J.Comp Physiol B (2004) 174: $41-S48 DO! 10.1007/s00360-004-0442-4 CEE Christian C. Voigt The power requirements (Glossophaginae: Phyllostomidae) in nectar-feeding bats for clinging to flowers ‘Accepted: 16 July 2004/ Published online: 17 August 2004 © Springer-Verlag 2004 Abstract Nectar-feeding bats are the heaviest pollinators exploiting flowers in a hovering foraging mode. As hovering flight is considered to be energetically costly, clinging to flowers would be beneficial from an energetic perspective. I examined the rate of oxygen consumption and carbon dioxide release during clinging flower visi- tation in two 10-g Glossophaga soricina (Glossophinae: Phyllostomidae) to evaluate the potential energetic benefit of clinging versus hovering. In addition, I mea- sured the duration of flower visits of free-ranging glossophagine bats to Markea neurantha (Solanaceae), & bat-pollinated plant that allows both hovering and clinging Rower visitation, After 20s of clinging to an artificial respiratory mask, the bats’ respiratory ex- change ratio did not significantly deviate from 1, indi- cating the combustion of sugar. The average oxygen uptake rate equaled 1.39 ml min“! (0.38 SD, STPD) and the carbon dioxide release rate ‘equaled 1.33 ml min“! (£0.20 SD, STPD) for feeder visits longer than 20 s (n=79). Converting the oxygen uptake rate into power input yielded 0.49 W, less than a third of the power requirements for hovering for a 10-g ba Free-ranging 10-g glossophagine bats exploited flowers of M. neurantha for, on average, 0.32 s ( +0.14 SD, n=273) during hovering and for 0.39 s (40.18 SD, n=152) during clinging visitations. A comparison between the power requirements of flower exploitation in differently sized bats indicates that clinging would benefit larger nectar-feeding bats to a greater extent than smaller species. Communicated by G. Helimaier C.C. Voigt (52) Research Group Evolutionary Ecology, Institute for Zoo and Wildlife Researe! Alfred-Kowalke-Str. 17, 10315 Berlin, Germany E-mail: voigt@izw-berlin.de Tel: +49-30-5168609 Keywords Clinging « Flower visit - Glossophaginae Bats Abbreviations M: body mass « RER: respiratory exchange ratio - Pegging: power input during clinging * Pyoyering: power input during hovering Introduction Bats are the heaviest of all aerial pollinators. As body size defines the metabolic requirements of an animal (Kleiber 1975), size is likely to be an important aspect of the co-evolutionary process between pollinating bats and plants (von Helversen and Winter 2003). The highly specialized neotropical nectar-feeding bats of the sub- family Glossophaginae usually exploit flowers during hovering flight. Recently, it was shown that glossopha- gine bats are the most efficiently hovering animals of three groups of pollinators with overlapping body size ranges; i.e. glossophagine bats expend the least meta- bolic power per unit of body mass (Voigt and Winter 1999; Dudley and Winter 2002), At a metabolic power input of 1.1 W, a glossophagine bat can generate the lift force to support 7g against gravitation, whereas a ‘fnummingbird can support only 4 g and a sphingid moth only 3 g of body mass with the same amount of power (Voigt and Winter 1999). Despite this high efficiency, the power requirements of hovering are still high, and it is therefore counterintuitive that nectar-feeding bats re- main airborne when clinging to the corollas is feasible. Thus far, the energetic costs of clinging flower visitation and, consequently, the relative energetic benefits of clinging versus hovering are unknown for bats. This may, however, be an important physiological aspect in the interaction between glossophagine bats and bat- pollinated plants, Pyke (1981) suggested that hovering birds could move more quickly between flowers than perching birds, thus compensating for the higher energy costs of hovering542 over perching (Hainsworth and Wolf 1972; Pyke 1981). Furthermore, it was argued that small individuals should hover and large individuals should perch, if the relative importance of energetic costs versus speed in- creases with increasing body size and if birds maximize their net rate of energy gain. Thus far, most studies on optimal foraging in nectarivorous vertebrates have been conducted with birds, The present study aims to stimu- late studies on optimal foraging in other nectarivorous vertebrate taxa by filing a gap in our knowledge of the energy costs of different foraging modes in flower-visit- ing bats. In this study, nectar-feeding bats were trained to ingest nectar from an artificial flower while clinging to the flower, Using the experimental setup described in Voigt and Winter (1999), respirometric measurements were performed with Glassophaga soricina as the model organism for neotropical glossophaginae bats. The study species, the 10-2 G. soricina, has a broad distribution range from northern Mexico to northern Argentina (Koopman 1981) and is considered to be a generalist, feeding mostly on nectar, but sometimes—depending on availability—also on fruits or insects (Herrera et al, 2001). Captive and free-ranging G. soricina mostly ex- ploit flowers while hovering, e.g. Bauhinia pauleria (Heithaus et al. 1974), Tetrascylis ovalis (Passifloraceac; Buzato and Franco 1992), Vriesea gladioliflora (Bro- meliaceae; own observation), Helicteres baruensis (Ster- culiaceae; von Helversen and Voigt 2002). Exceptions, for example, are flowers of Ochroma pyramidale (Bom- bacacene), and Bauhinia ungulata (Caesalpinaceae), which are either pollinated in a hovering or clinging mode, and Mucuna holtoni (Fabaceae), which is strictly exploited in a clinging mode (von Helversen and von Helversen 1999). To collect additional baseline data from free-ranging bats, foraging modes of nectar-feeding bats were studied in the bat-pollinated epiphytic herb Markea neurantha (Solanaceae). In M. neurantha approximately eight flowers are located at the tip of an inflorescence that hangs down from the tip of a branch on.aca, |-m petiole, Usually, only a single corolla opens during a night. M. neurantha exhibits typical traits of a chiropterophilous plant (Tschapka 1998; von Helversen et al, 2000); nocturnal anthesis, large nectar volume, large pollen volume, and flower scents that are attractive to bats. Tschapka (1998) described two main pollinators for M. neurantha: G. commissarisi and Hylonycteris un- derwoodi. Both species visit the plant in the hovering and clinging foraging mode (own observations). Material and methods Observations at Markea neurantha Field observations of flowers of Markea neurantha were performed during November and December 1994 and and September and October 1995 in the vicinity of La Sclva field station (Organization for Tropical Studies, Province Heredia, Costa Rica, N 10°25'/W 84°00’). T netted bats during 11 nights between 1800 and 2300 hours in the vicinity of flowering plants (2-10 m dis- tance) using mistnets which were 6-9 m long and 2,5 m high. Species were identified by size and dental charac- teristics using the key by Timm and LaVal (1998). The body mass was measured to an accuracy of 0.1 g with a handheld balance (Pesola balance). During 8 other nights, | documented the mode of flower visitation of bats at the same plants. I set up an infrared-sensitive camera (Sony) on a tripod and a portable VCR (Pana- sonic) in front of a total of 22 inflorescences of Markea neurantha. An electric flashlight covered with an infrared filter (750 nm) was used for illumination, A recordings were stored on VHS tapes and brought to the lab; the mode of flower exploitation, time of night and duration of flower visits were analysed. The duration of flower visits was measured by counting the video frames during which the bat clung to the flower or hovered in front of it with its snout reaching into the corolla. Each frame was equivalent to 40 ms. Laboratory experimental setup Respirometric measurements for this study were per- formed with two male Glossophaga soricina, Rehn (Glossophaginae: Phyllostomidae) originating from a captive colony that was maintained on ad libitum food in greenhouse facilites of the University of Erlangen-Niimn- berg (Germany). A week before the experiment, the ani- mals were transferred to a room with a light regime of” 12:12 h L:D, a constant temperature of 24 °C, and con- stant relative humidity of 75%. At the onset of the training phase, a single bat was introduced into a plastic flight tunnel (7 m longx1 m wide x2 m high). Within this fight tunnel, the only place the bat could rest was on a piece of | cork that was suspended from an electronic balance (Mettler PM 100, accuracy 0.001 g). Body mass and times, of arrival at and take-off from the roost were recorded with a computer connected to the electronic balance. The animal was fed with a 17% (w/w) sugar solution based on hummingbirds’ nutritional solution (Nektar Plus) pro- vided froma computerized nectar-feeder. During feeding, the bat interrupted an infrared light beam at the front edge of the feeder mask. This event was recorded by a computer that activated the nectar pump and recorded the duration of the hovering or clinging event (Winter et al. 1998; Winter and von Helversen 1998). The feeder opening was constructed as a flow-through respirometric mask, The base of the feeder mask had a diameter of 2.3 em. The opening was oriented with an inclination of 30° to the ground and ata height of about 100 em above the ground. Under captive conditions, the chosen study species usually exploited the feeder during hovering flights. To facilitate a clinging lower exploitation, a piece of silken cloth was wrapped around the feeder. To prolong the flower visitations to the feeder, the first sugar-waterreward (7.5 pl) was delayed for 1.2s after the initial interruption of the light beam, Subsequent rewards of 6.3 ul each were dispensed every 0.75 s. Indirect calorimetry Room air was drawn through mask—constituting the feeder—with a membrane pump ata rate of 1,000 ml min', As shown previously (Winter et al. 1998), this flow rate is high enough to collect all respiratory gases of G. soricina during a feeder visit. The gas-analyzing system was described in detail in Voigt and Winter (1999). Briefly, downstream of the pump, the air was dried using a molecular sieve (3 A) and the flow was regulated with a mass flow controller (Bronkhorst, F 201 C-FB). Next, the airstream was split and led to both the oxygen analyzer (Ametek, now AEI, S-3A/IL with a dual N-37M sensor) and the carbon dioxide analyzer (Hartmann and Braun, URAS 10 E). Due to the high sensitivity of the CO2 analyzer (0- 500 ppm), the sample air stream was diluted with dry, CO»-free air (NaOH as desiccant and CO, absorbant) by a factor of ten. The dilution factor was monitored with & mass flow controller (Bronkhorst F 201 C-FB) and a mass flowmeter (Bronkhorst F 111 C-HB) and held constant during the measurements. The analog signals of the gus analyzers were recorded with a 386 MS-DOS computer at a rate of 3 Hz, using Sable Systems hard- and software (16-bit A/D), The analog output signals of the three flowmeters were checked before and after an experiment with a digital volimeter to the nearest 0.001 V. the respirometrie Data analysis Due to the time delay and wash-out characteristies of the gas-analyzing system (especially the COs analyzer), 1 could not monitor changes in instantaneous rates of oxygen consumption and carbon dioxide release during a feeder visilution. Therefore, Z-transformations were performed for ull measurements according to Bartholo- mew et al, (1981) using the Datacan program (Sable Systems). On the basis of the Z-transformed data, I cal- culated the integrals of the O2 and COg signals for each feeding event and multiplied them by the sampling rate and the flow rate at the feeder mask. For the calculation of the CO; volume, the dilution factor was used. O> volumes were corrected using Eq. 3a in Withers (1977) as HzO absorbant, but no CO> absorbant was used for the sample gases, Gas exchange rates (ml s™!) were calculated by dividing the COs and Os volumes by the corresponding hovering durations of the animal. A correction for STP Was not necessary as mass flowmeters give rates in STP. Forconversion of oxygen uptake(ml O2 s~') to metabolic power input (W), data were multiplied by 21.1 J mi! Os, ming carbohydrate catabolism and a metabolic RQ 543 Calibration and validation The CO, analyzer was calibrated by injecting known concentrations of CO, (purity standard: 99.999) using a gus-mixing pump (Wéstholf, SA27/2). For the O» ana- lyzer, no zero calibration was necessary because of the technical characteristics of its zironium-oxide sensor. The system was regularly validated by simulating the O consumption and CO. production of a bat feeding at the artificial flower (see also Voigt and Winter 1999 for a de- tailed description). For the validation of the COs n surements, small volumes (100, 150 and 200 ul) of CO, were injected into the respirometric mask, Then, the measured CO» volume was compared with the injected C02 volume, For the validation of the O» analyzer, Fused the N2 dilution technique (Fedak et al, 1981). The average deviation between the measured and expected gas vol- umes was less than 3%. Results Observations of Markea neurantha In the vicinity of plants of Markea neurantha (Fig, 1), 1 captured a total of 20 Glossophaga commissarist; 8 males and [2 females. The body mass of G. commissarisi ig. 1 An inflorescence of Markew neuruntha (Solanaceae)sa averaged 9.7 g (£0.7 SD). Glossophagine bats started to exploit flowers of M. neurantha approximately 1h after sunset (Fig, 2A), shortly after the corollas opened. Between 1800 and 2000 hours, hovering and clinging fiower visitations were performed at the same rate. Later in the night, more hovering than clinging visitations were recorded (Fig. 2B). In four out of eight nights, flower visitations were followed by maneuvering flights during which the bat hovered over the inflorescence or corolla for approximately 0.5 s. During these maneu- vering Nights, bats approached within a few centimeters of the inflorescence and then moved away again. This was repeated several times. These aerial maneuvers were most often observed early in the night (Fig. 2C). The corollas usually fell off of the inflorescences between 0100 and 0200 hours in the morning. During the eight nights of video recording, 152 clinging and 273 hovering flower visits were noted at 22 flowers of Markea neurantha, resulting in, on average, 80 Oy 60 40 Flower visitation event 1 (8) flower visits Ratio of clinging to hovering 14 () Frequeney of| manoeuvering fights 16 16 17 18 1920212223 0 + Time of day 2345678 Fig. 2A Mode of flower exploitation at 22 flowers of Markea neurantha during the night. The black burs represent the clinging fand the gray bars the hovering flower visitations. The shaded area indicates the night, B Ratio of clinging to hovering flower visitation rate during the night. During the course of a night, hovering flower exploitation increases in relation to clinging. C Frequency of anewering flights in front of inflorescences. Arrow indicates ‘mean time when corolla fell ff of inflorescence co co : Sul, Sof a o| Ed oD _—— foverna Conaig ,. 3 Duration (6) of hovering and clinging fower visitation of sglossophagine bats at Markea neurantha. Bats spent significantly Tess time exploiting the flower when visiting it in a hovering foraging mode 6.9 clinging and 12.4 hovering visits per flower per night. On 309 occasions, a bat was recorded passing by the flower without visiting the inflorescence. The mean duration of flower exploitation at M. neurantha equaled 0.390 s (0.184 SD) for clinging flower exploitation and 0.315 (£0.135 SD) for hovering flower exploitation (Fig. 3). The mean values were significantly different (Student test, df 8, P<0.001). The differ- ence between the two flower exploitation modes equaled 75 ms on average. Oxygen uptake and carbon dioxide release rate during clinging flower visitation The two males that were trained to cling to the artificial feeder weighed on average 10.3 g (+0.6 SD; male 1) and I1.1 g(+0.3 SD; male 2). The mean total duration of all clinging feeder visits per night equaled 42.3 min per night (22.3 SD) and on average the feeder was visited 388 times per night (+135 SD). Gas exchange rates were measured during 90 feeder visits of male | and 210 feeder visits of male 2. OF the recorded visits, the first male spent on average 5.3 s (£4.2 SD; median 5.6 s; range 0.2-14.6 s) and the sec- ond male 18.0 s (+15.9 SD; median 15.2 s; range 0.3 60.4 s) clinging to the flower. The rates of oxygen uptake are shown in Fig. 4A and those of CO, release in Fig. 4B. After the first seconds, during which the gas exchange rates are probably most likely affected by metabolism (see Voigt and Winter 1999), the oxygen uptake and carbon dioxide release rate of male 1 de~ creased significantly (Vox: .=~0.55, P<0.001; Vos! r= 0.19, P=0.031; n= 128 feeder visits longer than 10s), but remained’ almost constant in male 2 (Vor! 15=0.03, P=0.92; Veor! fp=-0.4, P=0.219; n= 15 feeder visits longer than 10s). As this difference may possibly be caused by the individual difference in the length of recorded feeder visits, I additionally analyzed‘Oxygen uptake rate (mi min” STPD) (mi min"* STPD) Carbon dioxide release rate (ce) Respiratory exchange ratio Duration of feeder visit (s) Fig. 4 Oxygen uptake rate (A), carbon dioxide release rate (B) and respiratory exchange rate (C) during clinging flower visitation in two Glaxsophaga soricina (solid circles: male 1; open cireles: male 2). Ench data point represents a single feeder visit. The cashed line in (A) indicates the expected oxygen uptake rate during forward flight according to Voigt and Winter (1999) and the dashed line in (C) indicates a respiratory exchange rate of 1 (combustion of sugar). In C mean values (4: 1 SD) were calculated for 10-s intervals whether feeder visits of up to 15s in male | showed values similar to those of male 2. As in male 2, the gas exchange rates of male | did not change significantly for feeder visits of 10-15 s lengths (Vo2: r=0.32, P=0.12; Vcor: re=0.13, P=0.53; n=25 feeder visits). In male 1, the respiratory gas exchange ratio (RER) increased sig- nificantly with the length of the feeder visits (7,=0.49, P<0.001, n=128 feeder visits), bul not in male 2 (r,=0.20, P=0.48; n= 15 feeder visits). To evaluate at what length of feeder visit the gas exchange ratio comes to an equilibrium, I calculated mean values for 10-8 intervals and tested whether mean values were different from 1 (assuming combustion of carbohydrates in a steady-state situation). According to this analysis, the mean RER of feeder visits of more than 20 s did not 54s deviate significantly from 1 (Table 1). Therefore, 1 cal- culated the mean oxygen uptake and carbon dioxide release rate for male | (male 2 only performed feeder visits shorter than 20 s). The oxygen uptake rate aver- aged 1.39 ml min~' (0.38 SD, STPD) and the mean carbon dioxide release rate equalled 1.33 mi min”! (£0.20 SD, STPD) Discussion Flower exploitation of Markea neurantha Among the many flowers that are known to be bat- pollinated in the Neotropics (review in Dobat 1985), the majority are exploited by hovering bats (von Helversen and Winter 2003). However some plants, such as ‘Markea neurantha, offer pollinators the choice between a clinging or a hovering flower visit. At the study site M, neurantha has to main pollinators: Glossophager commissarisi and Hylonycteris underwood (Tschapka 1998), Capture success at flowering M. newrantha indi- cated that the study plants were more likely visited by G. commissarisi than by H. underwoodi. The discrepancy between this and Tschapka’s study is probably caused by the fuct that most plants in this study were located in a clearing outside the forest, and H. underwood! is more associated with dense primary rainforest. Video recordings of the foraging mode on flowers of Markea neurantha showed that bats hovered twice as often in front of the pendulant inflorescences than they clung to the corollas; hovering flower visits were shorter than the clinging flower visits by only 75 ms, Thus, consistent, with the arguments of Pyke (1981), a hovering foraging mode possibly enabled bats to exploit the nectar re- source more quickly than the clinging foraging mode, Nonetheless, whether bats sampled more nectar during hovering visits than during clinging visits and whether a time benefit of 75 ms is biologically meaningful needs to be tested. The start of flower visitation coincided with the opening of the corolla around 1900 hours and the ending, of flower visitation with the deattachment of the corolla from the inflorescence 1-2 h after midnight. A closer Took at the temporal pattern of the foraging behavior reveals that during the carly evening hours glossopha- gine bats clung to the flowers as often as they hovered. Nectar secretion rates of M. neurantha show a temporal Table 1 Mean respiratory gas exchange ratio (RER) for 10-8 intervals following the onset of clinging. Deviations of mean RER ng two-tailed Student t-tests sp of y oP 10-20 082 02166559 <0,0001 20-30 097 022 07434046 30-40 099 019022, 24 O83. 40-50 1071753 4S 50-60 hor 0.19 L077 0.32$46 pattern with a peak rate of ca. 60 ul between 1900 and 2100 hours (Tschupka 1998). Thus, glossophagine bats hovered relatively more often in front of Markea flowers when the energy gain was low. This pattern is counter- intuitive, as the power requirements of hovering are higher than those of clinging. Possibly, hovering flights are preferred later at night because bats inspect the flowers instead of exploiting them. Or, exploitation of late-flowering M. neurantha can only be performed by hovering bats, because the remaining nectar droplets are hidden deep inside the corolla, and the bats need to perform sophisticated maneuvering flights to reach it ‘The average total nectar volume of M. neurantha flowers equals 280 yl (Tschapka 1998), Thus, an average of 19 flower visits at a single flower of M. neurantha would result in an uptake of 15 jl nectar per visit. In addition to exploiting Markea during hovering and clinging visitations, glossophagine bats also ap- proached the flowers without contacting the corollas. Based on the video observations, it difficult to understand the funetion of these maneuvering flights, Several explanations are plausible: (1) glossophagine bats mark the inflorescences by spraying urine on top of them, (2) bats memorize the position of the inflores- cences, (3) bats evaluate the quality of the nectar re- source from above, or (4) bats check the inflorescence for nectar-robbing ants that could potentially interfere with the bat during a flower visit. The latter two hypotheses are weak as the maneuvering flights were performed after the bats had already visited the flowers. Marking and memorizing could be especially important in the carly night and the temporal pattern of this behavior is consistent with this idea (Fig. 2C). However, these hypotheses still need to be tested. Power requirements of clinging versus hovering flower exploitation. Both male Glossophuga soricina that were trained to feed in a clinging mode performed longer feeder visits than hovering conspecifies at the same type of feeder. Winter et al. (1998), for example, recorded that G. soricina exploited feeders in a hovering mode for a maximum of 4.5 s. This difference can best be explained by the di ferent power requirements of hovering versus clinging flower visitation. The performance of hovering flight is potentially limited by the high metabolic costs. During clinging visits of more than 20 s, oxygen uptake of male | averaged 1.39 ml Op min", which equals # power in- put of 0.49 W fora 10.3-g bat or 47.4 W kg"! assuming complete combustion of sugar. According to the scaling curve of hovering flight power (Voigt and Winter 1999), a 10.3-g bat expends 1.7 W during flower exploitation. Therefore, power requirements of hovering Glossophaga soriciner exceed those of clinging by a factor of more than three. The nightly resting metabolic rate (RMR) of glossophagine bats can be calculated according to von Helversen and Winter (2003) and Arends et al. (1995): RM RW) = 1.5x{2.98xM(kg)"7""]. Thus, a 10-g bat expends 0.17 W during active resting ut the roost According to this, the power requirements of clinging flower visitation exceed those of resting metabolic rate by 0.32 W or by a factor of approximately three. Hovering flights as well as clinging flower visits were short for free-ranging glossophagine bats collecting nectar at Markea aeurantha, Based on the measured duration of flower visits, the estimate of energy ex- pended for hovering at Markea equals ca, 0.51 J (1.6 Wx0.32 5), whereas that for clinging equals ca, 0.20 J (0.5 Wx0.39 s). Thus, the energetic benefit of clinging instead of hovering amounts to only 0.3 J per fiower visit. As a 10-g glossophagine bat visits approxi- mately 800 flowers per night (von Helversen and Reyer 1984), the daily energy savings of hovering over clinging equal approximately 0.24 kJ. This is only ca, 0.5% of the daily energy expenditure (Voigt 2003; von Helversen and Winter 2003) and, thus, probably not a substantial energy saving for a small glossophagine bat. This picture changes for larger members of the sub- family Glossophaginae. The power requirements of hovering equal 3.1 W for a 20-g Leptonycteris curasoae (Voigt and Winter 1999). Assuming that the power requirements of clinging flower visitation scales to the same exponent as basal und resting metabolic rates (ca. 0.75), the power input during clinging flower visitation would amount to 0.8 W for a 20-g nectar-feeding bat Thus, assuming that L. curasoae spends approximately 0.5 s at flowers (von Helversen and Winter 2003), irre- spective of whether they cling or hover, and that hov- cring flower exploitation costs 3.1 W in 20-g L. eurasoae (Voigt and Winter 1999), the energy savings per flower Visit would equal ea. 1.2.J (Fig. 5B). If an Ll-g Anoura candifer with a daily energy expenditure of 52k) day”! has to feed at roughly 1,000 flowers per night (von Helversen and Reyer 1984), a 20-g L. curasoae most likely has to feed at approximately 1,600 flowers to meet its relatively higher daily energy requirements of approximately 81 kJ day"! (von Helversen and Winter 2003). If'a L. curasoae clung to the 1,600 flowers instead of hovering in front of them, the total energy saving would equal ca, 1.9 kJ day”! (1,600x1.2 J), which is approximately 2.3% of the bats’ daily energy expendi- ture, Although this value still seems’ small, it demon- strates that heavy glossophagine bats may benelit to a greater extent from clinging than smaller glossophagine bats. This, in turn, may explain why larger glossopha- gine bats, ie. Lepionycteris, are more likely to exploit flowers in a clinging rather than in a hovering foraging mode. Co-evolution of small-sized glossophagine bats and delicate glossophagine flowers It has been suggested that the evolution of hovering flight enabled glossophagine bats to outcompete other pollinators like sphingid moths or birds for flowerFrequency P, Leptonyctonis curesoae (209) Power input (W) ~ Giossophaga soricina (10 g) ctnging 4 6 8 10 12 14 16 18 20 22 24 26 28 90 32 Body mass (g) A Froquency distribution of body masses for 25 of 32 feeding bat species (Glossophaginae: Phyllostomi- body size in nectar-feeding bats will shift the modal value towards larger size classes (sources: von Helversen and Reyer 1984; Eisenberg 1989; Molinari 1994; Silva and Downing 1995; Sahley and Baraybar 1996; Reid 1997; ‘Tychapka 1998; Voigt and Winter 1999), B Interspecific sealing of power input for clinging, Peng Golld tine), and hovering, Proving (dotted line) visitations. Th scaling relationship of hovering flight equals: Phovering (W)= 128M (kg) (Voigt and Winter 1999) and that for efinging flower visils was estimated 88 Paingig(W)= 16.8%M(Ka)"", Two size classes are compared: 10-g Glossophaga sorieina and 20-g Leptoniyeteris euvasoue (see arrows) nectar, which led, in turn, to the radiation of the subfamily Glossophaginae in the Neotropics (von Helversen and Voigt 2002; von Helversen and Winter 2003). Possibly, large ancestral phyllostomid bats cap- tured insects at sphingid- or bird-pollinated flowers and then switched to a strategy of nectar uptake in a clinging foraging mode. Visited plants, on the other hand, may have adopted their new mutualistic partners by developing specific morphologies that facilitated fiower visitation through bats, ic., large nectar volume or scent, and that excluded other competing pollina- tors, i.¢., nocturnal anthesis, or pendulous and fragile flowers. The latter Mower characteristic particularly could have forced bats to employ costly hovering flights instead of the less costly clinging foraging mode to access the nectar deep inside the corollas. As the power input of hovering flight increases substantially with body mass (Voigt and Winter 1999), selection could have favored the evolution of small-sized flower- visiting bats (Fig. 5A). Summarizing, [ argue that large, ancestral phyllostomid bats probably clung to a7 flowers and that the hovering foraging mode evolved later as a response to more sophisticated Mower mor- phologies. Due to energetic constraints, glossophagine bats experienced a radiation among smaller members. The sophisticated adaptations of these bats to the hovering foraging mode is refiected in the fact that, given the choice between exploiting a flower in a clinging or hovering foraging mode, most glossopha- gine bats will remain airborn while imbibing the nec- tar. Acknowledgements For helpful comments on this manuseript thank Sylvia Ortmann and Herbert Hofer. The treatments of animals during the experiments complied with national guidelines. References Arends A, Bonaccorso FJ, Gonoud M (1995) Basal rates of metabolism of nectarivorous bats (Phyilostomidue) from a semiarid thorn forest in Venezuela. } Mamm 76:947-956 Bartholomew GA, Vieck D, Vieck CM (1981). Instantaneous measurements of oxygen consumption during pre uupand post-fight cooling in sphingid and saturniid moth. J Exp Biol 90:17-32 Buzato , Franco ALM (1992) Tetrastiis oli: a second case of bat-pollinated passionllower (Passiloraceae), Plant Syst Evol 181-261-267 Dobut K (1985), Bldten und Fledermiuse, Waldemar Kramer, Frankfurt am Main Dudley R, Winter ¥ (2002) Hovering fight mechanies of Neo- tcopical flower bats (Phyllostomide: Glossophaginue) in nor- modense and hypodense gas mixtures. J Exp Biol 205:2325- 2336 Eisenberg JE (1989) Mammals of the Neotropies: vol 1. University ‘of Chicago Press, Chicago Fedak MA, Rome L, Seeherman HJ (1981) One-step Ne-dilution technique for calibrating open-circuit Vox measuring systems. J Appl Physiol Respir Environ Exercise Physiol 51:772-776 Hainsworth FR, Wolf LL (1972) Bnergeties of nectar extraction in ‘small, high allitude, tropical hunminghird Selasphorus flan ja. § Coinp Physiol 80:377-387 haus ER, Opler PR, Baker HG nation of Bauhinia pauierie. pl logy $5:412-419 i LG, Hobson KA, Miron L, Ramirez P, Mendez. CG, Sanchez-Cordero V (2001) Sourees of protein in two species of phytophagous bats in a seasonal dry forest: evidence from Stable-isotope analysis, J Mammal £2:352-361 Kleiber M (1975) The fie of life: an introduetion to animal ener- goties. Krieger, Boca Raton, pp 451 Koopman K (1981) The distributional pattern of New World nectar-feeding bats. Ann Missouri Bot Gare 68:352-369 Molinari J (1994) A new species of Anoura (Mammalit, Chirop tera, Phyllastomidue) trom the Andes of northein South America. Trop Zool 7:73-86 Pyke GH (1981) Why hummingbirds hover and honeyeaters perch ‘Anim Behav 29:861-867 Rayner IMY (1990) The mechanies of fight and bird migration performance, In: Gwinner TE (ed) Bind migration, Springer, Berlin Heidelberg New York, pp 283-299 Reid F (1997) A field guide to the mammals of Central America ‘and Southeast Mexico, Oxford University Press, Oxford Sahley CT, Baraybar LE (1996) Natura hislory of the longesnou- ted bat, Plural genovenstun (Phyllostomidae: Glossophiigi- ne) in southwestern Peru, Vid Silv Neotrop 5:101-109) Silva M, Downing JA (1995) Handbook of mammalian body mass. CRC Press, Boca Raton 974) Bat activity and polli- lant-pollinator coevalution,548 ‘Speakman JR, Racey PA (1991) No cost of echolocation for bats in flight. Nature 350:421-423 Suatez RK, Lighton JRB, Moyes CD, Brown GS, Glass CL, Ho- ‘chachkw PW (1990) Fuel selection in rufous’ hummingbicds: ‘ecological implications of metabolic biochemistry. Proc Natl ‘Acad Sei USA 87:9207-9209 ‘Timm RM, LaVal RK (1998) A ffelé key to the bats of Costa Rice, Qceas Publ Ser Center Latin Am Siud (University of Kansas) m2 Tschapka M (1998) Koexistenz. und Ressourcennutzung in ciner Artengemeinschall von Blumenfledermausen (Phyliostomidae: Glossophaginae) im atlantischen Tieflandregenwald Costa Ri- cas, PhD Thesis, University of Erlangen, Germany gt CC (2000) Intraspecific sealing of flight power in the bat Glossophaga soricina (Phyllostomidae). J Comp Physiol B 170:403-410 Voigt CC (2003) Reproductive energetics of the nectar-feeding bat Glossophaga soricina (Phyllostomidas). J Comp Physiol B 173:79-85, Voigt CC, Winter ¥ (1999) Energetic cost of hovering flight in nectar-feecing bats (Phyllostomidae: Glossophaginge) and its scaling in birds, bats and moths, J Comp Physiol B 169:38-48 von Helversen D, von Helversen © (1999) Acoustic guide ina bat- pollinated flower. Nature 398:759-760 vi von Helversen O, Reyer HU (1984) Nectar intake and energy expenditure in a ower visiting bat, Oecologia 63:178-184 von Helversen O, Voigt CC (2002) Glossophagine bat pollination in Helicteres barensis (Sterculiaceae). Ecotropica 8:23-30 von Helversen O, Winter Y (2003) Glossophagine bats and their flowers: costs and benefits for plants and pollinators. Ia: Kunz. ‘TH, Fenton MB (eds) Bat ecology. University of Chicago Press, Chicago, pp 346-397 von Helversen O, Winkler L, Bestmann JH (2000) Suiphur eon- taining “perfumes” attract Nower-visting bats. J Comp Physiol A 186:143-153 Winter Y (1998) Energetic cost of hovering flight in a nectar- feeding bat measured with fast-tesponse respirometry. J Comp Physiol B 168:434-444 Winter Y, von Heiversen O (1998) The energy cost of Might: do small bats fly more cheaply than birds? J Comp Physiol B 168:105-111 Winter Y, Voigt C, von Helversen O (1998) Gas exchange during hovering flight in a nectar-feeding bat, Glossophaga soricina. 3 Exp Biol 201:237-244 Withers PC (1977) Measurement of Yoo, Veo, and evaporative ‘water loss with a flow-through mask. J Appl Paysiol 42:120-123