1. A dispersed seed must remain in a microsite long enough to imbibe water and germinate. Here the post-dispersal movement caused by wind of four asymmetric samaras of Picea engelmannii, P. abies, Pinus contorta var. latifolia and P. strobus was studied on four surface roughnesses (a smooth board, c. 250 mum, c. 750 mum and 2.0 mm). A mechanistic approach is taken to determine the chance that a seed will remain on a surface long enough to germinate. 2. The threshold or lift-off velocities for all combinations of four samaras and four surface roughnesses were determined in a wind tunnel. These measured threshold velocities were incorporated with seed mass and wing area to define the non-dimensional resistance to seed movement (B). The resistance to movement is the ratio of the force due to gravity to that developed by the wind on a given reference area. 3. The resistance to movement was used in a logistic regression to predict the percentage of seeds unmoved (remaining) on a surface with increasing wind velocity. All regressions explained a significant proportion (P < 0.05) of the variation, with r2 greater-than-or-equal-to 0.64. Regressions had steeper slopes the greater the resistance to seed movement. The resistance equation should be species-independent, assuming a geometrically similar shape, and was so for the smoothest surface but not always so on the remaining surfaces. 4. The percentage of seeds that will be unmoved on different surfaces for different time intervals was estimated by determining the return time in a natural environment of the threshold velocities in the resistance to movement (B). Wind velocities in nature have empirically estimatable return times, i.e. lower velocities occur more frequently than higher velocities. These return times were calculated for Calgary, Alberta. Heavy seeds were found to move less frequently than smaller seeds, as would be expected. On rougher surfaces, seeds remained stationary longer because the threshold velocity is larger and also the return time of that velocity is longer. 5. Finally, the timing of seed movement determined above was compared to germination times (under ideal conditions) to determine if seeds, on surfaces of different roughness, were generally stationary long enough to imbibe water and germinate. On smooth surfaces (particle sizes up to c. 750 mum), all seeds moved before adequate time had elapsed for germination. On rougher surfaces (particle sizes up to 2 mm) the distribution of the percentage of seeds remaining was positively skewed and the tail of the distribution declined more slowly. This tail defines seeds which are caught in microsites that have the highest probability of germinating.
