Intron size and genome size in plants.

It has long been known that genomes vary over a remarkable range of sizes in both plants (Bennett, Cox, and Leitch 1997) and animals (Gregory 2001). It also has become evident that across the broad phylogenetic sweep, genome size may be correlated with intron size (Deutsch and Long 1999; Vinogradov 1999; McLysaght et al. 2000), suggesting that some component of genome size evolution takes place within genes. Examples include humans and pufferfish (Fugu), where comparisons of 199 introns in 22 orthologous genes showed that introns in Fugu were on average eight times as small as those in humans, consistent with their ratio of genome sizes (McLysaght et al. 2000). Similarly, Deutsch and Long (1999) tabulated intron sizes across a broad phylogenetic spectrum of eukaryotes and noted a general but weak correlation with genome size, with humans having the most and longest introns (mean of 3.4 kbp) among the 10 taxa studied. Intron size is also correlated with genome size in Drosophila (Moriyama, Petrov, and Hartl 1998), showing that the correlation may extend to more recent divergences. At present there is little information on the correlation between genome and intron sizes in plants, although there are suggestions that plants with small genomes have smaller introns (Deutsch and Long 1999; Vinogradov 1999). Whereas broad comparisons across widely divergent taxa are now possible given completed draft sequences for the rice (Goff et al. 2002; Yu et al. 2002) and Arabidopsis (Arabidopsis Genome Initiative 2000) genomes, the divergence time between Poaceae and Brassicaceae is so large that the influence of genome size on intron size may be confounded by numerous other, unstudied covariables. It seems likely that more informative studies will involve closely related taxa that vary significantly in genome size but which share recent evolutionary history and a broad suite of life-history features. An additional advantage of comparing close relatives is that orthology among genes, and hence introns, may be more readily established. This latter point may be especially important, given the relative lability of copy-number for many gene families (Small and Wendel 2000). To exemplify this approach, we studied the relationship between intron size and genome size for or-