Bat Diversity and Abundance as Indicators of Disturbance in Neotropical Rainforests RODRIGO A. MEDELLIN,*¢ MIGUEL EQUIHUA,+ AND MIGUEL A. AMIN* “Instituto de Beologia, Universidad Nacional AutGnoma de México, Ap, Postal 70-275, 01510, México, D.F #nsituto de Eeologia, AC, Ap. Postal 63, 91000 Xalapa, Verucruz, México, DF. Abstract: Evaluating the degree of disturbance of any region to determine is relative importance for conser ‘uation purposes requires procedures that are relatively inexpensive and that yield accurate results fast. Be ‘ease bats are abundant, diverse, and easy to sample, especially in the Neotropical rainforest, they fulfil sv ral of the requirements of indicator species as identified in the Iterature. For 10 months we sampled bat communities inthe Selea Lacandona in Chtapas, Mexico, at 15 sites representing five habitats. We aso meu sured 10 variables representing vegetation structure and diversity at each site, With fuzzy-set techniques we produced a gradient classification of disturbance forthe 15 sites based on the vegetation data, We explored ‘the relationship between vegetation conditions, desribed as the membership degrees in the construct “fuz=y forest set” (he complementary fuzzy set of “disturbance”), and four bat community variable. Bat species richness, number of rare bat species, and the bat diversity Index were positively correlated with the vegeta tion scores, and relative abundance ofthe most abundant bat species was negatively correlated with vegeta tion scores. A bigh number of phyllostomine spectes tn a community 18 a good indicator of low levels of dis turbance. Atbough a single indicator group will probably not be sufficient for decsion-matking processes tn ‘conservation, evaluating bat populations mary be a good fist step in assessing an area's conservation value, specialty in rainforest regions. Diversidad y Abundancia de Murciclagos como Indicidores de Perturbaciones en civas Hiedas Neoteopicales Resumen: La evatuacton det grado de perturbacion de wna regidn particular para determina su tmportan: ia relativa para propésitos de conservacion requiere que los procesos de toma de decisiones utiicen tec rnologias relatioamente baratas y que proporcionen resultados precios pronto. Puesto que los murciélagos son abundantes, diversos y féciles de muestrear, particularmente en las seteas buimedas del Neotroptc, ena varios de los requertmientos de las especies indicadovas. Muestreamos por 10 meses comunidades de ‘murcilagos en la Selva Lacandona de Cblapas, México, en 15 sitios representantes de cinco bsabitats. Tan ‘bien medimos 10 variables representantes dela estructura y dversidad de a vegetacién en cada sitio. Med ante o uso de téenicas de conjuntos difusos produjimos una clasificactin gradual de la perturbacton de la ‘vegetacion para los 15 sitios. Exploramos las relaciones entre las condiciones de. la vegetacton, deseritos ‘como tos grades de membresia en la construccién de tas conjuntos difusos del bosque (el conjunto difuso complementario dela perturbacin) y cuatro varlables de la comunidad de murcicagos La riqueza de espe ies de murciélagos, el nimero de especies de murciélagas raros, el indice de diversidad de murciélagos ex ‘uieron posttioamente correlactonados con los valores de la vegetacton y la abundancia relativa de la espe le mds abundante estuvo negativamente correlacionado con estos valores. Un alto ntimero de especies de Phyllostominae en la comunidad es un buen indlcador de bajos nlveles de perturbacin. Sin embrago, un solo grupo indicador probablemente no es suficiente en el proceso de toma de decisiones en conserac. La ‘evaluacion de poblaciones de murciagos puede ser un buen primer paso en la evatuaciin det valor de un ‘area para la conservactan, espectalmente en regiones de selva biimeda. toma! medelinmirandaccologia nam ms ‘Paper submiid Sepember 3, 198 revied manuscript accepted March 29,2000 1666Modena Introduction ‘Understanding the causes and consequences of variation in the distribution and abundance of life forms is pari ‘mount to the conservation of biological diversity. Many studies have sought to determine and measure the vari- ables influencing biodiversity at local, regional, and glo- bal scales (eg, May 1988, 1990; Arita 1993; Wilson 1993; Erwin 1997; Ceballos et al. 1998). Pristine ecosys- tems continue to dwindle at alarming rates, and the ef fects of human activities on biodiversity range from dre ‘matic to subtle. These anthropogenic effects have resulted ina global pattern of seminatural matrix (which consists of patches of pristine habitat among habitats with different levels and types of human-induced disturbances [Franklin 1992). Evaluation procedures need to be relatively inex: pensive and yield accurate results fist. Evaluating distur bance effects on ecological patterns and processes in spe- ‘cific areas yields important information on which to base conservation and management decisions. For conserva tion purposes, the use of adequate biodiversity indica: tors is particularly helpful in this sense (Mooney et al 1995). To be useful as indicators, taxa should be atleast abundant, and ecologically, taxonomically, and trophi cally diverse. They should have a substantial functional role in the ecosystem and respond to envisonmental changes in a quantitative, predictable way (Noss 1990; Mooney ct al. 1995). In addition, good indicators should have wide gcographic distributions and sampling tech- niques should be costefective (Noss. 1990). Indicator species should be surrogates for the state of health of ecosystems (Rapport et al. 1981; Rapport 1990; Keddy 1991). Using a single, traditional indicator esiterion, such as high habitat specificity or large size, is not useful in monitoring population trends or habitat quality CLan- res et al. 1988). Stating the level at which a given ind- cator is expected to work, from genetic to landscape scales, is critical for logical, reasonable use of the indica- tor (Noss 1990), Monitoring of populations or commun ties in the context of known environmental change, such as those due to human disturbances, can also pro- vide a basis for improved management decision making (Kremen 1992). We argue that bats are good indicators of habitat disturbance at the commu forest within a river floodpla Bats are ubiquitous in the vegetated terrestrial ecosys: tems of the world and are particularly diverse and abun- ant in the tropics (Fenton 1992), They are relatively easy to find and represent the second largest order of ‘mammals in terms of numbers of species (Wilson & Reeder 1993), Because of their dramatic ecological and evolutionary radiation, bats occupy virtually every trophic level, from primary to tertiary consumers. Bats feed on fruit, insects, nectar, pollen, fish, blood, vertebrates, and leaves, and in many cases they select Specific habitats (Hill & Smith 1985; Fenton 1992). They are important in aioe at Dery and Disturbance 1667 ecological processes through interactions such as seed dispersal, pollination, and insect population regulation Fleming 1988, 1993; Whittaker 1993; Medellin & Ga- ‘ona 1999). Their roosting habits are diverse, ranging from caves and crevices to hollow trees, to curled young Ieaves, to leaf roosts they construct (Timm 1987; Fenton 1992). Many bats have specialized needs ia diet, roost ing sites, and habjta selection (Fenton 1992). Diet and hhabitat specialization have been identified as correlates of ratty and are significant predictors of the probability of local extinction (Leck 1979; Laurance 1991). All these traits make bats good indicators of the status of several hnabitat features, such as particular structures (e.., spe cialized roosts), food variety, and an assortment of mi ccohabitats. These features are found in the pristine, in- tact conditions under which these species evolved, From this it follows that absence of these features will determine absence of these species. The relationship of an indicator species to environmental variables tells how this species or group of species, such as bats, may func- tion as a measure of habitat integrity (Blair 1999). In the Neotropics, bats are numerically more abun- ‘dant than all other mammalian groups and are equal to or greater than all frugivorous birds (Bonaccorso 1979; ‘Terborgh 1983). They are also diverse, with about 220 species (Emmons & Feer 1997) representing over 50% ‘of mammal species locally and 24% of the number of bat species worldwide. Literature on bat systematics, natu: nil history, and ecology is extensive, but we found only a few studies in which bats have been evaluated as ind ‘cators of habitat disturbance (Johns et al. 1985; Fenton et al, 1992; Wilson et al. 1996). The results of these stud- ies suggest that bats in the subfamily Phyllostominae are sensitive (0 habitat disturbance. High levels of abun- dance of certain species such as Desmodus rotwndus and Carollia perspicillata may also indicate habitat dis turbance. ‘The mammal fauna ofthe Selva Lacandona in Chiapas, Mexico, is the most diverse in the country, containing 116 species of mammals, including 64 bat species. This fauna is typical of many Neotropical rainforest sites and is as diverse as most (Medellin 1994). Proportions and pattems of disturbance inthe region are complex (Cuarén, this issue) and constitute a mosaic of human-ransformed habitats creating a disturbance gradient from pastures and ‘comnfields to oldfields and cacao plantations, ‘We studied the bat community to evaluate changes in species richness, composition, and abundance as they relate to vegetation diversity, structure, and complexity along a human-produced disturbance gradient. Our ob- jective was to show that the diversity and steucture of the bat community indicates how closely a given site re- sembles a rainforest, as evidenced by a compound vari- able composed of various raw vegetation variables, We show that bats can be used to assess the level of distur bance of a site beyond what can be perceived visually or conser BoySH | 1668 Rainforest at Perit and Disturbance by a single measure of vegetation diversity of structural variable. We distinguish berween several degrees of for- est disturbance and present some specific, quantitative ‘ways in which bats respond to disturbance Study Area and Methods (Our base was the Chajul Tropical Biology Station (lat 1677'N, long 90°56'W, 120 m above sea level) at the southern edge of the Montes Azules Biosphere Rescrve. ‘The forest in this reserve, embedded in the dona, is extensive, covering over 3000 kn 1994) The Lacantin River is its southern boundary and to the south is a complex mosaic of humandisturbed habitats. We focused on a disturbance gradient includ ing active cornfields, oldfields, cacao plantations, and rainforest. Active comfield was comdlominated agricultural field mixed with beans, squash, and other crops. It was cult vated and harvested during our study. Vegetation never ‘grew above 3.5 m in height and was not diverse or struc ‘urally complex. Oldtields were inactive agricultural fields covered with secondary vegetation that had been abandoned for 4 period between 8 and 18 years, Dominant tree species ‘were Cecropia spp., Schtzolobium parabybum, Ocbroma Pyramidale, and Tropbis mexicana. Oktiekds were char acterized by two vegetational strata: an abundant hesba- ‘ceousshrub layer up to 4 m tll and a discontinuous can- opy between 10 and 25 m with little complexity because termediate vegetation layers were poorly developed or absent and because there was almost a complete lack of It anas (Medellin & Equihua 1998) and low-tree species richness. We included a younger (8-12 years since aban- donment) and an older (>12 years since abandonment) representative of oldies, Cacao plantations are common in tropical America Cacao (Sterculiaceae: Theobroma cacao) is planted as an understory tree; the original, diverse canopy of trees of the forest—such as Spondias radikoferl, Lieanta Platypus, Inga spp., Ceiba pentandra, and others—is left as shade for the cacao trees, Cacao plantations were harvested during our study. We compared cach disturbance stage—com-domi nated agriculture, oldield, cacao plantation—against rainforest habitat. The rainforest vegetation was struc: turally complex and diverse, with many large trees and lianas of many species We selected three study areas within each habitat, for 4 total of 15 sites. Each site shared the following condi tions: forested on at least one side; 1-7 ha in area (only the rainforest was extensive); 500 m of more from the next study site; and in the Lacantun River flood plain. At each site, we Captured bats one night per month from April 1993 to April 1994, except in January, February, eden a. and March of 1994, when sociopolitical problems in the Selva Lacandona prevented us from visiting the site. We placed four, 9-mlong mist nets configured in two T lay: ‘outs 5 m apart near the cater of the site and worked them for 4 hours consecutively starting at sunset. In the cease of heavy rain (=3 times a month in June-Septen ben), we waited until the storm had passed, shook ex- cess water from the nets, and resumed counting the ‘-hour period. Fach captured bat was identified and individually ‘marked with a numerical combination of colored beads fon a plastic necklace. The bat was kept in captivity for up to I hour for another study and then released at the site of capture. As indicators of the bat diversity and abundance at each site, we used four variables: number of bat species recorded, diversity as measured by the Shan- snon:Wiener index (i; Magurran 1988), percentage of the ‘otal bats captured that were of the most abundant spe- cies, and number of rare species (those represented by <10 individual in all habitats), These four variables are ‘commonly used as indicators of community diversity and structure and are easily recordable. The number and repre- sentation of rare species is affected by sampling effor. Nevertheless, the log:normal characteristics of most spe- ciesabundance curves—a few species with many indi viduals and many species with a few individuals (Magar ran 1988; Krebs 1989)—leads to the conclusion that the length of the distribution tail how leptokurtic the distr bution is) is another good and simple indicator of its d- versity, The longer the tai of the distribution (.e., more rare species in the community among comparable com- ‘munities and equal sampling efforts), the more diverse the community Vegetation variables were measured in one rectangu: lar plot of 30 X 10 m G00 m?) per site, Diversity and stricture are good habitat descriptors because they rep- resent three dimensional complexity at the ground, sub- canopy, and canopy layers and describe in quantitative terms the variation and patterns of abundance of indivi ual species of plants, from trees to shrubs to herbaccous layers. In cach plot we measured diversity by the num: ber of species of vascular plants, the number of individ als per species, and the Shannon-Wienes diversity index, Vegetation structure was determined by herbaccous ‘cover and was measured as the percentage of a measur ng tape along the 30m length that intercepted herbs. ‘ceous cover, canopy cover (with a spherical densiome- ter; Lemmon 1956), tree basal area measured at breast height, and number of trees in each of four strata (2 5-10, 10-25, and >25 m tal), To characterize the diversity and structural variability in each habitat, we used a fuzzy-set description of the vegetation. This approach produces a flexible definition ‘of vegetation conservation status that is useful for mod- cling purposes. Thus, what we sought was a mathemati ccal map of the semantic notion of disturbance (KoskoTI Medlin a 1993; Medellin & Equihua 1998). To this end, we sum ‘murized the sampled vegetation variables through an ap- proach known as emeans fuzzy clustering (ezlek 1981; Equihua 1990), which produced two fuzzy sets ‘One set represented a gradient toward forest resem. bance (Iorest set) and the other a gradient toward a dis- tured area (disturbed set). One set was the comple- ment of the other, and both sets spanned the range of study sites from primary forest to active comtield, For these analyses we used the forest set. In any fuzzy classi- fication, the groups overlap cach other, and every site thas a degree of membership within the forest gradient ‘measured on a scale from zero, complete lack of mem bership in the forest set, to one, full membership in the forest set, or primary forest, Thus, membership be: ‘comes a new variable that maps, in this case, the struc tural and diversity attributes and degree of disturbance of the vegetation, providing both a gradient of habitat characterization and. a functional definition of groups. ‘These fuzzy groups encompassed the habitats originally recognized as forest, cacao plantation, oldfield, and corm field, which were then arranged in a gradient within the fuzzy sets, ‘To analyze the mapping of the bat community parame ters on the fuzzy sets of the vegetation, we applied the ‘same equation we used in the fuzzy clustering to calcu: late the fuzzy-set centroid! Sau Ea where us the value ofthe ah vale x hin ifoup Je uy the membenhip vac of ef rca gooey fed alle toe Sua anc proof gather wert a papel thee the fuzy ctcing ofthe vegetion The fa nes preter an ea vale beeen oneal “hur steaon and pli nity fey chssten ton with memibenip values I” everywhere, where isthe number of groups ‘Weanalyecd the mocaon between the fizy vegeta ton caateaton and the bat community by eeesion ofthe bat community vaables on the fret men Derip vara: Where counts were involved, mich a invchhessand numberof area pete, we ted ge crazed nex models suming Fouson buon We used the pacage GENSTAT 5 rease 3.2 Cawes Se elf Tas 1995) them Me Results ‘We captured 2415 individual bats (160 were recaptures fora total of 2573 captures) representing 34 species (Ta aijovet Bt Divers and turbance 1669 ble 1). CThe total bar fauna of the region includes at least 64 species [Medellin 1994)). Raw vegetation data indicated that the disturbance gradient was detectable in variables such as herbaceous cover (greater percentage in disturbed conditions), Shan- non:Wiener diversity index, number of plant species, number of trees (greater values in undisturbed cond tions), and others (Table 2). There was, however, nota ble difference between the lowest extreme of the gradi- ent (Corfield) and the subsequent habitats for example, canopy cover was much lower and grass cover was much higher in the comfield than in any other habitat ‘This is partially reflected in the scores ofthe sites along the fuzzy classification (Fig. 1). Also, the age of oldticlds did not seem to follow the gradient, because the older field scores were lower than those from the younger field; therefore, hereafter we do not discern between ‘young or old fields and refer to them collectively as old- fields ‘Table 1. Number of eptures of bat speces i four habitats inthe Selva Lacandona of Chias, Mevico Sbecies Forest Cacao Oldfield Cornfield Prevonotus parnelli 24 ‘Mormoops megalopbyila® “Micronyeteris megatous® 2 “Micronyeteris brachyotis® 1 ‘Mimion coxa 2 Phyllostomus discolor 2 5 Poyllostomus stenop! 3 Tonatia brasliense" 12 1 Tonatia evots® 2 Trachops cirrbosus 2 hrotopterus auritus* 5 Glossopbaga soricina 2 me Giossopbaga commissarist 15° 20387 Lichonycteris obscura” 1 Hylonyeteris underwood 2 Carolia brevicauda 72 1s 5045 Carolia perspietata 9 10 92 6 Sturnira iin 56 137 1023s “Antibes titratus 1 5166 Anibeusjamatcensit* =} 953830 Bnebisthenes bavi” 1 Dermanura phacotis 6 5 3 Dermanura watsont 7B BR Platyrebinus belert 6 0 7 ow Vampyresa pus’ 3 Vampyrodes major hiroderma vitiosum® 1 5 Uroderma bilobanum 7 402 4 CGenturiosenex 12 2 Desinodts rotundus" a. 5 1 Diplyita ecaudata 1 1 Thyroptera tricolor i Bauerus dublaquercus! 1 2 Myotis heayst™ 1 2 Total M469 os “Har species, represented by =10 tndividuale Moat abundant species each baba1670 infos a Divert and Disturbance Table 2 Average of vegetation varables foreach habitat a the Se eden tat a acandona, Chiapas, Mexico (range in parentheses). Na tndioulual Herbaceous Canopy No. trees No. tees No tees Totalbasal No.of species Shannon Wen savitat —" pante "cover (0) cover (%)_S-10 m tall 10-25 m tall “25 mallee (oe) of vascular plants index Corned 9518379 Ta ° ° ° 23 32 Lo ONEBID Ce ORO 6% aes oas-17) Young field S08 R36 ar ° Soa 5a C57) GS-66) 09) GA WH 025-618) 6035 odes 0 39 ose 2 Ws .26 2646 29) 7321865) 264465) (O86 3) IS) GH) @103-310.5) cre cacao 5.008 ‘e 38.2 33 Te ° a 22 1280-4970) 195-707) 65-994) 09) 25) GES 65-109) ais Forest 310 pe ei cir 8 5 34 70-1029) 229-500 C4199] G1 AHI dD GHEY GID, ees Most ofthe rare species recorded from each habitat (Table were phyllostomid bats (18 out of 22 rare species). Within thar family, the subfamily Phyllostominae was the most spe- Gestich (nine species) The forest had the lagest number of rare species (15), half of which were phyllostomines, and it had 6 species exclusively recorded in that habitat (four phy Jostomines), The comfield had the lowest number of rare species (6), only 2 of which were exclusive to the habi {at (both nectarivorous Glossophagines). The vampire bat (esmodus rowndus) was recorded in almost all habitats, bbutin numbers low enough to qualify tas rare species. The single most abundant species in each habitat was also indicative of disturbance. In the cacao plantation and oldficld sites Gntermedtiate disturbance levels), the ‘most abundant species were either Carolia brevicauda (ight sites) or C perspiciliata (one site). In all three cornfields, the most abundant species was Sturnira li jum. The most abundant species were different in each of the forest sites: Avtibeus jamaicensis, Sturnira Hi Forest set score ium, and Carolia brevicauda. ‘These three species ‘were abundant in all habitats, but the faet that the single ‘most abundant species varied among habitats suggests that this is useful indication of disturbance. In each of the forest sites, the most abundant species represented between 18% and 24% of all bats captured, whereas 8. Hl jum sepresented between 34% and 52% of those captured in the cornfields, The relative difference between the proportion represented by the single and next most abundant species showed a progressively skewed ten: dency. In the forest and cacao plantation, this difference ‘but it was large in oldfeld and comfield (Fig, 2). Rarefaction curves showing accumulation of species over time also show the trend in which the forest curve grew aster and higher than the other three habitats, and the cornfield habitat had the lowest accumulated number of species (Fig. 3). The results of the loglinear regression between bat species richness and the forestset scores of the veget il c3 Y2 YS O1 02 03 CAI CA2 CAS FI F2 Site oo Figure 1. Scores in the fuzzy forest set classification of vegetation in 15 sttes (three replicates per babitat) in Chia: pas, Mexico, indicating a disturbance gradient (C, cornfield; Y, young abandoned field [8-12 years since abandon. ‘ment}; O, old abandoned field [12 years); CA, cacao plantation; F, undisturbed forest)eine a Forest gea No. of individuals 13.5 7 9 11 1315 17 19 21 23 25 27 comfals 8 a , IIlh on 13 18 17 Species nfo at Drs ndistrbes 367A 180 160 Cacao plantation 140 1204 ‘00 80 so | 2 i 2 ° daa...--------- 138 7 8 113 15 17 19 21 Olarela a | 0 7 0 11 19 6 a7 10 ‘Species Figure 2. Relative abundance of bat spectes in three disturbed habitats compared with the original rainforest in Chiapas, Mexico. 0.03; was the tion was positive and significant (G = 4.31; p proportion of explained deviance, 0.375), the number of rare species and the degree of membership in the forest set ofthe vegetation 35 30 25 205 15 10 5 0 correlation herweet Bat species Month 0 1.2 3 4556 7 co 2 10 G = 62; p = 0.01; proportion of explained deviance 0.24; Fig 4), Similarly, the regression between the Shan: non-Wiener values for the bat community and the de- ‘afee of membership in the forest set of the vegetation Tecoma Figure 3. Rarefaction curves sbow- ing accumulation of species by ‘month in four different habitats in Chiapas, Mexico.18 1672 nfs a erty and Distrbance showed a postive, significant association (p = 0,01, 2 = 0.35), The percentage of individuals represented by the single most abundant bat species and the degree of mem- bership in the forest set of the vegetation showed a negative, significant association (p = 0.01, r? = 0.35; Fig. These four relationships (Fig. 4) showed a consistent pattern of association between the vegetation structure and the basic attributes of the bat community structure. ‘There isa wide spread of points, however, from the low to the high values, Low forest set scores (high distur: bance) co-occur with low values in the bat diversity pa rameters, except, obviously, in the case of relative abun: dance of the most abundant bat species, where high values cooccurred with high abundance. In contrast, high forest set scores co-oceurred with a wider range, including low and high values of the bat community pa rameters. Bat diversity dropped strongly below forest set 75) e Forest oa 20} acomtts . outes _. 20 . oo 02a o .. i” . =e] 3 1 . ~ oa . oo 2A SSC Maden ta scores of 0.7, which further supports the idea of bats be: ing highly sensitive to forest disturbance, Discussion Our results show a significant relationship between bat ‘community parameters and the fuzzy-set vegetation di versity and structure. This demonstrates the utility of this approach in indicating the degree of disturbance of specific areas in the Neotropics. Some key points arise from the literature (€.@, Johns et al. 1985; Fenton et al 1992; Wilson et al. 1996) which confer robustness on the concept of using bats as indicators of disturbance in the Neotropical rainforest region, especially ifthe sim- pling effort is standardized among sites within regions and if vegetation variables are quantified. In all the stud- ies we located, the subfamily Phyllostominae is a sensi ge 4, 6s na 7 02 on os 10 z ja] i ee Figure 4. Relationships between four bat community parameters and vegetation fuzzy forest set scores indicating degree of disturbance. On the sais, values clase to zero indicate habitat disturbance. Bat diversity ts measured by ‘the Shannon-Wiener index; rare species are those represented by a total of <10 individuals; and the single most ‘abundant species indicates the percentage of the total number of bats caught at that site represented by the most ‘abundant species,Modena tive group of species that tends to be represented poorly in disturbed areas, In this sense, they behave as “type I” species, those that indicate ecosystem integrity (Rap: port 1992), There is no single species exclusive to the ‘undisturbed forest in the four areas from which data are available (Quintana Roo and Chiapas, Mexico, the Peru: vian Amazon, and Panama), Chrotopterus auritus was caught only in undisturbed forest in Quintana Roo and (Chiapas CFenton etal. 1992; this study) but was present in both disturbed and undisturbed conditions in Peru (Wilson et al. 1996). Micronycteris brachyotts, Vamp odes major, and V. caraccioli (a sister species in South America) were caught exclusively in undisturbed forest in Peru and Chiapas; the first species was caught in dis turbed and undisturbed conditions in Quintana Roo. The most indicative parameter in this sense is the number of species of Phyllostominae caught in any site; these were al- ‘ways greatest in the undisturbed forest. In Peru (Wilson et al. 1996), 14 out of 18 phyllostomine species were ‘caught in undisturbed forest; 7 out of 9 in Chiapas (this study) and all species in Quintana Roo (Fenton etal. 1992) ‘were found in undisturbed forest. n Panama (Johns etal 1985), gleaning insectivores, a guild that inchides phyllo- stomine species, decreased from 15.3% ofall bats netted in an undisturbed area to 1.7% in a disturbed area. Why do phyllostomines disappear from disturbed ar ‘cas? This probably has to do with the degree of special iaation these bats have reached in niche dimensions such as diet (Some are exclusively carnivorous and ott cers glean calling insects from leaves or other surfaces; Belwood 1988) or roost (some species require unfurled Heliconia leaves, others termite nests, hollow trees, lange and shaded leaves for tent making, or widely vary: ing caves; Timm 1987; Fenton 1992), Another point of convergence between the Peru study (Wilson et al, 1996) and ours is with respect to the sin gle most abundant species. In the intermediate distur: bance sites of both studies, the dominant species was consistently Carollinae bat (an understory frugivore), ‘whereas in the undisturbed sites no consistency was evi dent and the dominant species varied, including an Arti beus, a Carollia, and in Chiapas a Sturnira 23%). The latter was consistently dominant in our three comfields GH, 39.696, and 52.3%), ‘The responses of these bat species to disturbance— and therefore their usefulness as indicators—can be ex: plained by their diet. Artibeus is a canopy wee specialist feeding heavily on figs and other forest fruits (Bonac- corso 1979; Handley & Leigh 1991; Gaona 1997). Carol lia is an understory species feeding heavily on succes- sional trees and tall shrubs such as Cecropia and Piper (Bonaccorso 1979; Fleming 1988; Gaona 1997). Sturmina is a generalist that eats a variety of fruits (Bonaccorso 1979), and in our study site half ofits diet was composed ‘of solanaccous small shrubs abundant only in the heavily disturbed sites (Gaona 1997). The use of the single most Rainforest at Diersty and Disurhnce 67 abundant species as an indicator has some important im plications because they represent the most common species and those that are easiest to detect and most likely to be represented from the start in any simple ‘Consequently, this approach does not focus on accurat detection of rare species, whose representation on sam- ples is strongly stochastic and therefore hard to stan- \dardize through sampling effort. In every case the most ‘common species was a frugivore, whereas nearly all pphyllostomines (al rare species) were insectivorous fol age gleaners. In our study sites vampire bats were not as abundant as in other areas (those with heavy cattle pres: fence), which may indicate theit original relative abun- dance levels. Providing protocols for determining indicator species and how they respond to anthropogenic factors has bbeen identified asa research priority for conservation bi ology (Soulé & Kohm 1989). Bats, as a community of consumers that occupies various trophic levels, are spe- ciestich, abundant, widespread, ecologically diverse, easy to sample, and responsive to disturtance in a pre: dictable way. They represent an important tool with which fo assess the functional integrity of a community ‘They are a better tool than a group that represents a sin ale trophic level, such as vegetation or top predators (the latter are rare and therefore hard to sample). It is ‘lear then, that bats are useful as indicators of a wide vie riety of disturbance conditions in Neotropical rainforests because (I) the value of an indicator is applicable only to factors related to the indicator itself (.e., a marsh bird ‘can only indicate the state of a marsh or a meadow; Beintema 1983); (2) bats consume such a variety of re- sources, in most cases specializing on a few species (see above; Fenton 1992), (3) bats depend on availability of a ‘wide range of resources and microhabitats; and (4) any type of disturbance in the Neotropical rainforests, from sclective logging to clearcutting for agriculture, from ¢a- ‘cao plantations that conserve the canopy to cattle ranch: ing, will by necessity affect atleast one of the attributes described above. Furthermore, bats are more numerous than all other mammal groups, and their abundance equals or exceeds that of all frugivorous birds (Bonac- corso 1979; Terborgh 1983). Bats have a significant ef: fect on ecosystem processes through ecological interac tions (Fleming 1988, 1993; Whittaker 1993). For example, bats disperse more seeds than birds in every step of the disturbance gradient (Medellin & Gaona 1999), These at tributes make bats a valuable tool as indicators of habitat disturbance. Our study also shows that bats display the at ributes established by Noss and Cooperrider (1994) for ‘community-ecosystem indicators, We present the first quantitative analysis ofthe param: ters of the bat community as they relate to vegetation structure and diversity in an anthropogenic disturbance gradient through fuzzy logic. The number of rare spe: cies, the total number of species in the community, the| GTA Rainforest at Dvr and Disturbance Shannon-Wiener value, and the relative abundance of the most common species are associated with the vene- tation disturbance gradient in forest set values of 0.7. ‘These findings show that these traits of Neotropical bat communities can serve as indicators of habitat distur bance and therefore can be useful in assessment of the conservation status of specific sites. ‘Our results support the idea that high diversity values, indicated by the number of total and rare species as well as Shannon-Wiener values, and low relative abundance Of the most common species are indicators of undis tured habitats. The reverse combination of values, how: ‘ever, may be recorded in both disturbed and undisturbed sites. Also, high number of phyllostomine species is an- other indicator of low levels of disturbance. ‘Our technique of using bats as disturbance indicators has been adopted by at least three biosphere reserves in tropical Mexico to evaluate the conservation state of dif ferent areas in each reserve (J. Bezaury, personal com munication), and the data are being analyzed to evaluate their usefulness in conservation and. reserve: manage- ‘ment decision making. A single indicator group does not provide a basis for decision making and management for conservation. Because bats are relatively easy to sample, and because the available evidence from three studies in the Neotropics (Quintana Roo and Chiapas, Mexico, and six sites in lowland Peru) agree on several important points, this ubiquitous, diverse, and abundant group of mammals may be used as one important indicator of habitat disturbance in Neotropical rainforests Acknowledgments We thank ©. Gaona for constant and enthusiastic help in the field and lab, and M. and R. Lombera, R. Guerrero, M, Villalobos, S. Sinaca, and the staff at the Chajul Trop cal Biology Station for logistic help. ©. Gaona, R. Dirzo, R, Frias, and C, Equihua provided fruitful discussion of these ideas, G. Ceballos, A. Cuaron, J. Malcolm, M. B. Fenton, and D. E. Wilson enthusiastically read and com mented on the manuscript. This work was partially sup- ported by the John D. and Catherine T. MacArthur Foun- ‘dation and by Conservation International Mexico. Literature Cited Ast, HLT 1995. Riqueaa de especies de la mastofuna de México. Pages 109-128 in R.A. Medlin and G. Ceballos, etors. Avances sel estudio de le mamifers de México. Publcaiones pe inles, Volumen 1, Asoclackn Mencia de Masiozoolog, A.C ‘eintema, A.J 1985. Meadow bids as indicator. Environmental Mon orig and Assessment 3591-398 etwood J.J, 1988 Forating behavior, prey selection, and echact ein ‘on in posting bas (Pyllostomida). Paes 61-605 iP. ‘Metall and P. W. B. Moore, editors. Animal sonar, Plenum Pres, New York Beadek, J.C. 1981, Patem recognition with fuzzy objective function slots. Plenum Press, New York ‘ia, HB, 199. Bins and ater logan wean gratin: surrogate tua for asesting baer? Ecological Appin 161-17. ‘Bonaccorso, FJ 1979. Foraging and reproductive ecology ina Pan ‘manian bat communty. Pages 359-08 in The Ponda Sate Mt ‘cum, Biological sciences bulletin 24 The Florida Sate Museum, (Cehullos, ., .Hodsiguer and KA Medel 1998. Assessing conser ‘ati pote ina megadversity country: patens of mammal iver, endemicty, and endangerment. Ecologial Aplications aa, eld pide. 2d edition. The Univerty of Chicago Press, Chica. quihua, 1990. Fuzzy clustering of eclogeal data. Jounal of Bea ogy 78519-534, Lewin, . 1997, Biodiversity atts utmost trope forest beetles Pages 27-40 in Mc L- Reaka Dad, D. F. Wikon, and EO. Wise, ‘eon, Biodiversity Joseph Henry Press, Washington, D Fenton, M. 8.1992 Bas Fes on Fle, New York Fenton, MH, L Achar, D- Aue, M. BC. Hickey Mesiman, St K.Obris,D. M. Syne, ana 8, Ans, 1992, Pllostomsd bats (Chi ropterPhylloxomie) 35 nictors of habia dption in he Neotropics Biorpea 2440-446. Reming, TH. 1988, The shore eu bat. University of Chicago Press, Chicago ming, TH. 1995, Plant isting bats. American Sclenst SHe461-647 Franklin) F, 1992, An ecologists view of ustsnailty, Defining tainable forestry. Island Press, Washington, B.C ‘Gaona, ©, 1997, Depersin de sails hibits alimenticio e mur ‘elgisfagivoron en i ta Lacandond, Chiapas. Undergrad thesis Universi Nacional Autonoma de México, MExko, DF Mandley, C. 0, Je. and E.G. Leigh Je 1991, Diet and food supply Pages 147-149 in CO. Handley Je, D.F. Wikon, and AL. Gi ner elton. Demography an natural history of de common Ft tari jmatconss on Baer Colorado sland, Panam. Coo ‘ebusons to zosogy 511. Smiisonan Iason, Washington, D.C ‘LJ. and). Smith. 1988. bats, a tua history. Texas Univer sity Press, Austin John, A.D, D.F- Wilson and RH Pine. 1985 ain forest bats: a un ern Fur Bat News 54-5. Key, PA. 1991. Biological monitosng and ecole prediion from aature reserve management to atonal state ofthe environ ment inaestors Pages 249-267 in FB Gosh editor Monitor ing for conservation and ecology. Chapman ana al, London. oak, I. 1995 Faey thinking he new science of faz fog. Hyper ects, €.,1989, coloical methodology. Harper & Row, New York Kemen, C1992. ASessing the indicor propels of specs sm ages for mtr aresmnioring. cokgil Apps 2208-217 andes, PB. J. Verner, ad JW Thomas. 1988, Eeolgial uses of etchrate indicator species: a cetgue. Conservation Biology 2: 316-329. {aurance, W. F 1991. Folica! concates of extinction pronenness ‘in Attala tropical rain forest mania, Conservation Biology Se 79.89. lawes Agricultural Tas 1995. GENSTAT Bre. Release 5. Reference nama Oxford University Pres, Oxford, United Kingdom eck, € F. 1979. Avan extinctions in an folate topes wet forest Preserve, Ector The Atk 96545-352 Lemmon, P1956. Aspherical densometer for estinating Forest over sor deny. Foret Selene 2814-320, on Univesity Pres, Princeton, New Jrsyeine May, RM. 1988, How many species ae thereon earth? Science 244 1441-149, May, RM. 1990. How many species? tilosophical Transactions of the Royal Sooty of London 8 330295-504, Medel, RA. 1994, Mammal diversity and conservation in the Sera Lacandona, Chiapas, México, Conservation Bolo 8788-799 “Medelin, KA, and Mt yuihua, 1998, Mammal species richness and ‘mbit use i ainforet and aadoned asic Bek in Chia is Mexico Journal of Ape Feoony 3513-25. “Medel, A. and O. Gaoaa. 1999 Seed pers by ats nd binds in forest and disturbed habitats ta Chaps, Mexico, Blotopic 31: Bea Mooney HAJ Inbehenco,R. Diz, and ©. F Sala. 1995 ver sity an ecosystem functioning: base principles. aes 275-325 V.H. Heywood, etor. Global biodverty assessnent, United Ne ‘ions Envonment Programme, Cambridge University Pres, Ca ridge, United Kingdom Noss, R199. Indians for monitoring bioiversty: a hierarchical ‘approach, Consenation Holy 4385-56, Nows, RF and AY. Comperidr, 199, Sng ate’ legac: Islnd Press, Washington, D.C apport, D.J. 1990, What constutes ecosystem eal? Perspectives 1 Bolo and Meine 33:120-132 apport D. J 1992. What clinic ecology? Pages 144-156 RCo 1675, ‘anc, BG. Nonon, and. Haskell eto. comptes es new sls or environmental management. and Press, Wasson, DC Rapport. JA: Reger and C. Thorpe. 1981. Diagoss rags and treatment of ecosystems under sess. Pages 121-137 In GW, Taret and R.Roseaber,etors Stes ects on natal econ: tems, Wiley, Chichester United Kingdom Soule, ME and KA Kohn. 1989. Research pronties for conserva ‘oa bilo and Press, Washington, DC ‘Tesborgh J. 1983. Five New Word pints. A stayin comparative ‘cology. Princeton Univerny Press, Princeton, New feney ‘Timm, RM. 1987. Tent consraction by bas of the genera Ariens and Uraderma Reina Zoaony (new sles) 39:187-212 ‘Whitaker JO, 1993. Bas, Beetles, and ups mor ig brown ats mean les agricul pests. ts 11(1)23, ‘ison, D. and D. Reeder, ers 1995. Mammal species of the ‘work atomic and peographe reference. 2nd etion. Smith: sonianInsttaton res, Washington, D.C Wilson, DE, C.F. Ascoen, and S. Sle. 1996, Rasa indicators of Ibis discrbance. Pages 615-625 in D. Wilson and A. Sandor al, editors. Manu, the biodiversity of southeastern Per. US. Nt onal Muscum of N:tural History Saithsonian Inston, Was ington, DC, and Editor! Horizont, Uma \Wikon, E10, 1993. The diversity of Me Belknap Pres of Hara Unk ‘verity Press, Cambri, Masset