Leitch, 1995; Bennett and Smith, 1991); this variation may reflect different amounts of retrotransposon proliferation or independent chromosomal duplications.

Studies of the Adh1 region by Bennetzen and coworkers (Springer et al., 1994; SanMiguel et al., 1996, 1998; Tikhonov, 1999) have provided invaluable insight into the structure and dynamics of maize intergenic DNA, but at least three important questions remain.

Question 1. Are retrotransposons distributed homogeneously among genomic regions? The Adh1 studies, as well as other studies (B. Meyers, personal communication), suggest that retrotransposon distribution may be roughly homogenous among regions of the maize genome. However, other lines of evidence suggest that such homogeneity is unlikely. For example, evolutionary theory predicts that transposable elements should gather in regions of low recombination, such as centromeres (Charlesworth et al., 1986, 1994). This prediction holds in Arabidopsis, where sequence data from chromosomes 2 and 4 indicate an increase in the frequency of transposable elements near centromeres (Copenhaver, 1999).

There are other reasons to suggest that retrotransposon distribution may not be homogeneous throughout the maize genome. One obvious reason is that there are heterogeneities in chromosomal structure, such as euchromatin, heterochromatin, nucleolus organizing regions, telomeres, centromeres, and knobs. Nonetheless, recent research indicates that retrotransposons constitute a substantial fraction of both heterochromatic centromeres and heterochromatic knobs (Ananiev et al., 1998a, b); for one chromosome 9 knob, retroelements comprise roughly one-third of knob-specific clones (Ananiev et al., 1998c). Many of the retrotransposons in knob and centromeric DNA belong to the element families found in the Adh1 region. Despite these commonalties, there are also substantive differences among knobs, centromeres, and the Adh1 region. For example, centromeres contain a centromere-specific retrotransposon (CentA; Ananiev et al., 1998b). Similarly, chromosomal knobs associate with 180-bp and 350-bp repeat elements that are otherwise sparse in the genome (Ananiev et al., 1998a). Altogether, the emerging picture is one in which some retroelement families are fairly ubiquitous, and other repetitive DNAs are heterogeneous in their distribution (e.g., Zhang et al., 2000).

The work of Bernardi and coworkers (Barakat et al., 1997; Carels et al., 1995) is an intriguing addition to this picture. They fractionated DNA by G:C content and hybridized each G:C fraction to 38 coding-region probes. The coding genes hybridize almost exclusively to a DNA fraction of very