What ever the mutational process, there does appear to be some biases in the mutation rate. An in vitro study has found evidence that repeat length and base composition affect the mutation rate, i.e. dinucleotide repeats mutate faster than tri's, and sequences with a high AT content mutate faster than those with a GC content (Schlotterer and Tautz, 1992). This indicates that template stability may affect the mutation rate, perhaps by reducing the frequency of strand-slippage events. Levinson and Gutman (1987b) also found evidence in prokaryotes that microsatellites with a larger number of repeats mutate at a higher rate than those with a smaller number. However, the in vitro study of Schlotterer and Tautz (1992) and a survey of human microsatellite variation (Valdes et al. 1993) have found no evidence for this bias. Finally, a comparison of human, chimpanzee, gorilla, orang-utan, baboon, and macaque microsatellite repeat length has shown that humans have significantly longer loci than those found in the other primates. Rubinstein et al. (1995) specualate that this indicates that there is a mutational bias in human microsatellites toward an increase in length.
Even though there may be bias towards an increase in repeat length, it is clear from empirical data that there is a size limitation on the number of repeats (Bowcock et al. 1994). Of 383 CA repeat microsatellites found in humans, only 45 had over 20 repeats (Valdes et al. 1993). The mechanism for limiting the number of repeats is not known, however, it may be related to the chromosomal instability associated with various human genetic diseases such as fragile X (Fu et al. 1991). The consequence of the observed selection on repeat length is that the number of homoplasies should increase over that predicted by the SMM, as the mean size of the microsatellite alleles increase. Although the effects of a size constraint on the linearity of distance measures has been investigated (Goldstein et al. 1995a) more research on the nature and variability of this constraint is needed in order to properly model its effects.
Empirical data also suggests that the structure of the microsatellite repeat may be very important in determining the operating mutational process (Estoup et al. 1995a, 1995b). Estoup et al. (1995a) examined the variation found at 2 loci with irregular microsatellite repeats in bee's to investigate the levels of homoplasy that exist between populations. Irregular repeats are core repeats that are interrupted by point mutations. The presence of these mutations allows alleles with similar sizes to be distinguished. Not surprisingly, as the taxonomic level of the comparisons increased (i.e. among species), the amount of homoplasy increased. Of interest, however, they also found that the frequency of multi-step mutational events had increased at both loci and the data was significantly different than that predicted by the SMM. An additional study of 7 honey bee microsatellite loci also found that the observed allele frequencies were better explained by the IAM ( Estoup et al. 1995b). In this survey the majority of loci examined were not simple repeats but composites made up of 2 or 3 core repeat units. The two phase model (TPM) was not tested in either study (see next section). These studies are of interest because they suggest that the more complex the repeat structure, irregular or composite, the lower the likelihood of single step mutational events.