Revista Chilena de Historia Natural 59: 135-150, 1986 The structure of Neotropical bat communities: a preliminary analysis
La estructura de comunidades de murcielagos neotropicales: un análisis preliminar
THEODORE H. FLEMING Department of Biology, University of Miami, Coral Gables, FL 33124, USA
ABSTRACT
I review the taxonomic composition and ecology of the bat fauna of Venezuela and examine the structure of several mainland and Canobean island phyllostomid bat communities in an effort to discern general rules in the assembly of Neotropical bat communities. The structure of species-rich phyllostomid communities appears to closely resemble the taxonomic, trophic, and morphological characteristics of the phyllostomid fauna. These communities contain clusters of morphologically similar species that tend to be less common and ecologically more specialized than a handful of abund- ant and broad-niched core species that are found in almost all faunas. As overall biological diversity decreases along rainfall and elevational gradients, specialist species drop out of communitiestwhereas the generalists persist. Phyllostomid generalists come from each of the major trophic and size adaptive zones into which this family has radiated. Mutualistic interactions with plants appear to have had a more profound influence on community structure in this family than have competitive interactions. Key words: Neotropical bats, community structure, competition, mutualism, zoogeography.
RESUMEN
En este trabajo reviso la composicion taxonomica y la ecologia de la fauna de murcielagos de Venezuela y examino la estructura de varias comunidades de murcielagos filostómidos en áreas continentales y de islas caribefias, con el prop6sito de discernir reglas generales en el ensamblaje de comunidades neotropicales de murcielagos. La estructura de comunidades de filostomidos con alta riq ueza de especies refleja claramente las caracteristicas taxonomicas, tr6ficas y morfologicas de la fauna de filostómidos en general. Dichas comunidades contienen especies morfologicamente similares que tienden a ser menos comunes y ecologicamente más especializadas que un pequefio grupo de especies nucleares abundantes y generalistas que se encuentran en casi todas las faunas. En la medida que la diversidad biológica decrece a lo largo de gradientes crecientes de pluviosidad y elevacion, las especies especialistas desaparecen de las comunidades, en tanto que las generalistas persisten. Los filostomidos generalistas provienen de cada una de las grandes zonas adaptativas troficas y de tamaiio dentro de las cuales la familia ha radiado. Las interacciones mutualistas con plantas parecen haber tenido una influencia más profunda que aquellas de tipo competitivas en la estructuracion comunitaria de esta familia de murcielagos. Palabras claves: Murcielagos neotropicales, estructura comunitaria, competencia, mutualismo, zoogeografia.
INTRODUCION tionarily (e.g., papers in Strong et a/. 1984 ). The role of competition in struc- Patterns of species diversity and the factors turing communities has been a central that influence the structure of communities focus in these studies. Both views (non- (i.e., species richness and relative abun- random vs. random assemblages; impor- dance of species) have been of long- tance vs. unimportance of competition) standing interest to ecologists (Elton 1927, have their strong advocates (e.g., Wiens Hutchinson 1959, Cody & Diamond 1975). 1983, Brown & Bowers 1984, Grant & Recent studies of vertebrate community Schluter 1984, Simberloff 1984). To re- structure have focused on determining the concile different viewpoints, most investi- degree to which communities differ from gators agree that detailed field observations random assemblages of sympatric species and, whenever possible, controlled field that do not interact ecologically or evolu- experiments are needed.
(Received 20 October 1985. Accepted 11 March 1986.) 136 FLEMING
A variety of approaches has been used to New World (Table 1). Members of eight of study the structure of bat communities. An the nine families are insectivorous (with the ecological approach aimed at documenting exception of Noctilio leporinus, which is the relative abundances of species and their primarily a fish-eater). As reviewed by food niche relationships has been used by Gardner ( 1977), members of the Phyllosto- Fleming et al. (1972), Black (1974), LaVal midae, in which six subfamilies are cu- & Fitch ( 1977), and Bonaccorso ( 1979). A rrently recognized (Jones & Carter 1976) primarily morphological approach has been and whose species make up one-half of the used by Tamsitt (1967), McNab (1971), total fauna, consume a diverse array of Fenton (1972), Findley (1973, 1976), foods, including insects, vertebrates, nectar Findley & Black (1983), and Schum and pollen, fruit, and blood. (1984). Both approaches have led to the The bat fauna of Venezuela has been conclusion that many bat communities extensively sampled by field crews directed contain species that are tightly packed in by C. 0. Handley, Jr. (see Handley 1976) ecological and morphological space. The and can be used to obtain an overview of extent to which competitive interactions general trends in the species richness, geo- influence (or have influenced) interspecific graphic distributions, and relative abun- relationships, however, is still unclear. dance of Neotropical bats. I will use the I have three goals in this paper. First, I results of this faunal overview to predict will review our current knowledge of the the detailed structure of bat communities structure of Neotropical bat communities under the assumption that community with particular emphasis being placed on structure merely reflects faunal structure the family Phyllostomidae, the New World (or vice versa). This assumption will serve fruit bat family. Second, I will attempt to as a. null hypothesis in this study. identify general trends in the "assembly" of phyllostomid communities that are indi- cative of general "assembly rules" in Neo- Distributions by Life Zones tropical bat communities. Third, I will compare the results of this study with As shown in Fig. I, bat species are previous studies of insectivorous bat com- unevenly distributed among life zones (as munities and faunas conducted by Fenton designated by Hoi bridge 1947 in Venezue- (1972), Findley (1973, 1976), and Schum la. A strong diversity peak occurs in tropi- ( 1984 ), and with studies of bird communi- cal dry and tropical moist life zones, and a ties and faunas conducted by Karr & James secondary peak occurs in subtropical or (1975) and Ricklefs & Travis (1980), in an lower montane moist life zones. The insec- attempt to identify general assembly rules tivorous families show this basic pattern in communities of volant vertebrates. An (Fig. I) except that species are equally underlying theme in this analysis will be represented in both the tropical dry and the degree to which interspecific inte- tropical moist life zones whereas they are ractions (e.g., competition and mutualism) better represented in the subtropical moist influence bat community structure. I want than in the subtropical wet life zone. to emphasize the preliminary nature of this Distributions of modal life zones differ inductive exercise. A complete analysis among phyllostomid subfamilies (Fig. I). requires a much more extensive set of Phyllostomines (which are insectivorous- ecological, behavioral, and morphological omnivorous) peak strongly in tropical data, and a more sophisticated statistical moist life zones; glossophagines (which are analysis. I strongly suspect, however, that nectarivorous) peak in the tropical moist the general patterns that I have found will and subtropical moist zones; no strong hold up under more detailed scrutiny. dominance of one life zone occurs in the Humphrey & Bonaccorso ( 1979) discussed frugivorous carolliines; and two peaks this subject from a different viewpoint. occur in the frugivorous stenodermines (tro- pical moist and subtropical wet-lower mon- An Overview of the Neotropical Bat Fauna tane moist zones). These distributional data indicate that faunal and community di- versity is highest in lowland moist forests in The Neotropical bat fauna consists of both insectivorous and plant-visiting bats approximately 282 species classified in nine and that a secondary radiation has occurred families, six of which are endemic to the in certain groups at mid-elevations. NEOTROPICAL BAT COMMUNITIES: A PRELIMINARY ANALYSIS 137
TABLE 1
Families of Neotropical bats and their general food habits. Data taken from Nowak & Paradiso (1983) Familias de murcielagos neotropicales y sus habitos alimenticios generales. Datos obtenidos de Nowak & Paradiso (1983)
FAMILY NUMBER OF SPECIES GENERAL FOOD HABITS
Emballonuridae 21 Insectivorous Noctilionidaea 2 Insectivorous/piscivorous Mormoopidaea 8 Insectivorous Phyllostomidaea 143 Phyllostorninae (Phyllos.)b 34 Many insectivorous, occasionally frugivorous; some carnivorous Glossophaginae 34 Nectarivorous, occasionally (Glosso.) frugivorous & insectivorous Carolliinae 7 Frugivorous, occasionally (Caroll.) insectivorous &·nectarivorous Stenod erminae 57 Frugivorous, occasionally (Steno.) insectivorous & nectarivorous Brachyphyllinae 8 Nectarivorous & frugivorous (Brachy.) Desmodontinae 3 Sanguinivorous (Desmo.) Natalidae8 4 Insectivorous Furipteridaea 3 Insectivorous Thyropteridaea 2 Insectivorous Vespertilionidae 63 Insectivorous Molossidae 36 Insectivorous ~ Neotropical endemics. Abbreviations used in figures.
Distributions by Elevation presentative ( Carollia brevicauda); and ( 4) the stenodermines are well-represented in From the preceding analysis, we expect bat the highlands with two genera (Stumira species richness to be inversely related to and Vampyrops) containing several upland elevation. Except in two groups (the Ves- species. These results imply that at the pertilionidae and Stenoderminae), lowland community level, species diversity in all forms greatly outnumber highland forms in trophic groups will decrease with increasing Venezuela (Fig. 2). Koopman ( 1978) and elevation. Some taxonomic turnover will Graham (1983) have documented a similar occur with increasing elevation. Lowland trend in the Peruvian Andes. With the taxa will be replaced by highland taxa exception of the Vespertilionidae and a few within insectivores, nectarivores, and steno- molossids, insectivores, regardless of their dermine frugivores. evolutionary history, are more-or-less res- Because insectivores of different phylo- tricted to the lowlands. The highland ves- genetic affinities and different thermoregu- pertilionids are represented by two genera, latory capabilities (McNab 1982) are con- Eptesicus (which is common in north tem- cordant in their elevational distributions, I perate regions) and Histiotus (which is hypothesize that this distributional simila- widespread in south temperate regions). rity has a trophic rather than a physiolo- The phyllostomid subfamilies display a gical basis. That is, reduced food availabi- variety of patterns: (I) like other insecti- lity, rather than reduced physiological ca- vores, the phyllostomines are strongly low- pacity to tolerate low nocturnal tempera- land; (2) the glossophagines are basically tures or cool daytime roosts, has restricted lowland, but two genera (Anoura and insectivores to the lowlands. McNab ( 1982) Lonchophylla) contain highland forms; (3) similarly concluded that, except for the the carolliines have one mid-elevation re- vampire Desmodus rotundus, food availa- 138 FLEMING
· Bats ( n = 138 bat- habitats)
40
(/)
-0 20
~ c <1J u L <1J 0.... Phyllos tom ids ~ Phyllos. (24) D Glosso. ( 13) f;3Caroll. (3) • Stene. (37)
8 9 Dry) low Wet, high Life Zones
Fig. 1. Distribution of Venezuelan bats by life zone. Data are based on the modal life zone of each species as determined from capture data presented in Handley(I976). Abbreviations ofphyllostomid subfamilies as in Table I. Life zones are as follows: (I) Tropical thorn forest, (2) Tropical dry forest, (3) Tropical moist forest, ( 4) Tropical wet forest, (5) Subtropical dry forest, (6) Subtropical moist forest, (7) Subtropical wet forest, (8) Lower montane moist forest, (9) Lower montane wet forest, (10) Montane moist forest. Distribucion de los murcielagos venezolanos por zona de vida. Los datos estan basados en la zona de vida modal de cada especie, determinada a traves de datos de captura documentados por Handley (1976). Las zonas de vida son las siguientes: (1) Bo~ue espinoso tropical, (2) Bosque seco tropical, (3) Bo~ue humedo tropical, (4) Bosque perhumedo tropical, (5) Bo~ue seco subtropical, (6) B~ue humedo subtropical, (7) Bo~ue perhllmedo subtropical, (8) Bo~ue humedo de montaila baja, (9) B~ue perhllmedo de montaila baja, (10) Bo~ue humedo de montaila alta. NEOTROPICAL BAT COMMUNITIES: A PRELIMINARY ANALYSIS 139
underlying factors; concordance, however, Elevation (x 103m) does not eliminate the possibility that 3 Insectivores different factors affect the distributions of bats and birds. In Fig. 3 I have plotted the elevational distributions of Panamanian birds, as reported in Ridgely ( 1976), by trophic group. Trophically analogous chi- ropteran groups include: canopy frugivores - stenodermines (especially Artibeus and Vampyrops); understory frugivores - ca- rolliines and certain stenodermines (espe- cially Stumira); nectarivores - glossopha- Moloss. gines; consumers of large insects - large Em ball. ___,_'o phyllostomines; understory insectivores - small phyllostomines, certain emballonu- 12 4 4 12 20 rids, and other small insectivores; aerial insectivores - certain vespertilionids and molossids. Phyllostomids ,o ~4 / ' Large Insects , ,....Canopy 40 P Fruits 40 (3, 25) I I (5 families, I 89 spp.) I / 20 I 20 / I tTurdidae Glosso. I P 0 (17) I I p / .~"' 40 40 u CIJ Understory 0. Fruits
Comparison of Figs. 2 and 3 reveals that fewer sites. Only the mormoopid Pterono the elevational distributions of Neotropical tus parnellii was caught at more that 30 bats basically resemble those of Neotropical sites. Phyllostomids were generally more birds. Except for the Trochilidae and three widespread than non-phyllostomids. Except additional families with strong temperate in the glossophagines, most phyllostomids zone representation (Turdidae, Parulidae, were captured at more than five collecting and Fumariidae), the species richness of sites. Geographically (and ecologically) each avian trophic group decreases with widespread phyllostomids included Micro elevation. The decrease is particularly rapid nycteris mega/otis, Trachops cirrhosus, and in canopy frugivores (but is not as rapid in Phyllostomus discolor (Phyllostominae); stenodermine bats) and in birds eating large Glossophaga soricina (Glossophaginae); Ca insects (as is the case in large phyllosto- rollia perspicillata (Carolliinae); and Stumi mines). Using more refined distributional ra !ilium and Artibeus jamaicensis (Steno- data, Terborgh (1977) and Stiles (1983) derminae). The vampire Desmodus ro reported similar trends in Peruvian and tundus (Desmodontinae) was also collected Costa Rican birds. Stiles further noted that at over 45 sites. Each of these widespread highest bird diversity occurs where the Venezuelan phyllostomids (plus others that humid lowlands meet the foothills (e.g. at were collected at somewhat fewer sites) Finca La Selva) and at the lower edge of ranges widely throughout the Neotropics cool wet "cloud" forest at an elevation of from Uexico to southern Brazil or northern 1000-1200 m, where insect diversity and Argentina (Koopman 1982). biomass tend to be high (Janzen 1973). Terborgh ( 1977) noted that the mid-ele- vation peak in Peru was caused by an increased density and diversity of insecti- Non-phyllostomids (n=66spp.) vorous birds. As documented below, bat 60 diversity is also very high at La Selva, but, unlike the avian situation, bat diversity is 40 not particularly high at a mid-elevation site in Costa Rica ( cf. sites LS and MV in Fig. 20 6). Therefore, precise one-to-one mapping of bat and bird diversity does not occur along elevational gradients in Costa Rica and probably elsewhere in Latin America. "' u Geographical .Ranges in Venezuela Phyllostomi ds ~ Phyllos. ( 23) a. 1/) 40 0 Glosso. ( 15) Some idea of the breadth of the ecological 0 .... tolerances of each bat species in Venezuela c 20 can be obtained by examining the number u L of collecting stations at which it was found. Because collections were made at many 0.. ( 100) sites in a variety of different life zones, bats collected at many sites can be considered to have broader ecological tole- 40 Phyllostomids ~ Caroll. (5) rances than those collected at only a few • Stene.• Desmo. (38) sites. This analysis assumes, of course, that 20 equal collecting efforts were made for all species at all sites. To judge from the great differences among sites in numbers of bats 5 15 25 35 45 >45 collected (approximate range = 10 - Number of Sites/Species 3800), however, this assumption is unrea- listic, but the extent to which collecting Fig. 4. Frequency distributions of the number of biases influence this analysis is unknown. collection sites per species of Venezuelan bats. Distributions of the number of collec- Data from Handley (1976). tion sites per species are summarized in Fig. Distribuci.i:m de frecuencias del numero de sitios de 4. A majority (65"to) of the non-phyllos- colecta para especies de murcielagos venezolanos. Los tomid bat species were caught at five or datos provienen de Handley (1976). NEOTROPICAL BAT COMMUNITIES: A PRELIMINARY ANALYSIS 141
Taken at face value, these results suggest ture of chiropteran communities. As we that non-phyllostomids have narrower ha- will see, these taxa represent "core" species bitat breadths than do phyllostomid bats, that are common elements of many low- and, because number of collection sites per land Neotropical bat communities. They species is positively correlated with the (or their close relatives) are the numerically number of individuals collected (r = 0. 79, dominant members of their feeding guilds P < 0.001 ), they are less common than in both low and high diversity commu- phyllostomids (see Brown, 1984). Howe- nities. ver, because of the much greater ease of capturing phyllostomid bats than other The Relative Abundances of Phyllostomid kinds of bats in mist nets (LaVal & Fitch Bats 1977), I believe these conclusions are open to serious question. Until unbiased To minimize biases caused by sampling sampling methods for all kinds of bats are problems, I will restrict my analysis of the devised, it will be impossible to compare relative abundance of Venezuelan bats to the abundance and distribution of different the Phyllostomidae. As is the case in families of bats solely from museum co- many groups of organisms, each phyllos- llections. tomid subfamily contains one to three The fact that certain species of bats are common species and a few to many uncom- abundant and geographically widespread mon species (Fig. 5). Because degree of has important implications for the struc- dominance of the one or two top-ranking
.70
~Carol!. : (5 spp., ~0 1 n=6067)
en Glosso. (15 2338) ro 1 ::J '0 I > I '0 I c I -0 I c .30 I 0 I +-' L •I Steno. ( 351 11 540) 0 I a. 0 .20 L 5 10 15 20 25 0..
Phyll os. ( 2 3 1 16 40)
5 10 15 20 25 Species Rank
Fig. 5. The relative abundances of species of Venezuelan phyllostomid bats by subfamilies based on mist net capture data in Handley ( 197 6). Abundancias relativas de especies por subfamilia de los murcielagos fJJ.ost6inidos venezolanos, basadas en datos de captura con redes de nieb la documentados por Handley (197 6). 142 FLEMING
(most common) species is inversely related to each subfamily's species richness, diffe- rences between subfamilies in this regard are statistical artifacts. Numerically dominant species include many of the same species mentioned above (P. discolor, T. cirrhosus, G. soricina, C. perspicillata, S. /ilium, and A. jamaicensis) plus Phyllos tomus hastatus and Glossophaga longiros tris. These species are the core elements of lowland phyllostomid communities. Trends in the Community Structure of Phyllostomid Bats - n
Although species of insectivorous bats com- 32 Steno. prise at least one-half of the Neotropical chiropteran fauna, their population and 28 community ecology is poorly known be- Car oil. ~ 24 cause, relative to phyllostomid bats, they ·c; ., Gloss c . are difficult to capture in mist nets set at ~ 20 ground level. Differences in ease of capture 0 16 undoubtedly also exist among phyllosto- .,<... .0 mids, but these have not been studied E 12 Phyflos. systematically. In the absence of data on z::J differential catchability, I will assume that 8
capture data accurately reflect the relative 4 abundance of phyllostomid bats, at least at
ground level. Because of their relatively MV high catchability, I will focus my attention in the rest of this paper on phyllostomid bats. Fig. 6. Proportional representation and species Mainland Communities richness of four phyllostomid subfamilies in six Neotropical communities as follows: Cerr = Brazi- Trends in species richness at six well- lian "edaphic" Cerrados (elevation < 1000 m, studied sites in Costa Rica, Panama, and annual rainfall < 1600 mm), Caa = Brazilian Brazil are summarized in Fig. 6. As ex- Caatingas (< 1000 m, < 1600 mm), SR = Parque pected, species richness increases along a Nacional Santa Rosa, Costa Rica, tropical dry rainfall/vegetational gradient in the low forest (300m, 1660 mm), BCI = Barro Colorado lands. The mid-elevation site contains fewer Island, Panama, tropical moist forest (50 m, 2600 species and lacks members of the Phyllosto- mm), LS = Finca La Selva, Costa Rica, tropical minae. Despite a three-fold difference in wet forest (100m, 4000mm), MV =Monteverde, species richness among the lowland sites, Costa Rica premontane moist forest (1200 m, 2500 the proportional representation of each mm). Data are from LaVal & Fitch (1977), subfamily does not differ significantly Bonaccorso (I 979), Willig (1983), and Fleming 2 among sites (X (6) = 3.98, p = 0.86, (unpubl. data). Glossophaginae and Carolliinae combined Representaci6n proporcional y riqueza de especies de into one group). Therefore, no differential cuatro subfamilias de filostomidos en seis comunidades enrichment by subfamily occurs with an neotropicales: Cerr = cerrados "edaficos" brasileiios (ele- increase in phyllostomid diversity. vacion < 1000 m, precipitacion anuall600 mm), Caa = caatingas brasileiias < 1000 m, 1600 mm), SR = Parque Trends in the distribution of species by Nacional Santa Rosa, Costa Rica, bosque seco tropical subfamily in two-dimensional morpholo- (300 m, 1660 mm), BCI =Isla Barro Colorado, Panama, gical space at the five lowland sites are bosque humedo tropical (50 m, 2600 mm), LS = Finca shown in Fig. 7. To ordinate species in La Selva, Costa Rica, bosque perhumedo tropical (100m, 4000 mm), MV = Monteverde, Costa Rica, bosque 2-space I chose one measure of overall size hlimedo premontano (1200 m, 2500 mm). Los datos (forearm length) and one measure of skull fueron extrafdos de LaVal & Fitch (1977), Bonaccorso size (breadth across the upper molars which (1979), Willig (1983) y Fleming (no publicado). NEOTROPICAL BAT COMMUNITIES: A PRELIMINARY ANALYSIS 143 should correlate with largest potential food variables so that multivariate statistical particle size) using data in Swanepoel & procedures can be used to yield more Genoways (1979). Future analyses should refined estimates of morphological rela- include a greater number of morphological tionships (Findley & Wilson 1982).
Phyllostominae Glossophag. Carelli i nae Stenoderminae ~:r :"'"" . 0 0
~ 2 ~ 0f Caatinga 0 0 .._ 10 X •• ~ .!: .. "' 0 +"' "0 ~ 20t Santa Rosa 0 0 co 10 • L • 20 La Selva .. 0 10 ., ... 10 10 • •• 10 eo 60 100 40 50 40 50 30 50 70 Forearm Length ( mm) Fig. Z Distribution of species of phyllostomid bats in two-dimensional morphological space in five main- land lowland communities. Within each subfamily, different symbols represent different genera. Distribucion de especies de murcielagos filostomidos en un espacio morfologico bidimensional en cinco comunidades de tierras bajas continentales. Dentro de cad a sub familia, simb olos diferentes representan generos distintos. Points that emerge from Fig. 7 include To document trends in relative abun- the following. ( 1) The range of sizes within dances with increasing diversity, I exa- subfamilies tends to increase with increased mined the capture data from two well- diversity (species richness) in the Glossoph- sampled Costa Rican sites, Santa Rosa aginae, Carolliinae, and Stenoderminae but (tropical dry forest) and La Selva (tropical not in the Phyllostominae. (2) Small wet forest). These data (Table 2) indicate species (ie., those with forearms smaller that members of the two frugivorous sub- than the mid-point of a subfamily's size families (Caro11iinae and Stenoderminae) range) are added to a community faster are more common than members of the than large species. (3) As a result, nearest other two subfamilies. Insectivorous (or neighbor distances in 2-space, measured as omnivorous) and nectarivorous phyllosto- the Euclidean distance between pairs of mids are only one-third to one-tenth as species, decrease with increased diversity common as frugivores. An expanded analy- (Fig. 8). That is, mean crowding in mor- sis encompassing more sites yielded the phological space increases with increased same results (Fleming unpublished data). diversity, but it does so differentially by Trends within subfamilies are summa- size in three of the four subfamilies. Small rized in Fig. 9. At both sites, one or two species are more crowded than large species species dominated captures in three of the irrespective of trophic adaptation. four subfamilies; strong numerical domi- 144 FLEMING ( Glossophaga commissarisi, Carollia brevi Phyllos. cauda, C. castanea, Artibeus jamaicensis • Mainland 15r- • and A. phaeotis) were numerically domi- 6 Islands nant. Except in the Phyllostominae, the • most common species were in the larger A E 101- half of each subfamily's size range; E phaeotis is also an exception to this pat- QJ tern. In the Phyllostominae, the most u c common species were in the smaller half of cu • +-' 51- the size range. This pattern reflects the Ill • 6 0 • trophic position of different phyllosto- L mines; small species are primarily insecti- 0 ..0 I I I I I vorous whereas large species tend to be .r::. Ol secondary carnivores (Freeman 1984) . '(ij z Steno. Caribbean Island Communities +-' 10 r- Ill QJ L 6 To see if diversity trends on islands are cu QJ similar to those in mainland communities, I z • 5r- examined the bat faunas of four represen- IX 6 • : tative Caribbean islands plus Trinidad using 6 .6 species lists published in Baker & Gena- ways (1978) and Carteret a/. (1981). The I I I I Caribbean bat fauna is taxonomically 4 8 12 16 20 Number of Species depauperate compared with the mainland fauna. Missing mainland families include Emballonuridae (except for Peropteryx ma Fig. 8. Relationship between mean nearest neigh- crotis on Grenada), Furipteridae, and bor distance in 2-space and species richness in two Thyropteridae. In the Phyllostomidae, the phyllostomid subfamilies in mainland and island Phyllostominae is represented only by Ma communities. crotus waterhousii in the Greater Antilles Relaci6n entre Ia distancia media a! vecino mas cercano en Micronycteris mega/otis un espacio bidimensional y Ia riqueza de especies en dos and on Grenada, subfamilias de fi!ost6midos en comunidades continentales and the Carolliinae is represented only by e insulares. Carollia perspicillata on Grenada. An ende- mic phyllostomid subfamily, Brachy- phyllinae, with three genera, has radiated in nance by a single species was absent in the the Greated Antilles, and one Phyllostominae. In the dry forest, three species(Brachyphylla cavernarum) is wide- species (Glossophaga soricina, Carollia perSL spread in the Lesser Antilles. Elsewhere picillata, and Artibeus jamaicensis) each in (Fleming 1982) I have examined diversity a different subfamily were numerically trends in West Indian bats and birds in dominant. In the wet forest, five species greater detail. TABLE 2 Relative abundance of four phyllostornid subfamilies in mist net samples at two Costa Rican sites. Data taken from La Val & Fitch (1977) and Fleming (unpublished). Abundancia relativa de cuatro subfamilias de filost6midos en capturas efectuadas con redes de niebla en dos localidades de Costa Rica. Los datos provienen de LaVal & Fitch (1977) y Fleming (no publicado) PROPORTION SAMPLE PHYLLOSTO- GLOSSOPHA- CAROLL- STENO- SITE SIZE MINAE GINAE IINAE DERMINAE Santa Rosa 6026 0.009 0.123 0.521 0.347 (dry forest) La Selva 1500 0.088 0.091 0.576 0.245 (wet forest) NEOTROPICAL BAT COMMUNITIES: A PRELIMINARY ANALYSIS 145 Santa Rosa Phyllos. (n=56) Steno. ( 2092) •Aj <1> 0.. E 40 ro Mh 1.}) •etv1b ePd 20 '+- • • 0 • • • ~ • c <1> u 40 60 80 100 30 40 30 50 70 L <1> La Selva 0.... 60- r- •Gc -(864) - ( 367) (132) (137) •cb 40 - r-- - - eAP • Cc Aj • • Tc 20- r-- • - - • • • • • • • • • .a\• l ''"'.• ... • I I I I •• I I I I I I 40 60 80 100 30 40 40 30 50 70 Forearm Length (mm) Fig. 9. The relationship between size and relative abundance of phyllostomid bats by subfamily at two Costa Rican sites. Within each subfamily, different symbols represent different genera. Abbreviations of common species are: Aj = Artibeus jamaicensis, Ap = A. phaeotis, Cb = Carollia brevicauda, Cc = C castanea, Cp = C perspicillata, Gc = G/ossophaga commarissarisi, Gs = G. soricina, Mb = Micronycteris brachyotis, Mh = M hirsuta, Pd = Phyllostomus discolor, Tc = Trachops cirrhosus. Relacion entre tamafio y abundancia relativa por subfamilia de murcielagos filostomidos en dos sitios de Costa Rica Dentro de cad a subfamilia, simbolos diferentes representan generos distintos. Abreviaturas de especies comunes son: Aj = Artibeus jamaicensis, Ap = A. phaeotis, Cb = Carollia brevicauda, Cc = C. castanea, Cp = C. perspicillata, Gc = Glossophaga commarissarisi, Gs = G. soricina, Mb = Micronycteris brachyotis, Mh = M. hirsuta, Pd = Phyllostomus discolor, Tc = Trachops cirrhosus. Trends in species richness and size are size in these bats (this is not always the shown in Fig. 10. Points to note include case in mainland phyllostomids, Fleming in the following. ( 1) In the least diverse fauna press), then generalists are more likely to (Anguilla), the single species present in occur in low diversity situations than each subfamily is relatively large. If diet specialists. (2) As in mainland communi- breadth is positively correlated with body ties, more small species are added to island 146 FLEMING faunas than large species as diversity in- Antilles. The reason for this taxonimic shift creases. (3) The Trinidad fauna is as is not clear because except for Brachyphy crowded in 2-space as is a diverse mainland lla, the brachyphyllines are ecologically community (cf. Figs. 7 and 11; also see Fig. more similar to glossophagines than they 8). (4) On Cuba (and elsewhere in the are to stenodermines. (5) As shown in Fig. Greater Antilles), phyllostomid diversity is 8, no clear relationship between nearest concentrated in the Brachphyllinae and not neighbor distance and species richness in the Stenoderminae as it is in the Lesser occurs in the Carib bean stenodermines. Phyllos. (*) or Brachy. Glosso. & Carol I.(* l Stene Anguilla c • v "t t I t St. Lucia • c " E v E "t t t .s:::. +' "0 Guadeloupe c ro • (!) " L v • (l) "t t t L ro 0 0 Cuba c 2 • ·t ·~ t v t Trinidad ( Ptlyllos. only) .. c • ... v c 10 "a • ... 111 •ll>li...... " :!'!'. •* 40 60 BO 100 40 50 40 60 Forearm Length (mm) Fig. I 0. Distribution of species of phyllostomid bats in two-dimensional morphological space in five Caribbean island faunas. Within each subfamily, different symbols represent different genera. Distribucion de especies de murcielagos filostomidos en un espacio morfologico bidimensional en cinco faunas insulares del Carl> e. Dentro de cad a sub familia, simbolos diferentes representan generos distintos. DISCUSSION omnivores (Trachops cirrhosus, Phyllos tomus discolor), small nectarivore-omni- vore ( Glossophaga soricina), medium The Assembly of Phyllostomid Commu Carollia perspicilla nities understory frugivore ( ta), small canopy frugivore (Artibeus pha eotis), and medium-large canopy frugivore Communities of phyllostomid bats (and (A. jamaicensis ). These species share several other bat families) are composed of rela- characteristics as follows: (I) they tend to tively few (one to three per subfamily) be primitive members of their subfamilies common and geographically widespread (this is certainly true of Micronycteris, species, and a variable number of uncom- Phyllostomus, Glossophaga, Carollia, and mon (but often widespread) species. Core Artibeus; Phillips 1971, Honeycutt 1981, species (sensu Hanski 1982) are rather Griffiths 1982, Hood & Smith 1982; (2) evenly distributed among body sizes and they tend to be intermediate-to-large mem- trophic niches as follows: small insectivore bers of their subfamilies (except for M (Micronycteris mega/otis), medium-large mega/otis and A. phaeotis); and (3) they NEOTROPICAL BAT COMMUNITIES: A PRELIMINARY ANALYSIS 147 tend to have generalized diets (Gardner presented ( 42.514 of the species cf. an 1977, Fleming in press) and are tolerant of expected 25.814) and glossophagines were a variety of ecological conditions. under-represented (12.414 cf. 25.814) in the As the diversity of phyllostomid commu- communities {X2 (3) = 11.84, P = nities increases along habitat gradients, less 0.0079). Carolliines (8.014) and stenoder- abundant (more specialized?) species are mines (37.214) closely resembled chance added to this core of generalists so that expectations. These taxonimic deviations total morphological space, which probably result, in part, from the geographic ranges reflects total niche space (Findley & Wilson of phyllostomines being broader than those 1982), increases. During this process, small of glossophagines (Koopman 1982). Be- species are added at a faster rate than large cause more phyllostomine than glossopha- species, and species (especially the small gine ranges overlap at any point in the ones) become more tightly packed in mor- lowland Neotropics, local communities phological space. Because we know relati- should contain a higher proportion of the vely little about the ecology and behavior phyllostomine fauna than the glossopha- of uncommon phyllostomids, trends in gine fauna. actual (rather than hypothesized) niche Significant taxonimic deviations also relationships between species in diverse occur up mountains and on Caribbean communities are currently poorly known islands, primarily as a result of the absence (e.g., Humphrey et al. 1983). Behaviorally of phyllostomine species. Reduced food oriented field studies rather than morpho- availability seems to explain the absence of logical studies are needed . to resolve phyllostomines in both of these situations. questions about niche relationships and the Why the abundant and widespread Carollia degree to which species compete for li- perspicillata is not present on Caribbean mited resources. Especially desirable would islands when two of its principal food taxa, be studies of microhabitat selection and shrubs of the genus Piper and Cecropia roost use. trees, are widespread in the West Indies is Because of the absence of members of currently unknown. the Phyllostominae and Carolliinae from In summary, at least two factors interact most Caribbean Islands, the assembly of to determine the taxonomic composition West Indian pl:tyllostomid communities is of phyllostomid communities: (I) the re- somewhat different from that on the lative abundance of different classes of mainland. Conspicuously missing are folia- food (which affects the Phyllostominae ge-gleaning insectivores and medium-sized more strongly than other subfamilies) and understory frugivores. Core members of (2) differences in the geographic ranges of island communities include one small nec- members of different subfamilies (which tarivore (Glossophaga longirostris or Mono affects the Glossophaginae more strongly phyllus spp.) and two medium-large frugi- than other subfamilies). vores (Brachyphylla spp. and Artibeus ja I searched for significant departures maicensis). As island size increases in the from random expectations in size (forearm) Greater Antilles, up to three additional distributions by comparing the size distri- brachyphyllines are added; up to three butions of portions of two communities stenodermines are added in the Lesser (the low diversity Caatinga and the high Antilles. diversity La Selva sites) with those in the To what extent does the community appropriate subfamilies in the entire structure of phyllostomid bats differ from mainland fauna using data in Swanepoel & random expectations as reflected in the Genoways (1979). The results of Kolmo- structure of the phyllostomid fauna as a gorov-Smirnov one sample tests indicated whole? To answer this question, I will that forearm distributions did not differ compare the taxonomic composition (which from random expectations in the Phyllosto- actually reflects trophic structure) and size minae and Stenoderminae at the Caatinga distributions of the five lowland mainland site (P's >> 0.05) nor did they differ communities with those of the entire significantly in the Phyllostominae, Glosso- phyllostomid fauna. Taxonomically, the phaginae, and Stenoderminae at La Selva five communities (whose data I combined (Ps >> 0.05). From these results I con- because they were not significantly hetero- clude that in these two communities (and geneous) differ significantly from the fauna by inspection of Fig. 7, in at least two as a whole. Phyllostomines were over-re- others, Santa Rosa and Barro Colorado 148 FLEMING Island) forearm distributions do not differ My second conclusion is that food availa- from those expected from a random draw bility, both in amounts and in kind, from the appropriate subfamilies. Contrary strongly influences the structure of phyllos- to Humphrey et al. (1983), I see no tomid communities. Insect availability evidence of "character displacement" by appears to play a crucial role in the overall size among members of the Phyllos- presence or absence of members of the tominae or any other subfamily. Fleming et Phyllostominae as does the availability of al. ( 1972) reached a similar conclusion appropriate flowers for the Glossophaginae. from an analysis of size (and food habit) Among mainland frugivores, the geographic distributions in two Panamanian and one and altitudinal ranges of many genera (e.g., Costa Rican bat community. Carollia, Stumira, Artibeus, and Vam It should be noted that this one-dimen- pyrops) correspond to the ranges of their sional analysis does not eliminate the most preferred kinds of fruit (Fleming in possibility that species within subfamilies press). Degree of food specialization (guilds) are non-randomly arrayed in n-di- appears to determine the presence and mensional morphological space. For exam- abundance of particular bats within phy- ple, McKenzie & Rolfe (in press) have llostomid communities. Generalist feeders found that guilds of insectivorous bats form the core elements of these commu- foraging in mangrove stands in northwes- nities over a wide range of ecological tern Australia are non-random regarding conditions, and they are joined by more two morphological features that influence specialized taxa where high rainfall and foraging behavior, aspect ratio and wing high plant productivity promote high biolcr loading. Syntopic species show less overlap gical diversity. in these parameters than expected by chance. Comparisons with other Bat and Bird Communities Despite this caveat, I reach two tentative conclusions regarding the evolution of com- munity structure in phyllostomid bats. How similar is the structure of commu- First, where appropriate resources are avai- nities of phyllostomid and non-phyllos- lable (particularly in the moist mainland tomid microchiropteran bats? Do the same lowlands) community structure basically general assembly rules apply to bats regard- reflects the taxonomic, trophic, and size less of their taxonomic and trophic affi- structure of the fauna as a whole. To the liations? Definitive answers to these extent that competitive interactions (which questions are not yet available because occur mostly within subfamilies) and mu- different techniques have been used to tualistic interactions have molded the ever study different groups of bats. Current lution of this bat family, they have had a data, however, seem to point to the exis- similar influence on ecological relationships tence of general morphological rules. The within communities. Of the two ecological studies of Fenton (1972), Findley (1976), interactions, evidence for the importance Findley & Black (1983),.and Schum (1984) of mutualism, (i.e., plant-pollinator inter- indicate that most species in temperate and actions and fruit-frugivore interactions) is tropical insectivorous bat faunas are clus- much more compelling than is evidence for tered together in 2- to n-dimensional the importance of competition. Plant- morphological space. Communities or fau- animal interactions have profoundly in- nas consist of a central group of morpho- fluenced the evolution of the Phyllost Similar morphological trends probably LITERATURE CITED are widespread in other vertebrate commu- BAKER RJ & HH GENOWAYS (1978) Zoogeography of nities (Findley & Black 1983 ). For Antillean bats. In: Gill F (ed) Zoogeography in example, Ricklefs & Travis ( 1980) docu- the Caribbean. Academy of Natural Sciences, mented similar trends in temperate commu- Philadelphia, Special Publication 13: 53-97. & BLACK HL (1974) A north temperate bat community: nities of grassland birds as did Karr James structure and prey populations. J oumal of (1975) in forest bird communities. In Mamrnalogy 55: 138-157. mainland phyllostomid communities, BONACCORSO FJ (1979) Foraging and reproductive clusters of similar-sized species occur in the ecology in a Panamanian bat community. Bulle- Phyllostominae, Glossophaginae, and tin of the Florida State Museum, Biological Sciences 24: 359-408. Stenoderminae (Fig. 7). These clusters, BROWN JH (1984) On the relationship between abundan- however, do not necessarily contain what I ce and distribution of species. American Natu- have called the "cere" members of phyllos- ralist 124: 255-279. tomid communities. My definition of BROWN JH & MA BOWERS (1984) Patterns and processes in three guilds of terrestrial vertebra- "core" species is based on abundance and tes. In: Strong DR Jr et al. (eds) Ecological broad ecological tolerances, and, as I have communities: conceptual issues and the evi- previously noted, these species tend to be dence. Princeton University Press, Princeton, larger in size than species in the morpholo- New Jersey: 282-296. CARTER CH, HH GENOWAYS, RS LOREGNARD, & gical core. This difference suggests that RJ BAKER (1981) Observations on bats from modal morphology does not always result Trinidad, with a checklist of species occuring in greatest ecological success, at least in on the island. Occasional Papers, Museum of phyllostomid bats. Texas Tech University 72: 1-27. CODY ML & JM DIAMOND, eds (1975) Ecology and In conclusion, the structure of species- evolution of communities. Harvard University rich Neotropical bat communities appears Press, Cambridge, Massachusetts. to closely resemble the taxonomic, trophic, ELTON CS (1927) Animal Ecology. Sedgwick & Jackson, and morphological characteristics of the London. FENTON MB (1972) The structure of aerial-feeding bat Neotropical bat fauna as a whole. This is to faunas as indicated by ears and wing elements. be expected because diverse communities Canadian Journal of Zoology 50: 287-296. contain a substantial fraction of the total FINDLEY JS (1973) Phenetic packing as a measure of fauna. Such communities contain clusters faunal diversity. American Naturalist 107: 58D-584. of morphologically similar species that tend FINDLEY JS (197 6) The structure of bat communities. to be less common and ecologically more American Naturalist 110: 129-139. specialized than a handful of abundant and FINDLEY JS & HL BLACK (1983) Morphological and broad-niched core species. As overall biolo- dietary structuring of a Zambian insectivorous gical diversity decreases along rainfall and bat community. Ecology 64: 625-630. fiNDLEY JS & DE WILSON (1982) Ecological signifi- elevational gradients, specialist species drop cance of chiropteranmorphology. In: TH Kunz out of communities whereas the generalists (ed) Ecology of bats. Plenum Press, New York: persist. In the Phyllostomidae, generalists 243-260. come from each of the major trophic-size FLEMING TH (1982) Parallel trends in the species adaptive zones into which these bats have diversity of West Indies birds and bats. Oecolo- gia 53: 56-60. radiated. Mutualistic interactions with FLEMING TH (1986) Opportunism vs. specialization: plants appear to have had a more profound the evolution of feeding strategies in frugivo- influence on community structure in this rous bats. In: A Estrada & TH Fleming (eds) family than have competitive interactions. Frugivores and seed dispersal. W. Junk, the Netherlands: 105-118. FLEMING TH, ET HOOPER & DE WILSON (1972) Three Central American bat communities: structure, reproductive cycles, and movement patterns. Ecology 53: 655-670. ACKNOWLEDGMENTS FREEMAN PW (1984) Functional cranial analysis of large animalivorous bats (Microchiroptera). Biolo- I thank the Costa Rican National Park Service for gical Journal of the Linnean Society 21: permission to conduct long-term research at Parque 387-408. Nacional Santa Rosa Financial support for this research GARDNER AL (1977) Feeding habits. In: Baker RJ, JK has come from National Science Foundation grants. I Jones Jr, & DC Carter (eds) Biology of bats of thank Marcia Fleming for writing a computer program the New World family Phyllostomatidae. Part and Brock Fenton, Jim Findley, and Mike Willig for their II. Special Publication Museum Texas Tech comments on a draft of this paper. This paper is University Number 13: 293-350. Contribution NO 201 from the Program in Ecology, GRAHAM GL (1983) Changes in bat species diversity Behavior, and Tropical Biology, Department of Biology, along an elevational gradient up the Peruvian University of Miami. Andes. Journal of Mammalogy 64: 559-571. 150 FLEMING GRANT P & D SCHLUTER (1984) Interspecific comp&- American Museum of Natural History Novitates tition inferred from patterns of guild structure. 2651: 1-33. In: Strong DR Jretal. (eds) Ecological commu- KOOPMAN KF (1982) Biogeography of bats of South nities: conceptual issues and the evidence. America. In: Mares MA & HH Genoways (eds) Princeton University Press, Princeton, New Jer- Mammalian biology in South America. Special sey: 201-233. Publication Series, Pymatuning Laboratory of GRIFFITHS TA (1982) Systematics of the New World Ecology, University of Pitt!burgh 6: 273-302. nectar feeding bats (Mammalia, Phyllos- LAVAL RK & HS FITCH (1977) Structure, movements tomidae), based on the morphology of the and reproduction in three Costa Rican bat hyoid and lingual regions. American Museum of communities. Occasional Papers Museum of Natural History Novitates 2742: 1-45. Natural History, University of Kansas 69: 1-27. HANDLEY CO Jr (1976) Mammals of the Smithsonian MCKENZIE NL & JK ROLFE (in press) Structure of Venezuelan project. Brigham Young University insectivorous bat guilds in mangroves. Journal Science Bulletin, Biological Series 20: 1-91. of Animal Ecology. HANSKI I (1982) Dynamics of regional distribution: the MCNAB BK (1971) The structure of tropical bat faunas. core and satellite species hypothesis. Oikos 38: Ecology 52: 352-358. 21D-221. MCNAB BK (1982) Evolutionary alternatives in the HEITHAUS ER (1982) Coevolution between bats and physiological ecology of bats. In: Kunz TH (ed) plants. In: Kunz TH (ed) Ecology of bats. Ecology of bats. Plenum Press, New Yorlc:: Plenum Press, New York: 327-367. 151-200. HEITHAUS ER, TH FLEMING, & PA OPLER (1975) NOWAK RM&JLPARADISO (1983)Walker'sMammals Patterns of foraging and resource utilization in of the world. Volume 1, 4th edition. Johns seven species of bats in a seasonal tropical Hopkins University Press, Baltimore, Maryland. forest. Ecology 56: 841-854. PHILLIPS CJ (1971) The dentition of glossophagine bats: HOLDRIDGE LR (1947) Determination of world plant development, morphological characteristics, formations from simple climatic data. Science variation, pathology, and evolution. Miscella- 105: 367-368. neous Publications University of Kansas Mu- HONEYCUTT RL (1981) Molecular evolution in New seum of Natural History 54: 1-138. World leaf-nosed bats of the family Phyllos- RICKLEFS RE & J TRAVIS (1980) A morphological tomidae with comments on the superfamily approach to the study of avian community Noctilionoidea. PhD. Thesis, Texas Tech Uni- organization. Auk 97: 321-338. versity, Lubbock, Texas. RIDGELY RS (1976) Birds of Panama. Princeton Univer- HOOD CS & JD SMITH (1982) Cladistical analysis of sity Press, Princeton, New Jersey. female reproductive histomorphology in phy- SCHUM M (1984) Phenetic structure and species richness llostomid bats. Systematic Zoology 31: in North and Central American bat faunas. 241-251. Ecology 65: 1315-1324. HUMPHREY SR & FJ BONACCORSO (1979) Population SIMBERLOFF D (1984) Properties of coexisting bhd and community ecology. In: Baker RJ, JK species in two archipelagoes. In: Strong DR Jr Jones Jr, & DC Carter (eds) Biology of bats of et al. (eds) Ecological communities: conceptual the New World family Phyllostomatidae. Part issues and the evidence. Princeton University III. Special Publication Museum Texas Tech Press, Princeton, New Jersey. University 16: 409-441. STRONG DR JR, D SIMBERLOFF, LG ABELE & AB HUMPHREY SR, FJ BONACCORSO & TL ZINN (1983) THISTLE, eds (1984) Ecological communities: Guild structure of surfac&-gleaning bats in conceptual issues and the evidence. Princeton Panama. Ecology 64: 284-294. University Press, Princeton, New Jersey. HUTCHINSON GE (1959) Homage to Santa Rosalia, or SWANEPOEL P & HH GENOWAYS (1979) Morphom&- why are there so many kinds of animals? trics. In: RJ Baker, JK Jones Jr, & DC Carter American Naturalist 93:145-159. (eds) Biology of bats of the New World family HUTCHINSON GE & RH MACARTHUR (1959) A Phyllostomatidae. Part III. Special Publication theoretical ecological model of size distri- Museum Texas Tech University 16: 13-106. butions among species of animals. American STILES FG (1983) Biros. In: Janzen DH (ed) Costa Rican Naturalist 93: 117-125. natural history. University of Chicago Press, JANZEN DH (1973) Sweep samples of tropical foliage Chicago: 502-530. insects: effects of seasons, vegetation types, TAMSITT JR (1967) Niche and species diversity in elevation, time of day, and insularity. Ecology Neotropical bats. Nature 213: 784-786. 54: 687-708. TERBORGH J (1977) Bird species diversity along an JONES JK Jr & DC CARTER (1976) Annotated cheklist, Andean elevational gradient. Ecology 58: with keys to subfamilies and genera. In: RJ 1007-1019. Baker, JK Jones Jr, & DC Carter (eds) Biology of bats of the New World family Phyllostoma- VAN VALEN L (1973) Body size and number of plants tidae. Part I. Special Publication Museum Texas and animals. Evolution 27: 27-35. Tech University 10: 7-38. WIENS JA (1983) Avian community ecology: an icono- KARR JR & FC JAMES (1975) Eco-morphological clastic view. In: Brush AH & GA Clark (eds) configurations and convergent evolution in Perspectives in ornithology. Cambridge Univer- species and communities. In: Cody ML & JM sity Press, Cambridge: 355-403. Diamond (eds) Ecology and evolution of com- WILLIG MR (1983) Composition, microgeographic varia- munities. Belknap Press, Cambridge, Massa- tion, and sexual dimorphism in Caatingas and chusetts: 258-291. Cerrado bat communities from northeastern KOOPMAN KF (1978) Zoogeography of Peruvian bats Brazil. Bulletin of the Carnegie Museum of with special emphasis on the role of the Andes. Natural History 23: 1-131.