Correlated variation and population differentiation in satellite DNA abundance among lines of Drosophila melanogaster

Tandemly repeating satellite DNA elements in heterochromatin occupy a substantial portion of many eukaryotic genomes. Although often characterized as genomic parasites deleterious to the host, they also can be crucial for essential processes such as chromosome segregation. Adding to their interest, satellite DNA elements evolve at high rates; among Drosophila, closely related species often differ drastically in both the types and abundances of satellite repeats. However, due to technical challenges, the evolutionary mechanisms driving this rapid turnover remain unclear. Here we characterize natural variation in simple-sequence repeats of 2-10 bp from inbred Drosophila melanogaster lines derived from multiple populations, using a method we developed called k-Seek that analyzes unassembled Illumina sequence reads. In addition to quantifying all previously described satellite repeats, we identified many novel repeats of low to medium abundance. Many of the repeats show population differentiation, including two that are present in only some populations. Interestingly, the population structure inferred from overall satellite quantities does not recapitulate the expected population relationships based on the demographic history of D. melanogaster. We also find that some satellites of similar sequence composition are correlated across lines, revealing concerted evolution. Moreover, correlated satellites tend to be interspersed with each other, further suggesting that concerted change is partially driven by higher order structure. Surprisingly, we identified negative correlations among some satellites, suggesting antagonistic interactions. Our study demonstrates that current genome assemblies vastly underestimate the complexity, abundance, and variation of highly repetitive satellite DNA and presents approaches to understand their rapid evolutionary divergence.