Using flowering data from rubber tree (Hevea brasiliensis; Fig. 1) plantations in Malaysia, Thailand, Indonesia, India, Sri Lanka, Vietnam, China, Nigeria, Cote d’Ivoire and Brazil, Yeang (this issue; pp. 283–289) takes a fresh look at the question of what signal tropical trees might use to synchronize their annual cycles of flowering. The external time giver proposed by Yeang is a meteorological event, namely the annually repeated cycles in incoming solar radiation and the associated changes in daily insolation. Why should these cycles, which after all are experienced by all organisms on Earth, be of special biological significance in the tropics? And just as curious, why should they have been overlooked as a time giver for rubber tree flowering until now? Flowering Hevea brasiliensis. (Figure kindly provided by Kazuo Yamasaki). The answer to the first question lies in a crucial difference between the temperate zone and the tropics. Everywhere on Earth, the amount of solar radiation that reaches the atmosphere is dependent on (i) the intensity of sunshine and (ii) its duration. Intensity is determined by the angle at which the sun's rays strike the surface of the Earth. In temperate latitudes, day length and irradiation intensity both contribute to the daily insolation, and the concurrent annual cycles in these two parameters are additive. Maximum radiation intensity – when the noon sun is at its highest above the horizon – and maximum day length occur on the same day, the summer solstice. Near the equator, however, seasonal change in insolation is dependent entirely on solar radiation intensity, as variation in day length is minimal. The seasonality is a result of the Earth's path around the sun and its tilt, with the sun passing directly over all tropical localities twice a year (Fig. 2 shows the latitudinal variation in the time course of insolation in the northern hemisphere from the equator to 50°N). The symmetric seasonal changes in insolation around the summer solstice (Fig. 2, vertical line) are the result of the declining (spring) and increasing (autumn) distance of the sun from its position at the summer solstice. In the equatorial belt proper, there are two distinct annual maxima of insolation, corresponding to the equinoxes (Fig. 2). The latitudinal variation in the time course of insolation in the northern hemisphere from the equator to 50°N. (figure kindly provided by R. Borchert). ‘Changes in radiation intensity as the explanation for how a tropical tree manages to synchronize its flowering may not convince everyone.’ What is the evidence that annual changes in insolation provide the external cue that synchronizes flower bud break in H. brasiliensis? First, rubber trees start flowering at the same time every year, making it unlikely that temperature, rainfall, cloud cover, or periods of drought could be the triggers as they do not occur with sufficient regularity. Secondly, data from rubber plantations in the northern and southern subtropics of South America, Africa and Asia show that flowering coincides with high insolation, not temperature or rainfall. The third line of evidence is the fit between the bimodal peak in incoming solar radiation (above and Fig. 2) and the bimodal flowering in H. brasiliensis in some localities (cf. Yeang's table 1). Most H. brasiliensis populations, however, flower only once a year, around the spring equinox (Yeang, this issue). As proposed by Yeang, high solar radiation intensity, and in particular bright sunshine (as distinct from prolonged diffuse radiation), induce flowering in Hevea, and he considers whether average radiation, very high radiation on a cloud-free days, or hours of high daily sunshine might be decisive. This fails to make explicit that bud break is not induced by some threshold value per se, but by the increase in daily insolation that precedes the two maxima in March and September (Fig. 2). Changes in radiation intensity as the explanation for how a tropical tree manages to synchronize its flowering may not convince everyone. First, given the large diurnal and day-to-day fluctuations of cloud cover, how can seasonal variation in insolation be a trigger for precisely timed flowering? Yeang elegantly refutes this argument with data on insolation near the ground that show that there is a consistent correlation with insolation above the atmosphere, such that one is a proxy for the other. Secondly, does Yeang's hypothesis obviate or refute the idea of Borchert et al. (2005), who linked the flowering times of 84 tropical tree species to the four annual periods at which changes in sunset times are greatest (as a result of the earth's tilt and elliptic orbit, the times of sunrise and sunset near the equator vary by 30 min over the course of the year)? Yeang argues (i) that the gradual time shifts of less than 1 min a day are too small to be perceived by plants, (ii) that they are ‘meaningful only when measured against an external reference chronometer’, and (iii) that a trigger linked to changes in day length would imply a single annual flowering (because no plant is simultaneously a long-day plant and a short-day plant). The first of Yeang's arguments is plausible, although the minimum change in day length required for flower induction is not yet known. The second does not hold, as there is 80 years’ worth of work (Bünning, 1936) showing that plants can perceive changes in light rhythms with the help of endogenous circadian clocks. The third argument, that photoperiodic control implies a single annual flowering, is unconvincing as some species could use the periods of advancing day length to flower, and others those of receding day length. Even individual trees that flower bimodally, as in some populations of Hevea (as mentioned above), do not present a problem for the hypothesis of Borchert et al. (2005) because each year there are two periods of receding and two periods of increasing day length. What are needed to resolve the relative roles of changes in insolation and day length as cues for the induction of flowering in tropical plants are more data on widespread species, such as H. brasiliensis, ranging from the equator to > 10°N. This will permit us to distinguish whether latitudinal changes in flower bud break in the same species parallel latitudinal changes in daily insolation or day length. Yeang's is among the few studies to have made use of the vast amount of phenological data available for commercially important tropical species. Besides rubber, these include durian (Durio zibethinus), mangosteen (Garcinia mangostana), rambutan (Nephelium lappaceum), avocado (Persea americana) and many others, and the obvious question now is to what extent their flowering also coincides with the yearly peaks in maximally changing daily insolation. Earlier studies of tropical phenology instead analysed natural communities (e.g. Opler et al., 1980; Wright & van Schaik, 1994) or species in a range of genera (Borchert et al., 2005). These studies show that synchronous flower bud break with minimal interannual variation characterizes not just H. brasiliensis but hundreds of tropical tree species in many different families, with some species flowering near the spring equinox, others near the autumn equinox, and a few bimodally (Spruce, 1908; Opler et al., 1980; van Schaik et al., 1993; Borchert et al., 2005). Given that van Schaik et al. (1993) and Wright & van Schaik (1994) explicitly drew attention to the annual cycle of insolation as a potential time giver for tropical trees (stressing that predictable irradiance seasonality occurs throughout the tropics), why did changing insolation receive so little attention before Yeang's investigation? One reason may be the temperate bias of plant physiology, especially modern Arabidopsis-centred work on the proximate mechanisms behind phenological changes (Simpson et al., 1999; Yanovsky & Kay, 2003). Other reasons may be that cloud cover and water stress were played up as modifiers or alternative triggers, and that irradiance was seen as a limiting factor (hence ‘insolation-limitation hypothesis’; van Schaik et al., 1993; Wright & van Schaik, 1994), rather than a precise external cue that allows tropical plants to synchronize their reproduction. Such synchronization must be under strong selection because stigmas and/or pollen are only viable for relatively short times, and successful reproduction therefore depends on the exact timing and synchronization of flowering. How trees perceive daily changes in insolation is not yet clear. Ultimately, the interaction of light absorbed by the various photoreceptors with the expression of circadian clock-dependent transcription factors must determine flowering (Simpson et al., 1999; Yanovsky & Kay, 2003). Yeang's study points to the need for experiments that address the effects of irradiance per se, independent of day length.