Wilson Bull., 113(1), 2001, pp. 73±77 EXPLOITATION OF A SEASONAL RESOURCE BY NONBREEDING PLAIN AND WHITE-CROWNED PIGEONS: IMPLICATIONS FOR CONSERVATION OF TROPICAL DRY FORESTS ALLAN M. STRONG1,2,4 AND MATTHEW D. JOHNSON1,3 ABSTRACT.ÐColumbids often exhibit irregular movement patterns in response to fruit abundance. We tested whether the abundance of nonbreeding Plain (Columba inornata) and White-crowned (C. leucocephala) pigeons was correlated with Thrinax parvi¯ora fruit production in a dry forest in southeastern Jamaica. Monthly, from November to March, we counted the number of pigeons leaving the forest to roost in an adjacent mangrove swamp. Within two days of each roost count, we also counted all fruits on ten T. parvi¯ora trees in the forest. Columbid and fruit counts showed similar patterns of temporal abundance, with increases from November to January and decreases from January to March. Peak (January) counts of White-crowned Pigeon, Plain Pigeon, and unidenti®ed columbids were 129, 77, and 151, respectively. The peak Plain Pigeon count was approximately three times greater than the highest counts previously recorded for Jamaica. These data indicate that the Portland Ridge dry forest may provide a critical resource for the Plain Pigeon, perhaps at a time when fruit abundance is low on other parts of the island. Given the globally signi®cant number of Plain Pigeons that use this site, protection from further development should be a priority. Received 8 June 2000, accepted 24 January 2001. The irregular temporal and spatial pattern These three species feed nearly exclusively on of fruit abundance has been well documented fruits, and are probably important seed dis- and is thought to be a result of variation in persers (Bancroft and Bowman 1994, Strong climatic conditions such as temperature and and Bancroft 1994, Baptista et al. 1997). The rainfall, as well as interspeci®c variation in Ring-tailed Pigeon (C. caribaea) is found at fruiting phenologies (Janzen 1967, Smythe relatively high elevations whereas the Plain 1970, Frankie et al. 1974, Crome 1975, Foster Pigeon (C. inornata) and the White-crowned 1982). This variation in resource abundance is Pigeon (C. leucocephala) generally prefer likely the ecological basis behind the irregular lower elevations (Downer and Sutton 1990). to nomadic distribution patterns of many fru- Of these two lower elevation species, the givores (Wheelwright 1983, Levey 1988, White-crowned Pigeon is relatively common Blake and Loiselle 1991, Loiselle and Blake on Jamaica, but little is known about the dis- 1991). Indeed, frugivory may be a precursor tribution and abundance of the Plain Pigeon, to migratory behavior (Levey and Stiles which was recently reported as possibly extir- 1992). pated from Jamaica (Miyamoto et al. 1994). Many columbiforms are frugivorous and In January 1996, we observed 16 Plain Pi- exhibit movement patterns that range from no- geons ¯ying from the dry limestone forest of madic to migratory, presumably in response Portland Ridge to a red mangrove (Rhizopho- to the temporal and spatial distribution of fruit ra mangle) swamp to roost. The next evening availability (Crome 1975, Frith et al. 1976, we saw 86 Plain Pigeons and large numbers Wiley 1979, Innis 1989, Lambert 1989, Mil- of White-crowned Pigeons ¯ying to the man- an-Rivera 1992). In the Caribbean, the genus grove roost. Columba is represented by four native species, We hypothesized that the use of the forest three of which occur regularly on Jamaica. by columbids was a seasonal response to in- creased fruit availability, since many trees in 1 Dept. of Ecology and Evolutionary Biology, Tu- Portland Ridge's dry forest produce ripe fruit lane Univ., 310 Dinwiddie Hall, New Orleans, LA during the dry season. In this paper, we doc- 70118. ument the seasonal use of a threatened Carib- 2 Present address: Univ. of Vermont, School of Nat- bean dry forest by Plain and White-crowned ural Resources, Aiken Center, Burlington, VT 05405. pigeons and examine their abundance in re- 3 Present address: Dept. of Wildlife, Humboldt State Univ., Arcata, CA 95521. lation to the fruiting phenology of Thrinax 4 Corresponding author; E-mail: astrong@nature. parvi¯ora, a common dry season fruiting spe- snr.uvm.edu cies. 73 74 THE WILSON BULLETIN x Vol. 113, No. 1, March 2001 TABLE 1. Mean counts (6 S.E.) of fruits from 10 Thrinax parvi¯ora trees and evening counts of col- umbids ¯ying from the Portland Ridge dry forest to a mangrove roost site during the 1996±1997 dry season. Fruit counts and columbid counts were conducted within 2 days of each other. Count date 19 Nov 5 Dec 30 Jan 27 Feb 13 Mar Fruit counts Unripe 1415 6 203 1373 6 215 582 6 136 90 6 87 32 6 32 Ripe 0 6 0 4 6 2 200 6 67 24 6 10 16 6 13 Desiccated 0 6 0 0 6 0 0 6 0 57 6 20 37 6 16 Columbid counts Plain Pigeon 17 52 77 34 6 White-crowned Pigeon 0 5 129 79 50 Unidenti®ed columbid 1 2 151 150 111 Total columbids 18 59 357 263 167 STUDY AREA AND METHODS 20 min prior to observing the ®rst birds ¯ying to the roost and continued until it was too dark to see birds, This study was conducted at Portland Ridge, Ja- at which time the number of birds leaving the forest maica, (178 449 N, 778 099 W; 12 km southeast of Lio- was near zero. Two observers conducted the counts nel Town), a peninsula that supports approximately from a lighthouse approximately 20 m above the can- 4,200 ha (D. B. Hay, pers. comm.) of relatively undis- opy. The birds ¯ew primarily from south to north, so turbed dry limestone forest, ranging from 5±120 m el- one observer watched to the east and one to the west. evation and characterized by ,125 cm rainfall/year One or two additional observers looked for birds from (Lack 1976). The canopy is dominated by Metopium the base of the tower (just above the canopy level) as brownii, Bursera simaruba, and to a lesser extent T. birds that ¯ew low over the forest canopy were some- parvi¯ora. The subcanopy was dominated by Ater- times dif®cult to observe from the top of the tower. amnus lucidus and Oxandra lanceolata. Birds were identi®ed as Plain Pigeon, White-crowned T. parvi¯ora is an endemic Jamaican palm that pro- Pigeon, or unidenti®ed columbids. Although birds of- duces large spikes of 6.5±7.5 mm diameter white ten were too distant to identify to species, we feel that drupes (Adams 1972). We conducted complete counts very few individuals passed undetected. of the fruit crops of 10 (6±8 m height) T. parvi¯ora trees on ®ve days during the 1996±1997 dry season (19 November, 5 December, 1996; 28 January, 27 Feb- RESULTS ruary, and 12 March, 1997). We chose T. parvi¯ora as a study species because we frequently ¯ushed pigeons Numbers of Plain Pigeons, White-crowned from these trees during ®eld work at the site. Addi- Pigeons, and unidenti®ed columbids all peak- tionally, the architecture of the tree was amenable to ed during the January count, with maximum complete fruit counts because the fruiting spikes typ- counts of 77, 129, and 151 individuals, re- ically hang well below the fronds. Although we were spectively (Table 1). Plain Pigeons dominated not able to document pigeon consumption of T. par- the ®rst two counts, but the proportion of vi¯ora by direct observation, White-crowned Pigeons consume fruits of two other species of Thrinax in south (identi®ed) Plain Pigeons in the total count Florida (G. T. Bancroft, pers. comm.), and both species dropped from 21.5% in January to 3.5% in consume royal palm (Roystonea spp.) fruits throughout March. We suspect that most unidenti®ed col- the Greater Antilles (Plain Pigeons, Puerto Rico umbids were White-crowned Pigeons, but [PeÂrez-Rivera 1978]; White-crowned Pigeons, Puerto counts from additional observation points Rico, [Wiley and Wiley 1979], Cuba [Godinez 1993], would be helpful to con®rm the identity of and Jamaica [AMS, unpubl. data]). We chose trees along trails maintained by the P. W. D. Gun Club of distant birds. Kingston with the criterion that all fruits were visible In November, T. parvi¯ora supported no without obstruction from leaves and branches. We ripe fruit, although the average crop was classi®ed each fruit as unripe (green), ripe (white), or .1400 unripe fruits/tree (Table 1). By early desiccated (brown). December, a few fruits had ripened, but .99% Within two days of each fruit count, we conducted of the crop was still unripe. In January, the evening roost counts of columbids leaving the forest and ¯ying to the mangrove roost. We began the counts percentage of ripe fruit peaked, coincident at 16:30 November through January, at 17:15 in Feb- with the peak pigeon abundance. However, the ruary, and 17:00 in March. All counts began at least total January (ripe 1 unripe) fruit crop was Strong and Johnson x PIGEON ABUNDANCE AND FRUIT PRODUCTION 75 50% lower than the November census, pre- resource during a period when the pigeons ex- sumably as a result of consumption by birds, perience food shortages in other parts of their as we observed few fallen fruits below the range (sensu Howe 1984). As such, these for- trees. Fruit abundance decreased further dur- ests should be protected from further devel- ing February, when about one-third of the opment, as the Plain Pigeon numbers we ob- crop was brown and desiccated. By March, served appear to be three times greater than the trees supported an average of only 16 ripe the highest counts documented previously in fruits.