Effects of shade tree removal on birds in coffee agroecosystems in Chiapas, Mexico

Effects of shade tree removal on birds in coffee agroecosystems in Chiapas, Mexico

Agriculture, Ecosystems and Environment 149 (2012) 171–180 Contents lists available at ScienceDirect Agriculture, Ecosystems and Environment journal...

962KB Sizes 0 Downloads 5 Views

Agriculture, Ecosystems and Environment 149 (2012) 171–180

Contents lists available at ScienceDirect

Agriculture, Ecosystems and Environment journal homepage: www.elsevier.com/locate/agee

Effects of shade tree removal on birds in coffee agroecosystems in Chiapas, Mexico Stacy M. Philpott ∗ , Peter Bichier Department of Environmental Sciences, University of Toledo, 2801 W. Bancroft St., MS 604, Toledo, OH 43606, United States

a r t i c l e

i n f o

Article history: Received 30 September 2010 Received in revised form 27 January 2011 Accepted 9 February 2011 Available online 5 March 2011 Keywords: Avian Agriculture Coffea arabica Functional diversity Insectivore Shade management

a b s t r a c t Coffee agroecosystems with complex shade canopies provide refuges for biodiversity, and reductions in complexity cause biodiversity loss. However, no studies directly compare farms before and after a management shift. We surveyed birds before and after a drastic canopy reduction. We compared abundance and richness of all birds, migrants and residents, and bird guilds, and examined impacts on functional group richness for insect-feeding birds and abundance of two key insectivore groups. Finally, we used confidence inference trees to examine which vegetation variables best explained bird abundance and richness for all birds and different groups. We observed 113 bird species from over 7700 individuals. Surprisingly, there were no changes in cumulative bird richness in the cut and uncut areas; however, bird abundance and mean richness was 3–6 times higher in uncut areas. Abundance and richness of all birds, migrants, residents, and individual guilds was higher in shaded areas, as was functional group richness of insectivores and abundance of key insectivore species. Canopy cover and canopy depth best predicted bird abundance and richness. Birds prey on arthropods including coffee pests. Most coffee farmers eliminate shade trees to increase yields; however, management changes that negatively affect insect-feeding birds may indirectly affect the coffee crop. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Agricultural expansion and intensification are two major reasons for decline of biodiversity. However, over the past 15–20 years, the benefits of some agricultural systems as refuges for biodiversity have often been examined (Perfecto et al., 1996). In particular, shaded coffee and cacao agroecosystems have become a major focus for conservation work due to their vegetatively complex shade canopies that provide needed resources in landscapes devoid of natural forests (Perfecto et al., 1996; Moguel and Toledo, 1999). One of the major reasons for the explosion in this field was the documentation that shaded agroforests provide important wintering habitat for migratory birds (Greenberg et al., 1997a). Since the initial work on birds, several have examined the potential for shaded coffee to support arthropods, bats, small mammals, and even amphibians (reviewed in Perfecto et al., 2007). Although the specific details of loss of bird richness described in individual studies vary, overall patterns strongly support the conclusion that coffee farms with diverse, dense, and thick shade canopies support high diversity of birds, and of forest specialists, and that as vegetation

∗ Corresponding author. Tel.: +1 419 530 2578; fax: +1 419 530 4421. E-mail address: [email protected] (S.M. Philpott). 0167-8809/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.agee.2011.02.015

in the canopy is simplified, bird richness is lost (Philpott et al., 2008). In addition, landscape level characteristics of coffee landscapes including nearness of forest fragments and abundance of edge habitats may affect birds (e.g. Tejeda-Cruz and Sutherland, 2004). Declines in bird richness in coffee agroecosystems are correlated with declines in tree richness and density, canopy cover, canopy height and depth, coffee density, lower height of understory plants, removal of epiphytes, increasing distance from forest fragments, and a lack of abundant fruit and nectar resources (Greenberg et al., 1997a; Cruz-Angón and Greenberg, 2005; Philpott et al., 2008; Peters et al., 2010). The effects of management intensification on bird abundance and richness are important from a conservation standpoint, and also because birds provide important ecosystem services to agricultural systems. Birds are important seed dispersers (Sekercioglu, 2006) and also play critical roles as pest control agents within coffee agroecosystems (Greenberg et al., 2000; Perfecto et al., 2004; Borkhataria et al., 2006; Kellermann et al., 2008; Van Bael et al., 2008; Johnson et al., 2009; Philpott et al., 2009). Birds reduce populations of arthropods in both the coffee plants and shade trees (Greenberg et al., 2000; Philpott et al., 2004) and limit outbreaks of potential pests (Perfecto et al., 2004). In addition, birds prey on specific coffee pests including the coffee berry borer, Hypothenemus hampei (Borkhataria et al., 2006; Kellermann et al.,


S.M. Philpott, P. Bichier / Agriculture, Ecosystems and Environment 149 (2012) 171–180

2008; Johnson et al., 2009), and the coffee leaf miner, Leucopotera coffella (Borkhataria et al., 2006). Abundance of birds, or of particular species may be important for this predatory role (Perfecto et al., 2004; Philpott et al., 2009). Likewise, increases in species or functional richness are important to increasing bird effectiveness as predators within agroforestry systems (Van Bael et al., 2008; Philpott et al., 2009). Changes in agricultural management and landscape composition impact bird community composition, and may affect functional richness. Some studies demonstrate significant declines for certain bird guilds with agricultural conversion or intensification. For example, richness of insectivores, frugivores, and nectarivores declined in maize and cacao fields in Sulawesi compared with nearby forests (Waltert et al., 2004). Landscape factors can also influence particular guilds. For example, frugivores are more common in low-intensity agricultural sites near to forest fragments (Luck and Daily, 2003). Larger reviews have summarized that globally, birds found in agricultural habitats are more often generalists, and are disproportionately frugivores and nectarivores (Tscharntke et al., 2008). Large frugivores are lost with conversion from forest to agriculture, but nectarivores, and small to medium sized insectivores and frugivores are sometimes more diverse and common in agricultural areas (Tscharntke et al., 2008). Insectivorous birds, in particular are often negatively affected by forest disturbance, and can be most sensitive to human impacts (Canaday, 1996). Nonetheless, insect-feeding birds are among the most abundant in coffee farms (Komar, 2006). Although loss of species may be reflected in services provided, the loss of functional groups of insectivores may more closely track pest control services, in particular. Broadly defined, a functional group is a grouping of species based on behavioral, morphological, physiological, or resource use traits, and can often better predict ecosystem function, compared with species richness (Philpott et al., 2009). As species richness declines, and as habitat intensification increases, both the number of functional groups and resilience within a community may decline (Fischer et al., 2007). Moreover, in a meta-analysis of several bird data sets (Flynn et al., 2009) documented that functional diversity of birds declined with a change from natural systems to semi-natural or agricultural habitats, and demonstrated that functional diversity actually declined faster than species diversity. Despite the large background literature on bird diversity in coffee agroforests, functional roles of birds in limiting potential and actual coffee pests, and the impacts of land use change on different bird guilds, there is relatively little information available about which specific characteristics of coffee agroecosystems drive changes in functional richness of insectivores within coffee agroecosystems. Furthermore, most studies of bird communities in agroecosystems rely on static differences among farms to examine the effects of the management on birds, rather than following an actual intensification process in action. Here, we take advantage of a large-scale manipulation of the shade tree canopy to examine the changes in the bird community after a large management shift and compare this to data available about the bird community before this management shift. Specifically, we investigated the effects of a dramatic shade tree thinning and pruning on the abundance, species richness, and composition of birds in coffee agroecosystems. We examined the impacts of the management shift on all birds, migrant and resident birds, and different bird guilds (nectarivores, granivores, insectivores, omnivores, and frugivores). We also examined the impacts on functional richness of insect-feeding birds (both insectivores and omnivores) and abundance of potentially effective insectivores. We then used permutation trees to identify those vegetation and site characteristics that most strongly related to abundance and richness of all birds, and for groups of birds potentially important to pestcontrol.

2. Methods 2.1. Study site We worked in an organic, shaded coffee farm in the Soconusco region of Chiapas, Mexico, Finca Irlanda (15◦ 20 N, 90◦ 20 W). Finca Irlanda was first certified organic (and biodynamic) coffee farm in the world (Giovannucci, 2001). For the past 50 years, the farm owners have promoted ecologically friendly farming techniques, and in the last decade received shade-certification from Rainforest Alliance and the Smithsonian’s Bird-Friendly program. There are >50 scientific publications from studies at the farm, thus there is a great deal of background information on the flora and fauna. Finca Irlanda contains ∼270 ha of coffee plus two forest fragments, is located between 950 and 1150 m elevation, and receives ca. 4500 mm of rain annually. Prior to the management shift there were ∼250 shade trees ha−1 that provided between 69 and 90% canopy cover. Trees were spaced approximately 5–8 m apart and were uniformly distributed, except along roads where they were clumped (Vandermeer et al., 2008). There were at least 100 shade tree species on the farm, with a canopy dominated by Inga spp. During April and May of 2007 and 2008, the farm owners carried out a large management shift during which time a large fraction of trees within certain areas of the farm were entirely cut or severely pruned. For example, during 2007, more than 3000 of 11,000 trees within a marked 45-hectare plot in the farm were removed (Perfecto et al., unpublished); many more trees were pruned. The stated goal of the farmers was to reduce the canopy cover from 75% to 50% in affected areas of the farm. We took advantage of the drastic management shift, and ample data on the vegetation and bird community before the management shift to examine the impacts on the bird community. We examined changes in bird richness and abundance, as well as several vegetation characteristics before the management shift (2001–2002) and after the management shift (2007–2009) to examine the direct and immediate effects of shade tree thinning and pruning on the birds. After the management shift, the vegetation in shaded and cut areas was visibly different (Fig. 1). We acknowledge that we deviate from a traditional study design by working in a single farm; however, we did not want to pass up the unique research opportunity provided by the farmer-driven management shift. 2.2. Bird and vegetation surveys We established plots throughout the farm to characterize the vegetation and bird communities. Each plot consisted of a 25 m radius plot, separated from other plots by at least 100 m. We sampled birds during three dry seasons and three wet seasons, but the number of points sampled, and the status of each point (shaded or cut) differed during each season (Table 1). Before the management shift, we established 64 point locations; all were sampled in the wet season, and some of these were sampled during the dry season. After the management shift, the same 132 point locations were sampled during each period; however, the number of shaded and cut points differed due to continued management changes. The 64 locations sampled in 2001–2002 were in the same general area of the farm, and close to 64 of the locations sampled during 2007–2009, but were not exact matches. The additional 68 locations sampled in 2007–2009 were in areas of the farm that were not sampled prior to the management shift. Thus, this study was not a completely controlled before–after experiment, and it is possible there could have been pre-existing differences among areas that were later cut and left uncut For clarity, we refer to points sampled before the management shift as shade before, and to points sampled after the management shift as shade after or cut, depending on their shade condition.

S.M. Philpott, P. Bichier / Agriculture, Ecosystems and Environment 149 (2012) 171–180


trees were given a unique morphospecies name. After the management shift, we also collected data on coffee height, and ground cover provided by dead vegetation, herbs and leaf litter, and bare ground. We estimated mean coffee height per plot. Ground cover (e.g. dead vegetation, bare ground, leaf litter, and herbaceous vegetation) in 25 m radius circles was estimated with visual estimates at 5% intervals. We created a vegetation complexity index (VCI) to summarize farm management strategy (Philpott et al., 2009). The VCI is an index that can be used to assess overall vegetation complexity of a site by equally weighting and taking mean values across several vegetation variables (Mas and Dietsch, 2003; Philpott et al., 2009). For the VCI computed here, we examined values for canopy cover, epiphyte score, mean tree height, number of trees, tree richness, coffee cover, bare ground cover, and percent of trees in the genus Inga for each plot. Data values for variables measured during each field season were converted from a scale of 0 (representing simple vegetation) to 1 (representing complex vegetation). To convert values of most variables (canopy cover, epiphyte score, tree height, number of trees, tree richness) to a scale from 0 to 1, we divided values recorded for a given plot by the highest recorded value for that variable in any plot. For coffee cover, bare ground cover, and percent of trees in the genus Inga (all variables associated with higher management intensity and less complex vegetation) we divided values by the highest recorded value in any plot and subtracted the product from one. We then took the mean of all eight converted variables to yield the VCI. We surveyed birds by sight and sound with 10-min point counts in each of the plots (Hutto et al., 1986; Petit et al., 1994). We sampled plots during both the wet and dry seasons in order to compare the effects on all birds, migrants, and tropical resident birds.

Fig. 1. Photographs of the study site, Finca Irlanda, showing cut (a) and shaded (b) areas of the farm in February 2008, after the management shift.

We collected vegetation data for all plots and sample periods except for the wet season of 2002 and for one forgotten plot in the dry season of 2009 (Table 1). Both before and after the management shift, we collected data on canopy cover, tree species richness, tree height, tree density, percent coffee cover and epiphyte density. We quantified epiphyte abundance within plots on a scale from 0 to 4 of percent of tree trunk surface area covered with epiphytes (0 = <1%, 1 = 1–25%, 2 = 26–50%, 3 = 51–75%, 4 = > 75% covered). Coffee cover was estimated as the fraction of the plot covered by coffee foliage. At five points per plot (at the circle center and 10 m to N, S, E, and W), we measured canopy cover with a convex spherical densitometer. We estimated tree heights (maximum, minimum, and mean), and sampled vegetation >18 m with a range finder. Canopy depth was calculated by subtracting minimum from maximum tree height for each plot. Tree identifications were made in the field and unknown Table 1 Bird and vegetation sample points in a coffee agroecosystem before and after a management shift. Sample date




December-01 June-02 July-07 July-07 February-08 February-08 July-08 July-08 February-09 February-09

Dry Wet Wet Wet Dry Dry Wet Wet Dry Dry

Before Before After After After After After After After After

Shade Shade Shade Cut Shade Cut Shade Cut Shade Cut

Total points

No. bird points

No. vegetation points

48 64 66 66 66 66 53 79 48 84

48 0 66 66 66 66 53 79 48 83



2.3. Data analysis We first compared vegetation characteristics in shade before, shade after, and cut areas. First, we calculated mean values for tree height and for canopy cover for each plot. We compared mean values for variables measured during each sample period (canopy cover, coffee cover, bare ground, epiphytes, tree height, tree number, tree richness, percent Inga) and for ground cover variables (dead vegetation, leaf litter, herbaceous vegetation) with two separate multivariate analyses of variance (MANOVA) followed by ANOVA and Tukey’s test to compare individual factors in shade before, shade after, and cut areas. We examined differences in coffee height and the vegetation complexity index (VCI) with ANVOA. Values for canopy cover, coffee cover, percent Inga, and ground cover variables were arcsine square root transformed, and coffee height, tree height, number of trees, and tree richness were natural log (+1) transformed to meet conditions of normality. We characterized the bird community according to migratory status (migrant and resident) and feeding guild (frugivore, granivore, insectivore, nectarivore, and omnivore) using standard sources (Ehrlich et al., 1988; Stiles and Skutch, 1990; Stotz et al., 1996). To compare bird richness in shade before, shade after, and cut areas, we generated sample-based rarefaction curves, scaled to the number of individuals (MaoTao estimates) with EstimateS 8.0 (Colwell, 2005). We statistically compared cumulative, rarefied richness in different areas for all birds, migrant and resident birds, and for birds in different feeding guilds by comparing the overlap in 95% confidence intervals. We also examined the mean species richness and abundance of all birds in individual points with analysis of variance (ANOVA) and for mean richness and abundance of migrants and residents with multivariate ANOVA. We followed MANOVAs with individual ANOVAs and Tukey’s tests to examine differences between specific groups and management areas. Data for bird abundance and richness were natural log (+1) transformed to meet conditions of normality. Repeated measures ANOVA was


S.M. Philpott, P. Bichier / Agriculture, Ecosystems and Environment 149 (2012) 171–180

not used for analysis because point locations before and after the management shift were not complete matches. We classified all insectivores and omnivores into functional groups based on body size (tiny, small, medium, large, and extra large), foraging strata (understory, canopy, both), diet (strict insectivores, insectivores, omnivores), and foraging strategy (foliage gleaner, bark gleaner, ground gleaner, hover and glean, hawks, and multiple strategies). Insectivores are those birds that entirely or feed mainly on insects, and were described in the texts as birds that (a) obtain less than 25–30% of their diet from non-insect items, (b) feed mostly or mainly insects, or (c) feed on insects and few berries or seeds; all other sometime insectivorous birds were classified as omnivores (Philpott et al., 2009). All other categories were classified exactly as in Philpott et al. (2009), and where birds were not included in that database, we went to the original sources to complement the data. We summed the number of functional groups present per point, and compared mean values across management types with ANOVA. We examined the abundance of three bird species and functional groups previously highlighted as important for pest control services (Philpott et al., 2009). We compared the abundance of small, strict insectivore, understory foraging, foliage gleaning species, and of tiny, omnivore, both strata, foliage gleaning birds, and of the Tennessee Warbler in shade before, shade after and cut areas with ANOVA. We used confidence inference trees to examine the relative importance of different vegetation and site variables in predicting the variation in bird richness and abundance (Strasser and Weber, 1999; Hothorn et al., 2006; Jha and Vandermeer, 2010). Permutation trees, such as confidence inference trees, are often used to examine ecological patterns, and to do so, split data for response variables into homogeneous groups using explanatory variables (De’ath and Fabricius, 2000; Olden et al., 2008). Generally, permutation trees are advantageous over multivariate regression because they have the ability to handle missing data, are easy to interpret, and allow explaining variation in a single response variable by one or more explanatory variables that may interact in a hierarchical fashion (De’ath and Fabricius, 2000; Olden et al., 2008). We created confidence inference trees using a binary recursive data-partitioning algorithm available in the package “party” in R (R Development Core Team, 2008). In contrast to other packages (e.g. “rpart” and “randomforest”) that are somewhat biased towards selecting variables with multiple splitting points, confidence inference trees require significant differences in the data to partition it, thereby lessening variable selection bias (Hothorn et al., 2006; Strobl et al., 2009; Thompson and Spies, 2009). Additionally, confidence inference trees allow the user to set the minimum p-value necessary for data splitting, thereby eliminating the problems associated with over fitting and tree pruning (Thompson and Spies, 2009). The ‘party’ algorithm first examines whether predictor variables are independent of each other and of the response variable. The package then selects the predictor variable with the strongest relationship to the response variable, and assigns the relationship a p-value. The data are then split into two groups of data or nodes that each subsequently compared to the predictor variables. The program will continue to split or partition the data into nodes based on significant relationships between the predictor and response variables significant to the assigned p-value. We included the following factors as potential explanatory variables (canopy cover, coffee cover, coffee height, number of tree individuals, number of tree species, canopy depth, percent of trees in the genus Inga, and percent ground covered by dead vegetation, leaf litter, and herbaceous vegetation). As response variables, we included bird richness and abundance, richness and abundance of omnivores and insectivores, functional richness of birds, and abundance of bird groups important to predatory function. We included data for individual points as replicates, and used total bird richness

and abundance per point for analysis. All variables included were transformed to meet conditions of normality as explained for vegetation and bird analyses. We used univariate tests, and because of the high number of tests performed, we selected a conservative critical value (p < 0.001) to reduce Type I error (Thompson and Spies, 2009). Although we could have investigated the relationships between the site characteristics and other bird groups, we limited the analysis to all birds, and for those birds associated with pestcontrol services within the coffee agroecosystem. Data from the wet season of 2002 were not included, as vegetation data were not taken during that sample period. 3. Results 3.1. Vegetation As expected, there were many significant changes in the vegetation of the farm before and after the management shift (Table 2). There were several differences in vegetation characteristics of both ground cover (MANOVA, F4, 522 = 31.25, p < 0.001) and for other vegetation factors measured (MANOVA, F18, 1130 = 80.03, p < 0.001). There was less than half as much canopy cover, many fewer epiphytes, twenty percent fewer trees, and fewer tree species in cut areas than in shade after area (Table 2). There was more dead vegetation but less leaf litter in the cut areas compared with shade after areas (Table 2). Coffee plants were on average shorter and coffee densities were lower in cut than in shade after areas, but it is unlikely that these changes were brought about by the management shift. Thus, the management shift resulted in lower vegetation complexity (measured as the VCI) in the cut area than in shade after area (Table 2). Interestingly, vegetation complexity in the cut area did not differ from vegetation in 2001–2001 (e.g. shade before). Between 2001 and 2007 (prior to cutting), farm vegetation grew more complex resulting in higher canopy cover, taller trees, more tree species, less dominance by Inga spp. trees, and less bare ground in shade after than in shade before (Table 2). The only major decrease in vegetation complexity in shade after (compared with shade before) was a dramatic loss of epiphytes (Table 2). Thus, vegetation in the farm before the management shift was only slightly more complex than in cut areas after the management shift. 3.2. Bird abundance and bird richness Overall, 7718 bird individuals from 113 species were seen or heard across all sample periods. We observed 67 bird species (1022 individuals) sampling before the management shift, and observed 101 species (4821 individuals) in shade after and 82 species (1875 individuals) in cut areas. Before the management shift, the most abundant bird species were the Tennessee Warbler (Vermivora peregrina) (206 or 20.2% of individuals), the Red-legged Honeycreeper (Cyanerpes cyaneus) (124 or 12.1%), and the Yellow-green Vireo (Vireo flavoviridis) (78 or 7.6%). The most common species in the shade after areas were the Red-legged Honeycreeper (1194 or 24.7%), the Clay-colored Thrush (Turdus grayi) (433 or 8.9%), and the Yellow-winged Tanager (Thraupis abbas) (183 or 3.8%) and the most common species in the cut areas were the Red-legged Honeycreeper (369 or 19.6%), the Rufous-Capped Warbler (Basileuterus rufifrons) (155 of 8.3%) and the Clay-colored Thrush (136 or 7.2%). There were no differences in cumulative richness before or after the management shift, but abundance and mean richness of birds and most bird groups per plot was higher in shaded than in cut areas. Based on accumulation curves generated with EstimateS and associated 95% confidence intervals, cumulative, rarefied species richness did not differ for all birds, for migratory birds, or for residents (Table 3). Similarly, there were no differences between

S.M. Philpott, P. Bichier / Agriculture, Ecosystems and Environment 149 (2012) 171–180


Table 2 Vegetation characteristics of the coffee farm before the management shift, and shaded and cut areas after the management shift.a

Canopy cover (%) Coffee cover (%) Epiphyte rank Tree height (m) Canopy depth (m) No. trees No. tree species Inga spp. trees (%) VCIb Ground cover Dead vegetation (%) Bare ground (%) Leaf litter (%) Herbaceous veg. (%) Coffee height (m)

Shade before

Shade after




36.67 ± 1.96b 73.62 ± 2.85a 0.79 ± 0.06a 8.23 ± 0.17b 9.82 ± 0.51c 31.56 ± 1.07c 8.27 ± 0.34c 53.24 ± 2.24c 0.37 ± 0.01b

43.44 71.94 0.42 9.51 18.0 50.64 15.27 67.5 0.48

± ± ± ± ± ± ± ± ±

0.64a 0.79a,b 0.33b 0.11a 0.33a 0.89a 0.27a 0.87b 0.01a

Cut 13.47 67.19 0.03 6.85 12.36 39.03 12.11 73.5 0.37

± ± ± ± ± ± ± ± ±

0.54c 0.85b 0.01c 0.07c 0.23b 0.61b 0.18b 0.75a 0.01b

2, 572 2, 572 2, 572 2, 572

<0.001 <0.001 <0.001 <0.001

2, 572 2, 572 2, 572 2, 572

563.452 10.286 76.389 238.339 155.343 99.311 116.888 21.244 306.475

na 14.23 ± 3.79a na na na

6.81 4.79 29.41 59.29 2.13

± ± ± ± ±

0.55b 0.40b 1.34a 1.78 0.02

11.16 9.1 21.33 58.24 1.96

± ± ± ± ±

0.32a 0.51a 0.79b 1.49 0.01

1, 525 2, 572 1, 525 1, 525 1, 525

26.22 19.652 23.44 0.442 62.813

<0.001 <0.001 <0.001 0.507 <0.001

<0.001 <0.001 0.001 <0.001

a Statistical results are from ANOVA followed by Tukey’s tests to compare differences between treatments. Small letters show differences (p < 0.05) between treatments for individual vegetation characteristics. b Vegetation complexity index where 1 shows complex vegetation and 0 shows simple vegetation.

the cumulative, rarefied richness of birds during the wet and dry seasons (Table 3). However, at the plot level, there were very significant differences in bird abundance and richness (Fig. 2). Bird abundance was more than twice as high in shaded after (20.7 + 0.7 individuals per plot) compared with shade before (9.1 + 0.5) and

cut areas (6.3 + 0.3) (F2, 638 = 267.8, p < 0.001; Tukey’s test for each pair, p < 0.001). Likewise, bird richness was twice as high in shade areas after (9.1 + 0.3 species per plot) than in shaded areas before (4.9 + 0.2) or in cut areas (3.9 + 0.1) (F2, 638 = 206.9, p < 0.001; Tukey’s test for each pair, p < 0.01). Both migrant and resident birds were

Fig. 2. Mean abundance (a and b) and species richness (c and d) per point for birds sorted by migratory status (a and c) and feeding guild (b and d) in a shaded coffee agroecosystem before and shaded and cut areas after the management shift. Letters show significant differences (p < 0.05) for particular factors within a given bird group.


S.M. Philpott, P. Bichier / Agriculture, Ecosystems and Environment 149 (2012) 171–180

Fig. 3. Mean insectivore functional richness per plot (a) and abundance of functionally important functional bird groups and bird species (b) in a shaded coffee agroecosystem before and shaded and cut areas after the management shift. Letters show significant differences (p < 0.05) within a given bird group.

much more abundant (MANOVA, F2, 4 = 93.2, p < 0.001) and species rich (F2, 4 = 76.8, p < 0.001) in shade after area compared with the cut area. There were two times more resident species and three times more resident individuals in shade after than in cut areas (Fig. 2a and c), and abundance and richness of migrants was twice as high in shade after than in cut areas (Fig. 2a and c).

after compared with cut areas (Fig. 2b). Insectivore and nectarivore abundance was higher in shade before compared with cut areas; abundance of frugivores, granivores, and omnivores was similar in shaded areas before and cut areas (Fig. 2b). 3.4. Functional richness and abundance of functionally important bird species and groups

3.3. Abundance and richness of different guilds According to species accumulation curves, and cumulative, rarefied richness values, there were no differences in species richness of any guilds across the habitats (Table 3). Richness for insectivores and granivores tended to be higher in shaded areas after the management shift than shade before or in cut areas and richness for nectarivores was slightly higher in cut areas than in the shaded areas before or after. But these differences were not significant. At the plot level, however, there were large differences in richness of each bird guild examined (MANOVA, F5, 10 = 36.5, p < 0.001). Mean richness of all groups (frugivores, granivores, insectivores, nectarivores, and omnivores) was roughly twice as high in one or more shaded habitat compared with the cut area (F2, 638 > 4.5, p < 0.001) (Fig. 2d). Similarly, bird abundance for different guilds depended on management area (MANOVA, F5, 10 = 44.8, p < 0.001). There were large differences in frugivore, granivore, insectivore, nectarivore, and omnivore abundance per plot (F2, 638 > 7.5, p < 0.001). For all groups, abundance was two to three times greater in shaded areas

Table 3 Rarefied bird species richness for all birds and for different bird groups before and after the management shift.a Shade before All birds Dry season Wet season Migrants Residents Insectivores Omnivores Granivores Nectarivores Frugivores

67.0 49.0 43.0 18.0 49.0 31.0 17.0 4.0 8.0 6.0

± ± ± ± ± ± ± ± ± ±

6.1 6.1 6.5 2.7 5.6 3.5 2.0 0 3.2 3.6

Shade after 70.6 56.1 41.6 24.0 46.5 38.5 15.3 4.5 4.9 6.7

± ± ± ± ± ± ± ± ± ±

10.1 9.0 8.4 5.6 8.5 7.1 4.4 1.4 2.3 3.3

Cut 71.1 53.8 42.2 24.9 47.5 34.9 16.8 4.6 7.3 6.9

± ± ± ± ± ± ± ± ± ±

7.7 5.1 7.5 4.2 6.7 4.7 3.3 0.5 3.7 1.8

a Numbers show cumulative species richness ± 95% confidence intervals (CI) for richness rarefied to the number of individuals in the area with the lowest abundance. There were no differences in rarefied richness between sites according to 95% CI.

The change in management had strong impacts on both the number of insectivore and omnivore functional groups present at the plot level and on the abundance of bird species potentially important to predatory function. Rarefied functional group richness was not different among habitat types with 36 (±4.4, 95% CI) groups in shade before, 39.3 (±4.6) groups in shade after and 37.2 (±4.3) groups in cut. However, there were more than twice as many functional groups per plot in the shaded areas after than in the shade before or cut areas, and significantly more functional groups in the shade before than in the cut area (Fig. 3a, F2, 637 = 163.7, p < 0.001). The two important functional groups were each represented by one or two species: the small, strict insectivore, understory forager, foliage gleaning species was the Spot-Breasted Wren (Thryothorus maculipectus) and the tiny, omnivore, both strata, foliage gleaner species were the Northern Beardless-Tyrannulet (Camptostoma imberbe) and the Tennessee Warbler. Spot-Breasted Wren abundance was three times as high in the shade after than in the cut, but there were no differences in the shade before and cut areas (Fig. 3b, F2, 637 = 9.8, p < 0.001). The tiny, omnivore, both strata, foliage gleaners, in contrast, were eight times more abundant before the management shift than in the cut areas, and twice as abundant in the shade after than in the cut (Fig. 3b, F2, 637 = 43.7, p < 0.001). Similarly, the Tennessee Warbler was much more abundant in the shade before with more than five times as many individuals than in the shade after or cut areas (Fig. 3b, F2, 637 = 53.2, p < 0.001). 3.5. Vegetation drivers of richness and abundance patterns Several vegetation and site characteristics were predictive of bird abundance and richness and abundance and richness of certain groups of birds. Higher canopy cover and canopy depth were predictive of higher bird abundance and richness, and higher canopies were also associated with higher bird abundance (Fig. 4a and b). Back transformed values show cutoff points for canopy cover around (26%), for canopy depth near (15 m) and for mean tree

S.M. Philpott, P. Bichier / Agriculture, Ecosystems and Environment 149 (2012) 171–180


4. Discussion

Fig. 4. Confidence inference tree showing the vegetation variables that significantly predicted bird abundance (LN + 1) (a) and bird species richness (LN + 1) (b) in a coffee agroecosystem.

height at 7.8 m for predicting the highest levels of bird abundance, and 26% cover and 15 m and then 22.9 m for canopy depth for highest bird richness. Insectivore abundance was best predicted by high canopy cover and high cover provided by leaf litter and dead vegetation (Fig. 5a), whereas insectivore richness was predicted only by canopy depth (Fig. 5b). Canopy cover and canopy depth were the two factors predictive of omnivore richness and abundance (Fig. 5c and d). Insectivore abundance was highest with above 26% canopy cover, 15% leaf litter cover, and 15% dead vegetation cover, and insectivore richness was higher with greater than 13.8 m canopy depth and highest over 22.5 m. Omnivore abundance and richness was highest over 34% canopy cover and 14.7 m canopy depth. The functional richness of insect-feeding birds was best explained by canopy depth and canopy cover. Functional richness was highest where the canopy exceeded 14.7 m thick, and over 34% canopy cover. The abundance of the tiny, omnivorous, both understory and canopy foraging foliage gleaners (Tennessee Warbler and Northern Beardless-Tyrannulet) was high under two combinations of values: either with canopy cover over 18% and tree richness under 10 species, or for high canopy cover, richness over 10 species, and dead vegetation cover great than 8% (Fig. 6b). Finally, the number of small, strict insectivore, understory foraging foliage gleaners was predicted only by canopy depth and was higher where canopies were more than 18 m thick (p < 0.001, low cover group, n = 471, high cover group, n = 104).

We found dramatic plot-level decreases in the richness and abundance of birds of different migratory status and guild in this study, but no differences in cumulative richness across the farm. This was somewhat surprising given the large number of studies that have documented declines in bird species richness with management intensification (reviewed in Komar, 2006; Philpott et al., 2008). Yet, many studies have found only small differences in cumulative richness of birds in coffee habitats varying in canopy characteristics (Greenberg et al., 1997b; Cruz-Angón and Greenberg, 2005) or have measured bird richness with speciessample curves (Thiollay, 1995; Tejeda-Cruz and Sutherland, 2004) or mean estimated richness (Waltert et al., 2004, 2005; Perfecto et al., 2003) making the results somewhat difficult to directly compare. Our results finding higher bird abundance in habitats with more complex shade were largely consistent with previous studies (Greenberg et al., 1997b; Johnson, 2000; Cruz-Angón and Greenberg, 2005; Philpott et al., 2008). Further, many of the same factors (e.g. canopy cover, canopy depth, tree richness) correlated with bird richness and abundance in our data set also predicted bird richness in other regions and studies (e.g. Parrish and Petit, 1996; Greenberg et al., 1997b; Johnson, 2000; Philpott et al., 2008). Other factors, such as an Inga-dominated canopy or the changes in epiphyte presence or abundance, were not significant predictors of bird abundance and richness for our data, even though these factors influence all birds and certain bird guilds elsewhere (Greenberg et al., 1997b; Cruz-Angón and Greenberg, 2005; but see Jones et al., 2002). It is noteworthy, however, that epiphytes are much less abundant in the Soconusco region of Chiapas than in other areas of Mexico and that in addition to trees, epiphytes provide canopy cover within agroforests (Cruz-Angón et al., 2008). Although the epiphyte measure in this study did not influence birds specifically, part of the influences of canopy cover may be attributable to changes in cover provided by epiphytes. One important caveat to our data is that we only examined the effects of such a management shift in one coffee farm, thus there is no spatial replication. Several studies have also examined the impacts of coffee management on different guilds of birds. Most studies focus on abundance of different guilds, or their relative abundance, and not species richness within each guild. Generally, there are more omnivores, frugivores, nectarivores, and fewer insectivores in coffee habitats compared with nearby forests (Komar, 2006). But results differ somewhat among studies and agroforests examined. For example, in one study, abundance of both insectivores and frugivores was lower in shaded monocultures than in rustic coffee and forested habitats, but did not differ among the latter (Tejeda-Cruz and Sutherland, 2004). In contrast to our results, they also observed increases in abundance of granivores and omnivores with increasing disturbance. Their study sites, however, were arranged along an elevation gradient with degree of disturbance negatively correlated with elevation, and our sites were all within a very similar elevation range. More generally, agroforests in Sumatra harbor fewer frugivores, specialists, and large insectivores, and more omnivores, nectarivores, and granivores compared with native forests (Thiollay, 1995). In agricultural landscapes near the Guinea–Congolian rainforest, frugivorous and omnivorous bird species richness did not differ between annual crop, agroforest, and forest habitats, granivore and nectarivore richness was higher in agricultural land-use types, and insectivore richness declined in agroecosystems compared with forests (Waltert et al., 2005). In contrast, we found consistent differences in abundance and species richness of all bird guilds with lower numbers in the cut areas than in the shaded areas. Thus many of the specific details of how guild structure, richness, and abundance changes may depend on


S.M. Philpott, P. Bichier / Agriculture, Ecosystems and Environment 149 (2012) 171–180

Fig. 5. Confidence inference tree showing vegetation variables that significantly predicted insectivore abundance (LN + 1) (a), insectivore species richness (LN + 1) (b), omnivore abundance (LN + 1) (c), and omnivore species richness (LN + 1) (d).

structural factors of the agroecosystem, or the landscape context in which the agroforests are embedded. Certainly other local and regional characteristics of coffee landscapes that we did not measure may affect bird richness and abundance. At the local level, high fruit abundance within coffee farms may increase frugivore and omnivore abundance, and also increase the foraging activity of fruit-feeding birds (Carlo et al., 2004; Peters et al., 2010). Predation risk might be one reason that birds are more abundant in more vegetatively complex agroforests with a higher density of trees and coffee plants (Johnson, 2000). Furthermore, reduced mean tree height, presence of certain microhabitats, lower variety of food resources, heavy hunting pressure, and competition by other birds or mammals may be important to birds in agroforests (Thiollay, 1995). At the landscape level, distance to nearest large or small fragments, habitat connectivity, amount of forest in the landscape, and thickness of edges may affect birds. Specifically within agricultural landscapes, species richness of birds increases with nearness to forests (Estrada et al., 1997), or is maintained near to forest fragments or with thick habitat edges (Hughes et al., 2002). Thus landscape context can strongly mediate local level responses such that effects of a manipulation at the farm scale might be more (or less) severe in a less forested landscape. Habitat connectivity may be important for increasing dispersal distances through the agricultural matrix (Castellón and Sieving, 2006). However, habitat connectivity may not always be important for bird richness in coffee landscapes, especially where most agroforests are relatively well connected (Jones et al., 2002). One important observation that was not explicitly examined in our study was that for the same period during which Finca Irlanda intensified their production, several nearby farms underwent intensification processes in certain areas (Pers. obsv.). Thus the overall matrix quality

may have decreased with potentially important impacts for bird foraging and survival in the region. In the Amazon forest fragments project, bird richness declined with time since fragmentation, and more so in small fragments; however, during the first few years after fragmentation, a larger number of bird species were seen even in small fragments, likely due to a temporary influx of birds from recently disturbed areas (Ferraz et al., 2003). As the entire valley was experiencing coffee intensification the remaining shaded areas of Finca Irlanda, a less-disturbed habitat, may have been similarly experiencing higher bird abundance. Some data from other studies support this hypothesis. For example, in Hispaniola, a coffee farm of relatively poor vegetation quality supported a relatively high abundance and richness of birds because it was surrounded by a poor-quality matrix (Wunderle and Latta, 1998). Likewise, in Puerto Rico, where coffee landscapes are somewhat dominated by abandoned plantations and second-growth forests, bird abundance in coffee farms, some rich in species of fruiting trees, was low. Thus patterns of bird abundance at the farm level may be better explained by landscape effects than by the local level vegetation characteristics of individual plots. A particularly novel aspect of this study was the analysis of the effects of a canopy management shift on functional groups of insectivores, and on specific bird species or groups likely important for insectivory. Further, we documented which aspects of the canopy likely influence functional groups richness. Very few studies have examined factors that influence abundance and richness of insectivores in coffee agroecosystems, or which vegetation characteristics correlate with increases in functional richness, or abundance of important insectivores. Foliage-gleaners are generally positively correlated with abundance of arthropods and increases in shade tree richness and density of crop structure (Johnson, 2000), two

S.M. Philpott, P. Bichier / Agriculture, Ecosystems and Environment 149 (2012) 171–180

Fig. 6. Confidence inference tree showing vegetation variables that significantly predicted number of insectivore and omnivore functional groups (LN + 1) (a) and the abundance of the tiny, omnivorous, both canopy and understory foraging foliage gleaners (LN + 1) (b). The latter is one of two functional groups singled out for a potential sampling effect with respect to arthropod removal.

factors that may enhance arthropod richness and reduce predation risk. Diversity and abundance of important predators of coffee pests may decline with increasing distance to forest patches (Kellermann et al., 2008), and such loses of diversity are correlated with less effective arthropod removal (Van Bael et al., 2008). Clearly, individual species can vary widely in their contribution to ecosystem services such that the loss of only a single or few species can disproportionately affect the system and services derived from it. This so-called sampling effect states that the mechanism driving diversity-ecosystem function impacts is via presence of such single important species (Hooper et al., 2005). In coffee agroecosystems, tiny, omnivorous, both strata, foliage gleaners, represented most notably here by the Tennessee Warbler, have been singled out as important abundant migratory insectivores whose abundance is correlated with higher arthropod removal (Greenberg et al., 2000; Philpott et al., 2009). Analyses of stomach contents from 42 Tennessee Warblers captured in the coffee layer of Finca Irlanda indicate these birds are indeed frequently preying on arthropods


(including herbivores such as lepidopteran larvae and seed predators like scolytids and curculionids) (T. Dietsch, unpublished data). Furthermore, foraging data from the same farm show Tennessee Warblers forage in the coffee layer at least 10% of the time, and given their abundance thus could have important impacts on coffee arthropods (T. Dietsch, unpublished data). The abundance of individuals in this functional group dropped strongly after the management shift (in both cut and shade after) where other insectivores were not largely affected. Thus understanding the specific factors resulting in large changes of Tennessee Warbler abundance may be particularly crucial for understanding provisioning of pest control services in coffee agroecosystems. The vegetation variables that best predicted its abundance were higher canopy cover, lower tree species richness, and more ground covered by dead vegetation. Other studies, however, have found that Tennessee Warblers are more abundant in Inga-dominated farms compared with farms dominated by other canopy trees (e.g. Gliricidia), perhaps due to a higher abundance of fruit or nectar resources, or potentially an increase in insects associated with fruits and flowers (Greenberg et al., 1997a). Here, relative abundance of Inga spp. trees did not predict the abundance of Tennessee Warblers, however other species of trees were likely flowering and fruiting within the plantations. We did not measure reproductive phenology of trees within the farms, nor what flowering and fruiting resources were available during surveys. What seems more likely is differences in Tennessee Warbler abundance may reflect differences in tree reproductive phenology during those time periods. Observations before the management shift were conducted in December, and observations during the dry season after the shift were conducted in February. It is possible, and even likely that a greater number of trees are flowering early in the dry season, thereby dramatically increasing Tennessee Warbler abundance. In sum, we did not observe impacts of the management shift on the cumulative species richness of birds, of migrants, residents, or birds of different feeding guilds, nor were there differences in functional richness of insectivores across larger scales. However, we did find highly significant decreases in abundance and richness of all groups of birds, including of functional richness of insectfeeding birds, and of some birds likely important for pest control functions at the plot level. Thus, our results show that bird species density was uniformly higher across the shaded areas of the farm; per plot richness was greater, but cumulative richness did not differ in shaded and cut areas of the farm. Although there are a dramatic number of new studies examining relationships between predator diversity and ecosystem services, few have examined the importance of distribution of predators at the local or more regional scales. The changes in abundance of insectivores and omnivores with management shifts at the local scale will likely be important for pest control services, as will changes in the number of functional groups observed with this drastic change in coffee management. Farmers generally intensify production to increase yields, and this was certainly the case for the owners of Finca Irlanda (Pers. comm.). However the decreases in insect-feeding birds within the farm may likely interact in a complex manner with other changes resulting from the management shift.

Acknowledgements We thank W. Peters and Finca Irlanda for allowing us to conduct this research on their farm. G. Lopez Bautista, B. Chilel, G. Dominguez, and A. De la Mora provided field assistance. G. Ibarra˜ and El Colegio de la Frontera Sur provided logical support. Nunez S. Jha, D. Jackson, and D. Allen helped with the R code. K. Ennis, R. Friedrich, D. Gonthier, L. Moorhead, C. Murnen, G. Pardee, I. Perfecto, J. Vandermeer, and Z. Liu contributed to discussions about the


S.M. Philpott, P. Bichier / Agriculture, Ecosystems and Environment 149 (2012) 171–180

manuscript. The Organic Farming Research Foundation, the University of Toledo, and NSF grant DEB-0349388 I. Perfecto and J. Vandermeer provided funding for this research. References Borkhataria, R., Collazo, J., Groom, M., 2006. Additive effects of vertebrate predators on insects in a Puerto Rican coffee plantation. Ecological Applications 16, 696–703. Canaday, C., 1996. Loss of insectivorous birds along a gradient of human impact in Amazonia. Biological Conservation 77, 63–77. Carlo, T., Collazo, J., Groom, M., 2004. Influences of fruit diversity and abundance on bird use of two shaded coffee plantations. Biotropica 36, 602–614. Castellón, T.D., Sieving, K.E., 2006. An experimental test of matrix permeability and corridor use by an endemic understory bird. Conservation Biology 20, 135–145. Colwell, R.K., 2005. EstimateS: Statistical Estimation of Species Richness and Shared Species from Samples. Version 7.5. User’s Guide and Application Published at: http://purl.oclc.org/estimates. Cruz-Angón, A., Greenberg, R., 2005. Are epiphytes important for birds in coffee plantations? An experimental assessment. Journal of Applied Ecology 42, 150–159. Cruz-Angón, A., Sillett, T., Greenberg, R., 2008. An experimental study of habitat selection by birds in a coffee plantation. Ecology 89, 921–927. De’ath, G., Fabricius, K., 2000. Classification and regression trees: a powerful yet simple technique for ecological data analysis. Ecology 81, 3178–3192. Ehrlich, P., Dobkin, D., Wheye, D., 1988. The Birder’s Handbook: A Field Guide to the Natural History of North American Birds: Including All Species that Regularly Breed North of Mexico. Simon and Schuster/Fireside Books, New York, New York, USA. Estrada, A., CoatesEstrada, R., Meritt, D.A., 1997. Anthropogenic landscape changes and avian diversity at Los Tuxtlas, Mexico. Biodiversity and Conservation 6, 19–43. Ferraz, G., Russell, G., Stouffer, P., Bierregaard, R., Pimm, S., Lovejoy, T., 2003. Rates of species loss from Amazonian forest fragments. Proceedings of the National Academy of Sciences of the United States of America 100, 14069–14073. Fischer, J., Lindenmayer, D., Blomberg, S., Montague-Drake, R., Felton, A., Stein, J.A., 2007. Functional richness and relative resilience of bird communities in regions with different land use intensities. Ecosystems 10, 964–974. Flynn, D., Gogol-Prokurat, M., Nogeire, T., Molinari, N., Trautman Richers, B., Lin, B.B., Simpson, N., Mayfield, M.M., DeClerck, F., 2009. Loss of functional diversity under land use intensification across multiple taxa. Ecology Letters 12, 22–33. Giovannucci, D., 2001. Sustainable coffee survey of the North American specialty coffee industry. A Report conducted for the Summit Foundation, The Nature Conservancy, North American Commission for Environmental Cooperation, Specialty Coffee Association of America, and The World Bank. Greenberg, R., Bichier, P., Angón, A., MacVean, C., Perez, R., Cano, E., 2000. The impact of avian insectivory on arthropods and leaf damage in some Guatemalan coffee plantations. Ecology 81, 1750–1755. Greenberg, R., Bichier, P., Angón, A.C., Reitsma, R., 1997a. Bird populations in shade and sun coffee plantations in central Guatemala. Conservation Biology 11, 448–459. Greenberg, R., Bichier, P., Sterling, J., 1997b. Bird populations in rustic and planted shade coffee plantations of eastern Chiapas, Mexico. Biotropica 29, 501–514. Hooper, D.U., Chapin, F.S., Ewel, J.J., Hector, A., Inchausti, P., Lavorel, S., Lawton, J.H., Lodge, D.M., Loreau, M., Naeem, S., Schmid, B., Setala, H., Symstad, A.J., Vandermeer, J., Wardle, D.A., 2005. Effects of biodiversity on ecosystem functioning: a consensus of current knowledge. Ecological Monographs 75, 3–35. Hothorn, T., Hornik, K., Zeileis, A., 2006. Unbiased recursive partitioning. Journal of Computational and Graphical Statistics 15, 651–674. Hughes, J.B., Daily, G.C., Ehrlich, P.R., 2002. Conservation of tropical forest birds in countryside habitats. Ecology Letters 5, 121–129. Hutto, R.L., Pletschet, S.M., Hendricks, P., 1986. A fixed-radius point count method for nonbreeding and breeding season use. Auk 103, 593–602. Jha, S., Vandermeer, J.H., 2010. Impacts of coffee agroforestry management on tropical bee communities. Biological Conservation 143, 1423–1431. Johnson, M., 2000. Effects of shade-tree species and crop structure on the winter arthropod and bird communities in a Jamaican shade coffee plantation. Biotropica 32, 133–145. Johnson, M., Levy, N., Kellermann, J., Robinson, D.E., 2009. Effects of shade and bird exclusion on arthropods and leaf damage on coffee farms in Jamaica’s Blue Mountains. Agroforestry Systems 76, 139–148. Jones, J., Ramoni-Perazzi, P., Carruthers, E., Robertson, R.J., 2002. Species composition of bird communities in shade coffee plantations in the Venezuelan Andes. Ornitologia Neotropical 13, 397–412. Kellermann, J.L., Johnson, M.D., Stercho, A.M., Hackett, S.C., 2008. Ecological and economic services provided by birds on Jamaican Blue Mountain coffee farms. Conservation Biology 22, 1177–1185. Komar, O., 2006. Priority Contribution. Ecology and conservation of birds in coffee plantations: a critical review. Bird Conservation International 16, 1–23.

Luck, G., Daily, G., 2003. Tropical countryside bird assemblages: richness, composition, and foraging differ by landscape context. Ecological Applications 13, 235–247. Mas, A., Dietsch, T., 2003. An index of management intensity for coffee agroecosystems to evaluate butterfly species richness. Ecological Applications 13, 1491–1501. Moguel, P., Toledo, V., 1999. Biodiversity conservation in traditional coffee systems of Mexico. Conservation Biology 13, 11–21. Olden, J.D., Lawler, J.J., Poff, N.L., 2008. Machine learning methods without tears: A Primer for Ecologists. The Quarterly Review of Biology 83, 171–193. Parrish, J.D., Petit, L.J., 1996. Value of shade coffee plantations for tropical birds: landscape and vegetation effects. In: Lockeretz, W. (Ed.), Environmental Enhancement Through Agriculture. Center for Agriculture, Food, and Environment. Tufts University, pp. 113–124. Perfecto, I., Armbrecht, I., Philpott, S.M., Soto-Pinto, L., Dietsch, T.V., 2007. Shaded coffee and the stability of rainforest margins in Latin America. In: Tscharntke, T., Leuschner, C., Zeller, M., Guhadja, E., Bidin, A. (Eds.), The Stability of Tropical Rainforest Margins, Linking Ecological, Economic and Social Constraints of Land Use and Conservation. Springer, Environmental Science Series, Heidelberg and New York. Perfecto, I., Rice, R., Greenberg, R., Van der Voort, M., 1996. Shade coffee: a disappearing refuge for biodiversity. BioScience 46, 598–608. Perfecto, I., Mas, A., Dietsch, T., Vandermeer, J., 2003. Conservation of biodiversity in coffee agroecosystems: a tri-taxa comparison in southern Mexico. Biodiversity and Conservation 12, 1239–1252. Perfecto, I., Vandermeer, J.H., Bautista, G.L., Nunez, G.I., Greenberg, R., Bichier, P., Langridge, S., 2004. Greater predation in shaded coffee farms: the role of resident neotropical birds. Ecology 85, 2677–2681. Peters, V., Mordecai, R., Carroll, C., Cooper, R., Greenberg, R., 2010. Bird community response to fruit energy. Journal of Animal Ecology 79, 824–835. Petit, D., Petit, L., Saab, V., Martin, T., 1994. Fixed-radius point counts in forests: factors influencing effectiveness and efficiency, in USDA Forest Service General Technical Report PSW-GTR-149. Philpott, S., Greenberg, R., Bichier, P., Perfecto, I., 2004. Impacts of major predators on tropical agroforest arthropods: comparisons within and across taxa. Oecologia 140, 140–149. Philpott, S.M., Arendt, W.J., Armbrecht, I., Bichier, P., Diestch, T.V., Gordon, C., Greenberg, R., Perfecto, I., Reynoso-Santos, R., Soto-Pinto, L., Tejeda-Cruz, C., Williams-Linera, G., Valenzuela, J., Zolotoff, J.M., 2008. Biodiversity loss in Latin American coffee landscapes: review of the evidence on ants, birds, and trees. Conservation Biology 22, 1093–1105. Philpott, S.M., Soong, O., Lowenstein, J.H., Luz Pulido, A., Tobar Lopez, D., Flynn, D.F.B., DeClerck, F., 2009. Functional richness and ecosystem services: bird predation on arthropods in tropical agroecosystems. Ecological Applications 19, 1858– 1867. R Development Core Team, 2008. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Sekercioglu, C., 2006. Increasing awareness of avian ecological function. Trends in Ecology & Evolution 21, 464–471. Stiles, F., Skutch, A., 1990. A Guide to the Birds of Costa Rica. Christopher Helm, London, UK. Stotz, D.F., Fitzpatrick, J.W., Parker, T.A., Moskovits, D.K., 1996. Neotropical Birds: Ecology and Conservation. University of Chicago Press, Chicago, IL, USA. Strasser, H., Weber, C., 1999. The asymptotic theory of permutation statistics. Mathematical Methods of Statistics 220, 250. Strobl, C., Hothorn, T., Zeileis, A., 2009. Party on! The R Journal 1, 14–17. Tejeda-Cruz, C., Sutherland, W.J., 2004. Bird responses to shade coffee production. Animal Conservation 7, 169–179. Thiollay, J.M., 1995. The role of traditional agroforests in the conservation of rainforest bird diversity in Sumatra. Conservation Biology 9, 335–353. Thompson, J., Spies, T., 2009. Vegetation and weather explain variation in crown damage within a large mixed-severity wildfire. Forest Ecology and Management 258, 1684–1694. Tscharntke, T., Sekercioglu, C., Dietsch, T., Sodhi, N., Hoehn, P., Tylianakis, J., 2008. Landscape constraints on functional diversity of birds and insects in tropical agroecosystems. Ecology 89, 944–951. Van Bael, S.A., Philpott, S., Greenberg, R., Bichier, P., Barber, N., Mooney, K., Gruner, D., 2008. Birds as predators in tropical agroforestry systems. Ecology 89, 928–934. Vandermeer, J., Perfecto, I., Philpott, S., 2008. Clusters of ant colonies and robust criticality in a tropical agroecosystem. Nature 451, 457–459. Waltert, M., Bobo, K.S., Sainge, N.M., Fermon, H., Muhlenberg, M., 2005. From forest to farmland: habitat effects on afrotropical forest bird diversity. Ecological Applications 15, 1351–1366. Waltert, M., Mardiastuti, A., Muhlenberg, M., 2004. Effects of land use on bird species richness in Sulawesi, Indonesia. Conservation Biology 18, 1339–1346. Wunderle Jr., J.M., Latta, S.C., 1998. Avian resource use in Dominican shade coffee plantations. The Wilson Bulletin 110, 271–281.