Transposable elements (TEs) are prevalent in plant genomes  and ubiquitous among eukaryotes . Although TEs comprise most of an average plant genome , their content varies markedly among populations [4, 5] and species [6, 7]. For example, TEs make up ~70% of the maize genome  but just 10% of the Arabidopsis thaliana genome . Moreover, TEs can accrue rapidly after polyploid and hybrid speciation events [10, 11]. These observations raise questions about the evolutionary forces that govern the distribution of TEs within plant genomes.
Population genetics has the potential to illuminate these forces, but our understanding of the population genetics of TEs has been based primarily on studies of Drosophila melanogaster. These studies have revealed that there are far fewer TE insertions in the D. melanogaster genome than possible insertion sites [12, 13] and that insertions tend to be at low population frequencies [12–14]. Both observations suggest that the spread of TEs is countered by natural selection [15–18]. However, the precise nature of selection against insertions is unclear. Some insertions may disrupt gene products or gene expression . Purifying selection against these deleterious insertions could be the driving force that governs observed TE distributions [15, 19–23]. Another possibility is that TEs facilitate deleterious chromosome rearrangements through non-homologous (or ectopic) recombination [18, 24–28].
The mating system of host species is likely to be an important factor that shapes TE dynamics [27, 29, 30]. For example, in highly homozygous selfing species most TEs have a paired homologous allelic partner, which reduces the probability of an ectopic recombination event [27, 29, 30]. If selection against TEs is primarily mediated by these ectopic events, then selfing species are predicted to have less efficacious selection and higher TE copy numbers than outcrossing species. Conversely, the deleterious effects of recessive TE insertions are expected to be stronger in a homozygous selfer, which may result in more efficacious selection and fewer TEs in selfers [20, 29, 30]. Thus, the effect of breeding system is difficult to predict precisely, but simulations of TE population dynamics provide evidence to support the possibility that both ectopic recombination and deleterious insertions will lead to differences in TE accumulation between selfers and outcrossers [29, 30].
Mating system influences the efficacy of selection against TEs in at least two other ways: First, the effective population size (N
) in a selfing species is expected to be half that of an otherwise identical outcrosser [31, 32]. Population size has a direct effect on the efficacy of selection, because efficacy is reflected in the compound parameter N
s, where s is the strength of selection. It is thus not surprising that empirical studies suggest that shifts in N
over time influence the number and frequency of TEs [5, 33]. Second, inbreeding reduces the effective recombination rate, which may lead to the accumulation of weakly deleterious TE insertions  via Hill-Robertson effects . Observations that TEs accumulate on non-recombining sex chromosomes support this conjecture [36, 37].
Despite predictions that TE population dynamics may differ markedly between selfing and outcrossing species, comparative data are quite rare. Recently, however, Dolgin et al.  documented that population frequencies of Tc1-like insertions are higher in selfing Caenorhabditis elegans than in outcrossing C. remanei. This pattern of diversity suggests less efficacious selection against insertions in the selfing species; indeed, Dolgin et al.  tentatively conclude that Tc-1 element insertions are effectively selectively neutral in C. elegans.
Plants are particularly well suited for inter-species comparisons of TE population dynamics because of broad diversity in mating systems. Studies of selfing and cultivated Lypersicon species have generally shown differences in TE complement that are consistent with less efficacious selection against insertions in selfing species. For example, the Lyt1 element family has higher copy numbers in the selfing members of the genus [27, 39], and copia-like insertions are generally found at higher population frequencies in selfers . In perhaps the best known study TE diversity between plant species with contrasting mating systems , Wright et al.  compared population diversity of Ac-like elements between selfing Arabidopsis thaliana and outcrossing A. lyrata. Ac-like insertions were slightly more numerous in A. thaliana but segregated at significantly lower frequencies in A. lyrata, consistent again with reduced efficacy of natural selection against insertions in the selfing lineage.
Although the limited data gathered to date suggests that selection against TEs is less efficacious in selfing lineages, it is difficult to determine whether extant patterns of TE diversity are due to the effects of selection or complicated by other factors that may differ between species, such as demographic history and transposition dynamics . How might one discriminate among these factors? One approach is to increase sampling to multiple TE families and multiple populations. If patterns of TE diversity vary across element families, transposition dynamics may play a major role in explaining differences between species like those observed for Ac-like elements . In contrast, if diversity patterns are consistent across TE families, forces that affect whole genomes (such as demography and breeding system) may be the primary determinants of TE diversity. Here we extend the study of Wright et al.  to contrast TE population genetics between A. thaliana and A. lyrata, generating polymorphism profiles from four populations of A. thaliana representing seven TE families. We compare these A. thaliana data to data gathered from four populations of the outcrossing congener A. lyrata . By contrasting TE frequencies and patterns across species, populations, and TE families, we gain insight into the relative roles of transposition, demography, and breeding system in shaping TE diversity.