets revealed a heavy utilization of forest tree resources by both P. samoensis and P. tonganus (Banack 1996).
Five plant species (Artocarpus altilus, Planchonella samoensis, Syzygium inophylloides, Palaquium stehlinii, and Inocarpus fagifer) represent the dominant plant taxa used for fruit by P. samoensis and P. tonganus in Samoa. These species were used sequentially throughout the year. Interestingly, only one of these species (Planchonella samoensis) was highly preferred on the yearly preference index although all showed high preference during at least one month. To these five species Ficus prolixa should be added as an important resource although the relative contribution to the diet and seasonal preference are not known.
It was not possible to determine the proportion of the total diet represented by flower resources or the various flower preferences in this study. However, a greater variety of flower species were used during the months when the most highly preferred fruit species were found to have little available fruit biomass. Thus, the number of flower species used was highest in February, October, and November when the available fruit from the primary forest represented 6.7, 11.0, and 12.0% of their highest yearly values respectively.
When considering both fruit and flower resources, five plant families - Sapotaceae, Myrtaceae, Moraceae, Combretaceae, and Fabaceae (Leguminosae) - were particularly important to flying foxes in Amalau Valley. Fruit from plant species within Sapotaceae and Combretaceae was highly preferred. Fruit from species within Myrtaceae, Moraceae, and Sapotaceae comprised a large portion of the diet of flying foxes throughout the year. Plant species within Sapotaceae and Myrtaceae were visited for both flower and fruit resources and thus form an important part of the resources used by flying foxes during a large portion of the year. Fruit from plant species in Moraceae was available for large portions of the year. Ficus spp. were available year-round (Trail 1994) and Artocarpus altilus was available for 10 mo of the year. Although Ficus spp. were not included in the analysis of preference due to inadequate sample sizes, they were heavily eaten by flying foxes when available. Leaves were also eaten, at least occasionally, from Ficus spp. (Banack 1996) and A. altilus (P. Craig, personal communication). Lastly, Fabaceae (Leguminosae) was an important source of nectar in July (Erythrina variegata), and an important source of fruit in February and March (Inocarpus fagifer). Leaves of both E. variegata (Banack 1996) and I. fagifer (Trail 1994) were also chewed. During the months of February and March when I. fagifer was heavily used by flying foxes, Pometia pinnata was fruiting in lowland forests in Western Samoa. P. pinnata represented the dominant indigenous resource in Western Samoa during this period. Because lowland forests have been virtually wiped out in American Samoa, I. fagifer (a naturalized species) was used as the dominant fruit resource during these critical months of low fruit availability in the primary forest. Prior to human colonization, flying foxes probably traveled to lowland forests to take advantage of fruiting P. pinnata in American Samoa.
Given this information on diet breadth and selection can we determine a set of "core plant taxa" for Pteropus spp. on Samoa? Fleming (1986: 109) indicated that, "core fruit taxa can be identified on the basis of their relatively high dietary representation and high degree of selectivity." The top two candidates for this designation in diet of Pteropus spp. on Samoa are Planchonella samoensis (17% of diet, [P.sub.i] = 0.34) and Syzygium inophylloides (17% of diet, [P.sub.i] = 0.09). However, each of these plants contributed to the diet for only three months of the year for a combined use for only half of the year. There were no fruit species that were consistently used in Amalau for the entire year despite four plant species that were available year-round (Palaquium stehlinii, Inocarpus fagifer, Cananga odorata, and Carica papaya). Of the five dominant plant species in the diet, selectivity was low for all but Planchonella samoensis and selectively was not consistently observed throughout the period when the fruit was available except for Planchonella samoensis (6/7 mo) (Artocarpus altilus, 1/10 mo; Syzygium inophylloides, 3/7 mo; Palaquium stehlinii, 1/12 mo; and Inocarpus fagifer, 2/12 mo).
This is very different from the case recorded by Fleming (1988) in which Piper species contributed 4547% of the diet of Carollia perspicillata through the year and were considered highly preferred species in several (but not all) studies. I argue that in fact a core plant taxa does not exist for Pteropus spp. on Samoa. Thus, although Neotropical species seem to rely on a set of core plant taxa supplemented by opportunistic consumption of other fruits as they become seasonally available, it would appear that Pteropus samoensis and P. tonganus on Samoa do not follow the same feeding strategy. Instead, they are closer to "sequential specialists" as suggested by Marshall (1985) and focus on preferred resources as they become seasonally available. However, these flying foxes do opportunistically add other species to the diet in a single foraging bout and do not strictly specialize on one plant taxa at a time. Thus, they would be considered "individual generalists" by Heithaus (1982).
These differences in degree of plant specialization between New World Phyllostomidae and Old World Pteropodidae likely reflect both the habitat and landscape characteristics of their respective environments. The oceanic-island homes of many Pteropus species have insular land areas, limited species pools, and periodic storms all restricting resource availability. Animals faced with comparably fewer choices of where to feed and what to feed on in a landscape of seasonally varying resources combined with unpredictable natural disasters (cyclonic storms) would be more likely to maintain a general feeding strategy rather than focusing on a set of core plant taxa. This supports Fleming's (1993) argument that phyllostomids are more specialized feeders than pteropodids because of differences in the spatio-temporal availability of plant food resources.
A community composed of a smaller species pool (as expected on oceanic islands) would not only have fewer prey species to select for possible inclusion in the diet but also fewer competitor species and fewer predator species. A single natural predator species was noted for the Samoan flying foxes (Grant and Banack 1995). High levels of both competition and predation have been noted to favor specialized diets. Oksanen (1992:19) stated, "In systems where the ability to use marginal forage is never tested while ability to escape predators is essential for daily survival, natural selection will thus favour specialization to high-quality forage." His arguments were twofold: (1) high-quality food allows an animal to obtain adequate nutrient levels from smaller amounts of food and thus spend a greater portion of time hiding; and (2) consumers of low-quality food must have larger gastrointestinal tracts, which reduces the ability of the animal to run away from predators. Similarly, intense competition pressures could induce specialization through competitive exclusion. Thus, dietary specialization would confer less of an advantage in systems with few competitors and low levels of predation. Therefore, the evolution of feeding strategies in mainland vs. island systems should have different trajectories due to differences in the spatio-temporal predictabilities of prey resources and differences in competition and predation pressures.
Correlates of dietary selectivity
Given the unpredictability of fruit resources on small oceanic islands, one would not expect frugivorous animals to focus on core plant taxa or key on year-round availability as an important characteristic determining fruit choice. Instead, local resource abundance and preference were significantly correlated with food choice of Pteropus spp. living in the Samoan Islands. These factors, however, did not predict the use of the three most highly preferred plant species. Therefore, other factors must be influential in determining diet selection. There likely were both ultimate and proximate factors contributing to fruit preferences in flying foxes. In addition to spatio-temporal availability, nutritional properties of the fruit may logically influence dietary choice from both an ecological and an evolutionary perspective.
Cox (1984) found that for P. samoensis the most attractive parts of the Freycinetia reineckei flower were higher in lipid, protein, and sugar than the less attractive parts. Baker and Baker (1983, 1990) found hexose sugars to be predictive of bat nectar choice. Cox (1984) also found a high ratio of hexose in bat-preferred floral parts.
Further research on all aspects of fruit quality will contribute to an understanding of fruit choice. Because flying foxes swallow mostly the juices extracted after mastication of the fruit pulp and discard most of the fiber (in the form of an ejecta pellet), an analysis of nutrients available in the fruit may not be sufficient. Instead, one should subtract the nutrients present in the ejecta pellets (and thus discarded by the flying fox) from the nutrients available in the fruit. This would add substantial information to the overall understanding of the nutrients actually available to the flying fox, and represents the next important step in understanding the correlates of diet selection.
In conclusion, flying foxes of the genus Pteropus in oceanic-island environments do not follow the same feeding strategies as do Neotropical fruit-eating bats and Paleotropical African pteropodids as suggested by Fleming (1982, 1986). Although both groups do eat a taxonomically nonrandom subset of fruits, Pteropus spp. in Samoa are sequential specialists and do not focus on core plant taxa. In addition, the year-round availability of fruit does not appear to be an important factor in diet selection by oceanic-island Pteropus species. Differences between the two groups of fruit-eating chiroptera likely reflect different evolutionary pressures in mainland and oceanic island systems.
This work was supported by the National Park of American Samoa, Bat Conservation International, Sigma Xi, the Museum of Vertebrate Zoology, University of California at, Berkeley, Department of Marine and Wildlife Resources, and the Department of Integrative Biology, University of California, Berkeley. I am very grateful to P. Meek, D. Meek, J. Richmond, and S. Vignieri who provided valuable field assistance. P. Trail generously shared his knowledge of the Samoan flora with me. P. Trail and W. A. Whistler kindly reviewed my list of forest trees in Amalau. I am grateful to P. Cox, C. D'Antonio, T. Elmqvist, T Fleming, G. Grant, W. Z. Lidicker, D. McCullough, E. Pierson, and two anonymous reviewers for helpful advice and reviews of earlier drafts of this manuscript. Maps were drawn by A. Stout, M. Sullivan, and B. Pendleton. Statistical help was provided by D. Banack, R. Cates, E. Linder, B. Mauer, and B. Schaalje.
[TABULAR DATA FOR APPENDIX OMITTED]
Banack, S. A. 1996. Flying foxes, genus Pteropus, in the Samoan islands: Interactions with forest communities. Dissertation. University of California, Berkeley, California, USA.
-----. 1998. Conservation implications of flying foxes and their interactions with forest communities. Conservation Biology, in press.
Baker, H. G., and I. Baker. 1983. Floral nectar sugar constituents in relation to pollinator type. Pages 117-141 in C. E. Jones and R. J. Little, editors. Handbook of experimental pollination biology Van Nostrand Reinhold, New York, New York, USA.
Baker, H. G., and I. Baker 1990. The predictive value of nectar chemistry to the recognition of pollinator types. Israel Journal of Botany 39:159-166.
Baker, J. R., and Z. Baker 1936. The seasons in a tropical rain-forest (New Hebrides) Part 3. Fruit bats (Pteropidae). Journal of the Linnean Society of London. Zoology 40: 123-141.
Chesson, J. 1978. Measuring preference in selective predation. Ecology 59:211-215.
Cox, P. A. 1981. Vertebrate pollination and the maintenance of dioecism in Freycinetia. Dissertation. Harvard University, Cambridge, Massachusetts, USA.
-----. 1983. Observations on the natural history of Samoan bats. Mammalia 47:519-523.
-----. 1984. Chiropterophily and ornithophily in Freycinetia (Pandanaceae) in Samoa Plant Evolution and Systematics 144:277-290.
Cox, P. A., T Elmqvist, E. D. Pierson, and W. E. Rainey. 1991. Flying foxes as strong interactors in South Pacific island ecosystems: a conservation hypothesis. Conservation Biology 5:448-454.
Cox, P. A., T. Elmqvist, E. D. Pierson, and W. E. Rainey. 1992. Flying foxes as pollinators and seed dispersers in Pacific island ecosystems. Pages 18-23 in D. E. Wilson and G. L. Graham, editors. Pacific island flying foxes. Proceedings of an international conservation conference. USFWS Biological Report 90(23). United States Fish and Wildlife Service, Washington, D.C., USA.
Craig, P., P. W. Trail, and T. E. Morrel. 1994. The decline of fruit bats in American Samoa due to hurricanes and overhunting. Biological Conservation 69:261-266.
Crome, F. H. J., and A. K. Irvine. 1986. Two bob each way: the pollination and breeding system of the Australian rain forest tree Syzygium cormiflorum (Myrtaceae). Biotropica 18:115-125.
Dobat, K., and T. Peikert-Holle. 1985. Bluten un fledermause. Waldemar Kramer, Frankfurt, Germany
Eby, P. 1991. Seasonal movements of grey-headed flying-foxes, Pteropus poliocephalus (Chiroptera:Pteropodidae), from two maternity camps in Northern New South Wales. Wildlife Research 18:547-559.
Ellis, J. E., J. A. Wiens, C. F. Rodell, and J. C. Anway. 1976. A conceptual model of diet selection as an ecosystem process. Journal of Theoretical Biology 60:93-108.
Elmqvist, T., P. A. Cox, W. E. Rainey, and E. D. Pierson. 1992. Restricted pollination on oceanic islands: pollination of Ceiba pentandra by flying foxes in Samoa Biotropica 24:15-23.
Elmqvist, T., W. E. Rainey, E. D. Pierson, and P. A. Cox. 1994. Effects of tropical cyclones Of a and Val on the structure of a Samoan lowland rain forest. Biotropica 26:384391.
Fleming, T. H. 1982. Foraging strategies of plant-visiting bats. Pages 287-325 in T. H. Kunz, editor. Ecology of bats. Plenum, New York, New York, USA.
-----. 1986. Opportunism versus specialization: the evolution of feeding strategies in frugivorous bats. Pages 105-118 in A. Estrada and T. H. Fleming, editors. Frugivores and seed dispersal Dr W. Junk, Dordrecht, The Netherlands.
-----. 1988. The short-tailed fruit bat: a study in plant-animal interactions. University of Chicago Press, Chicago, Illinois, USA.
-----. 1993. Plant-visiting bats. American Scientist 81: 460-467.
Fujita, M., and M. D. Tuttle. 1991. Flying foxes (Chiroptera: Pteropodidae): threatened animals of key ecological and economic importance. Conservation Biology 5:455-463
Grant, G. S., and S. A. Banack. 1995. Predation on Pteropus tonganus by a Barn Owl in American Samoa Australian Mammalogy 18:48-50.
Grant, G. S., P. Craig, and P. Trail 1996. Cyclone induced shift in foraging behavior of flying foxes in American Samoa. Biotropica 29:224-228.
Heithaus, E. R. 1982. Coevolution between bats and plants Pages 327-367 in T. H. Kunz, editor. Ecology of bats. Plenum, New York, New York, USA.
Jaeger, P. 1954. Les aspects actuels du probleme de cheiropterogamie. Bulletin of the Institut Frangais d'Afrique Noire, Serie A 16:796-821.
Koopman, K. F. 1993. Chiroptera. Pages 137-241 in D. E. Wilson and D. M. Reeder, editors Mammal species of the world: a taxonomic and geographic reference Second edition. Smithsonian Institution Press, Washington, D.C., USA.
Mabberley, D. J. 1993. The plant-book, a portable dictionary of the higher plants Cambridge University Press, Cambridge, UK.
Manly, B. F. J., L. L. McDonald, and D. L. Thomas 1993. Resource selection by animals: statistical design and analysis for field studies. Chapman & Hall, New York, New York, USA.
Marshall, A. G. 1985. Old world phytophagous bats (Megachiroptera) and their food plants: a survey. Biological Journal of the Linnean Society 83:351-369
Medway, L., and A. G. Marshall. 1975. Terrestrial vertebrates of the New Hebrides: origin and distribution. Philosophical Transactions of the Royal Society of London, B Biological Sciences 272:423-465
Mickleburgh, S. P., A.M. Huston, and P. A. Racey. 1992. Old world fruit bats. An action plan for their conservation International Union for Conservation of Nature and Natural Resources. (IUCN/SSC) Chiroptera Specialist Group, Gland, Switzerland.
Oksanen, L. 1992. Evolution of exploitation ecosystems. I. Predation, foraging ecology and population dynamics in herbivores. Evolutionary Ecology 6:15-33.
Pierson, E. D., T. Elmqvist, W. E. Rainey, and P. A. Cox. 1996. Effects of tropical cyclone storms on flying fox populations on the South Pacific Islands of Samoa. Conservation Biology 10:438-451.
Rainey, W. E., and E. D. Pierson. 1992. Distribution of Pacific island flying foxes. Pages 111-121 in D. E. Wilson and G. L. Graham, editors. Pacific island flying foxes. Proceedings of the International Conservation Conference. USFWS Biological Report Series 90(23), Washington, D.C.
Rainey, W. E., E. D. Pierson, T. Elmqvist, and P. A. Cox. 1995. The role of flying foxes (Pterododidae) in oceanic island ecosystems of the Pacific. Zoological Symposium 67:47-62.
Richmond, J. Q., and S. A. Banack. in press. Comparative analysis of wing morphology, flight behavior, and habitat use in flying foxes (genus: Pteropus). Australian Journal of Zoology.
Ridley, H. N. 1930. The dispersal of plants throughout the world. Reeve, Ashford, Kent, UK.
Rosenzweig, M. L., and P. W. Sterner. 1970. Population ecology of desert rodent communities: body size and seed-husking as bases for Heteromyid coexistence. Ecology 51: 217-224.
SAS Institute. 1989. SAS/STAT user's guide, release 6.09. Fourth edition, volume 2. SAS Institute, Cary, North Carolina, USA.
Thomas, D. W. 1988. Analysis of the diets of plant-visiting bats. Pages 211-220 in T. H. Kunz editor. Ecological and behavioral methods for the study of bats. Smithsonian Institution Press, Washington, D.C., USA.
Trail, P. 1994. The phenology of rainforest plants in Tutuila, American Samoa. Department of Marine and Wildlife Resources, American Samoa.
Vanderploeg, H. A., and D. Scavia. 1979. Calculation and use of selectivity coefficients of feeding: zooplankton grazing. Ecological Modeling 7:135-149.
Whitmee, S. J. 1874. Excerpts from an untitled letter. Proceedings of the Zoological Society of London 1874:666-667.
Wiles, G. J., and M. Fujita. 1992. Food plants and economic importance of flying foxes on Pacific Islands. Pages 2435 in D. E. Wilson and G. L. Graham, editors. Pacific island flying foxes. Proceedings of an International Conservation Conference, USFWS Biological Report 90(23), Washington, D.C.
Willson, M. F., and J. C. Harmeson. 1973. Seed preferences and digestive efficiency of cardinals and song sparrows. Condor 75:225-234.
Wilson, D. E., and J. Engbring. 1992. The flying fox Pteropus samoensis and Pteropus tonganus: Status in Fiji and Samoa. Pages 74-101 in D. E. Wilson and G. L. Graham, editors. Pacific island flying foxes. Proceedings of an International Conservation Conference, USFWS Biological Report 90(23), Washington, D.C.
Wodzicki, K., and H. Felten. 1975. The peka, or fruit bat (Pteropus tonganus tonganus) (Mammalia, Chiroptera), of Niue Island, South Pacific. Pacific Science 29:131-138.
Wodzicki, K., and H. Felten. 1980. Fruit bats of the genus Pteropus from the islands of Rarotonga and Mangaia, Cook Islands, Pacific Ocean. Senckenbergiana Biologica 61:143-151.
|Printer friendly Cite/link Email Feedback|
|Author:||Banack, Sandra Anne|
|Date:||Sep 1, 1998|
|Previous Article:||Gender and environmental sensitivity in nestling Collared Flycatchers.|
|Next Article:||Feeding preferences of a generalist salt-marsh crab: relative importance of multiple plant traits.|