I have a dataset composed of 63 individuals, all males of nine species of the cicada genus Tettigettalna. This dataset was sequenced in two separate RAD-seq lanes. Lane one contains 35 individuals belonging to 7 species, and lane two is composed of 28 individuals representing four species. The only species these two lanes have in common is the outgroup which is represented by one individual in each lane.

The original idea was to analyze the lanes together in order to rebuild the phylogeny of the genus to which the species belong. However, when I analyze individuals from both lanes together, I obtain a pattern of artificial separation between them: the individuals corresponding to lane one form a group and the individuals from lane two form a completely separate group. When I try to analyze the lanes separately, the patterns I obtain are apparently random and incongruent with all the previous literature I know.

I have already tried dozens of parameter combinations of the RAD-seq assembly programs ipyrad and Stacks.

I would like to know if someone has experience with a situation like this and can give me some advice in how to deal with this dataset.