Several conclusions arise from our whole-genome SDs analysis. First, there is an accumulation of relatively recent human SDs within some chromosomes that carry an evolutionary rearrangement between human and chimpanzees. Seven of the nine major inversions between humans and chimpanzees occurred in the chimpanzee lineage (HSA4, HSA5, HSA9, HSA12, HSA15, HSA16 and HSA17), thus inversions cannot be the cause of that accumulation. The classical explanation of the accumulation would be that some of these young SDs predate the split of humans and chimpanzees and, thus, that they originated the inversions via non-allelic homologous recombination, but this seems unlikely in the light of their location. Our observations are consistent with an alternative scenario in which both chromosomal rearrangements and SDs are consequences of a third factor, perhaps regions of high instability [29, 56]. This has been suggested in opposition to the idea that rearrangements and SDs are related only because highly similar regions promote rearrangements by non-allelic recombination [8–12]. A final possibility is that we are observing an excess of similar duplications in pericentromeric regions, specially in HSA5 and HSA9, in which there are an excess of young human SDs (> 98% ID) within regions that were inverted in chimpanzees. Even if we endeavored to remove the effect of centromeres, the possibility remains that particularly strong local effects were not accounted for. Only further research on primate SDs will allow to ascertain the involved phenomena and the order in which they occurred.
Several authors have found that the association among rearrangement breakpoints and segmental duplications is maintained between different lineages, but not within the same lineage [6, 9, 13]. For instance, primate segmental duplications occur at specific locations that are enriched for mouse-human synteny and mouse-rat synteny breaks. As the majority of synteny rearrangements have occurred in the rodent lineage, there cannot be a causal relationship between the two. Rather, it must be the case that primate segmental duplications tend to appear at the same locations in which rodent chromosomes have rearranged. Thus, instability would seem a long standing property of these genomes at these locations. In addition, She et al.  described a non-uniform distribution of intrachromosomal human SDs and highlighted nine autosomal human chromosomes with an excess of young human SDs, seven of which presented rearrangements between humans and chimpanzees (out of which five were chimpanzee specific). These observations provide evidence for a link between expansions of recent SDs in one lineage and chromosomal rearrangements in the other. Only deeper analysis of the two chimpanzee chromosomes that carry human-specific rearrangements (HSA1 and HSA2) will help to clarify any direct relationship among chromosomal rearrangements and expansion of SDs. This analysis, however, is beyond the scope of the present work and would require a higher quality sequence assembly of the chimpanzee genome.
Several explanations can be put forward as to why chromosomal rearrangements and young SDs should accumulate in sister lineages. The first one relates to the aforementioned instability regions. A recent change in the understanding of the evolution and behavior of SDs [56–58] poses that there are "core elements" that may act as sources for the dispersal of new SDs, by creating a large number of copies of themselves. These copies tend to cluster by means of local duplications. Thus, one explanation for our results would be that some core elements were present in the chromosomes ancestral to those that currently harbor inversions and SDs in humans and chimpanzees. As inversions decrease recombination between homologous chromosomes [31–33], core elements becoming active and expanding by local copies in a given class of chromosome, would be less likely to be eliminated by recombination from their source regions while rearrangements are still segregating in the ancestral population. Thus, these core elements would accumulate copies of themselves only in the lineage in which they appeared. Moreover, the reduction of recombination caused by inversions  may also prevent the dispersal of the other associated SDs (not just the "core" elements). SDs trapped within rearrangements would be more similar to the "original" state because they would be prevented from invading other regions or chromosomes that could affect mutation rates and thus produce highly divergent SDs copies.
A second possibility is that lower recombination rates themselves could help explain our results. As suggested in previous work [60–63], there is a positive correlation among low recombination rates, low diversity within species, and low divergence that can be explained by a mutagenic effect of recombination. While inversions are segregating, regions within rearrangements have lower recombination rates and, thus, they should present lower divergence (either inter-specific or intra-specific). Of course, this would only be the case if rearrangements had been segregating in the population for a long time, so that the reduction of recombination could have a detectable impact on mutation rates.
Finally, some of the pairwise alignments classified as young SDs may in fact not be young, but their high identity may have been maintained by gene conversion . Gene conversion is a homogenizing force that might erase differences among copies leading to underestimations of the age of SDs. It is possible that during the segregation of new rearrangements, the resolving structure of the few recombination events taking place within inversions would be biased towards increased gene conversion instead of the reciprocal exchange of chromatids. This would help explain the excess of highly similar tracks of SDs in one lineage together with inversions in the other lineage. However, this possibility implies that most gene conversion events ought to have happened before the separation of the two lineages and while the inversions were segregating in the population, which is unlikely. Moreover, She et al.  concluded that gene conversion events can not explain most of the high sequence identity of SD copies.
Secondly, we conclude that old and young SDs evolve at different rates when compared to single-copy DNA, hinting at different evolutionary trajectories for different SD classes. It is possible that young SDs are reflecting the history of recent primate evolution - which led to our species - while old SDs may reflect periods of duplication early during primate evolution. Our results, for example, support a recent expansion of young SDs or a more complex interaction among recombination and SDs. The latter appears to be the case for SDs in telomeres, where young SDs are marginally more divergent, but are significantly shorter than elsewhere in the genome, maybe as a result of telomeres having higher rates of recombination [64, 65]. In contrast, older SDs do not show this trend, which could be expected since telomeres are likely to have moved during primate evolution [66, 67].
Regarding centromeres, and probably as a result of their decreased recombination rates [64, 65], we obtained larger sizes of pairwise alignments of SDs. However, as centromeres have been reported to be prone to repositioning during evolution , this result could be reflecting some other cause rather than a direct recombination effect. SDs in HSAY are also longer, which could be related to the lack of recombination in that chromosome or with recent, HSAY-specific, SD expansions.
Our main conclusion regarding major rearrangements between humans and chimpanzees is that young SDs located in rearranged chromosomes are longer and exhibit greater sequence identity than SDs located in colinear chromosomes. This could be expected, since rearrangements are known to be either human or chimpanzee specific and, thus, old SDs should not be affected by such recent rearrangements. Still, both young and old paralogous copies of SDs tend to be larger within rearranged chromosomal regions. This is also the case for smaller rearrangements that have been detected in silico . These are puzzling patterns, hinting at some period of decreased recombination within rearranged regions. Finally, we observed higher levels of intraspecific divergence between SDs within smaller inversions . Altogether, these data suggest that chromosomal rearrangements might have affected SD divergence rates during primate evolution.
Our third and last finding is that interspecific SD divergence displays rates and patterns that are roughly equivalent to those of single-copy DNA. SDs located in telomeres and in HSA19 show higher levels of interspecific SD divergence. Also, SDs located in rearranged chromosomes show lower divergence between species. Still, there are some discrepancies between single-copy and duplicated DNA, such as the higher divergence between SDs located in centromeres or the lower divergence of SDs within small inversions. Finally, HSAY does not show the higher degree of divergence reported for single-copy DNA [40–42], perhaps as the result of the recent expansion of young SDs in that chromosome  or of extensive gene conversion .
As to individual inversions, HSA9 stands out as the only chromosome showing significantly higher human-chimpanzee divergence within its rearrangement. This suggests a burst of interspecific divergence within the inversion, that could perhaps predate speciation. Therefore, HSA9 is currently the best candidate to further study any potential relationship among SDs, rearrangements, divergence, and speciation. If chromosomes have played any role in any of the speciation events that led to humans and chimpanzees, it is clear that not all of them would have made the same contributions and, thus, would not bear the same molecular signatures. We should keep this in mind when trying to explain why HSA4, which presents high divergence of single copy DNA located within its inversion , does not present any particular pattern when considering its duplications. Also, certain chromosomes (such as HSA4, HSA5, HSA9, HSA15 and HSA16) have been pinpointed as the most dissimilar between humans and chimpanzees in terms of the expression intensities of their genes , findings which are only partially consistent with the results presented here.