They expand their genome size with the help of transposable elements - sequences of DNA that copy and insert themselves somewhere else in the genome.
Maize, for example, doubled its size in only three million years. That makes it the Oprah Winfrey of plants genomes.
But, luckily, there is a way for the plants to lose the extra pounds (i.e. megabases). They remove the extra DNA through recombination.
For three different species of cotton from the genus Gossypium (G. herbaceum, G. raimondii, and G. exiguum) and the sister group of the genus as an outgroup (Gossypioides kirkii), they used genomic shotgun sequences to estimate numbers of accumulation and deletion of a retrotransposon, Gorge3.
Surprisingly, Gorge3 copy numbers were more abundant in, pre-Gossypium than at any other time point, for all taxa. Copy number estimates from this oldest time point in G. kirkii (3,001 ± 2,445) and the D-genome [G. raimondii] (4,731 ± 2,725) were not significantly different from one another, but many retained ancient copies of Gorge3 were identified in the A-genome [G. herbaceum] (22,272 ± 6,331) and twice as many ancient copies were recovered from the K-genome [G. exiguum] (43,037 ± 9,063).A nitpick: I'm not sure what they mean. The second comma in the first sentence seems misplaced to me. Removing the comma (comma), it says that the outgroup, G. kirkii, has more Gorge3 copies than the three species, and yet the numbers in this paragraph tell a different story (I think, but I could be wrong). I wonder if text is missing before that comma? Generally it is not a carefully written paper. They write "posses" (and don't mean the sheriff's search party), and the paper is written in passive voice, which is generally discouraged (but often done). Just saying.
Using a simple growth model the authors were able to infer the changes in the rate of gain and loss for all four species. The results are that two species have lost a considerable amount of Gorge3, while two other have be gaining.
Fig. 3. Phylogenetic relationships and estimated rates of Gorge3 gain and loss among diploid members of Gossypium. Branch lengths are to scale. Numbers above the branches represent the estimate of the exponential rate of change in Gorge3 DNA with confidence intervals in brackets. Taxa are shown at tips with entire genome size as well as the amount (in Mb) of extant DNA from Gorge3 elements. (Click for larger image.)
Even among closely related species there is ample variation in genome sizes. Why do some genomes have a ‘‘one-way ticket to genomic obesity,’’ while others couldn't gain weight if they tried (sort of like Michael Phelps and other people with tapeworms)?
It is not known why genome expansion is differently tolerated in different lineages. One could imagine that some external, environmental factor is responsible; the three Gossypium species evolved on different continents. There can be benefits to adding extra DNA to the genomes, since this gives material that evolution can work with, but at the same time it is energetically costly to make all that extra DNA every time a cell is duplicated.
Something in one environment could favor large genome sizes (perhaps this could be investigated by comparing with different species in the same environment?) while there not being the same selection pressure in another. Or, the environment affects the fitness of the transposons differently (say, when it's really hot Gorge3 is more successful at jumping around). Or, Gorge3 is really different in the four different species, and that explains their different rates of proliferation.
Hawkins JS, Proulx SR, Rapp RA, & Wendel JF (2009). Rapid DNA loss as a counterbalance to genome expansion through retrotransposon proliferation in plants. Proceedings of the National Academy of Sciences of the United States of America PMID: 19815511