It’s not true for bacteria, but it’s very much true for vertebrates, and for most animals. Bacteria undergo lots of horizontal gene transfer, while vertebrates have very little. Vertebrates, not surprisingly, therefore yield very good, consistent phylogenetic trees when their DNA is analyzed.
I took another look at GULO, by the way, to see if some common mechanism really could explain the exons that were lost in both humans and guinea pigs. Your argument was that humans lost 7 out of 12 exons from the gene, and that it was unlikely that the exons guinea pigs happened to lose 2 and a half of the same exons by chance (exons 1, 5 and part of 6). Instead, they must have undergone similar deletions because they had similar genomes.
So what does the sequence actually look like when you compare the two? Conveniently, a pairwise alignment of the human and mouse genome is available online, as is an alignment of guinea pig vs mouse (yay, Jim Kent and UCSC). Since mouse has a functioning copy of the GULO gene, it provide a reference copy of the gene that has all of the exons intact. Here’s what the first half of the gene looks like:
In the plot, the x axis shows the position in the mouse chromosome and the y axis the equivalent position in the corresponding guinea pig chromosome. Blue segments show where the two still have similar DNA; where there is a gap, there has been an insertion or deletion in one of the two species. The green segments above that show the same thing for the human chromosome – the segments where it resembles the mouse gene. Along the bottom, the red blocks show where the GULO exons occur in the mouse. (They are numbered right to left because the gene lies on the reverse DNA strand.)
The first thing to note is that none of the genomes are very similar to one another, and that there are numerous insertions/deletions throughout the gene. Also, there is no sign that similar deletions are occurring at the same places. Exon 6 is completely missing from the human copy of the gene because a substantial deletion has taken out the entire region, while the guinea pig exon has been clipped by a different deletion.
Things are different in the case of the other two exons in question (exons 1 and 5): it turns out in both cases that one of the two species isn’t even missing the exon. Exon 1 is still present in guinea pigs, while exon 5 is still present in humans. I’m surprised that exon 5 was listed as missing in humans, since the sequence there isn’t that different from mouse. Here are the two lined up:
Exon 1 is much more heavily altered by mutation, probably because it’s almost all noncoding, and therefore mostly not constrained by purifying selection even for functioning copies of the gene. Here is that stretch of the gene in mouse and guinea pig:
You can sort of still see the resemblance, but it’s only because it’s part of a large block of similarity between gp and mouse that we can we sure they come from a common original state. For both of these exons, the matching sequence in the other species was completely removed by large deletions, a completely different mutation process.
Based on the actual sequence comparison, then, the argument is plainly wrong that humans and guinea pigs have lost overlapping exons because of common mutational processes in the two species.