INTRODUCTION: Seasons are an inevitable consequence of Earth's obliquity, and organisms everywhere have evolved mechanisms to time life history events to appropriate seasons. At temperate and boreal latitudes, plants use photoperiod, winter vernalization and growing degree-days as proximate cues to synchronize flowering with favorable environmental conditions. These potential cues are invariant (photoperiod at the equator) or irrelevant in tropical forests where the growing season never ends. We explore alternatives.

HYPOTHESES: We evaluate eight environmental signals hypothesized to be proximate cues for flowering in the tropical forest literature. The hypotheses include increasing or decreasing photoperiod; low temperature events; drought punctuated by heavy rainfall or broken by sustained rainfall; and low irradiance punctuated by a brief period of high irradiance or broken by a sustained period of steadily increasing irradiance or high mean irradiance.

METHODS: We exploit interannual variation in the timing of flowering and meteorological events to evaluate the eight hypotheses in a model competition framework. We formulate each hypothesis as a quantitative model to predict the probability of flowering for each day of the year. Model inputs are daily meteorological records. Model parameters include the duration of drought (or low irradiance), the increase in rainfall (or irradiance), the rate or duration of this increase, and a lag between flowering and the date the cue is realized. We evaluate predictions against species-specific flower records from 1,820 weekly censuses (35 years) of 200 passive traps at Barro Colorado Island, Panama.

RESULTS: Analyses are underway. Preliminary results follow. Photoperiod and low temperature hypotheses are not supported. Irradiance hypotheses are supported for the largest number of species. Rainfall hypotheses are supported for fewer species. None of the eight hypotheses adequately predict flowering times for a substantial number of species.

CONCLUSIONS: We have linked interannual variation in the timing of flowering to interannual variation in the timing of seasonal changes in irradiance in many species. Our models distinguish the aspect of seasonal change – more species respond to sustained periods of high mean irradiance then to sustained increases in irradiance – and quantify critical levels and durations of contrasting low and high levels of irradiance associated with flowering. Cloud cover largely determines irradiance levels on BCI, with irradiance averaging 49% higher in the four-month dry season than in the cloudy, wet season. The behavior of clouds remains one of the greatest uncertainties under global warming, with implications for plant phenologies in humid tropical forests.