I need to start keeping track of my references, here's one without any.
I once heard about a study where it was determined that gazelles that jump around more often are more likely to reproduce, even though they are more likely to get eaten or injured as a result of the constant bouncing. The bouncing alone was a negative trait. Initially, I figured this was some way to use sexual selection to self regulate fitness in a species that relied on being predated, so that the gazelles didn't get too hard to eat and ensure a balanced ecosystem. But then it occured to me that the bouncing might act as some sort of trainer gene. By imposing a behavior that made it more difficult to survive, it ensured that only exceptional animals who exhibited the behavior were the ones who would survive. This would no doubt put pressure on a host of traits, which would work their way into the entire population over several generations.
I love sexual selection.
Lady's choice ;)
