LAUSR

PREMISE In fleshy-fruited plants, fruit removal is widely used as a proxy for plant reproductive success. Nevertheless, this proxy may not accurately reflect the number of seeds dispersed, an assumed better proxy for total fitness (fruit removal × mean number of seeds dispersed per fruit). METHODS We examined under what circumstances fruit removal can be reliable as a proxy for total fitness when assessing bird-mediated selection on fruit traits. In three populations of the Blue Passionflower (Passiflora caerulea), we used the number of fruits pecked per plant as a surrogate for fruit removal to estimate phenotypic selection on fruit and seed traits, and simulations of the effect of the fruit-seed number trade-off on the number of fruits removed. KEY RESULTS Fruit removal was a good indicator of fitness, accounting for 55-68% of the variability in total fitness, measured as total number of seeds removed. Moreover, multivariate selection analyses on fruit crop size, mean fruit diameter and mean seed number using fruit removal as a fitness proxy yielded similar selection regimes to those using total fitness. Simulations showed that producing more fruits and a lower number of seeds per fruit can result in a negative relationship between fruit removal and total fitness. CONCLUSIONS Our results suggest that fruit removal can be reliably used as a proxy for total fitness when (1) there is a weak fruit number-seed number trade-off, (2) fruit crop size and fruit removal correlate positively, and (3) seed number variability does not largely exceed fruit number variability. This article is protected by copyright. All rights reserved.