Phyllostomid bat diversity in a variegated coffee landscape

Phyllostomid bat diversity in a variegated coffee landscape

BIOLOGICAL CONSERVATION Biological Conservation 122 (2005) 151–158 Phyllostomid bat diversity in a variegated coffee l...

600KB Sizes 0 Downloads 0 Views


Biological Conservation 122 (2005) 151–158

Phyllostomid bat diversity in a variegated coffee landscape Catherine Numa


, Jose´ R. Verdu´


, Pedro Sa´nchez-Palomino




Instituto de Investigacio´n de Recursos Biolo´gicos Alexander von Humboldt, Carrera 7 #35-20, Bogota´, DC, Colombia Centro Iberoamericano de la Biodiversidad (CIBIO), Universidad de Alicante, Ctra. San Vicente del Raspeig s/n, E-03080 Alicante, Spain c Estacio´n de Biologı´a Tropical ‘‘Roberto Franco’’. Cr 33 No. 33-76 Barrio Porvenir, Villavicencio, Meta, Colombia Received 8 March 2004; received in revised form 7 July 2004; accepted 7 July 2004

Abstract We examined bat diversity at two different spatial scales: habitat and matrix, in the Quindı´o coffee region in Colombia. Habitats were: forest, shaded coffee and associated coffee; and matrices were: associated coffee (M1) and shaded coffee (M2). Three sampling sites from each type of habitat were located at each matrix. The forest areas of the Quindı´o region are severely fragmented and less structurally complex than coffee patches. The shaded coffee habitat had patches that were larger and more complex. In spite of limited patch size and lower complexity, the forest remnants were those with greatest species richness and demonstrated clear similarities even between the two matrices. This was not observed in coffee plantations, neither in associated coffee nor shaded coffee. On the landscape scale, M2 showed lower b diversity and greater edge density (ED) than M1. This fact explains that greater connectivity between different habitats exists in M2 than in M1. Our results suggest that production and conservation are compatible, as maintenance of forest remnants in a mosaic structure by landowners of the vegetation is sufficient to conserve phyllostomid bats at landscape level.  2004 Elsevier Ltd. All rights reserved. Keywords: Coffee agroecosystem; a Diversity; b Diversity; Forest fragmentation; Landscape matrices; Northern Andes

1. Introduction Studies on biodiversity of fragmented landscapes in Neotropics have generally focused on natural vegetation itself and overlooked the fact that traditional human activities through the ages have increased landscape heterogeneity, and therefore its biodiversity. Recently, studies have become increasingly aware that the landscape matrix within which forest fragments or remnants exist may be as important for biodiversity as the forest fragments themselves (Laurance, 1991; Wiens, 1995; Vandermeer and Perfecto, 1997; Renjifo, 2001; Perfecto and Vandermeer, 2002).


Corresponding author. Tel.: 34 965903400. E-mail addresses: [email protected], [email protected] (J.R. Verdu´).

0006-3207/$ - see front matter  2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.biocon.2004.07.013

The influence of agricultural systems on biodiversity has traditionally been studied from a native forest perspective (Estrada et al., 1993; Estrada and CoatesEstrada, 2002) and a coffee shaded management perspective (Greenberg et al., 1997; Perfecto et al., 1996) in relatively well-conserved landscapes. That is, the original forest and forest remnants are a representative habitat in the territory (mainly a natural reserve), which have an important influence on agricultural habitats diversity depending on the distance between the main forest fragments and the agricultural patches (Ricketts et al., 2001). However, high production coffee regions in Colombia provide another scenario showing a good example of the degradation of coffee agroecosystems. As part of a strategy to increase production, the industry advocated the technification of coffee, whereby shade trees are removed and the use of chemical fertilizers is common


C. Numa et al. / Biological Conservation 122 (2005) 151–158

(Rice and Ward, 1997). In this sense, whereas twentyfive years ago, the majority of coffee production was associated with traditional practices (shaded coffee), in the 1990s 68% of coffee surface area was technified and 400,000 ton of chemical fertilizers were used (Rice and Ward, 1997). Coffee has been cultivated since the second half of the 19th century (Renjifo, 1999) in the Quindı´o department, and growers use both sun and shaded technified systems. In this area a few organic coffee farms have emerged lately but there are no rustic or traditional systems (similar to those described by Moguel and Toledo, 1999), and some patches of forest remain in areas that are difficult to cultivate, or in order to preserve water supplies. Thus, we could consider this landscape as an extreme scenario of coffee landscape. To analyse biodiversity in these habitats, we have used bats, a group proposed as promising indicators for analysing biodiversity and habitat disruption due to their high ecological, trophic and species diversity and their easy sampling methods (Fenton et al., 1992; Wilson et al., 1996; Medellin et al., 2000; Moreno and Halffter, 2000). We focused on bats to answer several questions: (1) What influence does relative habitat surface, complexity (measured as a fractal dimension) and connectivity (measured as edge dimension) have on bat diversity distribution? (2) How does bat diversity differ among forest, shaded coffee and associated coffee habitats?, and (3) What influence does matrix type have on bat diversity distribution at landscape scale?

2. Methods 2.1. Study area The study was carried out between October 1999 and February 2000 coinciding with the rainy season and the flowering and fruiting of coffee plantations in the Quindı´o department on the western slope of the Central Andes in Colombia (410 0 N:7535 0 W to 440 0 N:7550 0 W), at elevations ranging from 1100 to 1850 m. The mean annual temperature in this region is 20.5 C. Total precipitation between February 1999 and February 2000 was 2850 mm, exceeding the mean annual rainfall of 2262 mm (Federacio´n Nacional de Cafeteros de Colombia, 2000). The Quindı´o coffee region includes approximately 76,422 ha. The landscape is an agricultural mosaic dominated by coffee plantations (65.5%) interspersed with some patches of other crops and land use such as banana plantations (Musa paradisiaca L., and M. sapientum L.) and pastures for livestock (24.5%). Natural habitat remains in lower montane forest patches (3.35%) and bamboo (Guadua angustifolia Kunth) patches (1.14%).

We focused our study on the three main habitat types: (1) associated coffee, a sun growing method including coffee (Coffea arabica L.) in monoculture or mixed with banana, with no tree species or isolated trees, if present; (2) shaded coffee, comprising coffee plantations with shadow tree species such as Inga codonantha (Pittier); I. edulis (Mart); I. densiflora (Benth) and associated eventually with banana; and (3) forest, present as small fragments in the coffee region. The principal families of these fragments include Moraceae (Brosimum, Ficus, Sorocea), Lauraceae (Ocotea, Nectandra), Melastomataceae (Miconia) and Rubiaceae (Psychotria, Palicourea). At the landscape scale, two different matrix landscapes can be observed in the Quindı´o coffee region: M1 where associated coffee is the predominant habitat and M2, with shaded coffee as the main habitat. 2.2. Landscape analysis The Quindı´o coffee region map (Fig. 1) was analysed by GIS Unity of IavH (Villarreal, 2000) from a classification of CRQ (1997) based on a 1996 Landsat TM satellite imagery with a 23 m resolution (each pixel was 23 m by 23 m). In this map, 21,100 ha (100.03 SD 1.70 ha) spatial windows were selected at random for calculation of different spatial metrics with FRAGSTATS (McGarigal and Marks, 1995). We selected mean patch size (MPS) and mean patch fractal dimension (MPFD) to evaluate the spatial structure and complexity of different habitats studied. We examined fractal dimension using a perimeter–area method, because this is a good spatial metric to determine the fragmentation level resulting from anthropogenic activities (Krummel et al., 1987). This fact was evaluated using a simple linear regression model. The Durbin–Watson test for first order autocorrelation in regression residuals was used to examine independence of variables (Sokal and Rohlf, 1995). At landscape level, the type of matrix could determine the connectivity level between different habitats. The existence of two different coffee systems, one without shade trees and another structurally more similar to the forest, suggests the use of edge metrics to determine this effect. As an edge refers to the border between two different habitats, we calculated edge density (ED) as an estimator of connectivity between different patch types. ED (in m/ha) takes the shape and complexity of patches into account, being a measurement of the spatial heterogeneity of a landscape mosaic. We used a non-parametric Kruskal–Wallis ANOVA on ranks to evaluate differences in MPS of patches among habitat types and mean ranks to make pair-wise multiple comparisons using post hoc DunnÕs test. To examine differences in MPFD of patches between habitat types, one-way ANOVA analysis with FisherÕs PLSD test for pair-wise comparisons was used. For compara-

C. Numa et al. / Biological Conservation 122 (2005) 151–158


Table 1 Sampling effort for bat surveys in the 18 localities of the Quindı´o coffee region Habitat type

Number of sampling sites

Total number of survey nights

Total sampling effort a

Forest Associated coffee Shaded coffee Total

6 6 6 18

18 18 18 54

5587.8 5897.5 6133.5 17,618.8

a Total net meters multiplied by the total hours that the nets were open each night for a given site (Moreno and Halffter, 2001).

2.3. Sampling design and bat sampling We selected 18 sampling sites distributed throughout the three vegetation types in the Quindı´o coffee region: six forest fragments, six associated coffee and six shaded coffee plantations. Three sampling sites of each habitat were located at each matrix (M1 and M2). Bats were sampled with 50–80 m of mist nets (6–12 m long · 3 m high). Nets were exposed at dusk for 5–8 h and monitored every 45 min for three consecutive nights with no full moon at each location (Table 1). Each bat captured was identified to species level, sexed, measured (forearm length), marked and released. Each bat was temporarily marked with nail varnish to identify recaptures at each sample site. This study was restricted to species belonging to the Phyllostomidae family in order to avoid biases in the estimation due to species detectability, because mist nets tend to underestimate the presence of other bat families foraging at higher altitudes (Kalko et al., 1996). 2.4. Diversity analysis

Fig. 1. The study site in the coffee region of Quindı´o: (a) location of the windows (white squares) for spatial landscape analysis and coffee growing types (see the existence of two matrices); (b). location of sampling sites and distribution of forest remnants (black patches). Abbreviations: F, forest; Ca, associated coffee; and Cs, shaded coffee.

tive purposes of ED means between matrices, the Wilcoxon signed-ranks test was used (Sokal and Rohlf, 1995; Gardiner, 1997).

Species richness (S) was evaluated at two scales: at the habitat scale, comparing the three habitats localized in the coffee region; and at the landscape scale, comparing the same habitat when located in the two different landscape matrices. To compare species richness among habitat types, we calculated a species accumulation curve using rarefaction methods for each habitat type. The rarefaction algorithm generates expected species richness based on a specified number of individuals randomly drawn from a community sample. We calculated the expected species richness E(Sn) and generated a mean and variance of species richness for each abundance by 1000 randomly iterations with EcoSim 7.0 (Gotelli and Entsminger, 2003). For spatial b diversity measures, we used a complementarity index calculated as defined by Colwell and Coddington (1994) using EstimateS software (Colwell, 1997) based on bats assemblages along the spatial gradient. Pair-wise comparisons were made of the three habitats in each matrix type, resulting in a triangular matrix


C. Numa et al. / Biological Conservation 122 (2005) 151–158

with 15 data points. After analysis, we carried out a cluster analysis by WardÕs method for linkage (StatSoft, 1997). At landscape scale, we used the WhittakerÕs diversity index (Whittaker, 1972) and a new index based on biotic distinctness among a gradient of habitats and suggested as an overall measure of the level of continuity in species composition within a system of n habitat (‘‘sensu lato turnover’’ attributable to spatial heterogeneity). This formula was inspired by RapoportÕs Index of cosmopolitism (C) (Rapoport, 1982) and the related endemicity index (EN) (Verdu´ and Galante, 2002). Thus, from presence–absence matrices, we proposed the following formula: N  P nS n =S  1 bN ¼ 1  n¼1 ; N 1 where n is the number of habitats occupied by each species, Sn is the number of species which occur in nth habitat, S is the total number of species and N is the total number of habitats considered. Thus, the dimensionless b diversity measured by this formula varies from 0, when all species are present in all habitats (uniformity of spatial gradient), to 1, when all species occur in only one habitat (ÔuniqueÕ). Using this formula, we can compare the b diversity among different landscape units (especially in landscapes with distinct grades of human modification) showing the same habitat classification. In this study, habitat was defined as a discrete variable, but in practice it can also be quantified as a continuous variable using ranks or levels (e.g., temperature ranks, elevation, deforestation level, vegetation cover, etc.). Spatial c diversity was calculated in each matrix to compare the percentage explained by a and b diversity components. In this case, we used the formula proposed by Schluter and Ricklefs (1993), c diversity being the result of multiplying the average a diversity in the N habitats of the region, b diversity as the inverse of average number of habitats occupied by a species, and N as the total number of habitats considered.

with forest and associated coffee patches (Kruskal–Wallis H = 23.04, df = 2, p = 0.0001; DunnÕs test, Wij = 9.83, differences in average ranks, shaded/associated coffee = 12.99, p < 0.05; shaded coffee/forest = 16.82, p < 0.05), whereas forest and associated coffee patches did not differ significantly (difference in average ranks = 3.91, p > 0.05) (Fig. 2). Based on spatial structure, forest remnants showed the smaller fractal dimension on average (1.064, SD 0.016) and therefore were the less complex of the studied habitats (ANOVA F = 3.13, df = 2, p = 0.05) (Fig. 2). Only forest and shaded coffee patches differed significantly in their fractal dimension (Fisher PLSD = 0.029, p < 0.05). Thus, native forest is notably reduced to very small patches (average area 6.24 SD 3.04 ha), these being significantly smaller and simpler in the landscape studied. A positive and significant relation was observed between mean patch size and fractal dimension of forest patches (r2 = 0.73, Z = 2.51, p < 0.05) (Fig. 3). At landscape scale, M2 had greater ED (90.09 SD 22.69) than M1

Fig. 2. Spatial characterization of three habitat studied using mean patch size (in gray) and mean fractal dimension (in white). In all cases n = 11 and mean values were bracketed by their SD .

3. Results 3.1. Landscape analysis The coffee landscape of Quindı´o is severely fragmented, forming a mosaic structure dominated by coffee plantations. The vegetation map (Fig. 1(a)) corroborated the existence of two landscape matrices characterized by both associated (M1) and shaded (M2) coffee. In the Quindı´o, the lower montane forest was represented by small patches distributed along the river basins and included coffee matrices as small islets (Fig. 1(b)). The shaded coffee patches differed significantly in their areas

Fig. 3. Relation between mean patch size and mean patch fractal dimension of forest (r2 = 0.73, Z = 2.51, p = 0.022).

C. Numa et al. / Biological Conservation 122 (2005) 151–158


Table 2 Phyllostomid bat species captured in different habitat types at Quindı´o coffee region Species


Associated coffee

Shaded coffee

Total coffee region

Glossophaga soricina (Pallas) Artibeus jamaicensis (Leach) Artibeus lituratus (Olfers) Carollia perspicillata (L.) Sturnira lilium (E. Geoffroy) Carollia brevicauda (Schinz) Platyrrhinus vittatus (Peters) Sturnira ludovici (Anthony) Chiroderma salvini Dobson Anoura geoffroyi (Gray) Artibeus phaeotis (Miller) Phyllostomus discolor (Wagner) Desmodus rotundus (Geoffroy) Platyrrhinus helleri (Peters) Carollia castanea (H. Allen) Choeroniscus godmani (Thomas) Phyllostomus hastatus (Pallas) Vampyressa pusilla (Wagner) Platyrrhinus dorsalis (Thomas) Uroderma bilobatum Peters Mimon crenulatum (E´. Geoffroy Saint-Hilaire) Total individuals Total species

30 45 22 28 17 15 14 10 7 2 12 0 6 2 4 4 0 0 1 1 1 221 18

192 72 50 55 9 14 5 0 3 1 2 2 0 2 0 0 2 2 0 1 0 412 15

122 109 74 39 23 17 5 1 11 11 4 5 0 2 1 0 1 0 2 1 0 428 17

344 226 146 122 49 46 24 11 21 14 18 7 6 6 5 4 3 2 3 3 1 1061 21

(59.27 SD 38.19) (Wilcoxon signed-ranks, Z = 2.49, p < 0.05). 3.2. Bat diversity With an effort of 17,618.8 m/h at the 18 sampling sites in the Quindı´o coffee region, we recorded 1061 bats belonging to 21 species (Table 2). Rarefaction analysis showed that at n = 200 individuals, the forest was significantly the richest habitat in expected species (E(Sn) = 16.80, SD 0.41). The same analysis showed that the species richness between coffee growing types was similar to the sample for associated coffee (E(Sn) = 11.97, SD 1.13), which fell within 95% confidence interval of the rarefaction curve for shaded coffee (E(Sn) = 13.67, SD 1.14) (Fig. 4(c)). At landscape scale, the three habitats studied differ in accumulated species richness between landscape matrices. At M1, species richness showed a similar pattern with the habitat scale analysis. Rarefaction analysis showed that at n = 150, forest at M1 was the richer habitat (E(Sn) = 14.61, SD 0.56), with species richness being similar between the two types of coffee plantations, since shaded coffee richness (E(Sn) = 9.36, SD 0.69) fell within the 95% confidence interval of the associated coffee sample (E(Sn) = 7.86 SD 0.83) (Fig. 4(a)). These patterns are different at shaded coffee matrix (M2). At n = 34, species richness in all habitats were similar since forest (E(Sn) = 9.85, SD 0.36) and associated coffee (E(Sn) = 9.80, SD 1.18) fell in the 95% confidence interval of the shaded coffee sample (E(Sn) = 9.43, SD 1.26) (Fig. 4(b)).

In the dissimilarity cluster (Fig. 5) based on complementarity index, both matrices form distinct groups from the coffee habitat viewpoint. However, both forests were more similar to the shaded coffee matrix than associated coffee matrix. Associated coffee matrix (M1) showed greater b diversity (bW = 0.58; bN = 0.55) than shaded coffee matrix (M2) (bW = 0.32; bN = 0.36).c diversity in the shaded coffee matrix was higher (19.08 species) than that observed in associated coffee matrix (18.04 species). c diversity in the overall landscape was 21.53 species.

4. Discussion Coffee landscape in Quindı´o could be considered as a man-made landscape where agricultural patches dominate the territory and forest patches appear highly fragmented, less complex and immersed in the coffee matrices. From a spatial point of view, progressive fragmentation of the forest involves a reduction of the area of patches, and a progressive loss of structural complexity (Fig. 2). Fractal dimension has been suggested as a promising comparative measure of complexity and human disturbance in landscapes (Krummel et al., 1987; OÕNeill et al., 1988; Mladenoff et al., 1993). The lower values of MPFD and the positive and significant relation between MPS and MPFD found in forest patches of the Quindı´o landscape confirm a strong degradation process of forest, being reduced to small and disperse circular patches (Fig. 1(b)).


C. Numa et al. / Biological Conservation 122 (2005) 151–158

Fig. 4. Rarefaction curves for the three habitat and two landscape matrix studied. Comparisons were made at: (a) n = 150 at associated coffee matrix; (b) n = 34 for shaded coffee matrix; (c) n = 200 for the whole Quindı´o coffee region.

Fig. 5. Cluster analysis dendrogram (from complementarity matrix) showing differences in bat assemblage structure for the three habitat types at two landscape matrices. M1, Associated coffee matrix; M2, Shaded coffee matrix; F, forest; C, associated coffee; S, shaded coffee.

The structural differences between coffee system patches determined the complexity of the landscape matrices. In this sense, the landscape analysis showed the existence of two delimited affluent areas with partic-

ular characteristics. The shaded coffee matrix (M2) formed by larger shaded coffee patches was more complex in structure of edges between habitats (ED) than associated coffee matrix (M1), where associated coffee patches had more simple forms and less area. The fact that shaded coffee shows a structure of intermediate cover between the forest and associated coffee allows us to confirm that in M2 the connectivity between patches is greater. This fact was corroborated by the existence of a greater edge density (ED) between patches in M2, being an estimation of greater connectivity between habitats. In spite of spatial characteristics, forest patches at the coffee region of Quindı´o play an important role in biodiversity conservation. We found the highest species richness and some ‘‘unique’’ species in this habitat. Due to their spatial characteristics it is unlikely that forest fragments themselves may be suitable for sustenance and persistence of bat populations. Earlier studies of countryside landscapes have also found high diversity in relatively well-conserved forest fragments (Estrada et al., 1993; Estrada and Coates-Estrada, 2002; Daily et al., 2001). The bat species living in the Quindı´o coffee zone could be considered as an agricultural assemblage that can survive and breed in this landscape. The four most abundant species present in all habitats studied have been reported occurring in several habitats including agricultural, semi-urban and urban environments (Bredt and Uieda, 1996; Zorte´a and Chiarello, 1994; Estrada et al., 1993). Only three ÔuniqueÕ species, Desmodus rotundus, Choeroniscus godmani and Mimon crenulatum, seem to be highly dependent on the forest. The presence of refuges in the forest could be one of the factors that explain this behaviour because these species have been reported as using hollow trees, fallen logs and caves as refuges (Handley, 1976), which are not as readily available in coffee plantations. The occurrence of the hematophagous D. rotundus in the forest surrounded predominately by coffee also indicates the progressive process of the change from coffee to cattle (Sadeghian et al., 1998), especially in lower coffee zones (1000– 1200 m). This countryside bat assemblage had a low response in species richness to structural differences of coffee system management. As bat species can be segregated or benefited by the roost site selection (Kunz, 1982; Fenton, 1997), we suggest that in this case banana, the main species cultivated with coffee in associated coffee systems could provide temporal refuge and food resources for bats (Roubik, 1995; Fleming, 1986), allowing a similar co-occurrence as in the shaded coffee. This factor could explain the higher abundance of the nectarivore Glossophaga soricina at the coffee plantations. Diet and roost habits of this habitat generalist include pollen and nectar of Inga and Musa (Fleming, 1986) and a high variety

C. Numa et al. / Biological Conservation 122 (2005) 151–158

of roost sites, including man-made structures (C. N., pers. obs.). Landscape context plays a determining role for bat diversity distribution in this coffee region since the same habitats immersed in matrices with different structural complexity were different. While forest fragments at M1 were the richer habitat, at M2 this habitat was similar to both coffee systems, seeming to lose importance for concentrating bat species. This was confirmed by the lower b diversity values at M2 in relation to M1. The M2 area showed more complexity in patch forms but was more homogeneous in habitat structure landscape than M1, which had lower patch complexity and more heterogeneity of habitat structure. Moreover, the spatial distribution of bat species was more homogeneous in M2 than in M1. In this sense b diversity at the Quindı´o coffee region was more important to determine c diversity than in other agricultural regions with similar c diversity (Moreno and Halffter, 2001). In spite of high vagility and a high number of generalist species, the distribution of bat assemblages responded to the habitat type (Table 2). The importance of forests was confirmed when complementarity was examined: the forests at both M1 and M2 conserved their ‘‘species identity’’ in relation to the other habitat whereas coffee plantations were more influenced by matrix type than by the coffee growing system. The two coffee systems at M2 were more similar to each other than the same habitat at M1. This was also observed at the two coffee systems at M1, with the species composition of coffee habitat being more similar to the forest in M2 than in M1. Differences in species distribution of bat ‘‘agricultural’’ assemblage between matrix type and habitat type show the need to establish different strategies for biodiversity conservation of this agricultural region. The bat assemblage could survive in a landscape with very reduced fragments of forest that serve as refuge and provide food for many species, some unique to this habitat. This fact explains that in some groups, species richness does not differ significantly between very small forest fragments and larger areas (Sekercioglu et al., 2002). Our results suggest that the maintenance of the connectivity between the different patches from the mosaic is necessary to maintain intrinsic forest diversity. In this sense, although forest ÔuniquesÕ species should be considered for the conservation of forest diversity, the best determinant of the impact of disturbance and fragmentation could be bat species common to both forest and shaded coffee (see Table 2). At landscape scale, the maintenance of landscape structure in a variegated form is necessary for biodiversity conservation. The landscape dominated by coffee plantations is clearly suitable for the studied bat assemblage. We argue that production and conservation are not incompatible in this case. In order to conserve higher diversity in landscapes of this kind, high priority must be given to maintenance and protection of the forest


remnants that were preserved traditionally by land owners. We suggest that the key is to maintain spatial heterogeneity where the natural habitat and the countryside are alternated on different scales. Acknowledgements We thank the GIS unit of Alexander von Humboldt Institute for their support in providing maps and FRAGSTATS data. We are grateful to E. Galante, G. Halffter, C. Moreno and J.M. Lobo for their comments and suggestions of this manuscript. We are indebted to Yenyfer Mona´ and Pablo Zanabria for their assistance in field work. We thank Yaneth Mun˜oz-Saba for final determination of the bat reference collection and Kate Burke for checking the English manuscript. This work was funded by COLCIENCIAS through ÔBiodiversity and agricultural systems at coffee zone (Quindı´o) projectÕ from the Instituto de Investigacio´n Alexander von Humboldt. We thank the Corporacio´n Auto´noma Regional del Quindı´o (CRQ), UMATAS, Cenicafe´, landowners and farm administrators for research permission and collaboration. This paper is dedicated to D. Rodrı´guez (in memoriam). References Bredt, A., Uieda, W., 1996. Bats from urban and rural environments of the Distrito Federal, mid-western Brazil. Chiroptera Neotropical 2, 54–57. Colwell, R.K., 1997. EstimateS: Statistical Estimation of Species Richness and Shared Species from Samples. Version 6.0b. Available from Accessed on 6/6/03. Colwell, R.K., Coddington, J.A., 1994. Estimating terrestrial biodiversity through extrapolation. Philosophical Transactions of the Royal Society (Series B) 345, 101–108. CRQ., 1997. Mapa de coberturas vegetales y ocupacio´n del espacio en el departamento del Quindı´o (1:25000 scale). Corporacio´n Auto´noma Regional del Quindı´o. Armenia, Colombia. Daily, G.C., Ehrlich, P.R., Sa´nchez-Azofeifa, G.A., 2001. Countryside biogeography: utilization of human-dominated habitats by the avifauna of southern Costa Rica. Ecological Applications 11, 1–13. Estrada, A., Coates-Estrada, R., Merrit, D., 1993. Bat species richness and abundance in tropical rain forest fragments and in agricultural habitats at Los Tuxtlas, Mexico. Ecography 16, 309–318. Estrada, A., Coates-Estrada, R., 2002. Bats in continuous forest, forest fragments and in an agricultural mosaic habitat-island at Los Tuxtlas, Mexico. Biological Conservation 103, 237–245. Federacio´n Nacional de Cafeteros de Colombia, 2000. Centro de Investigaciones de Cafe´ ‘‘CENICAFE’’, Disciplina de agroclimatologı´a, archivos clima´ticos, Chinchina´, Caldas, Colombia. Fenton, M.B., 1997. Science and the conservation of bats. Journal of Mammalogy 78, 1–14. Fenton, M.B., Acharya, L., Audet, D., Hickey, M.B.C., Merriman, C., Obrist, M.K., Syme, D.M., Adkins, B., 1992. Phyllostomid bats (Chiroptera: Phyllostomidae) as indicators of habitat disruption in the neotropics. Biotropica 24, 440–446. Fleming, T.H., 1986. The structure of neotropical bat communities: a preliminary analysis. Revista Chilena de Historia Natural 59, 135– 150.


C. Numa et al. / Biological Conservation 122 (2005) 151–158

Gardiner, W.P., 1997. Statistics for the Biosciences: Data Analysis Using Minitab Software. Prentice-Hall, New Jersey. Gotelli, N.J., Entsminger, G.L., 2003. EcoSim: Null models software for ecology. Version 7. Acquired Intelligence Inc. and Kesey-Bear. Burlington, VT 05465. Available from Accessed on 6/10/03. Greenberg, R., Bichier, P., Cruz Angon, A., Reistma, R., 1997. Bird populations in shade and sun coffee plantations in Central Guatemala. Conservation Biology 11, 448–459. Handley Jr., C.O., 1976. Mammals of the Smithsonian Venezuelan project. Brigham Young University Science Bulletin 20, 1–89. Kalko, E.K.V., Handley Jr., C.O., Handley, D., 1996. Organization, diversity and long-term dynamics of a Neotropical bat Community. In: Cody, M.L., Smallwood, J.A. (Eds.), Long-Term Studies of Vertebrate Communities. Academic Press, San Diego, pp. 504– 555. Krummel, J.R., Gardner, R.H., Sugihara, G., OÕNeill, R.V., Coleman, P.R., 1987. Landscape patterns in a disturbed environment. Oikos 48, 321–324. Kunz, T.H., 1982. Ecology of Bats. Plenum Press, New York. Laurance, W.F., 1991. Edge effects in tropical forest fragments: application of a model for the design of nature reserves. Biological Conservation 57, 205–219. McGarigal, K.J., Marks, B.J.,1995. FRAGSTATS: Spatial pattern analysis program for quantifying landscape structure. General Technical Report PNW351. US Forest Service, Corvallis, Oregon. Medellin, R., Equihua, M., Amin, M.A., 2000. Bat diversity and abundance as indicators of disturbance in Neotropical rainforests. Conservation Biology 14, 1666–1675. Mladenoff, D.J., White, M.A., Pastor, J., Crow, T.R., 1993. Comparing spatial pattern in unaltered old-growth and disturbed forest landscapes. Ecological Applications 3, 294–306. Moguel, P., Toledo, V., 1999. Biodiversity conservation in traditional coffee systems of Mexico. Conservation Biology 13, 11–21. Moreno, C.E., Halffter, G., 2000. Assessing the completeness of bat biodiversity using species accumulation curves. Journal of Applied Ecology 37, 149–158. Moreno, C.E., Halffter, G., 2001. Spatial and temporal analysis of a,b and c diversities of bats in fragmented landscape. Biodiversity and Conservation 10, 367–382. OÕNeill, R.V., Krummel, J.R., Gardner, R.H., Sugihara, G., Jackson, B., DeAngelis, D.L., Milne, B.T., Turner, M.G., Zygmunt, B., Christensen, S., Dale, V.H., Graham, R.L., 1988. Indices of landscape pattern. Landscape Ecology 1, 153–162. Perfecto, I., Rice, R., Greenberg, R., Van der Voort, M., 1996. Shade coffee as refuge of biodiversity. Bioscience 46, 598–608. Perfecto, I., Vandermeer, J., 2002. Quality of agroecological matrix in a tropical montane landscape: ants in coffee plantations in southern Mexico. Conservation Biology 16, 174–182. Rapoport, E.H., 1982. Aerography Geographical Strategies of Species. Pergamon Press, Oxford. Renjifo, L.M., 1999. Composition changes in a subandean avifauna after long-term forest fragmentation. Conservation Biology 13, 1124–1139.

Renjifo, L.M., 2001. Effect of natural and anthropogenic matrices on the abundance of subandean bird species. Ecological Applications 11, 14–31. Rice, R.A., Ward, J., 1997. Coffee, conservation and commerce in the Western Hemisphere. Smithsonian Migratory Bird Center/Natural Resources Defense Council: Washington, DC. Ricketts, T.H., Daily, G.H., Erlich, P.R., Pay, J.P., 2001. Countryside biogeography of moths in a fragmented landscape: biodiversity in native and agricultural habitats. Conservation Biology 15, 378– 388. Roubik, D.W., 1995. Pollination of cultivated plants in the tropics. Food and Agricultural Organization Service Bulletin 118, Food and Agriculture Organization, Rome. Sadeghian, S., Rivera, J.M., Go´mez, M.E., 1998. Impacto de sistemas de ganaderı´a sobre las caracterı´sticas fı´sicas, quı´micas y biolo´gicas de suelos en los Andes de Colombia. FAO e-conference: ‘‘Agroforesterı´a para la produccio´n animal en Latinoame´rica’’. Available from . Schluter, D., Ricklefs, R.E., 1993. Species diversity. An introduction to the problem in species diversity in ecological communities. In: Ricklefs, R.E., Schluter, D. (Eds.), Historical and Geographical Perspectives. The University of Chicago press, Chicago, pp. 1–10. Sekercioglu, C.H., Ehrlich, P.R., Daily, G.C., Aygen, D., Goehring, D., Figeroa-Sandı´, R., 2002. Disappearance of insectivorous birds from tropical forest fragments. Proceedings of the National Academy of Sciences 99, 263–267. Sokal, R.R., Rohlf, F.J., 1995. Biometry, third ed. Freeman, San Francisco. StatSoft, Inc. 1997. STATISTICA for Windows computer program manual. Tulsa, Oklahoma. Vandermeer, J., Perfecto, I., 1997. The agroecosystem: a need for the conservation biologistÕs lens. Conservation Biology 11, 591–592. Verdu´, J.R., Galante, E., 2002. Climatic stress, food availability and human activity as determinants of endemism patterns in the Mediterranean region: the case of dung beetles (Coleoptera, Scarabaeoidea) in the Iberian Peninsula. Diversity and Distributions 8, 259–274. Villarreal, H., 2000. Ana´lisis de fragmentacio´n y uso actual del suelo en la zona cafetera del departamento del Quindı´o, pp. 31–36. Technical Report Number IAvH/130. Instituto de Investigacio´n de Recursos Biolo´gicos Alexander von Humboldt, Bogota´. Whittaker, R.H., 1972. Evolution and measurement of species diversity. Taxon 21, 213–251. Wiens, J.A., 1995. Landscape mosaics and ecological theory. In: Hansson, L., Fahrig, L., Merriam, G. (Eds.), Mosaic Landscapes and Ecological Processes. Chapman & Hall, London, pp. 1–26. Wilson, D.E., Ascorra, C.F., Solari, S., 1996. Bats as indicators of habitat disturbance. In: Wilson, E., Sandoval, A. (Eds.), Manu: The Biodiversity of Southeastern Peru´. Smithsonian Institution, Washington, DC, pp. 613–625. Zorte´a, M., Chiarello, A.G., 1994. Observations on the big fruit-eating bat, Artibeus lituratus, in an urban reserve of south-east Brazil. Mammalia 58, 665–670.