FOOD CACHING IN THE TROPICAL FRUGIVORE, MACGREGOR'S BOWERBIRD (AMBLYORNIS MACGREGORIAE) M. A. PRUETT-JONES1 AND $. G. PRUETT-JONES1 Museumof VertebrateZoology and Department of Zoology,University of California,Berkeley, California 94720 USA AI•STRACT.--Malesof MacGregor'sBowerbird (Amblyornismacgregoriae) cache fruit. In a study population in easternPapua New Guinea we found 437 cachesites at 39 bowers.Adult malesstored an averagetotal of 17.6 fruits (range 0-82) at 13.6sites (range 0-55), significantly more than immaturemales. All siteswere locatedin vegetationabove ground within 13 m horizontal and 9 m vertical distancefrom the males'bowers. For 3 malesstudied intensively, the rate of replacementof cachesaveraged 0.13 replacements.site-:.day-•, with no signifi- cant differencesamong males. Caching occurredonly during the seasonalperiods of bower attendanceand breeding activity. Femalesdid not store fruit and did not take fruit stored by males.We suggestthat by extendingthe time malescan remain at their bowers,caching may increaseinteraction time with femalesand decreaserates of bower maraudingby rival males. There was no evidence that cachedfruits served as bower decorations.Received 7 May 1984, accepted4 October1984. FOODstoring refers to the depositionof food changesin food abundancein temperatehab- items in a particular location for later con- itatsand the coolertemperatures that allow for sumption.It hasbeen observedin a number of storageof perishableitems over a longer peri- bird and mammal species(Roberts 1979, Van- od of time (Roberts 1979, Vander Wall and Bal- der Wall and Balda 1981, Smith and Reichman da 1981).Nonetheless, tropical jays (Turaek and 1984). In birds, food storage occurs as either Kelso 1968) and woodpeckers (Skutch 1969) a long-termstrategy, to yield foodduring times storefood, as do their temperatecounterparts, of scarcity,or on a short-termbasis for tempo- particularly in montane areas. rary retention or accumulationof items that In this paper we describefood-storing be- cannot be eaten at one time. havior in MacGregor's Bowerbird (Amblyornis Long-termfood storageoccurs predominate- macgregoriae),a frugivorous speciesfound in ly in nonmigratoryspecies and in habitatswith montane rain forest in New Guinea. This be- seasonal fluctuations in resource abundance havioris separatefrom fruit gatheringby males (Roberts 1979, Smith and Reichman 1984). for bower decoration.We present data on oc- The food storedis generally durable and par- currence,type, location, and replacementrates ticulate, such as seeds.Birds for which long- of storedfood items and discusscaching in re- term food storageis an important contribution lation to the species'social organization and to reproductivesuccess and winter survival in- bower-buildingbehavior. Food storing has not clude woodpeckers (Koenig 1978), corvids been describedpreviously in any bowerbird or, (Tomback 1977, Bossema1979, Vander Wall and to our knowledge, any tropical frugivorous Balda1981), and parids(Ulfstrand 1976,Sherry passefine. et al. 1982).Short-term food storageis known METHODS primarily in shrikes (Laniusspp., Craig 1974) and birds of prey (Newton 1979, Walter 1979) Our observations were made on the southwestern and in somecracticids, parids, and corvids(Piz- slopeof Mt. Missim, Kuper Range,Morobe Province, PapuaNew Guinea (7ø16'S,146ø47'E). The study area zey 1980, Smith and Reichman 1984). comprised750 ha of primary,midmontane rain forest Long-termfood storing apparentlyis more between 1,450 and 2,200 m altitude and included three commonin temperatethan in tropical avifau- drainagesand the ridge lines separatingthem. An- has (Smith and Reichman 1984). This may nual rainfall averaged about 2,000 mm, and daily reflectboth the greaterimportance of seasonal temperaturesvaried from 9ø to 25øC.More detailed descriptionsof this areaare presentedin Pruett-Jones and Pruett-Jones (1982) and Pratt (1983). • Present address:Department of Biology, C-016, Observationson MacGregor'sBowerbird were made University of California at San Diego, La Jolla,Cali- during 21 monthsof fieldwork betweenAugust 1980 fornia 92093 USA. and December1983. We first discoveredcaching in 334 The Auk 102: 334-341. April 1985 April1985] BowerbirdFruit Caching 335 this speciesin September1982. Most data presented As part of other studies(Pruett-Jones unpubl. data), here were gathered during the 6-week period from we monitored activity at bowers weekly or biweekly 17 October to 4 December 1983. throughout 1982 and 1983. These recordsprovided Caching behavior of 3 adult males that maintained data on the relative seasonal occurrence of fruit cach- adjacentbowers on a single ridge was studied in de- ing and its relation to bower maintenanceand breed- tail. We visited these 3 bowers 4 times over the course ing. of ! week and mapped and marked the cache sites before the start of regular sampling.We define a cache RESULTS site as a specificlocation where one or more fruits, of one or more species,were stored. If a structure (e.g. a small tree) had fruits cachedin several loca- Generalbiology and cachingbehavior.--On av- tions, each location was considered a separate site. erage, 47 (range = 43-54) male MacGregor's Changesin the number and speciesof fruits at each Bowerbirds were active on our study site each cache site were recorded from 3 to 22 November. season from 1980 and 1982. Adult males con- Sites were checked twice daily, between 1100 and struct a maypole bower (twigs piled up around 1300 and between 1600 and 1800. To facilitate an ac- a slender sapling) that is decoratedwith fruit, curate measure of replacement, all cached fruits fungus, charcoal, and insect frass. Immature within reach of the observer (up to 3-4 m vertical malesbuild rudimentary structuresthat may be height) were marked with small spots of nontoxic abandoned after a few weeks of use. paint, and those fruits out of reach were noted for color or variety. Bowers were regularly and linearly spaced A cachereplacement was defined as a decreasefol- along ridge-line habitat, with a mean inter- lowed by an increasein the number of fruits at the bower distanceduring the 1980 and 1981 sea- site, or vice versa,irrespective of the number of fruits sons of 182.8 m (n = 98, range = 60-423, SD = involved (i.e. a replacement occurredwhen the male 72.9). The local placement of bower sites ap- removed some or all of the fruit at a cache site and pears to be determined primarily by habitat then replacedit, or added fruit and then removed it). characteristicsof the ridge line (Pruett-Jones Fruit changerefers to the absolutenumber of remov- and Pruett-Jones 1982). Observations of marked als of marked fruits and replacementsof unmarked males in 1980 indicated that they spent an av- fruits at each site. Some new caches were discovered erageof 54%(range = 20-75%) of daylight hours near the study bowers after initiation of sampling. within a 20-m radius of the bower, and this In our analysisof replacementrates, we include only core area was aggressivelydefended. those sites that were monitored for at least 8 days (about the maximum length of time over which a A. macgregoriaewas primarily frugivorous; replacement might occur). Basedon direct observa- 95% of its diet consisted of medium to large tions of males eating cached fruits (see Results), we drupes and, to a lesser degree, arillate fruits. assumedthat only the resident male was responsible Arthropodsmade up the remainder of the diet. for the disappearanceor replacementof fruits at cache The birds foraged singly or in small groups, sitesat a given bower. with females and immature birds overlapping To see whether males would detect and use offered extensively in use of space and fruiting trees fruits, two experimentalsites were establishedat each with males. Males did not defend food re- of the three study bowers; one site was within 1 m and the other from 5 to 10 m from the bower. Fruits sources.Analysis of fecal samples showed at of the same speciesthe male stored were placed at least 130 speciesof fruit utilized by A. macgre- these sitesand their subsequentrecovery monitored goriaeon our study area; individual males ate during normal cachechecks for 7 daysfollowing first fruit from about 30 speciesof trees,shrubs, and placement. Each day, new fruits were placed at emp- vines (Pruett-Jones and Pruett-Jones unpubl. ty sites. data). On 29 November we removed all fruits cached at Dynamicsof fruit caching.--Fruitcaching was the 3 bowers. Once a day for the following 5 days observed only in males and only during the we recorded reestablishment of cache sites, initiation breeding season.Males maintained their bow- of new sites,and ratesof fruit change for thesemales. ers for about 9 months/yr (May-February); Between 23 November and ! December, the entire however, breeding was restricted to Septem- population of active bowers on the study area was checked for cache sites. At each bower we recorded ber-February.From May to August activity was the number of sites found in 30 observer-min of irregular while males rebuilt their bowers. search time and, for each site, the horizontal and ver- During this early phase males did not store tical distance from the bower. All cached fruits were fruit. Caching was observedonly during the collected,and identified if possible. period when males were in regular attendance 336 ?•u•rr-lo•s ^•D P•u•rr-Jo•s [Auk,Vol. 102 25 4 8 12 16 20 24 28 32 56 56 2 4 6 8 VERTICAL HEIGHT-m NUMBER OF CACHE SITES 25- Fig. 1. Frequencydistribution of number of cache sitesat 32 complete bowers of adult males (solid line) and 7 rudimentarybowers of immaturemales (dashed line). 15 of their bowersand the majority of intersexual interactions occurred. Late in the season, cach- es were maintained until residents abandoned their bowers.Males remainedin the vicinity of their bower site throughout the year, despite HORIZONTAL DISTANCE-m the markedseasonality of bower attendanceand Fig. 2. Distribution of vertical height and hori- abandonment. Females did not store food, nor zontal distance measurements of 421 cache sites at 31 did they take the food cachedby males.