Are the best dispersers the best colonizers? Seed mass, dispersal and establishment in Carduus thistles

Negative correlations between dispersal and establishment are often reported in the plant literature; smaller seeds tend to disperse better but germinate less well, and produce smaller seedlings. However, because dispersal capacity is often quantified using proxies, such as the settling velocity of wind-dispersed seeds, little is known about the exact shape of this negative relationship, and how it is modified by other plant traits and environmental conditions. We studied the dispersal-establishment relationship in two wind-dispersed thistles (Carduus nutans and Carduus acanthoides). We applied a mechanistic wind dispersal model (WALD) to seeds released under a range of environmental conditions, and tested germination and seedling growth under standardized conditions in a greenhouse. Dispersal distance and establishment (germination and seedling growth) were not significantly correlated, although in both species smaller seeds dispersed farther, and showed lower germination and lower seedling growth rates. This apparent paradox can partly be explained by the significant influence of other factors such as release height and environment (wind and vegetation), which explained more variation in dispersal than did terminal velocity. Another potential explanation is the variation in seed traits: germination is strongly positively related to seed mass, weakly positively related to plume loading, but not significantly related to terminal velocity. This weakening of the correlation with germination is due to additional layers of trait (co)variability: for instance, seed mass and pappus size are positively correlated, and thus big seeds partially compensate for the negative effect of seed mass with larger pappi. Our mechanistic approach can thus lead to a better understanding of both potentially opposing selection pressures on traits like seed mass, and diluting effects of other seed, plant and environmental factors.