Avian Resource Use in Dominican Shade Coffee Plantations

Wilson Bull., 110(2), 1998, pp. 271-281

AVIAN RESOURCE USE IN DOMINICAN SHADE COFFEE PLANTATIONS

JOSEPH M. WUNDERLE. JR.v3‘ AND STEVEN C. LATTA,2’

ABSTRACT-We quantified foraging behavior of 19 bird species in shade coffee plantations in the Domin- ican Republic to document and evaluate their use of food resources in the shade overstory relative to the coffee understory. All species were observed foraging in the Znga “era overstory, and 18 of the 19 species had median foraging heights significantly above the median maximum coffee height. Eight species (42%) foraged exclusively in the canopy or subcanopy and not in the coffee understory. No species foraged exclusively in the coffee, although the Narrow-billed Tody (To&s angustirostns)’ foraged mostly in coffee. A negative correlation was found between a species ’ median foraging height in our shade plantations and its abundance in nearby sun coffee plantations. Invertebrates and nectar were the most important food items in the Znga overstory where 95% of the species gleaned leaf surfaces, 63% probed flowers, 58% gleaned or probed wood, 47% used epiphytes (for invertebrates or fruits), and 26% gleaned or probed Inga fruit. In contrast, birds in coffee foraged primarily for invertebrate prey as 42% of all species gleaned leaf surfaces, 21% gleaned or probed wood, 21% gleaned or probed fruit, and 5% probed flowers. The Znga overstory was an important foraging site for most species suggesting that plantations without a shade overstory (i.e., sun coffee) will have a lower diversity and abundance of food and hence are less attractive to birds than traditional shade plantations. Received 5 May 1997, accepted 11 Nov. 1997.

SINOPSIS.--Cuantificamos el comportamiento alimenticio de 19 especies de aves en plantaciones de cafe de sombra en la Repdblica Dominicana para documentar y evaluar el uso de estas de1 recurso alimenticio en el dose1 en relacidn al sotobosque de cafetos. Todas las especies fueron observadas alimentdndose en el dose1 de Znga vera; y en 18 de las 19 especies la mediana de altura de forrajeo fuc significativamente mayor a la mediana de la altura maxima de1 cafe. Ocho especies (42%) se alimentaban exclusivamente en el dose1 y no en el sotobosque de cafe. Ninguna especie se alimentaba exclusivamente en 10s cafetales, alin cuando Todus angus- tirostris lo hacia primordialmente en el cafe. Una correlacidn negativa fue encontrada en la altura mediana de forrajeo y en su abundancia en nuestras plantaciones de cafe a la sombra y las plantaciones de cafe de sol cercanas. Los invertebrados y el nectar fueron las fuentes de comida mas importantes en el dose1 de Znga donde el 95% de las especies se alimentaban picoteando (“gleaning”) la superficie de las hojas, el 63% probaba las flores, el 58% probaba o picoteaba la madera, el 47% usaba las epffitas (para colectar invertebrados o frutas) y el 26% probaba o picoteaba en las frutas de Znga. Por el contrario, las aves en el cafe primordialmente cazaban invertebrados, ya que el 42% picoteaban en las superficies de las hojas, el 21% picoteaban o proban las madera, el 21% picoteaban y probaban las frutas, y el 5% probaban las flores por su nectar. El dose1 de Znga fue un importantisimo lugar de forrajeo para la mayorfa de las especies, suguiriendo clue las plantaciones clue carezcan de un dose1 con sombra (tal coma el cafe de sol) tendrdn una menor diversidad y abundancia de comida haciendolas asi menos atractivas para las aves clue las plantaciones tradicionales de sombra.

Coffee plantations with a shade overstory est reserves by extending the area covered by can have an abundance of birds and provide overstory canopy. Unfortunately, the current habitat for some forest species (e.g., Wetmore trend in coffee cultivation is to eliminate the 1916; Griscom 1932; Aguilar-Ortiz 1982; shade overstory as farmers convert to higher- Beehler et al. 1987; Wunderle and Latta 1996; yielding coffee grown in open sunlight (Per- Greenberg et al. 1997a,b). Given the attrac- fecto et al. 1996). This “sun coffee” does not tiveness of these agroforestry plantations for support forest species as often as is found in birds, Oldfield (1988) and Thiollay (1995) traditional shade coffee plantations (Borrero have suggested that such plantations could 1986, Wunderle and Latta 1996, Greenberg et function as buffer zones for some tropical for- al. 1997a), and requires higher inputs of fer- tilizers and pesticides (Greenberg 1994, Van- 1International Institute Of Tropical Forestry, USDA nini 1994, Wille 1994, Rice and Ward 1996). Forest Service, PO. Box B, Palmer, Puerto Rico The ecological implications of this change in 00721. coffee cultivation is especially important in * Current address: 110 Tucker Hall, Division of Bi- ological Sciences, Univ. of Missouri, Columbia, MO northern Latin America and the Caribbean 65211. where coffee plantations cover approximately 3 Corresponding author. 2.7 million ha at mid-elevation sites which

271 272 THE WILSON BULLETIN ’ Vol. 110, No. 2, June 1998 have alreadv been extensivelv deforested 8.7 ha). Within each plot we measured all stems at (Rice and Ward 1996). * least 3 cm at 1.3 m from the base (DBH). Stems less than 3 cm were measured in belt transects of 1.7 m To encourage coffee farmers to retain their width along the four cardinal directions from the plot shade overstory, an effort is underway to pro- center. Each belt transect was sampled by walking the mote the labeling of coffee packages desig- north, south, east, and west radii with outstretched nating a shade plantation origin (Greenberg arms and counting all vertical stems less than 3 cm 1994, Wille 1994, Rice et al. 1997). This la- contacted by the arms or chest at 1.3 m above the beling scheme would encourage environmen- ground. Basal area of shrub and tree stems was cal- tally conscientious consumers to purchase culated using the actual DBH measurements for stems at least 3 cm and an estimate of 1.5 cm DBH for each shade coffee, thereby using the marketplace to stem less than 3 cm. promote the maintenance of shade canopy Foliage heights were determined at 20 points located trees in some tropical agricultural landscapes. at 1.6 m intervals along the north, south, east, and west However, before such a program is initiated, radii of each circular plot. A 3-m tall pole (2.0 cm it is valuable to understand how wildlife use diameter) marked at 0.5 m intervals was placed ver- shade plantations and what attracts them to tically at each sampling point. We recorded the presence or absence of coffee foliage touching the pole plantations. within each height class. For height intervals above 3 Despite the potential value of shade coffee m, we sighted along the pole and recorded the pres- for bird populations, only Greenberg and co- ence/absence of foliage in each of the following height workers (1997a) have quantified some aspects intervals: 3-4 and 4-6 m. We used the upper limit of of resource use by birds in shade plantations. each height category to calculate the median maximum Therefore, the objective of this study is to de- coffee height based on the two highest categories in each plot. scribe where and how birds forage in shade Plantations were visited during three periods (24 coffee plantations to determine the food re- October-16 November 1994, 9 January-l February sources utilized by birds in shade coffee plan- 1995, 13 March-5 April 1995). Foraging observations tations in the Dominican Republic on the Ca- were made most often between sunrise and noon, but ribbean island of Hispaniola. These observa- some observations were made in the late afternoon. tions will allow us to evaluate the imnortance Observations were made by walking slowly though a I of the shade overstory relative to the coffee plantation until a foraging bird was located. A foraging event was recorded five seconds after an individual understory in attracting birds to coffee plan- was detected to avoid a bias toward the more conspic- tations. uous feeding techniques such as aerial maneuvers. To characterize foraging maneuvers, only a single forag- STUDY SITE AND METHODS ing event was recorded from an individual during a Fourteen shade coffee plantations (mean = 1.2 ha, morning to avoid the problem of auto correlation-u- range 0.1-8.7 ha) were sampled in the vicinity of Man- herent in sequence data (Wagner 1981). The height of abao (19” 6 ’ N, 70” 48 ’ W) and Jarabacoa (19” 9 ’ N, the location of the first maneuver was measured with 70” 39 ’ W), La Vega province, at an altitude of 560- an optical range finder. We tried to observe as many 840 m in the Cordillera Central of the Dominican Re- different individuals as possible, but for the rarer spe- public. The plantations are located in a zone receiving cies we estimate that a maximum of 20% of the total an annual rainfall of approximately 1200 mm per year observations were derived from the same individual. (Hartshorn et al. 1980) and classified as subtropical For the very abundant Bananaquit (CoerebaJlaveoZa), moist forest in the Holdridge life zone system (Anon- it was necessary to limit a mornings’ observation to a ymous 1967). One hundred meter transects run in the maximum of 3 individuals. Ovenbirds (&Gurus auro- four major cardinal directions from each plantation cupilus) were common ground foragers in the planta- (Wunderle and Latta, unpublished data) indicate that tions (Wunderle and Latta 1996), but difficult to ob- 75% of the land area surrounding the plantations is serve and therefore it was not possible to include non-forest (67% pasture and 8% cultivation of low ground foragers in this study. ground crops). Tree canopies cover 20% of the sur- Foraging maneuvers were classified into three major rounding area (13% wooded arroyo; 5% coffee plan- categories. We use the term “glean” to designate all tations; 2% pine) and the remaining 5% of the land “near-perch” maneuvers in which the forager remains area contains a miscellaneous mix of roads, buildings, on a perch and picks a food item from the substrate rivers, etc. surface. This category includes all forms of reaching, We used 27 16-m diameter circular plots (0.02 ha) hanging, and lunging. The term “probe” is used to to quantify vegetation distribution in the plantations. designate all “near-perch” maneuvers in which a for- A single plot was placed in the center of each small ager remains on a perch and obtains a food item from plantation (0.1-0.6 ha), and three to seven plots were the under surface by probing, gaping, or pecking. Fi- placed at 100 m intervals in the larger plantations (2.S nally, we use the term “hover” to include all aerial Wunderle and Lam l AVIAN RESOURCE USE IN SHADE COFFEE 273 maneuvers in which the forager obtains a food item by TABLE 1. Species composition, basal area, and leaving the substrate by leaping, or flying. While hov- importance value of plants in shade coffee plantations ering, a forager can obtain food from a surface or the near Manabao and Jarabacoa, Dominican Republic. air by gleaning or from the under surface by probing. The times between successive foraging maneuvers Density De$ity in the shade overstory and coffee understory were ob- of stems, stems, BaSd Impor- tained by continuously dictating foraging observations <3 cm 23 cm area tance DBH DBH of stems value of into a cassette tape recorder. Each forager was fol- (stems/ (stems/ 23 cm stems lowed for a maximum of 5 min or until it disappeared. Species ha) ha) (m%a) 23 cm Using a stop watch we calculated the average time Coffeea arabica 6907 744 2.26 14.74 between foraging maneuvers for each individual in Inga “era 50 206 10.73 69.99 which three or more successive foraging maneuvers Muss spp. 0 133 1.45 9.50 were observed. The median time between each for- Citrus spp. 58 52 0.15 0.98 aging maneuver was based on the means for each in- Persea americana 0 9 0.37 2.40 dividual in the particular habitat (coffee or overstory). Syzygiumjumbos 0 6 0.31 2.02 Nonparametric statistical tests were used throughout Roystoneasp. 0 2 0.04 0.26 the study because the data did not fit the assumptions Psidium guajava 0 2 0.01 0.07 of normality (Hollander and Wolfe 1973). Because of Guarea pidonia 0 2 0.01 0.07 the absence of normality we used medians to describe central tendency and in the analyses, although we do provide means for descriptive purposes (foraging heights). Mann-Whitney U-tests were used to compare understory, but constituted only a small pro- the median times between foraging maneuvers. A one- portion of the total basal area. Subcanopy tailed Wilcoxon Rank Sum Test was used to test the trees included Citrus spp. and guava (Psidium null hypothesis that the median foraging height of a guujava). Znga vera (Mimosoideae) predomi- species was not above the median maximum coffee height. The variation in a species ’ median foraging nated in the overstory and accounted for over height and its abundance in point counts or net cap- half the total basal area of the plantations. The tures (Wunderle and Latta 1996) was compared using maximum canopy height averaged 15.6 2 3.8 a Spearman Rank Correlation. Multidimensional scal- (SD) m and ranged from 9.2-22.0 m among ing (MDS) provided a quantitative method for defining the 27 plots. Other canopy trees such as rose- foraging guilds useful for comparisons with other sites. apple (Syzygium jumbos), royal palm (Roys- Multidimensional scaling was used to visualize the for- tonia sp.), American muskwood (Guarea gui- aging dissimilarities between 19 bird species in the plantations based on their proportional use of three dif- donia) and Caribbean pine (Pinus caribaea) ferent foraging substrates (&a “era, coffee, other) and were rare in the plantations, but were more three food types (nectar, fruit/seed, animal). Dissimi- common along the edges of some plantations. larities were computed indirectly by use of Euclidean We observed a total of 24 species foraging distances, which were standardized before MDS based in the plantations, but restricted our analysis on the Kruskal method with monotonic regression to 19 species for which we had adequate sam- (SYSTAT 1992). All analyses were conducted using ple sizes. Of the 19 species, 13 were perma- SYSTAT (ver. 5.3) on a Macintosh computer. nent residents, five were winter residents (Ne- arctic migrants), and one was a spring-sum- RESULTS mer resident (Neotropical migrant). As expected, coffee had the greatest stem Foraging rates.-Adequate foraging rate density in our plantations, although its contri- samples were obtained for only two species of bution to the total basal area was relatively Nearctic migrants. Median time between for- small because most stems were less than 3 cm aging maneuvers in Black-throated Blue War- DBH (Table 1). The predominant variety of blers (Dendroica caerulescens) did not differ coffee (Coffea arabica) in the shade planta- significantly (Mann-Whitney U = 177, P > tions was the traditional “tipica” variety, al- 0.05) between coffee (median = 6 sec., n = though “catorra” predominated in some of the 15) and Znga (median = 6 sec., n = 27). How- larger plantations and was used to replace ever, the American Redstart (Setophaga ruti- “tipica” in some of the smaller plantations. cilia) had significantly (Mann-Whitney U = The median maximum height of coffee in our 253.5, P < 0.001) shorter intervals in Znga plantations was 4.0 m, which defined the limit (median = 7 sec., n = 25) than in coffee (me- of the plantation understory. Bananas and dian = 11 sec., n = 12). plantains (Muss spp.) were also found in the Foraging heights.-The shade overstory 274 THE WILSON BULLETIN - Vol. 110, No. 2, June I998

EISPANIOLAN WOODPECKER GREATER ANTILLFAN PEWEE PALM CHAT

x =12.1* 3.5 X=8.0*3.3 x =I,.1 = 3.6 M=l2 M=8 hi=,* N=58 N-54 N=88

x = 10.7 * 3.0 x = 10.3* 3.1 x =11.0’28 x =10.2= 3.7 14-M M=ll M=l3 M=K? M= 10.8 u-14 N=32 N=49 N=41 N=116

FIG. 1. Foraging height and location of first foraging maneuver after at least five seconds of observation for species in shade coffee plantations in the Dominican Republic. Foraging locations are given in legend. x refers to mean k SD, M to median, and N to number of observations.

was the most important foraging site for al- (Phaenicophilus pulmurum). Only the Nar- most all species, with 18 of the 19 species row-billed Tody (Todus ungustirostris) for- having median foraging heights significantly aged mostly in the coffee understory (67% of above the median maximum coffee height of observations) as shown by its median foraging 4.0 m (one-tailed Wilcoxon tests, all P < height which was below the median maximum 0.01). All species were observed foraging in coffee height (one-tailed Wilcoxon test: Z = the Znga overstory, although they differed in -3.61, P < 0.001). Although it did not feed how they divided their time between Znga and in the coffee, the Greater Antillean Pewee the coffee understory (Fig. 1). Eight species (Contopus caribeu) sometimes perched on the (42% of 19 species) foraged exclusively in the coffee and fed in the airspace between the cof- canopy or subcanopy and were not observed fee and the Zngu canopy. in the coffee understory. These canopy dwell- Foraging height and presence in sun cof- ers included Hispaniolan Woodpecker (M&e- fee.-We expected that species which foraged nerpes striatus), Hispaniolan Lizard Cuckoo primarily in the shade overstory above the (Suurotheru longirostris), Palm Chat (Dulus coffee would be less abundant in sun coffee dominicus), Black-cowled Oriole (Zcterus plantations than species which foraged in the dominicensis), Black-whiskered Vireo (Vireo coffee below the shade overstory. We tested ultiloquus), Black-and-white Warbler (Mni- this expectation by comparing the median for- otiltu vuriu), Northern Parula (Purulu umeri- aging heights of the 19 foragers with their cunu), and Black-crowned Palm Tanager abundance in point counts and mist net cap- Wunderle and Lana l AVIAN RESOURCE USE IN SHADE COFFEE 275

PALM TANAGER x =10.5 * 3.5 = 10 =ba

x E 13.3 * 1.9 M=13 N=38

38 0 10 20 38 0 10 lo xl 10-20 PERCENT USE 1618

M-16

u-14 =7..8 *2.8

IO-12

8-10

6-8

4.6

z-4

o-2

0 10 20 30 40 50 60 FIG. 1. Continued. tures previously obtained in nearby sun coffee aging observations for Vermin Hummingbird plantations (Wunderle and Latta 1996). As (Mellisuga minima), Black-cowled Oriole, predicted, we found significant negative cor- Bananaquit, and Hispaniolan Emerald (Chlo- relations between median avian foraging rostilbon swainsonii). The next most com- height in our shade plantations and abundance monly used substrate was the Znga wood in sun coffee point counts (Spearman r = (trunk, branch, twig), upon which 58% of the -0.62, df = 17, P = 0.005) and mist net sam- birds foraged. Znga wood accounted for over ples (Spearman r = -0.58, df = 17, P = half the observations of Hispaniolan Wood- 0.009). Therefore, the higher a species forages pecker (branches and trunk) and Black-and- in shade coffee plantations the less abundant white Warbler (branches and twigs). Less than it is likely to be in sun coffee plantations. half (47%) the species were observed feeding Foraging substrate.-Within Znga trees, the on or from Znga epiphytes. Epiphytes were leaves were the most commonly used foraging gleaned or probed by 42% of all species but substrate, on which 95% of all species (n = only 16% of all species consumed epiphyte 19) foraged at least some of the time for in- fruits. Epiphyte foraging constituted only a vertebrate prey (Fig. 2). Znga leaves were the small portion of a species ’ foraging visits, as major foraging substrate (> 50% of observa- evident in the two species which foraged most tions) for Black-whiskered Vireo, Northern commonly on epiphytes: Stripe-headed Tana- Parula, Broad-billed Tody (To&s subulatus), ger (Spindalis zena; 36% of all observations, and Hispaniolan Lizard Cuckoo. Znga flowers all fruit) and Black-crowned Palm Tanager were the next most important foraging sub- (16% of all observations; 4 invertebrate, 6 strate visited by 63% of the species. Znga fruit). Znga fruits were not present during most flower visits accounted for over half the for- of the study. Perhaps as a result, they were the 276 THE WILSON BULLETIN l Vol. 110, No. 2, June 1998

HISPANIOLANL.zARn CucKoo HIsPANIOLANEMmALo AmxLEANMANoo V!ZR”AUi HUMMINGBIRD

N=47

BLACK-WHISKERED VIREO BLACK-m-warn WARBLER NORTHERN P.4R”l.A INGA FLOWER INGA FRUIT N=85 N=48 JNGA LEAF INGA WOOD INGA RPIPHYTE OTHER FLOWER OTHER FRUIT OTHER LEAF OTHRR WOOD

PERCENTAGE OF OBSERVATIONS

FIG. 2. Foraging substrates and three foraging maneuvers used by birds in shade coffee plantations in the Dominican Republic. Each observation (N) based on first substrate and maneuver used after at least five seconds of observation in a morning.

least commonly used substrate on the tree, were visited by 52% of the species, but at used by only 26% of the species (mostly for most constituted only about 10% of a species ’ invertebrate prey?), and at most, constituting diet (e.g., Antillean Mango, 11%; Bananaquit, only 9% of a species ’ foraging substrate (i.e., 10%). Syzygium flowers were the most com- Hispaniolan Woodpecker). monly used of the less abundant plant species Within coffee, the leaves were also the most (visited by 39% of the species). Fruits other commonly utilized substrate as evident in cof- than those of coffee and Znga were also con- fee leaf use by 42% of the 19 species. Coffee sumed by 52% of the species, but constituted leaves were especially important for Narrow- an appreciable portion of the diet in only two billed Tody (62% of observations), but con- species (Palm Chat, 39%; Stripe-headed Tan- stituted less than a quarter of the foraging ob- ager, 33%). The most important fruits ap- servations in the others (American Redstart, peared to be those of Roystonea (consumed 24%; Black-throated Blue Warbler, 14%; His- by 28% of the species) and Miconia (con- paniolan Emerald, 4%). The remaining parts sumed by 22% of species). In addition to of the coffee were used by only a relatively leaves of coffee and Znga, 52% of the bird small percentage of the species (coffee wood, species fed from the leaves of a variety of 21%; coffee fruit, 21%; coffee flower, 5%). other less abundant plants (8 plant species). In addition to substrates in Znga and coffee, At most these constituted only 12% of a spe- birds also foraged on substrates of the less cies ’ foraging time (American Redstart). The abundant plant species. For example, flowers most important leaves were those of Citrus of other species (excluding coffee or Znga) spp. which were used by 21% of the species. Wunderle and Latta l AVIAN RESOURCE USE IN SHADE COFFEE 277

CAPE MAY WARBLER BLACK-THR BLUE WAFSLER AMERICAN REDSTART UiGA FLOWER MGA FRuT N=179 PI=86 UsA LE4w R-GA WOOD INGA EPmlYTJt OTHER FLOWER 1 co~~~~

coFFEJ?,FlwlT maEAN COFPEELEAF PROBE E COFFEE WOOD b I’) STRIPEHEADED TANAGER BLK.-CROWNED PALM TANAGER BLACKCOWLED ORIOLE

INGA FRlm N=33 N=64 N=37 2 INGA LEAF INGA WOOD INGAEPIFWYTE OTDJIR FLOWER OTJJERFRwr OTHER LEAF OTHER WOOD

COFFEE FRvlT COFFEELEAF COFFEE WOOD b

0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 PERCENTAGE OF OBSERVATIONS

FIG. 2. Continued.

Foraging on woods other than coffee and Znga vertebrates on leaf surfaces and in the air, al- was uncommon and occurred in only 32% of though gleaning was also used. The leaps pre- the species and never constituted more than dominantly used by Hispaniolan Lizard Cuck- 2% of a species ’ foraging time. oos to capture both lizards and invertebrate Foraging maneuvers.-Individual species prey were classified as hovers, although cuck- were usually consistent in their foraging ma- oos also gleaned prey from wood and leaf sur- neuvers when feeding on similar substrates in faces. coffee and Znga (Fig. 2). The only exception Gleaning was used by 68% of the species, was the leaf probe maneuver used to obtain although it was the predominant maneuver in invertebrate prey from rolled or bound Znga only 32% of the species. Black-and-white leaves by Northern Parula, Black-throated Warblers gleaned mostly from wood surfaces Blue Warbler, Bananaquit, Black-crowned or epiphytes on wood surfaces (lichens). Both Palm Tanager, and Black-cowled Oriole. Leaf Black-whiskered Vireo and Northern Parula probing was absent in birds foraging in coffee, commonly gleaned Znga leaves and, less fre- because there were no rolled or bound coffee quently, hovered to take invertebrates from leaves. leaf surfaces. Gleaning prey from a variety of Species differed in their use of various for- substrates with some probing characterized aging maneuvers while feeding in the planta- Black-crowned Palm Tanagers. Gleaning was tions. For example, hovering was found in the predominant maneuver used by Palm 68% of the species, but it was the predominant Chats, mostly to remove fruit, but they some- maneuver in only 42% of the species. Some times probed Znga flowers. Although gleaning species used hovering exclusively, such as was the predominant maneuver used by hummingbirds which probed flowers or Black-throated Blue Warblers, they also hov- picked invertebrates from the air or leaf sur- ered and probed a variety of substrates and faces. Hovering also was the only maneuver were one of the most versatile foragers in the used by both species of todies as they gleaned plantations (with Cape May Warbler, Den- invertebrates from the undersides of leaves, droica tigrinum, see below). and was the predominant maneuver used by Probing while perched on the substrate oc- Greater Antillean Pewees to capture insects in curred in 53% of the species, but it was the the air. Hovering was the predominant maneu- predominant maneuver in only 26%. If the ver used by American Redstarts to capture in- flower probes of hovering hummingbirds are 278 THE WILSON BULLETIN * Vol. I IO, No. 2, June I998

1 INGA primarily in the “other” plant category from CANOPY INSECTNORES those foraging in Znga and coffee. Tight clus- ters were apparent for nectarivores, which fed mostly in the canopy (Antillean Mango, His- paniolan Emerald, Vervain Hummingbird, Bananaquit, Black-cowled Oriole) and canopy

OTHER insectivores (Black-crowned Palm Tanager, SUBSTRATES -1 Hispaniolan Woodpecker, Hispaniolan Cuck- B . oo, Broad-billed Tody, Black-whiskered Vir- R eo, Black-throated Blue Warbler, Northern Pa- COFFEE rula, Black-and-white Warbler). Palm Chats FRUIT , NECTAR ANIMAL -2 and Stripe-headed Tanagers clustered together -2 -l. 0 1 FIRST AXIS on the first two axes because of their tendency FIG. 3. Multidimensional scaling on two axis of to feed on nectar from Znga flowers and con- foraging birds in shade coffee plantations in the Do- sume fruit from other plants. minican Republic, based on three foraging substrates (coffee, other substrates, fnga vera), and three food DISCUSSION types (fruit/seed, nectar, animal). Letters refer to the Given the preponderance of foraging ob- following species: A. Stripe-headed Tanager, B. Palm servations in the Znga overstory, it is apparent Chat, C. Bananaquit, D. Vervain Hummingbird, E. Hispaniolan Emerald, E Antillean Mango, G. Black- that the characteristics of Znga veru strongly cowled Oriole, H. Cape May Warbler, I. Black- influence the foraging behavior and presence crowned Palm Tanager, J. Black-whiskered Vireo, K. of birds in these plantations. Our finding that Hispaniolan Woodpecker, L. Northern Parula, M. His- all species fed from Znga leaf surfaces at least paniolan Lizard Cuckoo, N. Black-and-white Warbler, some of the time, suggests that the Znga leaves 0. Broad-billed Tody, l? Black-throated Blue Warbler, in our plantations hosted an abundant inver- Q. American Redstart, R. Narrow-billed Tody, S. Greater Antillean Pewee. tebrate fauna. Indeed, Znga leaves are known to host a variety of invertebrates including fo- livores such as grasshoppers, katydids, lepi- included, then probing occurred in 68% of the dopteran larvae, and beetles and others that species, with 42% using it as the predominant bind together leaves including spiders, small maneuver. Probing was the predominant ma- orthopterans, skipper larvae, and microlepi- neuver for Black-cowled Orioles, Banana- doptera (Koptur 1983). Although the extraf- quits, and Cape May Warblers, primarily for loral nectaries on Znga leaves may attract var- nectar in flowers, but sometimes used for ob- ious ants, parasitic wasps, and flies that deter taining fruit juices. Hispaniolan Woodpeckers herbivores (Bentley 1977, Koptur 1983), no mostly probed wood, but also Znga fruit and invertebrates were detected during our inspec- flowers of Syzygium. tions of the nectaries on both young and old Foraging ordination.-Multidimensional leaves. Caterpillars and leaf damage were of- scaling (MDS) enabled us to visualize dissim- ten evident in the canopy and at times the ilarity among the 19 bird species based on frass rain was audible in some plantations. their proportional use of foraging site and Not surprisingly, several bird species were ob- food type (Fig. 3). The first MDS axis sepa- served feeding on caterpillars in the Zngu can- rated species primarily on the basis of diet, opy. with those species taking a high proportion of Flowers of Znga veru have nocturnal anthe- animal prey at one extreme on the axis in con- sis and are visited at night by bats and hawk- trast to those with a high proportion of plant moths (Salas 1967, 1974), but evidently nectar material on the other (nectarivory and frugi- remains in the morning when most bird visits vory). The second axis separated foragers on were observed. Znga vera flowers have been the basis of foraging site, with species forag- found to be attractive to a variety of avian ing mostly in coffee at one extreme of the nectarivores during restricted flowering peri- axis, in “other” plant category in the center, ods, at least in areas of seasonal rainfall (Wolf and in Znga at the other extreme. The third 1970, Greenberg et al. 1997b). In our planta- axis only weakly separated species foraging tions, some Zngu flowers were present Wunderle and Latta - AVIAN RESOURCE USE IN SHADE COFFEE 279 throughout the duration of our study, but flow- when captured. Thus, it appears that foraging ers were most abundant in March-April (pers. birds use coffee plants primarily to obtain in- obs.). Despite this seasonal variation in Znga vertebrate prey. flower abundance, the most abundant nectar- Despite the preponderance of foraging for ivores visited flowers in the same proportions invertebrates in the coffee, most birds foraged (relative to insectivory) throughout the study infrequently in coffee, and for American Red- period, although the numbers of nectarivores starts the foraging rates were lower in coffee increased with an increase in flowering (pers. than in the Znga overstory. These findings un- obs.). The March-April flowering peak in this doubtedly are due to relatively low inverte- study was later than in the previous two sea- brate densities on the coffee plants as noted sons when flowering peaked in January. As a by Greenberg and coworkers (1997a) who result of the later flowering, Znga fruiting was found arthropod biomass per 100 g leaf bio- also delayed and most fruits had not ripened mass to be approximately 6 times greater for by the end of our study. Therefore it is likely Znga than shade coffee. Evidence for low den- that Znga fruit consumption was under-repre- sities is also supported by our observations sented in this study relative to previous years that insect damage to coffee leaves was rare when we occasionally encountered parrots and in contrast to the obvious damage in the Znga parakeets feeding on Znga pods during the canopy (pers. obs.). Indeed, others have noted same time period. that insect damage to coffee leaves is unusual Epiphytes and plant parasites growing on (Le-Pelly 1973), in part because young coffee the Znga were not abundant in our plantations, leaves have an abundance of protective alka- which undoubtedly explains their relatively loids and the older leaves are very tough low levels of use by most birds. Lichens were (Frischknecht et al. 1986). Also, it is likely the most abundant epiphytes and were probed that the transfer of coffee to the Neotropics or gleaned for invertebrates by several species left behind many of the Old World herbivores including Black-and-white Warbler, Hispani- that evolved mechanisms to overcome its de- olan Emerald, Black-throated Blue Warbler, fenses, as often occurs with the transfer of cul- and Hispaniolan Lizard Cuckoo. Bromeliads tivated plants (Perfect0 et al. 1996). Our own were probed for invertebrates by Black- surveys of invertebrate abundance on coffee crowned Palm Tanagers, Hispaniolan Wood- leaves in these plantations (Wunderle and Lat- peckers, Bananaquits and Greater Antillean ta 1996) indicate levels of abundance three Grackles (QuiscuZus niger). Mistletoe fruits times lower than are found in native, moist were consumed most frequently by Stripe- broadleaf forest elsewhere in the Caribbean headed Tanagers, but were also taken by (Askins and Ewert 1994). Black-crowned Palm Tanagers and Black- Given the importance of the overstory to cowled Orioles. Also, Antillean Euphonias avian foragers in our plantations, it is not sur- (Euphonia musica) probably would have been prising that canopy dwellers were less abun- more abundant in our plantations had mistle- dant in the absence of a canopy in nearby sun toe been more common given the reliance of coffee plantations (Wunderle and Latta 1996). euphonias on mistletoe fruit (Snow 1981). For instance, of the 9 species that foraged ex- Almost all of the foraging maneuvers in clusively in the Znga overstory, seven (Black- coffee were associated with insectivory (in- cowled Oriole, Black-whiskered Vireo, Great- cludes insects and spiders). This preponder- er Antillean Pewee, Hispaniolan Lizard Cuck- ance of insectivory is consistent with mist net oo, Palm Chat, Northern Parula, Black-and- results from the same plantations (Wunderle white Warbler) were absent from point counts and Latta 1996). These results, based on diet in sun coffee; Hispaniolan Woodpeckers were classification from the literature, indicate that present in the sun coffee, but significantly less 50% of the 358 birds captured in the coffee frequently; and Black-crowned Palm Tanagers were primarily insectivorous compared with were less frequent in sun coffee, but not sig- 43% nectarivores and 6.7% fruit/seed eaters. nificantly. Our observations indicate that nec- Given our foraging observations, it is likely tarivores (Vervain Hummingbird, Hispaniolan that most of the captured nectarivores were Emerald, Antillean Mango, Bananaquit, Cape actually feeding on invertebrates in the coffee May Warbler) mostly fed on Znga flowers in 280 THE WILSON BULLETIN l Vol. 110, No. 2, June I998 the shade plantations, and as expected in the species. Fruits of the Melastomataceae are absence of flowering, each showed signifi- also known to be important for a variety of cantly lower abundance in sun coffee. Insec- bird species (Snow 1981), but their rarity too tivores that fed mostly in the Znga, but de- may have limited our observations to only 4 scended into the coffee were also found in the species, all of which fed elsewhere in the sun coffee at counts equivalent to (American plantations. Syzygium flowered throughout the Redstart, Broad-billed Tody), or significantly duration of the study and was visited by 7 higher (Black-throated Blue Warbler) than species, all of which were common at Znga counts in shade plantations. Finally, the insec- flowers. tivorous Narrow-billed Tody foraged mostly The results of our study suggest that a cof- in coffee and was also significantly more fee plantations’ attractiveness to birds is en- abundant in sun than shade plantations. Thus, hanced with the abundance, variety, and con- species most dependent on the shade oversto- sistency of food resources in a plantation. This ry for foraging are most likely to decrease in finding applies mostly to the shade overstory the absence of a shade overstory in sun plan- or plantation border, as the coffee plants alone tations, while those that can forage in the cof- provide limited food resources or foraging fee may be present, if not abundant, in some substrates for most species in this study. A sun plantations (assuming low or no pesticide diversity of plant species in the overstory or use). border attracts more birds, particularly if the Other cultivated crops in our plantations plants are asynchronous in flowering and provided foraging opportunities, but they did fruiting and extend the time period over which not appear to be utilized disproportionately in resources are available in a plantation. Given relation to the plants ’ abundance. For exam- these traits, shade coffee plantations can pro- ple, bananas and plantains, despite their rela- vide food resources for a variety of bird spe- tive abundance in the plantations were infre- ties in agricultural areas, and thereby help quently used by foraging birds. The leaves maintain avian abundance and diversity in ag- were only rarely gleaned for invertebrates and ricultural regions. Furthermore, these findings the flowers sometimes visited for nectar (An- suggest that shade coffee plantations could tillean Mango, Bananaquit, Cape May War- serve as an effective buffer between forest re- bler). Other plants such as citrus, avocado, serves and surrounding agricultural land- and guava appeared to be utilized by foragers scapes. in direct proportion to their abundance in the plantation. Although these crops may not have ACKNOWLEDGMENTS attracted additional bird species to the plan- We gratefully acknowledge the fine assistance pro- tations, they may contribute to a plantations’ vided by Teodoro Lara, Esteban Terranova, and Ed- attractiveness to birds by providing resources uardo Vgzquez. We thank the numerous coffee plan- out of synchrony with the Znga overstory. In tation owners for permission to work in their planta- addition such sub-canopy trees as citrus and tions. The manuscript benefited from the constructive comments of John Francis, Russell Greenberg, John avocado provide additional foliage layers be- Rappole, Fred Scatena, and an anonymous reviewer. low the Znga canopy that serve as additional Funding was provided by the National Fish and Wild- foraging substrate for insectivores. life Foundation and the John T and Catherine C. Mac- Some of the non-agricultural plants in the Arthur Foundation. plantations appeared to be important in attracting birds to the plantations. For example, LITERATURE CITED royal palms attracted Palm Chats and Black- AGUILAR-ORTIZ, E 1982. Estudio ecoldgico de las cowled Orioles. Both species were present aves de1 cafetal. Pp. 103-128 in Estudios Ecol& only where palms occurred in the immediate gicos en el agroecosistemacafetalero (E. Avila Ji- vicinity, presumably because of nesting re- menez, Ed.) Inst. National de Investigaciones so- quirements, although Palm Chats frequently bre Recursos Biciticos, MCxico. ANONYMOUS. 1967. Reconocimiento y evaluaci6n de fed on palm fruit. Royal palm fruits are con- 10s recursos naturales de la Reptiblica Domini- sumed by a diversity of species (Wunderle, cana.Unidn Panamericana,Secretaria General Or- unpubl. data), and the rarity of palms limited ganizacidnde 10s Estados Americanos, Washing- our fruit consumption observations to only 5 ton D.C. Wunderle and Luffa * AVIAN RESOURCE USE IN SHADE COFFEE 281

ASKINS, R. A. AND D. N. EWERT. 1994. Wildlife re- ical moist forests. Case studies and guidelines. In- search, 1993-94. Annual letter, Int. Inst. Tropical ternational Union for Conservation of Nature and For., Rio Piedras, Puerto Rico. Natural Resources, Gland, Switzerland BEEHLER, B. M., S. R. KRISHNA RAJU, AND S. ALI. PERFECTO,I., R. A. RICE, R. GREENBERG,AND M. E. 1987. Avian use of man-disturbed habitats in the VAN DER VOORT. 1996. Shade coffee: a disap- eastern Ghats, India. Ibis 129: 197-211. pearing refuge for biodiversity. Bioscience 46: BENTLEY, B. L. 1977. Extrafloral nectaries and pro- 598-608. tection by pugnacious bodyguards. Annu. Rev. RICE, R. A. AND J. R. WARD. 1996. Coffee, conser- Ecol. Syst. 8:407-427. vation, and commerce in the western hemisphere. BORRERO,J. I. 1986. La substitution de cafetales de Smithsonian Migratory Bird Center, Washington, sombrio por caturrales y su efecto negativo sobre D.C. la fauna de vertebrados. Caldasia 15:725-732. RICE, R. A., A. M. HARRIS, AND J. MCLEAN (Eds.). FRISCHKNECHT,P. M., J. V. DUFEK, AND T W. BAU- 1997. Proceedings of the the first sustainable cof- MANN. 1986. Purine alkoloid development in fee congress. Smithsonian Migratory Bird Center, buds and developing leaflets of CofSeeaarabica: Washington D.C. an expression of optimal defense strategy. Phyto- SALAS, S. 1967. Znga sp. as a bat pollinated tree in chemistry 25:613. Caiias, Costa Rica. OTS project report, mimeo- GREENBERG, R. 1994. Coffee and birds. Smithson. graphed. San Jo&, Costa Rica. Mag. 25(11):24-27. SALAS, S. 1974. Anrilisis de1 sistema de polinizacidn de Znga vera ssp. spuria. M.S. Thesis, Univ. de GREENBERG,R., P BICHIER, A. CRUZ ANGON, AND R. Costa Rica, San Jo&, Costa Rica. REITSMA. 1997a. Bird populations in shade and SYSTAT 1992. Statistics, version 5.2 edition. SYS- sun coffee plantations in Central Guatemala. Con- TAT, Evanston, Illinois. serv. Biol. 11:448-459. SNOW, D. W. 1981. Tropical frugivorous birds and GREENBERG,R., l? BICHIER, AND J. STERLING. 1997b. their food plants: a world survey. Biotropica 13: Bird populations in rustic and planted shade. cof- l-14. fee plantations in eastern Chiapas, Mexico. Bio- THIOLLAY, J. M. 1995. The role of traditional agro- tropica 29:501-514. forests in the conservation of rain forest bird di- GRISCOM, L. 1932. The distribution of bird-life in versity in Sumatra. Conserv. Biol. 9:335-353. Guatemala. Bull. Am. Mus. Nat. Hist. 64:1-439. VANNINI, J. P. 1994. Nearctic avian migrants in coffee HARTSHORN, G., G. ANTONINI, R. DUBOIS, D. HAR- plantations and forest fragments of south-western CHARIK, S. HECKADON, H. NEWTON, C. QUESADA, Guatemala. Bird Conserv. Int. 4:209-232. J. SHORES,AND G. STAPLES. 1980. The Dominican WAGNER, J. L. 1981. Visibility and bias in avian for- Republic, country environmental profile. U.S. aging data. Condor 83:263-264. Agency for International Development, Washing- WETMORE, A. 1916. Birds of Port0 Rico. U.S. Dept. ton, D.C. of Agric. Bull. No. 326. HOLLANDER, M. AND D. A. WOLFE. 1973. Nonpara- WILLE, C. 1994. The birds and the beans. Audubon metric statistical methods. John Wiley & Sons, 96:58-64. New York. WOLF, L. L. 1970. The impact of seasonal flowering KOPTUR, S. 1983. Znga. Pp. 259-261 in Costa Rica on the biology of some tropical hummingbirds. natural history (D. H. Janzen, Ed.). Univ. of Chi- Condor 72: l-14. cago Press, Chicago. WUNDERLE, J. M., JR., AND S. C. LATTA. 1996. Avian LE-PELLY, R. H. 1973. Coffee insects. Annu. Rev. abundance in sun and shade coffee plantations and Ent. 18:121-142. remnant pine forest in the Cordillera Central, Do- OLDFIELD, S. 1988. Buffer zone management in trop- minican Republic. Ornitol. Neotrop. 7:19-34.