2. I'm very interested to know the answer to your second point myself. As displayed in figure 2, it looks like it wasn't exactly simultaneous or equal in magnitude everywhere. It does look like most geographic clusters did see something of a drop, at some point. Eyeballing it, it appears that the near-East and the Caucasus had the relatively most-significant drop. The African cluster appears to have had a more modest drop and at a later date. However, I wouldn't rely too heavily on trying to read more into the figures than the authors did (and as I just tried to do).
I suspect the resolving power their data sample gave them was insufficient to resolve your question. While it was enough to demonstrate the existence of such a drop, I would be skeptical of its power to demonstrate the relative severity or timing between each geographical cluster. (If I was the author, this finding would be begging for more grant money to do exactly that: gather more data and nail down exactly how much, when, and where the Y diversity was lost.)
Final point: it's not necessary to assume everyday life was brutal or violent. I'm not an anthropologist, so take my reference to Napoleon Chagnon with a grain of salt. The Amazonian Yanomamo people he lived among had high rates of violent death compared to other causes of death, yet everyday life was peaceful. Most of the bloodshed and conflict he documented occurred over very few days in quick moments. When a population is low, it doesn't take much to move the % up a lot.
Whatever caused the drop in Y diversity, if it happened over several thousand years, the yearly attrition rate could have been very low also.
