Family trees are funny things for anything within the history of ideas. The semantic space they map has always got an approximate time-linearity on one axis, but the other axis remains ambiguous. Most of them don't show the nature of the links, and very few reflex the nature of the language as a discourse. The LISP genealogy for HOPL2 is actually the only one that properly matches the complexity - but this was slightly tongue-in-cheek.
My favourites are those localised trees in Baron's "Computer Languages: a guide for the perplexed". I have adopted her approach and used local ones as well.
There are lists aplenty about on the net, but I found many of them to be unsourced and contradictory. A lot of people have taken the standard list started by Bill Kinnersley and continued by various people, and largely incorporated into FOLDOC. The list is particularly difficult to use when it comes to the earlier languages, and many versions of it exist solely to drag the viewer to a commercial site.