Although flowers are considered to be relatively invariant within species, considerable intraspecific variation in floral size traits is often observed along moisture gradients.  This variation may be due to plasticity, whereby floral size increases with increasing moisture availability.  Additionally, this variation may be due to genetic variation.  Floral transpiration can be substantial and can deleteriously impact plant physiological processes.  Therefore, natural selection may favor smaller flowers in drier environments in order to reduce evaporative surface area.  The objective of this study was to observe variation in floral size within a species across a naturally occurring precipitation gradient.  This variation was observed in conjunction with leaf physiological traits and fitness to identify possible sources for any observed environmental variation in floral size.  In this study, four populations of Leptosiphon androsaceus (Polemoniaceae) were studied for four years (2005-08) across a naturally occurring precipitation gradient in the Coast Range of Central California.  Morphological measures of floral size, leaf size, and plant size were collected in conjunction with measures of physiological rates (water-use efficiency, δ¹³C) and fitness (seed size and number).   
Results/Conclusions

Most floral size traits, such as corolla diameter, increased across populations (P < 0.001) and years (P < 0.001) as precipitation increased.  However, the length of the floral tube, which should be under strong pollinator-mediated selection, did not vary with moisture availability (P = 0.21).  Flower size increased with other plant size traits, such as height (P= 0.001) and leaf area (P < 0.001).  But when simultaneously controlling for variation in precipitation, floral size varied independently of both plant height (P = 0.57) and leaf area (P = 0.21), suggesting a direct effect of precipitation on floral size.  In contrast, in 2007, a drought year, floral size increased with leaf area, even when controlling for variation in precipitation (P = 0.001).  These results suggest that under especially dry conditions, there are also indirect effects of precipitation on flower size, as mediated via leaf responses to the environment.  In conclusion, precipitation affects floral size both directly and indirectly via leaves, but the relative contribution of each varies across space and time.