Of the world's 250000 angiosperm species, only about 200 are recognized as important weeds. 4C nuclear DNA amounts were estimated for 39 such species. Success for many important weeds is suggested to reflect several traits known to correlate with low DNA C-value, so such weeds may have smaller DNA C-values than other species. Our work tests this hypothesis, comparing DNA amounts in 156 species recognized as important world weeds or British garden weeds, with 2685 other species. DNA amounts did not differ significantly between the two weed samples, but weeds showed highly significant differences from other species. For example, nuclear DNA amount in weeds (mean 11.74 pg) was smaller than in other species (mean 28.13 pg), and restricted to the lowest 20 % of their range. Similarly, DNA amount per genome in weeds (mean 3.79 pg) was smaller than in other species (mean 12.14 pg), and restricted to the lowest 10 % of their range. As significant differences between weeds and other species remain for almost all subsamples tested, this contrast is widely distributed. So it is important to ask how selection against high nuclear DNA amount and genome size in weeds operates. The probability of a species being a weed fell significantly with increasing nuclear 4C DNA amount, and mean genome size, reaching zero just above 100 pg, and 19 pg, respectively. Moreover, polyploidy was significantly more frequent in weeds (51 %) than in other species (27 %), increasing with nuclear DNA amount in both, reaching 100 % in weeds with the highest 4C DNA amounts, but only 41 % in other species. Thus, selection for polyploidy in weeds may partly reflect their increased genetic variability, independent of DNA amount. However, such selection pressure grows strongly with rising nuclear DNA amount, and this may act mainly on correlated factors including faster development. (C) 1998 Annals of Botany Company.
