Shade Coffee: a Disappearing Refuge for Biodiversity Shade Coffee Plantations Can Contain As Much Biodiversity As Forest Habitats

Shade Coffee: A Disappearing Refuge for Biodiversity Shade coffee plantations can contain as much biodiversity as forest habitats

Ivette Perfecto, Robert A. Rice, Russell Greenberg, and Martha E. Van der Voort

Within the expanding agri- cal biodiversity. In this article we cultural frontier in the trop- discuss the role of shade coffee plan- ics, one can find a variety of The importance of tations in protecting biodiversity. We small, managed forest patches and shade coffee as a refuge focus on northern Latin America, an traditional agricultural systems, area encompassing the Caribbean is- which provide a refuge for forest- for biodiversity may lands, Mexico, Central America, and dwelling organisms. These managed the Andean countries of South habitats are frequently overlooked not be in the total land America. However, many of the is- as potential areas of biodiversity con- sues and conclusions discussed here servation (Pimentel et al. 1992). Fur- it involves, but in its also apply to coffee-exporting coun- thermore, the conservation biology tries throughout the tropics. literature often refers to forest re- location in areas that serves as islands in a sea of devasta- have been particularly The economic importance tion, in which the sea is formed by of coffee agriculture. Although chemically in- hard hit tensive monocultural systems may Coffee was introduced into the New fit well with this perception of low by deforestation World by the Dutch in 1723 (Wrigley 1988). During the twentieth century Ivette Perfecto is an assistant professor it has reached considerable impor- in the School of Natural Resources and biodiversity, many other agro- tance in the world market as an ex- Environment, University of Michigan, ecosystems, especially in the tropics, port crop. Production has tripled in Ann Arbor, MI 48109-1115. Her re- are characterized by high vegetational northern Latin America since World search interests include the transforma- diversity. One such agroecosystem is War II, and area under cultivation tion of agriculture and associated coffee, when managed with tradi- has nearly doubled (UNFAO Pro- changes in biodiversity. Robert A. Rice tional cultural practices. This popu- duction Yearbooks). works on policy issues at the Smithsonian lar beverage, used worldwide for cen- It is hard to overestimate the im- Migratory Bird Center, National Zoo- turies, constitutes a major source of portance that coffee production and logical Park, Washington, DC 20008. household income and foreign ex- exportation has had for northern His interests include land use changes in Latin America and the environmental change for many tropical countries, Latin America. More than 32% of consequences of agricultural modern- especially in Latin America. the world's coffee comes from this ization. Russell Greenberg is the direc- Coffee is traditionally grown un- region, where it is the leading source tor of the Smithsonian Migratory Bird der a canopy of shade trees. Because of foreign exchange. Although cof- Center, National Zoological Park, Wash- of the structural and floristic com- fee is produced on only 7.4% of the ington, DC 20008. His research inter- plexity of the shade trees, traditional total arable land, coffee lands at ests include the ecology of migratory coffee plantations have relatively present take up approximately 44% birds in tropical ecosystems. Martha E. high biodiversity. However, coffee of the area of permanent cropland Van der Voort was a staff researcher at plantations increasingly are being (UNFAO Production Yearbook Smithsonian Migratory Bird Center, transformed into industrial planta- 1991). In northern Latin America, National Zoological Park, Washington, DC 20008. She is currently studying tions with little or no shade (Figure coffee plantations cover approxi- nontimbef forest products at the De- 1). The way that coffee production mately 2.7 million ha. This total partment of Wildlife, West Virginia evolves in the coming decades is likely includes roughly 700,000 ha in University, P.O. Box 6125, Morgantown, to have a tremendous impact on its Mexico, 300,000 ha in the Carib- WV 26505. ability to provide a refuge for tropi- bean, 750,000 ha in Central

598 BioScience Vol. 46 No. 8 Figure 1. (Lefr) A sun plantation near San Jose, Costa Rica (photo by I. Pcrfecro). (Right) A traditional shade plantation near Tapachula. Chiapas (photo courtesy of M. Van der Voort).

America, and 1,000,000 ha in Co- agrochemicals, modernization inten- lombia. sified in the 1970s. We estimate that almost half of the area in coffee pro- Coffee cultivation techniques duction in northern Latin America had been converted by 1990. The egy for the coffee farmer. Siman Coffee cultivation systems fall along speed and extent of conversion, how- found that modern farms out- a continuum, ranging from the "tra- ever, have been uneven. The percent- produced semi-modern (a combina- ditional" ro the "modern" (Table age of land converted in the region tion that includes some shade reduc- l).1 The modern system is character- varies from as low as 15% in Mexico tion, a change to new coffee varieties, ized by a reduction in shade, in- to more rhan 60% in Colombia (Fig- and at least some use of agrochemi- creased reliance on new high-yield- ure 2). cals) and traditional farms, with ing varieties, and an increase in Modernization was initially seen yields of 1397, 953, and 317 kg/ha, chemical inputs, pruning, and coffee as a way of combating fungal dis- respectively.2 However, the levels of plant density (Coyner 1960,DeGraaf eases, particularly coffee leaf rust production had considerably differ- 19S6). The removal of shade in cof- [Hemilcia vastatrix). The role that ent costs as well; in absolute terms, fee farms helps establish a suite of coffee leaf rust played in plantation the cost (in US dollars) of production characteristics of a coffee cultivation modernization is significant, because for a hectare of modern, semi-mod- system aimed at incteasing yields, at the disease ranks as the most feared ern, and traditional coffee was least over the short run. However, obstacle to production in most cof- $1738.94, S1092.00, and $269.47, with the loss of canopy cover, mod- fee- producing areas (Agrios 1982). respectively. The cost to produce 1 kg ern plantations, also known as sun Early on, coffee leaf rust provided of coffee was thus $1.24 for modern plantations, become more prone to the hook on which the entire coffee coffee, $1.14 for semi-modern cof- water and soil runoff, threatening modernization process hung its hat. fee, and SO.85 for traditional coffee. the long-term snstainability of the Phytopathological reasoning main- Actual profits, of course, vary with system (Rice 1990). tained that less shade would allow world coffee price fluctuations. The One of the most striking features moisture on coffee plants to dry more traditional technology, with a much of the conversion from traditional to readily, therefore reducing fungal lower use of chemical inputs, repre- modern coffee cultivation is the ra- germination success. The arrival of sents a passive production system pidity with which it has occurred. rust in Brazil in 1970 and in Central in which the coffee unit receives After a largely unsuccessful attempt America in 1976 brought to life the little attention in the way of labor in the 1950s to modernize coffee agronomic nightmares that had and/or capital. Traditional produc- growing using new strains and more plagued coffee growers and govern- tion devotes 2% of its expenditures ments in the Old World for genera- to chemical inputs, whereas semi- tions. However, coffee leaf rust has modern and modern production >W* CM* the term modern here, a system also not been as problematic as predicted, spend 19% and 25% on chemical referred to as ;'«*atl«f^fifarrd technified.Tbk and the major motivation behind inputs, respectively. In addition, last term, although cumbersome, is used by modernization has since become in- development agencies and local institution; creased production. : arid describes quire well thj technical and J. J. Siman 1991, unpublished manuscript. industrial approach TO what has heretofore Modernized coffee represents a Tropical Agronomic Center for Research and been a traditional production sjrsrem. major departure in economic strat- Teaching, Managua, Nicaragua,

September 1996 599 Table 1. Distinguishing characteristics of traditional and modern coffee production technologies.

Characteristic Traditional Modern

Coffee variety Tipica, bourbon, maragogipe Caturra, catuai, Colombia, Guarnica catimor Coffee height 3-5 m 2-3 m Shade cover Moderate to heavy, 60%-90% coverage None to moderate, up to 50% coverage Shade trees used Tall (15-25 m), mixed forest trees, Short (5-8 m), legumes; often monocultures legumes, fruit trees, bananas Density of coffee plants 1000-2000/ha 3000-10,000/ha Years to first harvest 4-6 3-4 Plantation life span 30+ years 12-15 years •< Agrochemical use None to low High, particularly fertilizers, herbicides, fungicides, nematocides Pruning of coffee Individualized pruning or no pruning Standardized stumping back after first or second year of full production Labor requirements Seasonal for harvest or pruning Year-round maintenance with higher demands at harvest Soil erosion Low High (particularly on slopes) nonharvest labor accounts for the lower total income, organic coffee Shade coffee and biodiversity single largest cost in modernized sys- production resulted in a significantly conservation—an overview tems because it entails an array of higher net revenue (approximately intense cultivation practices such as $350.00/ha), in part because of lower On a geographical scale, the impor- standardized pruning, fertilization, production costs. Furthermore, when tance of shade coffee as a refuge for and insecticide, fungicide, and nema- externalities generated by environ- biodiversity may not be in the total tocide applications to individual mental costs associated with coffee land it involves, but in its location in plants. production (e.g., pesticides and/or areas that have been particularly hard Comparisons have also been made soil erosion) were incorporated into hit by deforestation. In fact, the total between modernized and organic cof- the analysis, the differences in net land area planted in coffee is moder- fee production by Akkerman and revenue between organic and non- ate compared with some other land Van Baar (1992) and Boyce et al. organic production increased (Boyce uses (particularly pasture). However, (1994). They reported that despite et al. 1994). plantations tend to be localized on relatively high-quality soils and in the mid-elevation (500-2000 m) ecological zone. Natural habitats in these zones, which include pine-oak woodland and premontane tropical forest, are often highly fragmented and de- .So homos graded, and few reserves have been established to protect the many organisms endemic to these habitats. The Pacific slope of the central Cor- dillera in Central America has been particularly devastated. In areas where deforestation is high and coffee is still produced on traditional shade plantations, these plantations are likely to be a critical refuge for the forest biota. In fact, coffee plantations may already have served as a critical refuge during a human-caused habitat bottleneck. Brash (1987) suggested that the relatively low rate of avian extinction experienced on Puerto Rico during recent periods of deforestation may Figure 2. A map of northern Latin America with proportional circles depicting be due in part to the presence of relative coffee area by country (figures in thousands of hectares; based on UNFAO shade coffee plantations. Similarly, Production Yearbook 1990) and approximate area modernized (white portion of circle) and nonmodernized (shaded portion of circle). The modernized area does not Nir (1988) argued that many rare include lands designated as "semi-modernized." Question marks indicate no data orchids survived deforestation in available for modernized coffee area. Data are from a variety of sources for the early Puerto Rico on shaded coffee farms. 1990s, except area estimates from Mexico and Nicaragua, which are for the early By the turn of the nineteenth cen- 1980s (see Rice 1993 for details). tury, 99% of the original forest cover

600 BioScience Vol. 46 No. 8 on Puerto Rico had been lost, with Table 2. Number of species of beetles, ants, wasps, and spiders in the canopy of essentially no second-growth forest shade trees and coffee plants in different types of coffee farms, based on fogging replacing it. However, shaded coffee with Pyrethrin-based insecticides.* plantations still covered 9% of the island. As the rural economy has Species Type of farm Beetles Ants Non-ant Hymenoptera Spiders been abandoned, forest is returning Shade trees to much of the island, and the "seed" Erythrina Traditional 126 30 103 NA* for its regrowth is often the aban- poeppigiana doned coffee estates (Weaver and Erythrina fusca Traditional 110 27 61 NA Birdsey 1986). Annona sp. Traditional NA 10 63 NA E. poeppigiana Technified with 48 5 46 NA The plant diversity of coffee plan- shade tations results from two distinct pro- Coffee plants cesses. First, a small percentage of Coffea arabica* Traditional 39 14 34 44 plantations ("rustic") are planted in Technified with 29 9 31 NA shade forest cleared of its understory. Technified without 29 8 30 29 Therefore, the diversity of the natu- shade ral forest canopy is preserved in a modified form. In addition, rustic *I. Perfecto, 1996, manuscript in preparation. •Data not yet available. plantations are often unweeded dur- *Coffee based on ten plants per treatment. ing periods of low coffee prices, which adds to the maintenance of biodiversity as well. Purata and Throughout northern Latin America, canopy trees from the impact of rain Meave (1993) found that rustic plan- it is common to find banana (Musa and wind, and they also lack the tations provide the only habitat for spp.), citrus (e.g., tangerine, orange, input of canopy leaf litter (Beer several forest trees in the mid-eleva- grapefruit, and lemons), or other fruit 1988). Therefore, even structurally tion zone of Oaxaca, Mexico. The trees (e.g., avocados, mamey, man- equivalent layers of shade and sun term rustic is also often conferred goes, and zapotes) mixed in with the plantations are dramatically differ- on highly diverse indigenous agro- coffee, filling out the multistrata sys- ent habitats. Finally, shade trees pro- forestry systems that incorporate tems in which coffee itself forms the vide a high diversity of food items coffee production. One well-studied shrub layer. Some fuel wood- and for herbivores, frugivores, and example is the te'lom, or managed timber-producing trees, including nectarivores. Where there is a diver- forests, of the Huastec Maya of Cedrela mexicana, Cordia alliodora, sity of canopy species, differences in Tamaulipas, Mexico. Te'lom forests and Swietenia macrophylla, are also the timing of fruit and flower pro- contain more than 300 plant species found in diverse coffee farms. Tradi- duction are likely to reduce pheno- and cover a quarter of the Huastec's tional planted coffee farms com- logical gaps (periods when no fruit agricultural land (Alcorn 1984). The monly have more than 40 tree spe- or nectar resource for a particular incorporation of coffee into such a cies. Larger plantations tend to be taxa is available). traditional agroforestry system is not less diverse, planted with one or a surprising because cacao (chocolate) few species of native legumes, which Arthropod diversity was cultivated in this manner for are often heavily pruned. In many 2000 years before the Conquest plantations, exotic trees are used, Studies that have compared arthro- (Bergmann 1969). particularly Grevillea robusta, pod diversity in coffee plantations In the second and more common which grows well at higher eleva- with that in forests have reported process, the shade canopy is of sub- tions and survives low temperatures. either similar or higher diversity in stantially lower diversity than in rus- The high structural complexity of plantations. In Sulawesi, Stork and tic plantations and is maintained the traditional coffee plantation is a Brendell (1990) found the number of through deliberate planting. The result of the various vegetative lay- arthropod species in coffee planta- overstory species found in traditional ers in the agroecosystem. This struc- tions to be almost double that of coffee plantations vary from country tural complexity offers living and mid-elevation forests. In a compara- to country and regionally within each nesting sites for a variety of organ- tive study in Puerto Rico, Torres country (see Fuentes-Flores 1979 for isms. In addition to increasing pri- (1984) reported a more diverse ant a classification for Mexican systems), mary structural diversity of foliage fauna in a coffee plantation than in as does the intensity with which shade layers, the canopy of plantations can an upland tropical forest in the same trees are pruned. Nitrogen-fixing le- support secondary structures com- region. Similar high ant diversity has gumes such as Ittga spp., Erythrina prised of epiphytes, parasites (e.g., been reported for cacao (Theobroma spp., and Gliricidia sepium form an mistletoes—Lorarithaceae), mosses, cacao) plantations, which are struc- important component of many cof- and lichens, which in turn support a turally similar to coffee plantations fee farms. The latter two genera lose community of arthropods, amphib- but are typically found at lower el- most or all of their leaves during the ians, and other creatures. The canopy evations (Majer 1978, Room 1971, dry season, which renders planta- also affects the microclimate of the 1975). tions on which they are grown simi- coffee understory. Sun coffee planta- Studies of arthropod assemblages lar to sun plantations for this period. tions lack the protection provided by in the canopy of shaded plantations

September 1996 601 attest to the high diversity of pend on closed canopy forest. A more these guilds (Greenberg et al. in arthropods in these systems (Stork detailed study in the Dominican Re- press), a figure that is significantly and Brendell 1990). The most dra- public supports this finding (Wun- higher than for other forest habitats matic finding to date is that of Per- derle and Latta in press). In addi- in the region. Shade coffee also sup- fecto et al.3 from a shaded plantation tion, Wunderle and Latta (1994) ports a high concentration of in Heredia, Costa Rica (Table 2). By found that individuals of several nectarivores or partial nectarivores fogging with pyrethrin-based insec- migratory species in shade coffee in the Dominican Republic (Wun- ticides, in a manner similar to some plantations survived the winter at a derle and Latta in press). The high tropical forest studies (Erwin and rate comparable with those in natu- concentration of euphonias (small Scott 1980, Stork and Brendell 1990), ral forest habitats. Greenberg et al. tanagers that eat mistletoe berries) Perfecto and colleagues sampled the (1995) classified the migratory avi- in taller, less-pruned plantations sug- arthropods in the canopy of four fauna of eastern Chiapas, Mexico, as gests that parasitic plants support shade trees and ten coffee bushes. forest specialists, forest generalists, additional diversity as well. Ants, other hymenopterans, beetles, and scrub/open species, and then de- In general, migratory birds seem and spiders were sorted into morpho- termined that shade coffee planta- to fare better than resident birds in species. In the canopy of a single tions support a high number of shade plantations, perhaps because Erythrina poeppigiana they recorded species of forest migrants (both gen- migratory birds have less stringent 30 species of ants, 103 species of eralists and specialists) compared habitat requirements than those spe- other hymenopterans, and 126 spe- with other habitats in the region, cies committed to breeding in the cies of beetles. A second tree yielded and often at higher densities than region. Resident birds may be af- 27 species of ants, 61 species of other natural forests. fected by a variety of local ecological hymenopterans, and 110 species of Shade coffee plantations may also and landscape factors. The small size beetles. Although the two sampled be an important dry season refuge, of the average coffee plantation trees were less than 200 m apart, the providing fruit and nectar for birds makes it susceptible to fragmenta- overlap of species was only 14% for when insect populations are other- tion, whose effects are known to be beetles and 18% for ants. These pre- wise dwindling. Several of the com- severe in tropical areas (Lovejoy et liminary results suggest that shaded monly planted shade trees are native al. 1986). Structural modifications plantations can have a local species species that produce flower crops that remove foraging and nest sites diversity within the same order of favored by omnivorous birds. It is for some species probably account magnitude as undisturbed forest. For likely that the movements of latitu- for the loss of many forest special- example, Wilson (1987) reported 62 dinal and altitudinal migrants ists. In addition, larger birds, such as and 47 ant species from two trees (Vannini 1994) are timed to take cracids, parrots, and raptors, may be each in upland rain forest in Peru, advantage of the asynchrony of susceptible to hunting pressures. and Adis et al. (1984) reported 38 flower crops available in shade plan- Nevertheless, Aguillar-Ortiz ant species in one Dipterix alata and tations. Greenberg et al. (in press) (1982), Corredor (1989), and two Eschweilera cf. odora in an up- examined the seasonality of bird use Greenberg et al. (in press) found that land rain forest in Brazil. of coffee plantations and other habi- the species richness of birds in coffee tats in eastern Chiapas through the plantations with a structurally and Birds and other vertebrates repeated censusing of transects. Of floristically diverse canopy compares the 23 habitats censused in eastern well with other natural forest habi- With the possibility of deforestation Chiapas, only the two plantation tats with which many species are causing declines in several species of types (rustic and planted with Inga) shared. Greenberg et al. (in press) birds that migrate from North showed an overall significant increase also showed that diversity of birds in America to northern Latin America in bird numbers over the winter. coffee plantations (and other forest (Askins et al. 1990), many studies Both the number of individual birds types) is considerably higher than in have focused on the status of over- and the number, of species nearly other agricultural habitats. In part, wintering populations in different doubled, a pattern that held for resi- the high diversity of shade planta- agricultural and natural habitats. dent tropical species as well as mi- tions results from the number of edge Coffee plantations have often been grants. This increase was specific to and second-growth species that oc- singled out for their ability to sup- omnivorous species; insectivores cur along with a smaller number of port numbers of forest migrants, showed stable or slightly declining true forest birds (Corredor 1989, those species most likely to be af- populations through the winter. In Greenberg et al. in press). Greenberg fected by conversion of forest to farm- the 7«g<3-dominated plantations, the et al. (in press) showed that in land. Wunderle and Waide (1993) increase consisted largely of flower- Chiapas, the avifaunal similarity conducted a regional survey of the feeding species such as Baltimore between pine-oak woodland and Greater Antilles and concluded that orioles (Icterus galbula) and Tennes- planted and rustic coffee is high shade coffee plantations support high see warblers (Vermivora peregrina). (75%-80%). Similarity with meso- densities of certain species that de- Shade coffee plantations are par- philous forest, however, was low, ticularly well represented by canopy and several of the more specialized 3 species found commonly in premon- I. Perfecto et al., 1995, unpublished omnivores and nectarivores. During mansucript. University of Michigan, Ann the dry season in Chiapas, more than tane forest, such as the spectacled Arbor, MI. 45% of the individual birds were in foliage-gleaner (Anabacerthia varie-

602 BioScience Vol. 46 No. 8 gaticeps), were never recorded in cof- plantations in Veracruz. Estrada et matic reduction in hymenopterans.6 fee plantations. The bird diversity of al. (1994) did not find such mam- Aside from the loss of food provided heavily pruned shade plantations mals in plantations in Las Tuxtlas by the trees, the habitat structure dominated by a single canopy spe- but did regularly observe howler becomes simplified through loss of cies, a common plantation type, was monkeys (Allouata palliata). And canopy foliage layers, tree trunks, only two-thirds of that of the more working in Guatemala, Seib (1986) and associated epiphytes. Canopy forestlike coffee plantations.4 In par- reported that mixed-shade planta- trees provide a host of poorly known ticular, forest frugivores, bark-glean- tions can support up to 50% of the microhabitats. The effect of canopy ers, and understory species were original forest snake fauna. loss is likely to be severe for trees poorly represented in the more mono- with a specialized canopy fauna. For specific and heavily trimmed shade Biodiversity and the impact of example, the high diversity of arbo- plantations. real beetles, with more than 100 spe- Shade coffee plantations support coffee conversion cies in a single tree, is undoubtedly a high diversity of other vertebrate The few direct comparisons between lost in systems that lack canopy trees groups as well as birds. Estrada et al. sun and shade plantations focus on (Table 2). Preliminary studies7 sug- (1993) found that, compared with the ground or coffee strata, and gen- gest that a large percentage of the other agricultural habitats, a high erally they show a decrease in diver- ants found in the canopy of shade diversity and abundance of bats use sity with the conversion from shade trees are also exclusively arboreal. various shade plantations with di- to sun types. Perfecto et al.5 (Table 2) For example, in a shaded plantation verse canopies. However, diversity showed that arthropod diversity was in Costa Rica, an average of 72% of was considerably lower than in low- lower in the ground strata of mono- the ants were found exclusively in land tropical forest. As with birds, a specific shade farm and shadeless trees. large proportion of individuals and coffee monoculture than in that of The loss of the shade in coffee species were partially frugivorous and shaded canopy. Although Hanson plantations also means the loss of nectarivorous, feeding on the flow- (1991) reported a high hymenopteran resources for many species in the ering and fruiting trees of the canopy. diversity in coffee monoculture (80 detritivore food chain, particularly Estrada et al. (1993) argue that the species of parasitoids) in the same saproxilic and leaf litter arthropods. mobility of bats (like some birds) region of Costa Rica, species rich- Shaded plantations in Costa Rica allows them to forage over shade produce between 5000 and 20,000 ness was still lower than on tradi- 1 1 plantations and other forest patches tional plantations. Perfecto and kg ha" • yr of leaf litter and prun- scattered over a large area. A strong Snelling (in press) surveyed ant spe- ing residues (Beer 1988)—values that relationship between the presence of cies diversity using bait transects on fall within the range for tropical for- a structurally diverse canopy and a 16 coffee farms and found a positive ests (Vitousek 1984). Shaded planta- high diversity of small terrestrial and correlation between species diver- tions, particularly rustic ones, con- scansorial mammals was found by sity and vegetational complexity. tain old and dead trees that provide Gallina et al. (1992) in Veracruz, Once again, the highest diversity was habitats for a diverse saproxilic ar- Mexico. In this study, species depen- found in the traditional farm and the thropod community. In Mexico, a dent on canopy trees made up more lowest in sun coffee plantations. Per- single coffee and cacao plantation than half of the fauna of the planta- fecto and Vandermeer (1994) and was reported to contain 78 families tions, and more than 40% of the Perfecto (1994) suggested that both of saproxilic invertebrates, with 93 % species were omnivores that com- direct (e.g., loss of nesting sites) and belonging to the orders Coleoptera, monly fed on fruit. indirect (e.g., changes in competitive Diptera, Hymenoptera, or Collem- A similar high abundance and di- interactions) mechanisms are respon- bolla (Moron and Lopez-Mendez versity, as well as proportion of sible for reductions within the ant 1985). Although studies directly com- omnivores, was found for nonflying community in the coffee monocul- paring saproxilic communities be- mammals in the Las Tuxtlas region ture. Working in Mexico, Nestel and tween shade and nonshaded planta- of Mexico (Estrada et al. 1993, colleagues (Nestel and Dickschen tions are apparently lacking, the 1994). This observation is not sur- 1990, Nestel et al. 1993) reported a reduction in decaying wood and leaf prising, considering that many of the reduction in the diversity of ant and litter suggests that these assemblages trees managed in coffee plantations macrocoleopteran assemblages in sun are greatly reduced along with shade produce fruit that is eaten not only coffee plantations as compared with elimination. by humans but also by other mam- shaded coffee plantations. Changes in the coffee stratum it- mals. Gallina et al. (1992) also re- Studies restricted to the coffee lay- self due to loss of microclimate buff- ported that in addition to the many ers are likely to greatly underesti- ering are profound. The conversion omnivorous species, some more spe- mate the difference in overall diver- of coffee plantations invariably re- cialized mammals, such as small cats sity between plantation types. The sults in an increase in the amount of and otters, can be found in shade elimination of trees, with their foli- solar radiation reaching the ground, age, flower, fruit mesocarp, and with concomitant increases in temperature and wind speed, direct im- 4R. Greenberg, P. Bichier, A. Cruz, and R. extrafloral nectaries, results in a dra- Reitsma, 1996, manuscript in review. Smithsonian Migratory Bird Center, Wash- *See footnote 3. ington, DC. 5See footnote 3. 7See footnote 3.

September 1996 603 pact of precipitation, and a decrease species are probably the most spe- in relative humidity (Beer 1987). In cialized migratory species on shaded sun coffee plantations, fluctuations plantations (although smaller popu- of both temperature and humidity lations may occur in urban gardens): become more extreme. Perfecto and the Baltimore oriole and the Tennes- Vandermeer8 demonstrated that by see warbler in Mesoamerica, and the experimentally increasing shade, the Cape May warbler {Dendroica diversity of ground-foraging ants in tigrina) in the Antilles.10 Data from coffee plantations increased, at least the Breeding Bird Survey of the US partly as a result of changes in mi- National Biological Service indicates croclimate. that all three species have experi- The modernization of coffee plan- enced sharp and statistically signifi- tations frequently includes a sub- cant population declines from 1980 stantial increase in agrochemicals (De to 1994, corresponding to the period Graaf 1986). Insecticides are known of intense coffee modernization. The to decrease biological diversity in Figure 3. Annual population indices of annual estimated declines are 2.2%, agroecosystems (Jepson 1989). Fun- the Baltimore oriole from 1978 to 1994 5.7%, and 4.2% for the oriole, Ten- gicide applications are also more based on 1351 Breeding Bird Survey nessee warbler, and Cape May war- common in modern plantations than routes throughout North America (cour- bler, respectively (Figure 3). The three in traditional ones. Certain fungi- tesy of US National Biological Service). species experienced significantly ex- cides also are known to have insecti- panding populations during the pre- cidal activity and can have a detri- vious 16 years, which would also be mental impact on insect diversity Wunderle and Latta (in press) found predicted from the continuing in- (S otherton and Moreby 1988). More- a reduction in overall diversity and a crease of area under coffee cultiva- over, it is likely that fungicide appli- significant shift from forest to tion. Other factors may be respon- cations adversely affect the decom- matorral (shrubby second growth) sible for population declines, for position of leaf litter in modern species when comparing monotypic example, habitat fragmentation, systems. The shade of traditional Inga shade and sun plantations in parasitism, and long-term cycles in plantations reduces weed growth the Dominican Republic. prey abundance (Askins et al. 1990). (Nestel and Altieri 1992), so intense Greenberg et al.,' working in Gua- The two warblers, for example, are herbicide applications are necessary temala, determined that the density boreal forest "spruce budworm" spe- to reduce the ground cover of forbs and diversity of birds in sun coffee cies and may be tracking long-term and grasses in sun plantations. The plantations is approximately half that caterpillar cycles on the breeding removal of shade has also been shown in traditional coffee plantations. grounds. However, the sharp decline to disrupt the natural nitrogen cycle Furthermore, sun coffee plantations in orioles is less likely to be caused by associated with litter decomposition support few individual birds and bird these breeding season factors because and with the actions of nitrogen- species than adjacent areas of this species is not known to respond fixing bacteria associated with com- matorral, and many common mator- to insect cycles, it breeds successfully monly planted leguminous shade ral species avoid sun plantations. in edge situations, and it is rarely trees, therefore requiring the addi- This finding probably reflects the parasitized by cowbirds. tion of chemical fertilizers (Babbar low density of arthropods associated 1993). These compounds can pol- with coffee plants and the high de- Predator-prey interactions and lute local water supplies. gree of weed control associated with Even the most cursory observa- sun plantations. Birds make relatively coffee pests tion in sun plantations shows them little use of coffee flowers or berries, Coffee, at least when grown in the to be almost devoid of birds. Borrero and consequently the nectarivorous Western Hemisphere, is well known (1986) first noted the dramatic de- and frugivorous species so prevalent for its lack of insect pests. Although crease in bird diversity in planta- in shade coffee plantations largely many herbivores can potentially dam- tions in Colombia. In part, birds disappear. Gallina et al. (1992) esti- age coffee plants, only a few are respond to the same loss of food mated that approximately half of the economically important (Le-Pelley resources, structural complexity, and species diversity of nonflying mam- 1973). Coffee's resistance to herbi- microclimate buffering that is re- mals is lost due to coffee conversion. vores may lie in the fact that it is a sponsible for changes in arthropod An even higher percentage of reptile chemically well-defended plant assemblages. Most observers have and amphibian diversity appears to (Frischknecht et al. 1986), with noted that the high abundance and be lost (Seib.1986). young leaves containing high quan- diversity of birds in coffee planta- Migratory birds often occur in a tities of alkaloids. In addition, older tions isassociated primarily with the range of habitats, so the impact on leaves are tough. Furthermore, there canopy trees. Not surprisingly, then, them of coffee modernization is dif- may simply be no native species in ficult to assess. Three largely nectarivorous migratory songbird 10 *I. Perfecto and J. Vandermeer, 1995, unpub- J. Wunderle, 1995, personal communica- lished manuscript. University of Michigan, tion. Institute for Tropical Forestry, US For- Ann Arbor, MI. 'See footnote 4. est Service, Palmer, Puerto Rico.

604 BioScience Vol. 46 No. 8 Latin America that have evolved ings varies considerably from coun- harvest represents an equivalent of mechanisms to overcome coffee's try to country. For example, Mexico 100 work days at minimum farm defenses—a common phenomenon is dominated by small holdings on wage. in plant introductions. private and ejido land—91% of the A shaded coffee farm displays two In addition, it has been argued holdings are less than 5 ha. In Co- distinct types of biodiversity, man- that the structurally complex and lombia this value is only 49%, and aged and natural. The choices made floristically diverse traditional cof- 5% of the holdings are greater than by growers to use a variety of shade fee plantation supports a high den- 100 ha. species yields an array of useful prod- sity and diversity of predators and For small farmers, committing ucts and at the same time provides parasitoids, which are ultimately re- oneself to total dependence on coffee cover with varying degrees of struc- sponsible for the reduced number of puts one at great risk, not only with tural diversity. This intentionally insect pests in traditional plantations the vagaries of local weather and managed biodiversity by growers al- (Ibarra-Nunez 1990). The few com- pest outbreaks, but with the often lows for higher levels of natural parative studies in coffee plantations dramatic and unpredictable fluctua- biodiversity as well in these shaded support this assertion (Benitez and tions of the global market. However, systems than is found in sun coffee. Perfecto 1990, Nestel and Dickschen the traditional coffee farm sustains The issue now becomes how best to 1990, Perfecto and Snelling in press, the grower beyond simply generat- make use of the relationship between Perfecto and Vandermeer 1994). For ing an income at harvest time be- managed and natural biodiversity. example, of the arthropod taxa cause of the noncoffee products as- sampled by Perfecto et al.,11 ants, sociated with the shade treees. For Promoting biodiversity on other hymenopterans, and spiders example, overstory species provide coffee farms were all more diverse in shaded plan- fruits, fuelwood, and construction tations than in sun plantations. materials for household consump- Defining an environmentally friendly Robinson and Robinson (1974) esti- tion, as well as a potential source of coffee. Research to address this issue mated spider abundance in shade income derived from the local mar- is in its infancy; however, we believe plantations in Papua New Guinea ket. Honey production is a common that the broad aspects of a biologi- and suggested that the spiders have rural industry in northern Latin cally diverse coffee farm can be out- considerable insecticidal effects. America. Chazaro (1982) found at lined. Clearly the presence of a shade Web-building spiders, for instance, least 90 species of bee-pollinated canopy is essential. Furthermore, the consume 40 million insects per hect- plants in the shade coffee planta- greater the structural and floristic are per year (Robinson and Robinson tions near Xalapa, Veracruz. Wood diversity of this canopy, the greater 1974). Perfecto etal.12 reported 34% from natural and human prunings the likelihood that resources will be more spiders in the coffee bushes in provides a steady supply of fuel and, provided for a greater array of or- a traditional plantation than in a in the case of larger shade species, ganisms. A greater variety of animal- coffee monoculture. Ants, which construction materials for the home pollinated and -dispersed plants will show high diversity in traditional and household furnishings. Selling support the diverse guild of omnivo- plantations, are effective predators surplus wood also brings in added rous species that populate traditional as well (Carroll and Risch 1989). income. Where precious hardwoods shade plantations throughout the These observations suggest that the are mixed within the shade trees, year. The canopy needs to provide elimination of shade may ultimately single trees can be sold to local saw- sufficient coverage throughout the result in increased pest problems as mills or other buyers when times are year to buffer the microclimate of well. tough economically. In situations in the understory from rain and desic- which a more managed coffee plan- cating winds. Tree species selection Diversity and economic tation system is possible, Somarriba and pruning practices should have (1990,1992) has shown that timber minimal impact on the epiphytic risk reduction production and harvesting based on plants, mosses, and lichens as well as Biological diversity can provide im- C. alliadora as a shade species can on dead trunks and limbs that pro- portant economic returns to coffee occur. Production of various fruit vide homes for so many canopy spe- growers. Because of the larger num- provides a household with a con- cies. ber of products derived, the diverse tinuous supply of nutritious prod- On-farm presence of shrub veg- plant community within a traditional ucts for consumption and for the etation along arroyos or on steep farm fits much better into the risk- local market. A single, well-tended slopes will protect streams from ero- averse mentality of many small farm- avocado tree, for instance, can yield sion and provide an additional ha- ers (Reeves and Lilieholm 1993). between 2000 and 3000 fruits per ven for understory species unable to Although much coffee is grown on a season. Aside from household use, cope with a coffee monoculture. Sun relatively small number of large es- such produce can fetch $0.18 per drying, or using more energy-effi- tates, in most coffee-producing coun- fruit. For the two months of the cient technology, will reduce the need tries the average coffee plantation is avocado harvest, a producer can gain for harvesting trees to provide fuel small. The size distribution of hold- as much as $360." This single tree's for coffee dryers used in bean pro- cessing. Finally, reduced or no use of 11 See footnote 3. 13R. Rice, field notes and personal observa- pesticides, herbicides, and chemical '-See footnote 3. tions. fertilizers should be promoted. The

September 1996 605 use of natural or managed mulch fee modernization process and aug- difficult for the majority of small, will additionally foster a rich soil ment projects that promote prod- traditional coffee farmers to com- flora and fauna. Clearly, some of uction of organic and other environ- pete with more industrial produc- these recommendations, such as mentally sustainable coffee. It should tion units. Probably the least ex- maintenance of a diverse shade struc- be noted that PROMECAFE (a Cen- plored set of policy tools with ture and protection of epiphytes, are tral American USAID-supported pro- potential to influence coffee cultiva- inconsistent with what are consid- gram aimed at coffee modernization) tion techniques is incentives that ered to be the most productive or held a workshop in February 1995 could be provided to traditional expedient agronomic practices (Beer that sought to explore a sustainable farms through tax easements, access 1987, Boyce et al. 1994). The chal- coffee sector. Subsequently, this to credit, and technical and market- lenge will be to develop cultivation IICA-sponsored endeavor has begun ing assistance. The rationale for such systems that are a workable compro- to question the strictly modern incentives would be that farmers mise between what is good for the model. Moreover, recent develop- employing traditional shade tech- farmer and what will truly benefit ments and advances in organic cof- niques are providing a long-term biological diversity. fee production in El Salvador point stewardship service of protecting to USAID's burgeoning interest in topsoils, pure water supplies, and Foreign assistance. Because coffee alternatives to modernization. worker safety. This approach is at plantations are managed primarily least partly addressed under pro- for export commodity production, Marketing environmentally friendly grams participating in the Interna- the motivating force behind their coffee. Perhaps market forces can be tional Coffee Register. The register existence is a powerful international harnessed to provide economic in- certifies "fair trade" practices, in market. Therefore, their continued centives to farmers producing "envi- which roasters provide small farm- ecologically sustainable management ronmentally friendly" coffee. Fortu- ers and producer cooperatives with, is likely to require the use of nontra- nately, because coffee functions as a among other things, access to credit, ditional policy tools. One possible segmented market, the possibility of prices above production cost indexed approach is to influence the institu- providing market incentives for en- to world prices, and technical assis- tional programs providing assistance vironmental coffee is better than for tance to increase productivity using for rural development in the region. many export commodities. An in- recently developed organic tech- A principal institutional link for the creasing number of consumers are niques and to diversify commodities projects involving modernization of prepared to pay premium prices for produced. Currently, banks often tie the coffee sector of the region is the so-called specialty coffees—the fast- access to credit, which is critical for US Agency for International Devel- est growing segment of the coffee farmers to bring coffee to market opment (USAID), although for some market. If shade-grown coffee can be during periods of both low and high countries, such as Colombia and marketed in this context, thereby prices, to certain technological pack- Mexico, USAID has played no role. providing higher prices to the pro- ages that include the use of agro- (Influential national institutions in ducer, this could compensate for chemicals, rather than to more eco- these two countries precluded any lower levels of production. Perhaps logically sustainable technologies. US involvement in their coffee sec- the closest that most US consumers tors.) Working sometimes through can come to purchasing coffee with a Internalization of environmental its own (now-defunct) Regional Of- high probability of coming from costs. Finally, growing coffee in fice on Central America and Panama shaded plantations is to purchase modern plantations outcompetes and sometimes with regional institu- "certified organic." The organic cof- more traditional systems in part be- tions like the Inter-American Insti- fee sector, still a minuscule fraction cause associated environmental costs tute for Cooperation on Agriculture of the total US coffee market (ap- are paid by the state or people in (IICA) of the Tropical Agronomic proximately 0.5 % ),14 has grown con- other sectors of the economy, rather Center for Research and Teaching, siderably. Promotion of organic pro- than by the coffee producer. These both located in Costa Rica, USAID duction has been embraced by costs include the cleanup of polluted has played a major role in promoting grassroots institutions working with water supplies or the development of modernized coffee. Since 1978, at small cooperatives, particularly in alternative sources of water, produc- least eight projects totaling US $81 Mexico, Costa Rica, and Nicaragua. tion declines associated with long- million have targeted small coffee Many producers have enthusiasti- term pesticide use or soil erosion, the producers for modernization through cally embraced organic production treatment of workers exposed to pes- reduced shade, high-yielding variet- because it brings higher prices. How- ticides, and the loss of fish produc- ies, and increased chemical applica- ever, the certification process can be tion in streams suffering sedimenta- tion (Rice and Ward 1996). Coffee time consuming, expensive, and bu- tion. Establishing policies to ensure modernization continues in at least reaucratic. that some of the environmental costs three USAID-sponsored projects in are borne by the local producers the region through 1997 (Haiti, El Farmer incentives. Without addi- would encourage more environmen- Salvador, and Guatemala; Rice and tional long-term support, it may be tally benign coffee production. One Ward 1996). example could be an environmental At a minimum, USAID should re- 14M. Rozyne, 1994, personal communication. or health fund supported by taxes on duce or eliminate its role in the cof- Equal Exhange Coffee Co., Stoughton, MA. pesticides.

606 BioScience Vol. 46 No. 8 Conclusions thrum-fogging techniques. Studies on Neo- tienen Documentatie Dienst Landbouw tropical Fauna and Environment 19: 223- Onderzoek. Coffee lands within northern Latin 236. Estrada A, Coates-Estrada R, Merrit D Jr. American coffee-producing countries Agrios GN. 1982. Plant pathology. Orlando 1993. Bat species richness and abundance in (FL): Academic Press. tropical rain forest fragments and in agri- are undergoing fundamental changes. Aguilar-Ortiz F. 1982. Estudio ecologico de las cultural habitats at Los Tuxtlas, Mexico. For the landscapes involved in this aves del cafetal. Pages 103-128 in Avila- Ecography 16: 309-318. transformation, these changes trans- Jimienez E, ed. Estudios ecologia en el . 1994. Non flying mammals and land- late into a reduced vegetative cover, agroecosistema cafetalesa. Xalapa (Mexico): scape changes in the tropical rain forest Instituto Nacional de Investigaciones Sobre region of Los Tuxtlas. Ecography 17: 229- lowered species diversity of the plant Recursos Bioticos. 241. community and its associated fauna, Akkerman A, Van Baar X. 1992. El cafe Erwin TL, Scott JC. 1980. Seasonal and size and the application of agrochemi- organico: la sostenible de ungramo de oro. patterns, trophic structure, and richness of cals onto lands that previously re- San Jose (Costa Rica): Centro de Estudio Coleoptera in the tropical arboreal ecosys- para el Desarrollo Rural, Universidad Libre tem: The fauna of the tree Leuhea seemannii ceived little or no such inputs. Al- de Amsterdam. Triana and Planch in the Canal Zone of ready, 1.1 million ha of coffee within Alcorn J. 1984. Development policy, forests, Panama. Coleopterist Bulletin 34:305-322. the countries in northern Latin and peasant farms: reflections on Huastec- Frischknecht PM, Dufek JV, Baumann TW. America qualify as modernized. The managed forests' contribution to commer- 1986. Purine alkaloid formation in buds total potential area that could be cial production and resource conservation. and developing leaflets of Coffea arabica. Economic Botany 38: 389-406. Phytochemistry 5: 613-617. modernized is just more than twice Askins RA, Lynch JR, Greenberg R. 1990. Fuentes-Flores R. 1979. Coffee production sys- that, at 2.7 million ha. What little Population declines in migratory birds in tems in Mexico. Pages 60-72 in de Salas G, work has been done on the environ- eastern North America. Current Ornithol- ed. Agroforestry systems in Latin America. mental impact of the landscape modi- ogy 7: 1-57. Turrialba (Costa Rica): The Tropical Agro- Babbar LL. 1993. Nitrogen cycling in shade nomic Center for Research and Teaching. fications suggests that, unless steps and unshaded coffee plantations in the Cen- Gallina SE, Mandujaro S, Gonzales-Romero A: are taken, many of these coffee zones, tral Valley of Costa Rica. [Ph.D. thesis.] 1992. Importancia de los cafetales mixtos characterized by high rainfall and University of Michigan, Ann Arbor, MI. para la conservation de la biodiversidad de broken terrain, are likely to suffer Beer J. 1987. Advantages, disadvantages and Mamiferos. Boletin Soc. Ver. Zool. 2: 11- environmental degradation in the desirable characteristics of shade trees for 17. coffee, cacao and tea. Agroforestry Systems Greenberg R, Salgado-Ortiz J, Warkentin I, coming years. This degradation is 5:3-13. Bichier P. 1995. Managed forest patches likely to include a severe loss of bio- . 1988. Litter production and nutrient and the conservation of migratory birds in logical diversity in areas where cof- cycling in coffee (Coffea arabica) or cacao Chiapas, Mexico. Pages 178-190 in Wilson fee plantations currently provide the (Theobroma cacao) plantations with shade M, Sader S, Santana E, eds. The conserva- trees. Agroforestry Systems 7:103-114. tion of migratory birds in Mexico. Techni- last refuges. Actions that might re- Benitez J, Perfecto 1.1990. Efecto de diferentes cal Publication. Orono (ME): University of verse this loss include working with tipos de manejo de cafe sobre las comuni- Maine, School of Natural Resources. small farmers to market ecologically dades de hormigas. Agroecologia Neotropi- Greenberg R, Bichier P, Sterling J. In press. Bird sustainable coffee and reduce the cal 1: 11-15. populations and planted shade coffee plan- support for technification in favor of Bergmann JF. 1969. The distribution of pre- tations of eastern chiapas. Biotropica. Columbian America. Annals of the Associa- Hanson P. 1991. Los parasitoides asociados al policies that reward land steward- tion of American Geographers 59: 85-96. cafeto en Costa Rica. Manejo Integrados de ship. Borrero H. 1986. La substitucion de cafetales Plagas (Costa Rica) 20-21: 8-10. de sombrio por caturrales y su effecto Ibarra-Nunez G. 1990. Los artopodos asociados negative sobre la fauna de vertebrados. a cafetos en un cafetal mixto del Soconusco, Acknowledgments Caldasia 15: 725-732. Chiapas, Mexico. Variedad y abundancia. Boyce JK, Fernandez Gonzales A, Furst E, Segura Folia Entomologica Mexicana 79:207-231. We thank L. Petit, M. Marvin, P. Bonilla 0.1994. Cafe y desarrollo sostenible: Jepson PC, ed. 1989. Pesticides and non-target Rosset, J. Vandermeer, D. Weiden- del cultivo agroquimico a la produccion invertebrates. Andover: Intercept. field, J. Wunderle, and an anony- organica en Costa Rica. Heredia (Costa Le-Pelley RH. 1973. Coffee insects. Annual Rica): Editorial Fundacion UNA. Review of Entomology 18:121-142. mous reviewer for commenting on Brash AR. 1987. The history of avian extinc- Lovejoy TE, et al. 1986. Edge and other effects the manuscript. Figures were pre- tions and forest conversion on Puerto Rico. of isolation on Amazon forest fragments. pared by Ellen Seefeldt and John Biological Conservation 39: 97-111. Pages 275-285 in Soule ME, ed. Conserva- Sterling. Support for research on Carroll CR, Risch SJ. 1989. An evaluation of tion biology: the science of scarcity and coffee was provided to Ivette Per- ants as possible candidates for biological diversity. Sunderland (MA): Sinauer. control in tropical annual systems. Pages Majer JD. 1978. The maintenance of the ant fecto by National Science Founda- 30-46 in Gleisman SR, ed. Agroecology. mosaic in Ghana cocoa farms. Journal of tion grant RII-9012484, and to New York: Springer-Verlag. Applied Ecology 13:123-144. Russell Greenberg by the National Chazaro M. 1982. Flora epicola de la zona Moron MA, Lopez-Mendez JA. 1985. Analisis Geographic Society, Scholarly Stud- cafetales de Coatepec, Veracruz. Pages 95- de la entomofauna necrofila de un cafetal en 102 in Jimenez E, ed. Estudios ecologicas en el Soconusco, Chiapas, Mexico. Folia ies Program of the Smithsonian In- al agrosistema cafetalero. Xalapa (Mexico): Entomologica Mexicana 63:47-59. stitution, and the US AID Projecto Instituto Nacional de Investigaciones Sobre Nestel D, Altieri MA. 1992. The weed commu- Ambiental de Centro America pro- Recursos Bioticos. nity of Mexican coffee agroecosystems: ef- gram. Corredor G. 1989. Estudio comparative entre fect of management upon plant biomass and la avifauna de un bosque natural y un cafetal species composition. Acta Oecologia 13: tradicional en el Quindio. [Professional the- 715-726. References cited sis.] University del Valle, Cali, Colombia. Nestel D, Dickschen F. 1990. The foraging Coyner MS. 1960. Agriculture and trade in kinetics of ground ant communities in dif- Adis J, Lubin YD, Montgomery GG. 1984. Nicaragua. Washington (DC): Foreign Ag- ferent Mexican coffee agroecosystems. Arthropods from the canopy of inundated riculture Service. Oecologia 84: 58-63. and Terra Firme forests near Manaus, Bra- De Graaf R. 1986. The economics of coffee. Nestel D, Dickschen F, Altieri MA. 1993. Di- zil with critical considerations on the pyre- Wageningen (the Netherlands): Publika- versity patterns of soil macro-coleoptera in

September 1996 607 Mexican shaded and unshaded coffee in southern Nicaragua. [Ph.D. dissertation.] insect fauna of Salawesi trees with season, agroecosystems: an indication of habitat University of California, Berkeley, CA. altitude, and forest type. Pages 173-194 in perturbation. Biodiversity and Conserva- _. 1993. New technology in coffee pro- Knight WJ, Holloway JD, eds. Insects and tion 2: 70-78. duction: examining landscape transforma- the rain forests of South East Asia Nir MA. 1988. The survivors: orchids on a tion and international aid in northern Latin (Wallacaea). London (UK): The Royal En- Puerto Rican coffee finca. American Orchid America. Report to the Smithsonian Migra- tomological Society of London. Society Bulletin 57: 989-995. tory Bird Center, Washington, DC. Torres JA. 1984. Diversity and distribution of Perfecto 1.1994. Foraging behavior as a deter- Rice R, Ward J. 1996. Coffee, conservation, ant communities in Puerto Rico. Biotropica minant of asymmetric competitive interac- and commerce in the Western Hemisphere. 16: 296-303. tions between two ant species in a tropical Washington (DC): Smithsonian Migratory [UNFAO Production Yearbook] United Na- agroecosystem. Oecologia 98: 184-192. Bird Center. tions Food and Agriculture Production Year- Perfecto I, Snelling R. In press. Biodiversity and Robinson M, Robinson B. 1974. A census of books. Rome (Italy): United Nations. tropical ecosystem transformation: ant di- web-building spiders in coffee plantations Vannini JP. 1994. Nearctic migrants in coffee versity in the coffee agroecosystem in Costa at Wau, New Guinea, and an assessment of plantations and forest fragments of south- Rica. Ecological Applications. their insecticidal effects. Tropical Ecology western Guatemala. Bird Conservation In- Perfecto I, Vandermeer JH. 1994. Understand- 15:95-107. ternational 4:209-232. ing biodiversity loss in agroecosystems: re- Room PM. 1971. The relative distribution of Vitousek PM. 1984. Litterfall, nutrient cycling duction of ant diversity resulting from trans- ant species in Guana's cocoa farms. Journal and nutrient limitation in tropical forests. formation of the coffee ecosystem in Costa of Animal Ecology 40: 735-751. Ecology 65: 285-298. Rica. Entomology (Trends in Agriculture) . 1975. Relative distribution of ant species Weaver PL, Birdsey A. 1986. Tree succession 2: 7-13. in cocoa plantations in Papua New Guinea. and management opportunities in coffee Pimentel D, Stachow U, Takacs DA, Brubaker Journal of Applied Ecology 12:47-61. shade stands. Turrialba 36: 47-58. HW, Dumas AR, Meaney JJ, O'Neil JAS, Seib R. 1986. Feeding ecology and organization Wilson EO. 1987. The arboreal ant fauna of Onsi DE, Corzilius DB. 1992. Conserving of neotropical snake faunas. [Ph.D. disser- Peruvian Amazon forests: a first assessment. biological diversity in agricultural/forestry tation.] University of California, Berkeley, Biotropica 19: 245-251. systems. BioScience 42: 354-362. CA. Wrigley G. 1988. Coffee. New York: John Purata S, Meave J. 1993. Agroecosystems as an Somarriba E. 1990. Sustainable timber produc- Wiley & Sons. alternative for biodiversity conservation oi tion from uneven-aged shade stands for Wunderle JM, Latta S. 1994. Overwinter turn- forest remnants in fragmented landscapes. Cordia alliodora in small coffee farms. Agro- over of Nearctic migrants wintering in small Page 9 in Forest remnants in the tropical forestry Systems 10: 253-263. coffee plantations in Dominican Republic. landscapes: benefits and policy implications. . 1992. Timber harvest, damage to crop Journal fuer Ornithologie 135:477. Symposium abstracts. Washington (DC): plants and yield reduction in two Costa . In press. Avian abundance in sun and Smithsonian Migratory Bird Center. Rican coffee plantations with Cordia shade coffee plantations and remnant pine Reeves LH, Lilieholm RJ. 1993. Reducing fi- alliadora shade trees. Agroforestry Systems forest in the Cordillera Central, Dominican nancial risk in agrof orestry planning: a case 18: 69-82. Republic. Ornitologia Neotropical. study in Costa Rica. Agroforestry Systems Sotherton NW, Moreby SJ. 1988. The effect of Wunderle JM, Waide RB. 1993. Distribution of 21:169-175. foliar fungicides on beneficial arthropods in overwintering nearctic migrants in the Ba- Rice RA. 1990. Transforming agriculture: The wheatfields. Entomophaga 33: 87-99. hamas and Greater Antilles. Condor 95: case of coffee leaf rust and coffee renovation Stork NE, Brendell MJDI1990. Variation in the 904-933.