The distances between 'groups' will be based on some sort of 'group averages', which may be expected to produce genetic distances that may be smaller as compared to genetic distances measured using the 'individuals' as basic units. The averaging process leads to some loss of information.   

There will be differences between individuals due to common polymorphisms which would be averaged-out when comparing groups

Ok. Thank you very much.

Splended

There are differences from one individual to another. Depending on the individuals sampled, estimated differences between populations can be larger or smaller than the true differences. When comparing two samples each made up of several individuals representative of the respective population variability you obtain an estimate closer to the real genetic distance

As distâncias entre "grupos" está baseada na "médias do grupo ', no entanto o processo de cálculo da média leva a alguma perda de informação.

Thank you all for the valuable contributions!