Evaluation of seven indices of on‐plant seed storage (serotiny) shows that the linear slope is best.

On‐plant storage of seeds (serotiny) is a feature of many fire‐prone dominant trees and shrubs in North America, Mediterranean Basin, South Africa and Australia. Understanding how it has responded to the prevailing fire regime and recruitment/growing conditions, and its genetic basis and adaptive significance, depends on the use of an accurate index of the level of serotiny.Seven indices, three measuring 'apparent' serotiny and four measuring 'inherent' serotiny, were evaluated for their ability to reflect the extent to which seeds are stored in the crown for later release. The most widely used indices ignore the temporal aspects of seed attrition and are vulnerable to plant size, age, fecundity and local growing‐condition effects, while age of the oldest crop year with closed cones/fruits depends on the accuracy of a single observation. Presence/absence of serotiny at the individual tree level ignores the effect of plant age on the expression of serotiny.The slope of the line through the percentage of cones/fruits that still retain their seeds in progressively older crop years, the 'degree of serotiny' (100/slope), is shown to be unbiased by annual fluctuations in crop size, differences in plant age or variations in local growing conditions. It also corresponds to a probability, and thus more accurate, estimate of age of the oldest unopened structures. Since seeds have a lower % viability than the % of supporting cones/fruits closed at any age, there is a case for applying the slope method to the fraction of viable seeds retained to arrive at an estimate of the age of the oldest viable seeds on the plant, as the ultimate index of serotiny.If the level of serotiny is to be related to the fire regime, then care needs to be taken to use records over a sufficiently long time to have had an evolutionary impact, unless this is being used against controls to examine a genetic response to the new fire regime.Synthesis. Among seven currently available indices of serotiny, the slope measure is shown to be the most accurate for studies that relate the level of serotiny to the fire regime and habitat conditions under which the species evolved, as well as for examining its heritability. Attention needs to shift from just recording cones/fruits closed to include the viability of seeds that remain as a better basis for understanding the role of serotiny in post‐fire recruitment. [ABSTRACT FROM AUTHOR]