JOURNAL OF FIELD ORNITHOLOGY Publishedby Associationof Field Ornithologists

VoL. 60, No. 3 SUMM•.V•1989 Pacts 289-420 j. Field Ornithol., 60(3):289-295

COLONY SIZES AND NEST TREES OF MONTEZUMA OROPENDOLAS IN

ROSENDO M. FP.^•^ SmithsonianTropical Research Institute Apartado2072, Balboa,Panamd Abstract.--I examined 36 coloniesof Montezuma Oropendolas(Psarocolius montezuma) to checkwhether colonysize was related to characteristicsof nesttrees. Colonies were spread over 46 trees and containeda total of 1109 nests(range 3-172 nestsper colony,median 21.5). Numbersof nestsin multiple- and single-treecolonies were not significantlydifferent. Thirteen species(11 dicots,2 palms)were used as nesting trees, but 4 nativespecies contained 82% of all nests.Above 500 m coloniesin palms containedfewer nestsper tree, but more trees per colony than coloniesin dicot trees. At lower elevationsthe number of nestsin palms was similar to the number in dicot trees. Six nestingtrees (of 2 species)had wasps.Trees of the samespecies with or without waspshad similar numbersof Oropendola nests.

TAMAlqlOS DE COLONIAS Y CARACTERiSTICAS DE LOS J•RBOLES EN DONDE ANIDA PSAROCOLIUS MONTEZUMAE EN COSTA RICA Resumen.--Examin• 36 coloniasde la orop•ndola?sarocolius montezumae para relacionar el tamafio de las colonias con caracteristicas de los firboles en donde anidaron. Las colonias utilizaron 46 firboles,yen estosconstruyeron 1109 nidos(rango 3-172 nidospor colonia, mediana21.5). No hubodiferencias significativas en el nfimerode nidosentre coloniasque usaronuno o variosfirboles. Trece especiesde plantas(11 dicotiled6neasy 2 palmas)se utilizaron para anidar; 4 especiesnativas contuvieronel 82% de los nidos. Las coloniasen palmascontentan menos nidos por firbol, pero usualmentemas firbolespor colonia.Las especiesde palmaso dicotiled6neasno difirieron en el numero de nidos por firbol. En 6 firboles(de 2 especies)habla avisperosde Synoeca.No hubo diferenciasen el numero de nidosentre firbolesde la mismaespecie con y sin avispas. Ecologicalfactors influencing coloniality and colonysize in oropendolas (Psarocoliusand relatedgenera) are not well known, with little quanti- tative information on colony size for most species.Colony size varies considerablyamong species and P. angustifronsmay nestsolitarily (Schafer 1957). Characteristicsof nest trees may affect colonysize. The idea is plausible,because some species of oropendolasand the related caciques (Cacicus)nest in different speciesof trees,at least locally (Drury 1962, Koepcke1972, Oniki and Willis 1983, Schafer1957). Presenceof wasps

289 290] R. M. Fraga J. Field Ornithol. Summer 1989 and beesin nesttrees has beenreported for severalspecies of oropendolas and caciques(Feekes 1981, Koepcke1972, Skutch 1954, Smith 1968). These insectsdeter predatorsand ectoparasites(Robinson 1985, Smith 1968). Nevertheless,there is no informationon how the presenceof wasps and beesinfluences colony size. The literature suggestsselectivity of tree speciesby nestingMontezuma OropendolasPsarocolius montezuma (Alvarez del Toro 1980, Crandall 1914), and at least one sourceindicates regular associationswith un- identifiedwasps (Richmond 1893). I report in this paper characteristics of Montezuma Oropendolacolonies in CostaRica (particularlynumbers and speciesof nestingtrees and presenceof wasps)and examinewhether these characteristicsaffected numbers of nests. Effects of geographical locationand habitat on colonysize were also investigated.

STUDY AREA AND METHODS MontezumaOropendola colonies were observedalong 315 km of roads in easternCosta Rica. I locatedcolonies while drivingslowly or by asking local residents.All colonieswere within 300 m of a road. Roads (195 km) in which slowdriving was difficult were scanned6-11 times.Vocal- izations of male oropendolashelped locate colonies.Colonies were ex- amined with the help of 8 x 30 binoculars.Only four nest trees could not be reached on foot. The study area extendedfrom Rio Frlo (provinceof Heredia) south to Bribrl (provinceof Lim6n) andwest to Turrialba (provinceof Cartago), and up to an altitude of 800 m. I dividedthe study area into four areas: (1) Turrialba, which included all colonieslocated 500 m above sea level (all in the provinceof Cartago) (2) Cahuita, including all locationsin the provinceof Lim6n southof the Rio Estrella (3) Lim6n, all locations in that provincebetween the Rio Estrella and Rio Reventaz6nand (4) Gufipiles,with the locationswest of the Rio Reventaz6n,and including the easternmostpart of the provinceof Heredia. I locatedcolonies at various dates between 26 May and 21 August 1987. The breedingseason of the speciesextends from January to Sep- tember (Skutch 1954). The first colonieswere seen at Turrialba, where I residedfor 10 d between26 May and 12 June 1987. Variables measuredfor each colonyincluded the habitat, speciesof tree, number of nestsand presenceof wasp nests.Trees were identified from severalsources (Allen 1977, Holdridge and Poveda1975, Sfinchez Vindas 1983), and from commonnames given by local residents. As in the literature, colonysize in this studywas the number of nests in the tree. Nests found on fallen branches were not counted. Females with nestson broken brancheswill renestin the sametree (Schafer 1957, Skutch1954). In three large coloniesthere were clustersof inactivenests that lookeddiscolored and worn, by comparisonwith the remainingnests. These nestscould have beenbuilt in pastbreeding seasons and were also excluded;they represented11 to 17% of the total number of nestsin the colonies.The low number of old nestswas probably related to the fact Vol.60, No. 3 MontezumaOropendola Colonies [291 that I startedmy study 7 mo after the end of the previousseason (see above).

RESULTS AND DISCUSSION Colonysize, geographicallocation and habitat.--I found 36 colonies, spreadover 46 trees,containing 1109 nests.The mean linear densitywas one oropendolacolony every 8.7 km. The highest density occurredat Cahuita, with six colonies (and 341 nests) in 19 km. The mean number of nestsper colonywas 30.9 (SD = 35.2), the median number 21.5, and the range 3-172 nests.The largestdifference in colonysize occurredbetween Cahuita (median = 34 nests)and Tur- rialba (median = 13 nests),perhaps reflecting a gradientbetween coastal plain and mountains.However, there were no significantdifferences in colonysize between the four subareas(Kruskal-Wallis statistic = 6.77, df = 3, P -- 0.08). The range of colonysizes reported from Costa Rica was 9-88 nests (Crandall 1914, Skutch 1954). Four coloniesin Co16n, , contained 10-41 nests (period 1964-1975), with a median of 23 nests(Neal Smith, pers. comm.). Within probablyno oropendolasurpasses P. monte- zuma in the size of its colonies.The maximum colonysize for P. wagleri in Central Panama was 132 nests(Neal Smith, pers. comm.). The commonesthabitats for colonieswere small plantations,mostly of cacaoor bananas(n = 13 colonies)or cattle pastures(n = 11). I found no significant differencesin colony size between these habits (Mann- Whitney, U = 56.5, 2-tailed, P • 0.10). Other locationsincluded large suburbangardens and lawns. I observed3 coloniesnear rivers only in Gufipiles. Six of 36 colonies(16.7%) were spreadover two or three trees (with a total of 16 trees). The averagedistance between the trees was 13.7 m (range6.5-22 m). One might expectmore nestsin multiple-treecolonies. Single tree colonieshad a median of 19.5 nests,multiple tree colonies 24.5 nests,but the differencewas not significant(Mann-Whitney U test, z = 1.02, 1-tailed, P -- 0.15). Oropendolasnesting in palms often formed multiple tree colonies(see below). A multiple-tree colonyof Montezuma Oropendolaswas reportedby Skutch(1954). This type of colonyoccurs in other oropendolas(e.g., ?. wagleri in Panama and Costa Rica, pers. obs.). Multiple tree coloniesspread over 100-200 m are particularly commonin P. angustifronsin ;this loosecoloniality probably affectsthe socialorganization of this oropendola(Schafer 1957). My informantsinvariably statedthat colonysites of Montezuma Oro- pendolaswere traditional. One colony I found at Cahuita was photo- graphed before 1983 (SfinchezVindas 1983). Four coloniesin Panama were in the sametrees for at least 7-10 yr (Neal Smith, pers. comm.). This site constancysuggests selection of specialcharacteristics of nest trees. Characteristicsof nest trees.--As reported in the literature (Crandall 1914, Skutch1954) treesselected by MontezumaOropendolas were usu- 292] R.M. Fraga J.Field Ornithol. Summer 1989

TABLE 1. Tree speciescontaining nests of Montezuma Oropendolas, and numbers of coloniesand nestsper species.Two trees were identifiedonly with commonnames.

Single tree Multiple tree Tree species colonies colonies Total nests

DICOTS Bursera sirearuba 8 3 271 Terrainalia oblonga 4 0 232 Ficus werckleana 4 2 229 Cordia alliodora 4 2 175 Licania arborea 1 0 75 Andira inerrnis 1 0 18 Albizzia caribaea 2 0 13 Rollinia microsepala 1 0 8 Cedrela mexicana 0 1 2 "Pejibayito" 1 0 12 "Guatarrama" 1 0 8

PALMS Bactrisgassipaes 2 3 41 Roystonearegia 1 5 23

ally isolated (I saw only two exceptions),had the shape of an umbrella, lackedlower limbs or branches,and mostlyranged 12-30 m high. I found little or no variation in these characteristics.The only spiny tree used wasthe pejibayepalm (Bactrisgassipaes). Richmond (1893) mentionsuse of deadtrees for nesting.I did not observethis, but five of six Cordiaused for nestingwere leaflessor almost so in May-August. I found Montezuma Oropendolasnesting in 13 speciesof trees(Table 1) 2 of which could only be identifiedby commonnames. Two species of palms were used, and both were cultivated. Otherwise no introduced trees were used. Most nests(82%) were found in only four speciesof trees (Bursera, Terrainalia, Ficus and Cordia, Table 1). The data would suggesthigh selectivityby oropendolas,but a comparisonwith published censusesof tree speciesin easternCosta Rican forestswould be misleading. Oropendolasnest in man-modifiedenvironments, and the speciesof large treesremaining in pasturesand plantationsare probablynot a random sampleof thoseoccurring in forests.Allen (1977) lists31 speciesof large trees(canopy and subcanopyspecies in Hartshorn and Poveda1983) as commonin pasturesin Costa Rica. Five speciesin the list were usedby oropendolas(including Cordia,Ficus, and Terrainalia,Table 1). Sometree specieswere usedfor nestingonly in restrictedareas. Colonies in Ficuswerckleana were seenonly at Cahuita, althoughthis fig is wide- spread in Costa Rica below 800 m (SfinchezVindas 1983). Pejibaye palms are commonlyplanted throughout Costa Rica (Vandermeer1983, pers.obs.) but I observedcolonies in this plant only at Turrialba. Skutch (pets. comm.)found in 1941 a colonyin this palm near Turrialba, sug- gestingpersistence of local traditions. Vol.60, No. 3 MontezumaOropenctola Colonies [293

Crandall (1914) mentions Ceibapentandra as the main nest tree at Curlpiles. I saw several isolatedCeiba, but only once was I told that oropendolashad usedone of them. I couldnot verify the information, as no nests remained. Bursera sirearuba and unidentified Ficus are often used by MontezumaOropendolas in Chiapas, (Alvafez del Toro 1980) agreeingwith my sample from Cahuita. Within the oropendolasthere are specialistsas regards nest trees (Psarocoliuslatirostris in Cecropia, Koepcke1972) and generalistswhich freely use exotictrees (P. angusti- frons, Schafer 1957). P. montezumaseems to occupy an intermediate positionin this spectrum. In Turrialba, palms had fewer nestsper tree (median = 6, n = 11) than dicot trees (median = 18, n = 5) (U = 10, 2-tailed, P < 0.05). Possiblyas a result of this, palms were more frequently part of multiple tree colonies (8 of 11 cases)than dicot trees (8 of 35 cases,P = 0.004, Fisher exacttest). The only multiple-tree colonyreported in the literature (Skutch 1954 and pets. comm.)was built on sevenpejibaye palms. Within broad taxonomiccategories (palms and dicot trees) I did not detectrelationships between tree speciesand numbersof nestsper tree. There were no significantdifferences in numbersof nestsper tree between Bursera,Ficus, Terrainalia, or Cordia (Kruskal-Wallis statistic = 1.51, df = 3, P = 0.68). The result applies even within subareas:I found no differencesin numbers of nestsbetween Ficus (n = 6) and Bursera (n = 6) in Cahuita (U = 10, P = 0.20). In Turrialba the palms Bactris(n = 5) and Roystonea(n = 6) containedcomparable numbers of nests(U = 24, P = 0.12). Tree specieswas not a major determinantof colonysize in this oropendola. Organizationofcolonies.--Montezuma Oropendolas nesting in 11 dicot treeswith wide crownsclustered their nestsin 35 separateclumps, leaving mostbranches unoccupied (cf., Skutch 1954). Clump size ranged from 1 to 48 nests,with a mean of 10.9 nests.In palm treesthere were usually 1-2 nestsper frond and the nestswere not clustered. I have little information on the behavioralsignificance of nest clumps. Femalesthat were building late nestsin May-June within three colonies at Turrialba and Cahuita were segregatedin discreteclumps (cf., Schafer 1957, Skutch 1954). Nest clumpingblurred the differencesbetween single and multiple-tree colonies,as clumpsin the sametree couldbe as distant as those in different trees. I obtainedinformation on numberof malesper colonyfor four colonies at Turrialba. Two colonies with three and 14 nests had one male, two colonieswith 26 and 47 neststwo and three males respectively.I saw only one male singingin the nest tree at any time in the last colonies. The other males were sometimessinging in peripheral trees. In three sequencessongs of central and peripheral males occurredat random (P > 0.05, runs test). Waspsin nesttrees.--Six nestingtrees (the two Albizzia and four of the 11 Bursera)had large and activevespiaries built by the "guitarr6n" wasp (Synoecaseptentrionalis, Richards 1978). The samplesare too small 294] R.M. Fraga J.Field Ornithol. Summer 1989 for a significancetest, but the presenceof wasps in nest trees was ap- parentlynot independentof the speciesof tree. If the six Synoecavespiaries were distributedat random in the 46 nesttrees (with probability = 0.13) their expectedoccurrence in Albizzia and Burserawould have been 0.26 and 1.43, respectively. MontezumaOropendolas could still be selectingtrees with wasps(e.g., selectingAlbizzia, if Synoecawas present).The spatial arrangementof oropendolanests and vespiarieswithin the treesdid not suggesta strong association.Only once(in Albizzia) was the vespiarywithin 3 m of the nearestnests. In one of the four Burserathe nestsand the waspswere in oppositebranches. In the other two Burseraonly a minority of the nests (33 and 37%) were at the distal end of a branch with wasps (i.e., a climbingpredator could have reached the remainingnests without meeting the wasps). The hypothesisof protectiveassociation would predicthigher numbers of nestsin the presenceof Synoeca.Within Bursera,nest trees with wasps had a median of 27 nests(n = 4) and thosewithout waspsa median of 18 nests(n = 7). Nevertheless,the differencein numbersof nestsbetween the samplesis non-significant(U = 15, 1-tailed,P = 0.46). The data of Crandall (1914) do not suggesta significantassociation between Mon- tezumaOropendolas and wasps,and Skutch(1954, and pers.comm.) did not observewasps in his colonies.There is no evidenceof a protective associationbetween Synoeca wasps and Montezuma Oropendolas. Colony size in Montezuma Oropendolaswas largely independentof the characteristicsof nest trees. Palms containedfewer nestsper tree, but mostcolonies in palms were spreadover many nest trees.in the related icterinesCacicus cela and C. haemorrhouscolony size is probably related to foodsupply and predationpressure (Feekes 1981, Robinson1985).

ACKNOWLEDGMENTS

j. Barborak,D. Duffy, S. Harding, A. Marin, L. Poveda,J. Sfinchezand G. Stileshelped me during the research.Many Costa Ricans gave me indicationsabout oropendolasand their nestingtrees. S. Patonprovided a statisticalpackage. N. Smith andJ. Karr commented on a draft of the manuscript.D. Windsor helpedme in wasp identification.The editor, J. Greenlaw and an anonymousreviewer providedhelpful comments.

LITERATURE CITED

ALLEN, P. H. 1977. The rainforests of Golfo Dulce. Stanford Univ. Press, Stanford, California. ALVAREZDEL TORO, M. 1980. Las avesde Chiapas.Universidad Aut6noma de Chiapas, Tuxtla Gutierrez, Mexico. CRANDALL,L.S. 1914. Notes on CostaRican birds. Zoologica(NY) 1:328-343. DRURY,W.H. 1962. Breedingactivities, especially nest building, of theyellowtail (Ostinops decumanus)in Trinidad, West Indies. Zoologica(NY) 47:39-58. FEEKES,F. 1981. Biologyand colonialorganizations of two sympatriccaciques, Cacicus celaand C. h. haemorrhous(Icteridae, Aves) in Suriname. Ardea 69:83-107. HARTSHORN,G. S., AND L. J. POVEDA. 1983. Checklistof trees. Pp. 158-183, in D. H. Janzen, ed. Costa Rican natural history.Univ. of ChicagoPress, Chicago. Vol.60, No. 3 MontezumaOropendola Colonies [295

HOLDRIDGE,L. R., ^ND L. J. POVEDA. 1975. Arboles de Costa Rica. Vol. 1. Centro CientificoTropical, San Josg,Costa Rica. KOE?CKE,M. 1972. Uber die Resistenzenformender Vogelnesterin einem begrentzen Gebiet des tropischenRegenwaldes in Peru. J. Ornithol. 113:138-160. ONI}CI,Y., ^ND E. O. WILLIS. 1983. Breeding recordsof birds from Manaus, Brazil: V. Icteridae to Fringillidae. Rev. Bras. Biol. 43:55-64. RICH^RDS,O.W. 1978. The socialwasps of the Americasexcluding the Vespinae.British Museum (Natural History), London. RICHMOND,C.W. 1893. Notes on a collectionof birds from Eastern Nicaragua and the Rio Frio, Costa Rica, with a descriptionof a supposednew trogon. Proc. U.S. Nat. Mus. Bull. 16:479-532. ROBINSON,S. K. 1985. Coloniality in the Yellow-rumpedCacique as a defenseagainst nest predators.Auk 102:506-519. SKNCHEZVIND^S, P.E. 1983. F16ruladel ParqueNacional Cahuita. Editorial Universidad Estatal a Distancia, San Jos& Costa Rica. SCHAFER,E. 1957. Les conotos.Etude comparativede Psarocoliusangustifrons et Psarocolius decumanus. Bonn. Zool. Beitr. 5:1-151. SIqUTCH,A. F. 1954. Life histories of Central American Birds. Vol. 1. Pacific Coast Avifauna No. 31, Berkeley, California. SMITH, N. G. 1968. The advantageof being parasitized.Nature 219:690-694. VANDERMEER,J. 1983. Pejibaye palm. Pp. 98-101, in D. H. Janzen, ed. Costa Rican natural history. Univ. of ChicagoPress, Chicago. Received5 Oct. 1987; accepted2 Oct. 1988.

NOTES AND NEWS

AFO MISTNETS are availablethrough Manomet Bird Observatory(for completedescription of sizesand pricessee J. Field Ornithol.60:286-287). For additionalinformation please use MBO's new phonenumber: 508/224-6521.