Does somebody know/use a method for phylogeny calibration of groups without fossil records? I've read about calibration using genetic distance of sequences to estimate approximate divergence times but I really don't know how it works.
That is called a molecular clock. Basically, at some point, someone will have generated a series of time-calibrated divergence dates and measures of sequence divergence for some group of organisms for some gene. If you examine the same gene and are willing to assume that your group is evolving at the same rate as the calibrated group, then you can use the calibrated "rate" (% divergence per million years) to estimate the age. There are more sophisticated methods available, such as a Bayesian program called BEAST, but it still requires some sort of calibration estimate to infer the ages of divergence of the nodes on the tree.
You need to either a known date or a rate in order to date your tree. The date does not have to come from a fossil, but could instead come from, for example, the age of a clade found only on one island, using the age of the island as the calibration age of the node (using a prior that says 'this node is MAXIMALLY this old because this is how old the island is, but it could be younger because the taxa may have colonised it at any point after its emergence'). Alternatively, you can use a rate inferred for a closely related group for the genes you are using (or at least genes of the same organelle - mitochondrial and nuclear rates differ, as do chloroplast rates in plants). It really depends on the genes and taxa with which you are working.
I highly recommend using BEAST to do your dated analysis.
For example i'm working with populations of fishes and my calibration points are not fossils, i'm dating my molecular clock with the uplift of one mountain that separates two basins.
Thanks Nicole, which software are you using for molecular clock analysis? Finally, I've found some fossils records of the group and other related taxa, so I'm using both, fossils and geological events that separated the two island where the species occur.