The evolutionary ecology of seed size.

Seed mass is a trait that occupies a pivotal position in the ecology of a species. It links the ecology of reproduction and seedling establishment with the ecology of vegetative growth, strategy sectors that are otherwise largely disconnected (Grime et al., 1988; Shipley et al., 1989; Leishman and Westoby, 1992). There is a startling diversity of shapes and sizes of seeds among the plant species of the world. Seeds range from the dust seeds of the Orchidaceae and some saprophytic and parasitic species (around 10 6 g), across ten orders of magnitude to the double coconut Lodoicea seychellarum (104 g) (Harper et al., 1970). Within species, seed size typically spans less than half an order of magnitude (about fourfold: Michaels et al., 1988). Most within-species variation occurs within plant rather than among plants or populations (Michaels et al., 1988; Obeso, 1993; Vaughton and Ramsey, 1998), indicating environmental effects during development rather than genetic differences between mothers. This chapter is concerned with the differences in seed size among species, and the consequences for vegetation dynamics and community composition. During the last 10–15 years, there has been considerable progress in the ecology of seed mass. Unlike many other areas of comparative plant ecology, we have substantial published information from several different scales and research styles. As well as field experiments and demographic studies with a few species at a time, we have simple experiments with larger numbers of species (ten to 50), quantification of seed mass and its correlates in whole-vegetation types (hundreds of species) and tests of consistency across different continents. The wide-scale quantification began as early as Salisbury (1942) and Baker (1972), but has been much added to and consolidated over the past 10 years (e.g. Mazer, 1989, 1990; Leishman and Westoby, 1994a; Leishman et al., 1995; Eriksson and Jakobsson, 1998). The work spanning large numbers of species is complementary to detailed experiments involving only a few species, giving a stronger sense of how widely the results from particular experiments can be generalized. Much of the literature examines how natural selection on seed size might be influenced by various environmental factors. In this context, it is at first glance surprising that seed size varies within communities across a remarkable five to six orders of magnitude (Leishman et al., 1995; Fig. 2.1). Further, there is strong overlap of seed-size distributions between quite different habitats. Within the temperate zone,