Within-genus size distributions in angiosperms: Small is better

Abstract: The composition of natural vegetation typically includes a wide range of species sizes, but usually, relatively small size classes are represented by the greatest number of resident species. This is the case even within habitat types where traditional theory predicts that the intensity of plant competition should be most severe. Here we show that distributions of species size metrics are generally right-skewed also at the within-genus level. Our analyses are based on published reports of maximum plant height, leaf size, and seed size for an unbiased sample of fourteen angiosperm genera, involving eleven families – including representation of both endemics and cosmopolitans, plus genera that differ widely in both size (number of species) and growth form diversity. Pair-wise relationships for these three traits were, in most cases, also generally positively correlated for con-generic species. These results indicate that the right-skewed species size distributions commonly reported for natural vegetation reflect a pervasive signal for adaptive size metrics, essentially unstructured by phylogenic relationships. From these data therefore, we infer that most episodes of species origination within virtually all angiosperm lineages must have generally involved favouring effects of natural selection associated with relatively small plant body size, leaf size, and seed size. This represents a striking disconnect with the conventional view that superior competitive ability in plants requires relatively large body size/biomass production. Our interpretation of why ‘small is better’ focuses on two recently developed hypotheses – the ‘physical-space-niche’ hypothesis, and the ‘fecundity allocation premium’ hypothesis. In future research, detailed phylogenies at the species level might be used to test the prediction that relatively small size metrics are especially well represented in the most recently evolved species within angiosperm lineages. [Copyright &y& Elsevier]