1 Stage projection (Lefkovitch) matrices for 21 species of woody plants and 45 herbaceous perennials were extracted from the plant demographic literature or compiled from published data. 2 Each matrix was divided into six regions representing: 1, recruitment of seeds to the seed pool
2, recruitment of seedlings or juveniles from current seed production
3, clonal growth
4, retrogression, due to plants decreasing in size or reverting in stage
5, stasis, (survival from one year to the next in the same stage class)
6, progression to later stage classes. 3 Matrix analysis was used to calculate the finite rate of increase lambda for each population and to calculate the elasticities of each transition coefficient in the matrices. Elasticities were summed within each of the six regions of the matrix to give measures (E1 - E6, respectively) of the importance of each component of the life cycle to lambda and fitness. 4 Herbs as a group differed significantly from woody plants in most of these components. Seedling recruitment was more important in herbs than woody plants. Retrogression occurred only in herbs, particularly those with a tuber. Stasis occurred in nearly all species, but was most important in woody plants. Progression was more important than fecundity in almost all species. 5 Trade-offs among life cycle components were determined from correlation matrices of r (= In lambda) and elasticities E1 - E6 for the whole sample and for herbs and woody plants separately. As a whole, r was positively correlated with elasticities for fecundity (E1 + E2) and growth (E3 + E6) and negatively correlated with survival (E4 + E5). In clonal herbs, fecundity and clonal growth were negatively correlated. 6 The division of elasticities into three major components (growth, G = E3 + E6
fecundity, F = E1 + E2
and survival, L = E4 + E5) allowed us to construct triangular plots in G-L-F space. This was done separately for iteroparous forest herbs, iteroparous herbs from open habitats, semelparous herbs and woody plants. Each of these four groups occupied a distinct position in G-L-F space. Within woody plants, shrubs of fire-prone habitats occupied the end of the distribution with the lowest survival elasticity. 7 It is argued that the demographic approach to the classification of distinct ecological groups offers new insights into the relationship between life history and habitat.
