We present results concerning the power to detect past population growth us
ing three microsatellite-based statistics available in the current literatu
re: (1) that based on between-locus variability, (2) that based on the shap
e of allele size distribution, and (3) that based on the imbalance between
variance and heterozygosity at a locus. The analysis is based on the single
-step stepwise mutation model. The power of the statistics is evaluated for
constant, as well as variable, mutation rates across loci. The latter case
is important, since it is a standard procedure to pool data collected at a
number of loci, and mutation rates at microsatellite loci are known to be
different. Our analysis indicates that the statistic based on the imbalance
between allele size variance and heterozygosity at a locus has the highest
power for detection of population growth, particularly when mutation rates
vary across loci.