Evolution, 32(4), 1978, pp. 740-751 RELATIVE BRAIN SIZE AND FEEDING STRATEGIES IN THE CHIROPTERA JOHN F. EISENBERG AND DON E, WILSON National Zoological Park, Smithsonian Institution, Washington, D.C., and National Fish and Wildlife Laboratory, National Museum of Natural History, Washington, D.C. Received July 22, 1977. Revised December 1, 1977 Pirlot and Stephan (1970) demonstrated that the relative brain size varied greatly when nine families of Chiroptera were compared. Some SI species were studied and an average index of encephalization for each family was calculated. They pointed out that the Pteropidae, Desmo- dontinae, and Noctilionidae had the high- est indices of encephalization for their sample. Stephan and Pirlot (1970) expand- ed the above observations by comparing the relative volumes of 11 brain structures for 18 species drawn from eight families. These volumes were then compared with equivalent volumes derived for a series of "basal" insectivores (Suncus, S or ex, Ten- t?e, Hemicentetes). Although specific brain structures evolve at different rates within each family with some structures declining in relative size and others in- creasing, a generalized correlation of brain size with feeding habits could be made. The relative size of the neocortex showed a strong positive correlation with the de- gree of encephalization. This correlation in turn could be related to dietary spe- cializations. They proposed that the Noc- tilionidae, Desmodontinae, and Pteropi- dae showed the highest development of the neocortex, but were broadly over- lapped by certain species of the Phyllosto- matidae. The order Chiroptera includes two dis- tinct lineages which have been separated since the Eocene: the Pteropidae or Mega- chiroptera and the Microchiroptera. The ancestral Pteropidae may have at- tained the capacity for flight after diverg- ing from a flightless common ancestor which in turn gave rise to the ancestral microchiropteran. The species of the mi- crochiropteran family Phyllostomatidae are rather unified morphologically, but their origins are somewhat obscure (Smith, 1976). No doubt they separated from an- cestral microchiropterans before the Oli- goc?ne and radiated in the Neotropics without competition from the Megachi- roptera. In their exploitation of feeding niches in the Neotropics through adaptive radiation, the phyllostomatid bats exhibit the principle of complementarity (Darling- ton, 195 7) in that they parallel the niche subdivision seen in the radiation of the Megachiroptera in the Palaeotropics. We measured the cranial capacities of 225 species of bats from 14 families in- cluding all families and genera previously studied by Stephan and Pirlot (1970). We wished to confirm their conclusions about encephalization and refine their first order correlation with dietary specializations. We did not use actual brain weights but estimated brain volume from a measure- ment of cranial capacity, using lead shot and volumetric correction constants. METHODS Specimens were selected from the re- search collections at the National Museum of Natural History which had both an in- tact skull and a field recorded body weight. To eliminate weight changes in- duced by pregnancy, we used only adult males in the computations. We selected at least four specimens of each species and determined the cranial capacity of each specimen twice, according to the following procedure. The skull was weighed to the nearest 0.01 g. If the freshly collected specimen weighed less than 150 g. No. 3 dust shot was introduced into the cranial cavity through the foramen magnum and the skull reweighed. The procedure was repeated after emptying out the original shot. Larger specimens were treated in the 740 BRAIN SIZE AND FEEDING STRATEGIES IN BATS 741 TABLE 1. Comparisons of absolute and estimated brain weight for 11 species of bats. Known brain Estimated brain weight (g)' weight' (g) Taxon Range X Rhinolophus hipposideros 0.150 0.11-0.20 0.15 Glossophaga soricina 0.358 0.38-0.41 0.40 Phyllostomus discolor 0.836 1.03-1.11 1.08 Carollia perspicillata 0.442 0.55-0.60 0.58 Stumira lilium 0.487 0.S3-O.S9 0.56 Artibeus jamaicensis 0.87 0.91-1.15 1.04 Artibeus lituratus 0.98 1.12-1.35 1.26 Desmodus rotundus 0.78 0.89-1.00 0.94 Noctilio leporinus 1.189 1.23-1.35 1.29 Cynopterus brachyotis 0.803 0.95-1.19 1.05 Cynopterus horsfieldi 1.144 1.06-1.31 1.20 ' Pirlot and Stephan (1970). ^ See text for method of estimation same manner with the exception that No. 10 dust shot was used. We then calculated species averages based on at least eight measurements for each species. The weight (Wt.) of the shot was converted to volume (F) in cubic centimeters (cm'*) by using an empirically derived constant (K) and the formula Wt. ^ K = V. For No. 10 shot, K = 6.0; for No. 3 shot, K = 6.4. The accuracy of the method was tested by comparing the calculated volume of the cranial capacity with known brain weights, as published by Pirlot and Stephan (1970) (see Table 1). Since the brain nearly fills the cranial cavity and the specific gravity of the brain nearly equals the specific gravity of water, cranial volumes in cm' should approximate the brain weight in grams. Pirlot and Stephan analyzed only single brains for each species and our volumes represent a series of at least eight mea- surements of four different specimens, thus it is difficult to compare the two sam- ples. The data for actual brain weights of Rhinolophus hipposideros and Cynopterus horsfieldi fall within our volumetric ranges for these species. The actual brain weights for Glossophaga soricina, Arti- beus jamaicensis, and Noctilio leporinus fall within 0.04 g of our low range. This implies that our volumetric values could be from 4 to 6% higher than the actual brain weights. Since our technique takes no account of the space occupied by the meninges, this discrepancy seems plausi- ble. The striking fact is that the corre- spondence is rather good. We decided to proceed with the comparison of the cranial volumes to body weight ratios, bearing in mind that a slight overestimate of actual brain size might be involved. RESULTS We examined 225 species from 88 gen- era and 14 families (see Fig. 1). The av- erage cranial capacity of each species has been regressed on the average body weight by using a double logm plot. Given the general equation V = bX" this can be ex- pressed in log form: log V = log b + a log X where Y = average cranial capacity; A' = average body weight; a = slope of the log regression and b = the Y intercept. In surveying the data the following state- ments can be made: over a series of body weights, the cranial volumes for almost all families show similar slopes. The Pterop- idae, Mormoopidae, Molossidae, and Rhinolophidae have the highest correla- tion coefficients. The Vespertilionidae show the most variation within a family (see Table 2). The Rhinopomatidae, Mo- lossidae, Emballonuridae, Mormoopidae and Vespertilionidae have the lowest b values (logioF intercept) and show on the average only a slight increase in average cranial capacity over that of the "basic in- 742 + 096- + 075- + 0.55- + 0.34- +0.14 - 0.07- - 0.28- -048- - 0.69- -Q89 -1,10. J. F. EISENBERG AND D. E, WILSON o o XO X o oox OO X ? o ooo OO o OO OO? OO o o o X ox o o o X OO o o o o xoo?o? o o o OO o o? OO o o o ooo o X o ooo oooxo X o OO ox o o ooo o oO oox o o o o o o X ooooo ox o o o OO o XO OO ? OO ? o o o ? ox X ox o o o X o o o o OO X ooo o o o o o o OO X o o= 1 x= 2 ?= 3 0= 4 040 0.66 093 119 145 1,71 198 224 250 277 3.03 FiG. 1. Distribution of log,,) average cranial volumes (ordinale) regressed against logio average body weights (abscissa) for 225 species of Chiroptera. Where points overlap appropriate symbols have been used to indicate the numbers of species involved. 1,2,3, and 4 = numbers of species. sectivores" (Bauchot and Stephan, 1966). The Megachiroptera (Pteropidae), Phyl- lostomatidae, Megadermatidae, Noctil- ionidae and Nycteridae display the high- est cranial capacities for their weight classes. The Rhinolophoidea are interme- TABLE 2. Summary of regression and correlation analysis, using log Y = log h + a log X. Taxon Slope (a) Intercept log,? (f>) SE Chiroptera Pteropidae Emballonuridae .802 .661 .697 -1.368 -1.0186 -1.323 .04 .07 Rhinolophidae Hipposideridae Mormoopidae Phyllostomatidae .597 .565 .737 .683 -1.195 -1.136 -1.332 -1.120 .04 .08 .02 .04 Vespertilionidae Molossidae .589 .686 -1.300 -1.359 .09 .05 diate. (When the families Hipposideridae and Rhinolophidae are referred to as a unit, we employ the superfamilial cate- gory Rhinolophoidea.) Given that the Pteropidae show many adaptations differing from the other fam- ilies of the Chiroptera, it is interesting to note that many phyllostomatid bats attain a cranial capacity equal to that shown by the pteropids in the same weight class. The data suggest that aerial insectivores that capture prey by flying and rely almost completely on echolocation have the low- est brain size to body weight ratios. On the other hand, insectivorous bats that discretely sample microhabitats (e.g., Hipposideridae) or utilize a complex for- aging strategy involving vision and sonar to exploit micro-niches in the vegetation BRAIN SIZE AND FEEDING STRATEGIES IN' BATS 743 + 90- + .30- 292 320 FIG. 2. Regression analysis comparing the slopes and intercepts (in logs) for frugivores (higher slope) and aerial insectivores (lower slope), according to Wilson (1973). Scales as in Fig. 1. have "intermediate" brain size to body weight ratios (e.g., some Vespertilionidae and some Phyllostomatidae). The strong correlation (r = 0.9S) for the data in Fig- ure 1 encouraged us to examine subsets of the data in order to suggest functional rea- sons for the variation demonstrated. In order to test the hypothesis that rel- ative brain size is a reflection of feeding strategy, we used only species which Wil- son (1973) had unequivocally allocated as either "Frugivores," or "Aerial Insecti- vores." A regression analysis comparing these two groups (Fig. 2) showed that fru- givores of the family Pteropidae and the phyllostomatid subfamily Stenoderminae had consistently larger brain volumes for a given body weight than aerial insecti- vores drawn from the families Mormoop- idae, Emballonuridae, Vespertilionidae, and Molossidae. That there is no overlap between these two groups is central to our hypothesis: relative cranial capacity is a reflection of feeding strategy. Once regression lines for aerial insecti- vores and frugivores are established, it was possible to plot other species of the Chiroptera with different feeding strate- gies in comparison with those lines (Fig. 3). Carnivores, piscivores, sanguivores, ? and foliage gleaners fall between the aerial insectivore and frugivore regression lines. For some of these species for which food habits and feeding strategy are poorly known, it may be possible to infer some- thing about their habits from their brain size. For example, Nycieris arge and Mi- cronycteris megalotis both have relatively larger cranial volumes than their conge- ners, and in fact fall on the regression line for frugivores. Nycteris arge and A', gam- biensis are exactly the same size, but N. 744 J. F. EISENBERG AND D. E. WILSON + 90- / + 70- X + 50- XX +.30- /o 0 y^ +.10- ylo X y^ -10- / 0 / -30- -50- J' ?Ng o? X * Foliage gleaners ? Carnivores X Piscivores -70- /^ y ? Sanguivores -90- X -11 X 1 1 1 1 1 lilil? .96 1.24 152 2.08 236 264 292 320 FIG. 3. The average cranial volumes for foliage gleaners, carnivores, piscivores and sanguivores shown with the regression lines for frugivores and aerial insectivores from Fig. 2. Na = Nycteris arge; Mm = Micronycteris megalotis; Ng = Nycteris gambiensis. arge has a larger volume; that increase may well relate to differences in feeding strategy and habitat selection. Nectarivorous bats (Fig. 4) fall close to the frugivore line, and might be consid- ered to use the same strategy for locating flowers as do frugivores for locating fruit. The two species of pteropid nectar-feed- ers, Megaloglossus and Nanonycteris, fall above the frugivore line and most of the glossophagine and phyllostomatine nec- tarivores fall slightly below it. Using the regression lines generated in Figure 2 as standards for comparison, we examined various families and subfamilies of bats as phylogenetic subsets. Figure S shows the family Pteropidae which is clus- tered around the regression line for fru- givores, except that very large forms seemed to be displaced slightly below the line as might be expected from allometric considerations (Jerison, 1973). Hypsigna- thus monstrosus, which fell below the line, has been reported to be partially carniv- orous (Van Deusen, 1968), and this de- parture from frugivory may affect its dis- placement, although the teeth of this bat do not seem adapted for carnivory. The larger species of the subfamily Stenoderminae and the family Vespertili- onidae (Fig. 6) also are displaced slightly downward, but this may be a simple result of general allometric trends. Referring to Figure 6 the Vespertilionidae appears to be one of the most variable groups of all, and causes much of the scatter around the regression line for aerial insectivores in Figure 2. The gtmis Myotis is responsible for much of this scatter. Therefore we ex- amined the genus Myotis more closely. BRAIN SIZE AND FEEDING STRATEGIES IN BATS 745 3.20 FIG. 4. Relative cranial volumes of nectarivorous bats plotted against the regression lines for frugivores and aerial insectivores from Fig. 2. M = Megaloglossus; N = Nanonycteris. and we were fortunate to obtain the data set assembled by James S. Findley for his study of phenetics in the genus Myotis (Findley, 1973). Findley (1973) suggested that one group of species in the genus Myotis might be adapted to "pick prey from various sub- strata, ..." or, in our terminology, to glean foliage. All species in this group had high positive loadings for the first factor in his multivariate analysis and included Myotis auriculus and M. thysanodes, the two species farthest from the aerial insec- tivore line in Figure 6. These two species are rather small. Although Findley's fac- tor / contains loadings for size, it also con- tains an entire suite of characters which are unrelated to size but do correlate with hovering flight. We regressed Findley's factor / scores against a ratio of cranial volume to body weight (x wt. shot/x body wt.) for nine species of Myotis to see if part of the cranial volume variation with- in the genus might be explained by mor- phological features which Findley thought reflected actual differences in feeding strategies. The results of that analysis in- dicate that there is a positive correlation (r = 0.87) between relative cranial vol- ume and Findley's factor / phenetic mea- sure. This relation strengthened our inter- pretation that the variation shown in Figure 6 is a reflection of different forag- ing strategies within the genus Myotis. Larger mammals generally have a lower brain to body weight value than do small- er ones. The allometry of brain to body weight ratios as a function of size has been analyzed by Jerison (1973). In his analysis of the class Mammalia he determined that the slope for the regression of logio brain weight against logio body weight was J. F. EISENBERG AND D. E. WILSON 3.20 FIG. S. Cranial volumes for the Fteropidae shown with the regression lines for frugivores and aerial insectivores from Fig. 2. The dotted line is the regression for the Pteropidae. Hm = Hypsignathus monstro- 0.66. His sample had few bats. Using our series of bats and other mammal data col- lected by Eisenberg (unpubl.), we recal- culated a regression for the entire class Mammalia which gave a slope of 0.74 and an intercept of logio = ?1.26. We then recalculated encephalization quotients {EQ) for each species of the Chiroptera and for those members of the Insect?vora for which we had data. EQ is the ratio between the observed brain volume and expected brain volume. EQ values were determined using the general formula from Jerison (1973) with our newly deter- mined values for the mammalian regres- sion line. The expected volume {Ev) is cal- culated from the known mean body weight (Wt.) and the formula: Ev = O.OSS: Wt''''*. Table 3 presents the EQ av- erage values for four insectivore families and IS chiropteran families. Utilizing EQ values, we avoid the effects of negative allometry when comparing large and small forms. Clearly the Pteropidae, Phyl- lostomatidae, and Thyropteridae have rel- atively large cranial volumes. The Megad- ermatidae, Nycteridae and Natalidae are not far behind. The bat families with the lowest EQ values are far in advance of the values for the "basal insectivores" and higher than most insectivore families with the exception of the fossorial and semi- aquatic Talpidae. Table 4 presents the average EQ value for each trophic role as defined by Wilson (1973). In composing this table we includ- ed those species which seemed to fall un- equivocally in a given trophic role. In con- formity with the regression analysis, nectarivores and frugivores have the high- BRAIN SIZE AND FEEDING STRATEGIES IN BATS 747 + 90 + 70- + .50 +.30- + 10- -.10- ^,30- <^ Stenoderminae o Vespertilionidae 2.92 3.20 FIG. 6. Relative cranial volumes of the Stenoderminae and Vespertilionidae plotted against the regression lines for frugivores and aerial insectivores from Fig. 2. Note the scatter in the Vespertilionidae. Ma = Myolis auriculus; Mt = Myotis thysanodes. est EQ values closely followed by the san- guivores. Carnivores come next followed by foliage gleaners and piscivores. The aerial insectivores have the lowest en- cephalization quotients. The classification scheme of Wilson was based on a survey of the literature com- bined with first hand knowledge of some phyllostomatids and vespertilionids. The scatter around the regression lines and the standard deviations for EQ values suggest some diversity of feeding strategies by the species of selected genera or subfamilies. Encephalization quotients when calcu- lated for the subfamilies of the Phyllosto- matidae indicate that the Phyllostomati- nae fall near the mean for foliage gleaners. Mimon crenulatuni {EQ = 0.86) and To- natia sylvicola {EQ = 0.92) are thought to be foliage gleaners (Wilson, 1973; Gard- ner, 1977). Micronycterismegalotis (EQ = 1.28) is thought to be a frugivore (Good- win and Greenhall, 1961). The Desmo- dontinae have encephalization quotients nearly as high as the frugivores. The Emballonuridae and Molossidae have EQ values near those of the aerial insectivores but some emballonurids with high EQ values may well be foliage glean- ers, e.g. Taphazous longimanus (EQ = 0.88) and Diclidurus isabellus (EQ = 0.8S). The Rhinolophidae have EQ values near the aerial insectivore average, but the Mormoopidae and Hipposideridae are somewhat higher. Some members of the Hipposideridae may forage in a manner similar to classical foliage gleaners such as Hipposideros cyclops (EQ = 0.86). The Natalidae fall near the foliage gleaner average and the Rhinopomatidae 748 J. F. EISENBERG AND D. E. WILSON TABLE 3. Encephalization quotients for insectivores and bals. Range Chiroptera Megachiroptera Pteropidae Microchiroptera Rhinopomatidae Emballonuridae Noctilionidae Nycteridae Megadermatidae Rhinolophidae Phyllostomatidae Vespertilionidae Molossidae Hipposideridae Natalidae Furipteridae Thyropteridae Mormoopidae Insect?vora Solenodontidae Tenrecidae Soricidae Talpidae "Basal" Insect?vora Tenrec ecaudatus Hemicentetes semispinosus Setifer setosus Echinops telfairi S or ex murinus Sorex minutus Sorex araneus Erinaceus europeaus 0.9S-1.47 O.S9-0.62 0.64-1.23 0.74-0.93 0.94-1.28 0.93-1.14 0.73-0.93 0.79-1.41 0.39-1.00 0.47-0.88 0.S4-1.31 0.7S-0.87 O.SS-0.69 0.31-0.98 0.37-0.80 0.78-1.34 1.20 ? 0.16 0.61 ? 0.01 0.76 ? 0.13 0.84 1.08 1.06 ? 0.11 0.79 ? 0.09 1.13 ? 0.13 0.66 ? 0.14 0.6S ? 0.09 0.82 ? 0.21 1.00 0.84 1.20 0.81 ? O.OS 0.S9 ? 0.08 0.S3 ? 0.20 0.61 ? 0.13 1.02 ? 0.26 0.42 0.43 0.4S 0.40 0.47 O.SS 0.62 O.Sl fall near the low range for the aerial in- sectivores. The Noctilionidae contains two quite different feeding strategies. Noctilo leporinus feeds on fish and its EQ is 0.928. Noctilio albiventris (=labialis) on the other hand, is a known aerial in- TABLE 4. role. Encephalization quotients trophic EQ Trophic role* Range Mean ? SD ? Carnivore 0.8-1.20 1.05 ? .13 Sanguivore 1.10-1.22 1.17 Piscivore 0.74-1.20 0.94 ? .20 Foliage Gleaning 0.61-1.13 0.93 ? .21 Frugivore 0.81-1.47 1.18 ? .14 Nectarivore 0.88-1,43 1.22 ? .14 Aerial Insectivore 0.39-1.1 0.70 ? .14 * According to Wilson, 1973. sectivore (Wilson, 1973), and \is EQ value is also as expected (0.74). In the Furipter- idae, Furipterus horrens was suspected by Wilson (1973) to be an aerial insectivore, but its EQ value (0.84S) suggests that it is a foliage gleaner. Thyroptera tricolor (Wil- son, 1973) is an aerial insectivore, but its high EQ (1.26) is at variance with other members of this trophic class. This ap- parent contradiction may be due to the unusual roosting habits of this species. These bats roost in the rolled leaves of Heliconia plants and are forced to change roosts about every night (Findley and Wilson, 1974). Thus, the searching behav- ior used in censusing the environment for developing leaves to be used as future roosts may be similar to that of a frugivore censusing fruit trees. BRAIN SIZE AND FEEDING STRATEGIES IN BATS 749 DISCUSSION Findley (1969), who measured the chi- ropteran cranial capacity from a sample almost equivalent to ours, established a correlation between cranial capacity and the area of the foramen magnum. His published plots are not directly compara- ble with our data, but he could find no specific trends with respect to cranial ca- pacity and phylogeny. Eisenberg (1975) noted that the cranial capacities of bats exceeded those of "basal" insectivores of a comparable weight, but the data plotted in his Figure 4 exaggerate the statement because they were all drawn entirely from the Phyllostomatidae. At the time he pre- pared the figure, it was not appreciated how unique the phyllostomatid bats were with regard to the remainder of the Mi- crochiroptera. Jerison (1973) brie?y considers the Chi- roptera, but he included no specific treat- ment of the trends in brain-body weight correlations for this taxon. His admirable survey of the class Mammalia permitted the conclusion that through time the phy- logenetic trend has been an increase in rel- ative brain size for all higher taxa of mam- mals on the contiguous continental land masses. Given the differences in relative cranial volume and presumed brain size, we might ask how these differences re?ect variations in the relative size of discrete brain structures. As early as 1903 Dr?seke noted that the brain of a megachiropteran resembled that of a strepsirhine primate and differed markedly from the brains of other chiropterans. Henson (1970) wrote that the microchiropterans have enlarged the eighth cranial nerve and all major sub- cortical ganglia associated with it. This specialization correlates with the evolu- tion of their echolocating ability. On the other hand, the Megachiroptera show lit- tle specialization of the structures associ- ated with the eighth cranial nerve, but do show a high development of the visual and olfactory centers. Lutgemeier (1962) indi- cates that the optic system accounts for about 9.4% of the brain stem in Pteropus but only 5.8% in Pipistrellus. Mann (1963a) pointed out that the olfactory bulbs accounted for 57% of the "funda- mental encephalon" in Pteropus, but only 21.6% in Myotis. Artibeus was interme- diate with a value of 49%. Nevertheless, reduction in the relative size of neuronal structures associated with olfaction and vision in the Microchiroptera does not im- ply that these sensory modalities are un- involved in orientation (Suthers, 1970). What is implied is that vision and olfac- tion are supplementary to echolocation. As a general rule, the Megachiroptera show a well developed cerebellum com- pared with most microchiropterans al- though the Phyllostomatidae show a trend toward an enlarged cerebellum (Henson, 1970), especially the Stenoderminae and Desmondontinae (McDaniel, 1976). It seems possible then to conceive of the chi- ropteran brain evolving in a mosaic fash- ion as natural selection acts on the various populations. Where feeding strategies converge, we find morphological conver- gence in brain structure and relative size. Although the relatively large size of the brain found in the Pteropidae and Phyl- lostomatidae may be related to the need for maintenance of a large mass of neural tissue devoted to olfaction and vision, one may well ask if the neocortex shows any trends in relative size across taxa. Wirz (1954) noted that the neocortex of the Mi- crochiroptera was better developed than that of the "basal" insectivores. Mann (1963?) noted that the neocortex of the Pteropidae as well as of the phyllostoma- tid genera Artibeus and Desmodus was far better developed than that of the insecti- vorous Microchiroptera which he studied. The insectivorous phyllostomatid genera Mimon and Lonchorina have the least modified brain structure of the family Phyllostomatidae (McDaniel, 1976). Large cranial volumes with their larger brain size imply a relatively large amount of neocortex, and these data further sug- gest that any departure from aerial insec- tivory as a trophic strategy may be accom- panied not only by modifications in the 750 J. F. EISENBERG AND D. E. WILSON visual and olfactory projection areas but also in the relative amount of neocortex itself. Stephan and Pirlot (1970), who made discrete volumetric measurements on the brains of several chiropterans, con- cluded that, although the neocortex was somewhat advanced relative to the basal insectivores in the Hipposideridae, Molos- sidae, Rhinolophidae and Vespertilioni- dae, the neocortex showed profound en- largement in the Pteropidae, Phyl- lostomatidae and the Noctilionidae. Again, it is suggested that a relatively large brain is not only a correlate of large olfactory bulbs or relatively large eyes but also of a relatively larger neocortex. Relatively large cranial volumes and larger brains with a large neocortical vol- ume have evidently evolved independ- ently several times within the order Chi- roptera. The stenodermine and glossophagine bats have replicated a trend in brain structure comparable to that of the Pteropidae. The most specialized chi- ropterans for flight, the Molossidae and some Vespertilionidae, do not exhibit pro- found increases in relative brain size when compared to the non-volant Insect?vora. Either such specialized aerial insectivores have been under selective pressure to maintain a reduced organ mass (including the brain) to facilitate active flight or spe- cialization for echolocation as the prin- ciple means of locating aerial prey does not require a large mass of neocortical material. Thus, such specialization per- mits a reduction in relative brain tissue devoted to connections and projections from the other sensory inputs compared to foliage gleaners and other trophic spe- cialists. We submit that a foraging strat- egy based on locating relatively large pockets of energy-rich food that are un- predictable in temporal and spatial distri- bution necessitates the use of a complicat- ed information storage and retrieval system involving input from several sense organs. The enlarged neocortex and as- sociated subcortical structure reflects this adaptive strategy for nectarivores, frugi- vores, carnivores, piscivores, and sangui- vores. SUMMARY Brain weights were estimated from the cranial volumes of 225 species of Chirop- tera. The log of the average cranial vol- ume was regressed against the log of the average body weight and the results ana- lyzed for correlations with phylogenetic affinity and foraging strategy. It is con- cluded that the family Phyllostomatidae shows a strong convergence of brain to body weight ratios toward the patterns shown by the Pteropidae. Foraging strat- egies involving the location of rich food resources which are isolated in small pockets seem to require a large brain weight relative to body mass. If we as- sume the ancestral chiropteran had a brain structured more like a terrestrial in- sectivore, then the highly specialized ae- rial insectivores with the lowest relative brain to body weight ratios of the extant chiroptera reflect an evolutionary tendency to maintain the brain mass at a minimum weight. ACKNOWLEDGMENTS We thank Ronald Giegerich who assist- ed with the volumetric measurements; R. Thorington who originally developed the technique for volumetric measurements and assisted us throughout; C. O. Hand- ley who generously made the Smithsoni- an-Venezuelan Project phyllostomatid collection available to us; J. S. Findley who allowed us the use of his data on Myotis phenetics; and M. A. Bogan, A. L. Gardner, and G. G. Musser, who pro- vided editorial criticism. LITERATURE CITED BAUCHOT, R., AND H. STEPHAN. 1966. Donees nouvelles sur I'encephalisation des Insectivores et des Prosimiens. Mammalia 30:160-196. DARLINGTON, P. J., JR. 1957. Zoogeography. J. Wiley, New York. 675 p. DRASEKE, J. 1903. Das Gehirn der Chiropteren. Monatsschr. Psychiat. Neurol. 13:448-463. EISENBERG, J. F. 1975. Phylogeny, behavior and BRAIN SIZE AND FEEDING STRATEGIES IN BATS 751 ecology in the Mammalia, p. 47-68. In W. P. Luckett and F. Szalay (eds.), Phylogeny of Pri- mates. Plenum, New York. FiNDLEY, J. S. 1969. Brain size in bats. J. Mam- mal. 50(2):340-344. . 1973. Phenetic relationships among bats of the genus Myolis. Systematic Zool. 21:31-52. FiNDLEY, J. S., AND D. E. WILSON. 1974. Obser- vations on the neotropical disk-winged bat, Thy- roplera tricolor. J. Mammal. 55:562-571. GARDNER, A. L. 1977. Feeding Habits, p. 293-350. In R. J. Baker, J. K. Jones, Jr., and D. C. Carter (eds.). Biology of Bats of the New World Family Phyllostomatidae, Part II. Spec. Publ. The Mu- seum, Texas Tech. University. GOODWIN, G. G., AND A. H. GREENHALL. 1961. A review of the bats of Trinidad and Tobago. Bull. Amer. Mus. Nat. Hist. 122:187-302. HENSON, O. W., JR. 1970. The central nervous system, p. 57-152. In W. A. Wimsatt (ed.). Bi- ology of Bats, Vol. II. Academic Press, New York. JERISON, H. 1973. The Evolution of the Brain and Intelligence. Academic Press, New York. L?TGEMEIER, F. 1962. Histologische Sonderheiten der Gehirne von Chiropteren Korpergrosse und Verschiedener Orientierungsweise. Z. Morphol. Oekol. Tiere 50:587-725. MANN, G. 1963O. The rhinencephalon of Chirop- tera. Invest. Zool. Chil. 9:1-93. . 1963fc. Phylogeny and cortical evolution in the Chiroptera. Evolution 17:589-591. MCDANIEL, V. R. 1976. Brain anatomy, p. 147- 200. In R. J. Baker, J. Knox Jones, and D. Car- ter (eds.). Biology of Bats of the New World Family Phyllostomatidae, Part I. Special Publi- cations, the Museum, Texas Tech. University # 10. PiRLOT, P., AND H. STEPHAN. 1970. EncephaU- zation in Chiroptera. Can. J. Zool. 48:433-444. SMITH, J. D. 1976. Chiropteran evolution, p. 49- 70. In R. J. Baker, J. Knox Jones, and D. Carter (eds.). Biology of Bats of the New World Family Phyllostomatidae, Part I. Special Publications, The Museum, Texas Tech. University # 10. STEPHAN, H., AND P. PIRLOT, 1970. Volumetrie comparisons of brain structures in bats, Z. Zool. Systematik und Evolutionsforschung 8:200-236. SUTHERS, R. A. 1970. Vision, olfaction and taste, p. 265-309. In W. A. Wimsatt (ed.). Biology of Bats, Vol. II. Academic Press, New York. VAN DEUSEN, H. M. 1968. Carnivorous habits of Hypsignathus monstrosus. J. Mammal. 49:335- 336. WILSON, D. E. 1973. Bat faunas: A trophic com- parison. Systematic Zool. 22:14-29. WiRZ, K. 1954. Zur Quantitativen Bestimmungen der Rangordung bei S?ugetieren. Acta Anat?mica 9:134-196.