The canonical paper on this topic is Edmund Schulman's classic "Longevity under Adversity in Conifers" (1954)[1]
When I worked in a tree-ring dating lab, we always sought out the harshest local environments in a region and then at a given site, the harshest micro-environments (typically water limited but fire protected) because that's where you find the most ragged, pathetic, and correlatively oldest trees around with the most climate sensitivity in their ring patterns. This makes the ideal trees to date and build climate chronologies from.