PREFERENTIAL CONSUMPTION OF LARGER FRUITS OF Piper arboreum ( Piperaceae ) BY Carollia perspicillata ( Phyllostomidae ) IN THE BRAZILIAN TROPICAL DRY FOREST

Frugivorous animals may use morphological traits of food items such as size, hardness, shape, color and smell as cues that allow them to assess cost-benefit relationship of foraging activity. Fruit size is an important trait that influences feeding behavior of most frugivores, since there is a functional correlation between fruit size and frugivores’ body size. Therefore, size-based preference is fundamental to understand plant-frugivore interactions and seed dispersal. In this sense, we tested the hypothesis of preferential consumption of larger fruits of Piper arboreum by the short-tailed fruit bat Carollia perspicillata in the Brazilian tropical dry forest. Results showed considerable variation in fruit size among and within plants. There was also significant difference in fruit ripening time among plants, which was not related to fruit size. Average size of remaining fruits reduced as they were removed from plants by the bats, which indicates preferential consumption of larger fruits. On the other hand, plant phenology constrained consumption of bigger fruits. Only a small fraction of plant crop was available for consumption each night, regardless of fruit size, which probably coerced bats to feed on smaller fruits as the availability of larger ones decreased. Results suggest that bat preference on fruit size is mediated by plant phenological strategies.


INTRODUCTION
According to optimal foraging theory, animals should optimize energetic costs of foraging activities by reducing energy spent in searching, acquiring and processing food (PYKE, 1984).To be able to do this, foragers may use phenotypical traits of food items such as size, hardness, shape, color and smell as cues that allow them to assess costbenefit relationship of resources, which leads to differential consumption of these varying traits (GAUTIER-HION et al., 1985).
The 'syndrome hypothesis' states that seed dispersers choose fruits based on some particular traits (e.g.morphology, color, smell) and, thus, exert selective pressure on the evolution of fruit characteristics (JANSON, 1983;GAUTIER-HION et al., 1985).Despite the extensive literature on the influence of fruit traits on frugivores foraging behavior, there is considerable debate regarding the 'syndrome hypothesis' (CAMARGO et al. 2013).Fruit bearing plants and frugivores interaction networks are asymmetric and mainly generalistic (BASCOMPTE; JORDANO, 2007), and because of this low degree of specialization, most of fruit characteristics do not necessarily respond to directional selective pressure or constrain consumption by different frugivores (HERRERA, 2002, FLÖRCHINGER et al., 2010).For instance, a colorful fruit, which is a trait basically associated with visually oriented seed dispersers such as birds, may be consumed by smell-oriented animals as well.Therefore, frugivore choices that rely on fruit and plant traits remain an issue that requires further investigation (LOMÁSCOLO; SCHAEFER, 2010).
Among fruit traits, one that doubtlessly influences consumption is fruit size (JORDANO, 1995;FLÖRCHINGER et al., 2010).There is a functional correlation between frugivore's body mass and fruit size (FLEMING, 1991).For instance, plants bearing large and heavy fruits may limit their consumption by small-bodied animals, which usually do not have the necessary morphological and behavioral adaptations to remove, handle and consume them efficiently (WHEELWRIGHT, 1985).On the other hand, small fruits may not be energetically suitable for large-bodied frugivores, which need to spend more time foraging to achieve their energetic requirements, leading to increased energy expenditure and susceptibility to predation during foraging (MUSCARELLA;FLEMING, 2008).Thus, differential consumption of varying fruit sizes by frugivores is an important feature to understand plant-frugivore interactions since it may affect seed dispersal and the fitness of both parties involved (HERRERA, 2002).Most studies have assessed frugivore size-based preference on fruits of different plant species (FUENTES, 1994;STANLEY et al., 2002, LOMÁSCOLO et al., 2010), but others have approached this selectivity on plant population and individual levels (JORDANO, 1995, MELLO et al., 2005, NUNES et al. 2007).
Fruit size may influence the foraging behavior of frugivorous bats (MUSCARELLA;FLEMING, 2008), but there is no specific pattern for these animals, since some species display a sizebased preference (KORINE; KALKO, 2005;NUNES et al., 2007), while others choose fruits irrespective of their size (DUMONT; IRVINE, 1998;WENDELN et al., 2000).According to optimal foraging theory (PYKE, 1984), it would be reasonable to assume that bats choose fruits in order to maximize their net energy intake and minimize risks by reducing exposure to predators such as owls and domestic cats, and so, foraging on bigger fruits may be the most profitable strategy because it would require fewer visits to the fruiting plant, whereas plants would benefit from enhanced seed dispersal per fruit removed.Here, we tested the hypothesis of preferential consumption of larger fruits of Piper arboreum Aubl.(Piperaceae) by the short-tailed fruit bat Carollia perspicillata (Linnaeus 1758) (Phyllostomidae).

Study site
The study was conducted from August 2010 to August 2011 at Parque do Sabiá (hereafter PS), (48°14'02" W, 18°54'52" S), in Uberlândia city, Brazil.The PS has ca.35-ha and is mainly composed of a semideciduous forest, but it also includes some Brazilian Cerrado formations (GUILHERME et al., 1998).

Species description
Piper arboreum is a common understory tree species in gallery forests from Southeastern Brazil, and its fruits are consumed mainly by Carollia perspicillata (Phyllostomidae), a mediumbodied fruit-eating bat that is considered a Piper specialist (BIZERRIL; RAW, 1997).The shorttailed fruit bat forages using different sensory abilities and by means of echolocation and vision it distinguishes variations on Piper fruit size and morphology (THIES et al., 1998).Fruit smell is used as an imprecise cue to fruit location by bats (KORINE; KALKO, 2005).In PS, P. arboreum fruits did not exhibit evident differences in other morphological traits such as color, shape or accessibility, so it is reasonable to assume that if there is any preference based on fruit characters it would be on fruit size.The term 'fruit' was used in its ecological and functional rather than in its botanical meaning, since P. arboreum has actually infructescences.Nonetheless, bat choice is based on the whole infructescence.Moreover, P. arboreum fruits grow mainly in length, so it is referred as 'size' from hereafter.

Methods
To assess if fruit size varied in P. arboreum population, 14 trees (about 2 m high; similar number of stems and phenological state) were tagged with small, inconspicuous plastic bands.Individuals were selected so that a ten meters minimum distance was kept from one another.In each tagged plant, we labelled ten fruits of different sizes (N =140).We also measured fruit size (mm), weight (mg) and mean seed size (15 seeds/fruit; mm) of two unlabeled fruits taken from 14 nonexperimental neighboring plants.We used a 0.1 mm precision digital caliper and a 0.01 g precision weighing balance to record fruit measurements.Fruits from tagged plants were measured again every week until they were removed because they could have gotten bigger before reaching the maximum size.We recorded ripening time of each labeled fruit daily.Fruits of P. arboreum ripe late in the afternoon and are mostly consumed by bats on the same day (pers.obs., see also BIZERRIL; RAW, 1998).Therefore, average ripening time can be used as a synonym of fruit removal rate per plant.
To record fruit selection by bats, labeled fruits were visually searched on plants daily.When a tagged fruit was not found, the floor below plant perimeter (1 m radius) was inspected to avoid considering a fallen fruit (for example, by rain or wind) as a record of removal by bats.If a fruit was found, we checked for bat bite marks (small, evenly spaced perforations caused by teeth cusps).Therefore, a fruit was only recorded as having been chosen when: (a) it was neither found on plant nor on the floor; or (b) it was not found on plant, but it was on the floor within plant perimeter and presented conspicuous bat-bite marks.
In order to assess if fruits of P. arboreum were consumed by diurnal frugivores (e.g.birds, white-eared opossum), we visually checked plants twice a day (from 0730h to 1000h; and from 1600 h to 1830h), during the first two weeks of study, accounting for 70 h of observation.From the following week onwards, plants were checked only early in the morning, from August to October 2010 and from July to August 2011.To confirm P. arboreum consumption by bats, we conducted two hour night sessions observations (28 h) of all tagged plants.Finally, to verify P. arboreum consumption by C. perspicillata (and possibly other bat species), we performed captures using four mist nets (9 x 3 meters, at ground height) opened near non-tagged fruiting individuals of P. arboreum for five hours after sunset.Captured bats were kept in cotton bags for 30 min, after which they were released and the bag checked for feces and seeds.

Data Analysis
A Kolmogorov-Smirnov test was used to assess normality in the distribution of fruit sizes.The Mann-Whitney U test was performed to compare the sizes of what was considered smaller (< median fruit size) and bigger fruits (> median fruit size).Fruit size variation among individuals was assessed using the Kruskall-Wallis test and variation within individuals was assessed through the average coefficient of variation (CV).The Pearson's correlation test was used to analyze the relationship between fruit size and fruit width, weight and seed size.Additionally, in order to assess if there was indeed fruit preference by bats regarding to fruit size and that this preference was not simply a consequence of the removal of earlier matured fruits, we also tested (ANCOVA) if ripening time and fruit removal per plant was dependent of fruit size among plants.Finally, the analysis of relationship between fruit removal and fruit size was performed using a Spearman's correlation test.Statistical analyses were performed in Systat 10.2, following Zar (1999).Descriptive statistics are presented as mean ± 1 standard deviation (SD), unless specified otherwise.
A statistical significant difference was found in fruit ripening time (fruit removal rate) among plants (F 13,113 = 3.67; P < 0.001, ANCOVA), although this difference was not related to mean fruit size per plant (F 1,113 = 2.08; P = 0.152, ANCOVA).At PS, 128 (91.40%) of the labeled fruits were considered to have been removed by bats.Mean fruit size available for selection decreased as fruits were removed from plants (r s = -0.69;N = 9; P < 0.05) (Figure 4).Variation in fruit removal explained about 50 percent (R² = 0.48) of variation in the size of remaining fruits.
Four individuals of Carollia perspicillata were captured.From these, four fecal samples were obtained, of which three contained seeds, exclusively from P. arboreum.

DISCUSSION
Our findings support the hypothesis of preferential consumption of larger fruits of Piper arboreum by Carollia perspicillata and also suggest that feeding behavior and plant phenological strategies are important in explaining fruit consumption and seed dispersal.
A specific fruit trait must vary among individuals within a population in order to allow frugivores to choose upon it (GAUTIER-HION et al., 1985).Fruit size varied among and within individuals of P. arboreum from PS, thus allowing bats to choose fruits in population and in individual levels (among plants and within an individual parental tree, respectively).The hierarchy of foraging decisions is important to understand plantfrugivore interactions (SALLABANKS, 1993).
Body size may limit the set of resources frugivores are able to consume, mainly because small-bodied animals would not be able to remove or handle large and/or heavy fruits efficiently without more sophisticated behavioral repertory (MARTIN, 1985, MUSCARELLA;FLEMING, 2008).In the present study, bigger fruits of P. arboreum did not restrict consumption by C. perspicillata, since these bats consumed both small and large fruits.Bizerril and Raw (1998) did not observe fruit removal by C. perspicillata as we did, but rather the consumption of fruits on the plant.This difference in fruit consumption behavior may be due to the much heavier and larger fruits reported in the previous study (9.4 g 169 mm,, respectively).Bats may need to use different behavioral strategies when foraging on different fruit traits (DUMONT, 1999), even if it is on the same plant species, and so it is reasonable to expect behavioral adjustments regarding the consumption of fruits differing in size (MOERMOND; DENSLOW, 1983).
Fruits of P. arboreum were systematically consumed by C. perspicillata, as bigger fruits were removed sooner than smaller ones, which corroborated our expectations.The preference for larger fruits may be explained by a cost-benefit relationship, as larger fruits hold more pulp than smaller ones and, consequently, more available energy (WENDELN et al., 2000).P. arboreum fruits are low quality resources, with many small seeds (BIZERRIL; RAW, 1997).Thus, foraging on bigger fruits may be the most energetically profitable strategy adopted by C. perspicillata, since it would attend its energetic requirements with fewer visits to fruiting plants, while reducing foraging time and predation risks (NUNES et al., 2007).
Despite the observed preference of bats for larger fruits, a considerable percentage (ca.50%) of fruit removal was not explained by bat choice.This is due to the fact that smaller fruits were also consumed during foraging activity.This result implies that other factors may limit fruit choice by bats (FLEMING et al. 1993, CARLO;MORALES, 2008, PIZO;ALMEIDA-NETO, 2009).Some of these factors are variation in nutritional quality, plant structure, crop size and distance among fruiting individuals (FOSTER, 1990, WENDELN et al., 2000, CARLO;MORALES, 2008).Additionally, results showed that P. arboreum reproductive strategy is one of these factors.In the study site, P. arboreum was basically a "steady state" plant (sensu GENTRY, 1974) regarding its fruiting pattern, which means that some individuals ripened only a few fruits late in the afternoon, asynchronously and independently of fruit size.This pattern is similar to those described in the studies of Bizerril and Raw (1998) and Thies and Kalko (2004) for the same plant species.Hence, fruit availability was restricted per night, limiting bat choice to ripened fruits.
However, if larger fruits are more attractive to C. perspicillata, why does P. arboreum spend energy on the production of less preferred fruits and also restrict fruit availability?This could be explained by the presence of other frugivores that choose fruits of P. arboreum differently, but it could also be approached in terms of mutualism evolution.Plant-frugivore interactions are mutualistic associations (BASCOMPTE; JORDANO, 2007).Nevertheless, mutualisms are mutual exploitation interactions in which there are costs involved in their maintenance (BRONSTEIN, 2012).The amount of energy that plants allocate to produce larger fruits is greater than that allocated to smaller ones (HOWE;VANDE KERCKHOVE, 1981).Thus, P. arboreum can reduce mutualism costs and maintain the benefits of seed dispersal by constraining larger fruit availability, coercing bats to remove also the smaller fruits.Limitation in fruit availability by plants has also been investigated in previous studies (e.g.ORTIZ-PULIDO; RICO-GRAY, 2000).
Interactions with frugivores have certainly played an important role in the diversification of plant structures (TIFFNEY, 2004).However, in order to clearly elucidate the coevolutionary patterns of plant-frugivore interactions, we suggest that studies should take into account not only the role of frugivore behavior in shaping plant and fruit diversity, but also plant characteristics, such as phenology, that constrain frugivore activity and, thus, selective pressure upon fruit traits.

CONCLUSION
Carollia perspicillata chose fruits of Piper arboreum on a size-based preference in a tropical dry forest fragment from Southeast Brazil.However, this behavior was strongly influenced by plant phenology, since fruit availability was constrained each night.Although plant-frugivore interactions have long been studied in community ecology, there is no consensus on what are the determinants of fruit choice, if there is any.Further insight is needed on this subject in order to understand the ecological and evolutionary implications of fruit choice.

Figure 1 .
Figure 1.Frequency distribution of fruit size (mm) in Piper arboreum population from Parque do Sabiá, municipality of Uberlândia, State of Minas Gerais, Southeastern Brazil.

Figure 3 .
Figure 3. Pearson's correlations between fruit size and (A) fruit weight and (B) seed size of Piper arboreum in Parque do Sabiá forest, municipality of Uberlândia, State of Minas Gerais, Southeastern Brazil.
Four other phytophagous bat species -Artibeus lituratus (Olfers 1818), A. planirostris (Spix 1823), Platyrrhinus lineatus (E.Geoffroy 1810) -captured in the same study site (from which we obtained 19 fecal samples) and kept under the same conditions were not observed or recorded consuming fruits of P. arboreum.Moreover, no diurnal consumption of fruits of P. arboreum was observed, and the only nocturnal frugivore spotted approaching and consuming the fruits was the short-tailed fruit bat C. perspicillata.Bats always approached the fruitbearing plants more than once before removing the whole fruit in flight and never landed on plant to feed.Preferential consumption... PIRES, L. P.; DEL-CLARO, K.; UIEDA, W; Biosci.J., Uberlândia, v. 31, n. 2, p. 634-642, Mar./Apr.2015

Figure 4 .
Figure 4. Spearman's correlation between proportion of fruits removed from plants (%) and average size (mm) of remaining fruits in Piper arboreum individuals per week.
The authors would like to thank Natália O. Leiner, Wesley R. Silva and Marco A. Pizo for helpful comments on earlier versions of the manuscript; Giâncarlo A. Ferreira, Diego P. C. Teodoro, Vanessa F. Gonçalves, Carolina M. Freitas, Alexandre O. Rezende and Adriano M.