In this study, we examined the extent to which between-species leaf size variation relates to variation in the intensity of leaf production in herbaceous angiosperms. Leaf size variation has been most commonly interpreted in terms of biomechanical constraints (e.g. affected by plant size limitations) or in terms of direct adaptation associated with leaf size effects in optimizing important physiological functions of individual leaves along environmental gradients (e.g. involving temperature and moisture). An additional interpretation is explored here, where adaptation may be more directly associated with the number of leaves produced and where relatively small leaf size then results as a trade-off of high ‘leafing intensity’—i.e. number of leaves produced per unit plant body size.
The relationships between mean individual leaf mass, number of leaves and plant body size were examined for 127 species of herbaceous angiosperms collected from natural populations in southern Ontario, Canada.
In all, 88% of the variation in mean individual leaf mass across species, spanning four orders of magnitude, is accounted for by a negative isometric (proportional) trade-off relationship with leafing intensity. These results parallel those reported in recent studies of woody species. Because each leaf is normally associated with an axillary bud or meristem, having a high leafing intensity is equivalent to having a greater number of meristems per unit body size—i.e. a larger ‘bud bank’. According to the ‘leafing intensity premium’ hypothesis, because an axillary meristem represents the potential to produce either a new shoot or a reproductive structure, high leafing intensity should confer greater architectural and/or reproductive plasticity (with relatively small leaf size required as a trade-off). This greater plasticity, we suggest, should be especially important for smaller species since they are likely to suffer greater suppression of growth and reproduction from competition within multi-species vegetation. Accordingly, we tested and found support for the prediction that smaller species have not just smaller leaves generally but also higher leafing intensities, thus conferring larger bud banks, i.e. more meristems per unit plant body size.
Species of herbaceous angiosperms ranging widely in leaf size were collected from natural populations in southern Ontario, Canada, in the vicinities of Queen's University and the Queen's University Biological Station near Kingston, Ontario (44°16′N, 76°30′W), the Royal Botanical Gardens in Hamilton, Ontario (43°16′N, 79°54′W) and the Agriculture and Agri-Food Canada Research Station in Harrow, Ontario (42°02′N, 82°54′W). A total of 127 species from 31 families were included (Appendix). Collection sites spanned a wide range of habitat types, including woodlands, meadows, farm fields, lakeshores, marshes, roadsides and other disturbed habitats. As many species were sampled as time permitted within the growing season. Since leaf size variation was of specific interest, care was taken to include species with the largest and smallest leaves that could be found. Sampling focussed on dicots because of their wide variation in leaf size and architecture and because of their more discrete leaf size at maturity; hence, grasses—where young leaves are often not visible and hence difficult to count—were specifically excluded from sampling. Cultivated species were not collected, and vines were avoided because of difficulty in collecting entire plants. Clonal species were generally avoided unless it was possible to easily distinguish individuals as distinct ‘rooted units’ (Aarssen 2008) based on minor soil excavation (this is another reason why grasses were impractical to collect). Aside from these criteria, species were sampled as they were encountered based on their availability and ease of access to local populations within the study region.
For each species, three to five replicate plants were collected, depending on local abundance and specimen quality. Only reproductive individuals (showing flower buds, flowers or fruits) were selected to ensure that the plants had attained typical adult size, as well as to aid in positive identification; otherwise, plants were selected randomly from within the local population of each species. Each plant's height (maximum vertical extent) was measured and the entire above-ground biomass from the rooted unit was collected and transported to the laboratory. Samples were stored in a freezer until processing was possible.
For each sampled plant, the total number of leaves produced was determined by counting intact green leaves, dead (withered or brown) leaves and leaf scars (representing lost leaves) associated with the current year's shoot growth. The intact green leaves (including petioles) were collected and dried at 60°C for at least 48 hours, after which their total mass was recorded and mean dry mass per leaf was calculated. Total per-plant leaf dry mass was adjusted to take account of lost leaves (indicated by leaf scars). Dry mass was also recorded for all remaining (non-leaf) above-ground tissue. The vast majority of this was supporting shoot biomass but also included some reproductive biomass (the latter was not recorded separately because of the inordinate time required to separate it). Leafing intensity, analogous to Kleiman and Aarssen's (2007) measure, was calculated as the total number of leaves divided by the total biomass of supporting tissue—i.e. the total remaining (non-leaf) above-ground biomass.
To assess the effect of phylogeny on the leaf size/number trade-off, we constructed a phylogenetic tree based on the family-level phylogeny of Stevens (2001), developed from work of the Angiosperm Phylogeny Group. Published data on phylogenies were used to resolve the tree from the family level to genus level, and where these were not available for the species of interest, taxonomic groups were assumed to be monophyletic and used as a surrogate for phylogeny (Garland et al. 1992). Where more than one species shared a genus, mean trait values were assigned. The full tree is given in the supporting information (See online supplementary material). All tree branch lengths were set at 1.0. Unresolved nodes in the phylogeny were left as soft polytomies. Consequently, we opted for a conservative approach when testing hypotheses with independent contrasts, subtracting one degree of freedom for each unresolved branch (Garland and Diaz-Uriarte 1999, Purvis and Garland 1993). To test for the statistical adequacy of branch lengths, we performed the diagnostic analyses suggested by Garland et al. (1992). We computed standardized contrasts for log (leafing intensity) and log (mean leaf mass) across species using phylogenetically independent contrasts using the PDTREE module of the phenotypic diversity analysis programs (Midford et al. 2005) with Mesquite software (Maddison and Maddison 2006).