Vol. 143, No. 1 The American Naturalist January 1994 LIGHT AND THE PHENOLOGY OF TROPICAL TREES "Smithsonian Tropical Kesearch Institute, Unit 0948, APO AA 34002-0948, U.S.A., or Apartado 2072, Balboa, Republic of Panama; ?Department of Biological Anthropology and Anatomy, Duke University, 3705 B Erwin Road, Durham, North Carolina 27705 Submitted March 18, 1992; Rer'ised June 23, 1992; Accepted June 26, 1992 Abstract.-Tree phenologies from eight disparate tropical forests are consistent with the hypoth- esis that leaf and flower production have been selected to coincide with seasonal peaks of irradiance. Plants that produce new organs when irradiance is maximal will realize two advantages. First, assimilation is greatest for a crop of new leaves before herbiv- ory and senescence reduce leaf area and photosynthetic potential (Larcher 1973; Field 1987). Second, it is most efficient to transfer assimilates directly into grow- ing organs rather than store them in different tissues first and mobilize and translo- cate them later (Chapin et al. 1990). Strong phenological selection results for light-limited understory plants, and in many deciduous forests understory plants produce leaves and flowers when the overstory is deciduous and understory irra- diance is maximal (Janzen 1972). Seasonal variation in irradiance also occurs above forest canopies because of changes in cloud cover, day length, and solar angle. The effect on the phenologies of taller forest plants is unexplored. This article evaluates this effect for tropical forest trees. Light-limited trees are predicted to produce new leaves and flowers during the reason of maximal irradiance. Seed dispersal is not considered because its timing is colzfounded by selection for germination and seedling survival. Where rainfall seasonality is pronounced, dry-season drought constrains phenologies (Reich and Borchel-t 1984). Predictions are therefore evaluated separately for forests with strong rainfall seasonality. Tropical insect seasonality also increases with rainfall seasonality (Wolda 1988). Minimal insect activity and maximal irradiance often cs-occur in the drier season, and the alternative hypothesis that reduced pest pressure selects for dry-season production is evaluated. First, however, two req- uisites for phenological selection are considered. IRRADIANCE IN TROPICAL FOREST CANOPIES Phenological selection requires predictable temporal variation in a limiting envi- ronmental factor. Correlations between plant performance and irradiance suggest Am. Nat. 1994. Vol. 143, pp. 192-199. O 1994 by The University of Chicago. 0003-014719414301-000B$0ZZ0OOAll rights reserved193 TROPICAL TREE PHENOLOGY that light limits many tropical forest trees. Seed set increases with irradiance in Borneo (Wycherley 1973), the likelihood of a mast flowering increases with irradi- ance in peninsular Malaysia (van Schaik 1986), and trunk radial growth increases with irradiance in Surinam (Schulz 1960) and possibly Costa Rica (D. A. Clark and D. B. Clark, personal communication). These correlations are consistent with light limitation; however, co-varying environmental factors cannot be discounted. Photosynthetic measurements also suggest that light limits many tropical forest trees. Photosynthesis by in situ canopy leaves becomes saturated with light at photosynthetic photon flux densities (PPFD) of 450-600 ymol m-2 s-' for a wide variety of tropical forest trees (Oberbauer and Strain 1986; Pearcy 1987; K. Ho-gan, A. Smith, K. Winter, and 6.Zotz, personal communication). Lower PIC'FD will limit photosynthesis. Radiant energy is extinguished exponentially with dis- tance beneath forest canopies. Photosynthetically active radiation was reduced by 94% just 5 m into the canopy of a Puerto Rican wet forest (Johnson and Atwood 1970), and global radiation was reduced by 53% just 6 m into the canopy of a Malaysian rain forest (Yoda 1974; Aoki et al. 1975). More than 70% of PPFD measurements made for canopy leaves in tropical forests in Costa Rica and Australia were less than 400 ymol m-' s-' (Oberbauer and Strain 1986; Doley et al. 1987), and low PPFD associated with heavy cloud cover frequently limited wet-season photosynthesis by the canopy emergent Ceiba pentandra in central Panama (G. Zotz and K. Winter, personal communication). Self-shading, lateral shading by neighbors, and shading by lianas limit all but the uppermost leaves in tropical forests, and light interception by clouds can limit all leaves. Predictable irradiance seasonality occurs throughout the tropics. Near the equator seasonal changes in cloud cover are a more important cause than seasonal changes in day length and solar angle (Miller and Feddes 1971; Hastenroth 1985). In central Panama, for example, global radiation averages 31% greater on the March equinox than on the September equinox because of seasonal changes in cloud cover (Windsor 1990). Global radiation averaged 50% greater in the highest than in the lowest month in a pantropical survey of 24 sites where forest is the potential vegetation (annual rainfall of more than 1,000 mm; compiled from Miiller 1982). Irradiance seasonality controls seasonal variation in net primary produc- tion in tropical evergreen forests (Raich et al. 1991) and should be a strong selec- tive pressure on phenology. WEAKLY SEASONAL FORESTS Weakly seasonal forests were defined to have minimum mean monthly rainfall of more than 60 mm. This criterion is widely used to identify tropical forests where seasonal water stress is unimportant (Schulz 1960; Nieuwolt 1977; Whit- more 1984) and is close to Walter's (1971) criterion of twice the average tempera- ture. The null hypothesis that 25% of tree species center production in the three sunniest months was tested for four weakly seasonal forests for which species- level phenologies have been published. For the Ducke Forest Reserve in Brazil and Itabuna, Brazil, numbers of individuals with new organs were presented by month (Alvim and Alvim 1978; Alencar et al. 1979), and the the angle of the mean194 THE AMERICAN NATURALIST vector was calculated. For La Selva, Costa Rica, and Piste St. Elie, French Guyana, the population-level presence or absence of new organs was presented by month (Frankie et al. 1974; Opler et al. 1980; Sabatier 1985), and the month on which production centered was