Labs are, and always have been, under selective pressure. They evolve.
Or go extinct.
In recent times we've seen an increase in this pressure. Fewer TT positions. Decreasing funding rates. Perhaps even a push to eliminate single R01 (or equivalent) labs. It would appear labs/PI's need to evolve more rapidly than in the recent past. What traits will be most beneficial in this rapidly changing landscape?
I would suggest two: efficiency and flexibility.
This is the ability to get top-notch work done in the most rapid, cost-effective, and yet thorough, manner. Note that I said "top-notch." One can certainly churn out the research rapidly if one is willing to be second-rate. But you don't want to be second-rate, do you? Do it right or bugger off.
How to be efficient? Think ahead. Focus. Plan out what you're doing and how carefully. Be organized.
Recognize that you'll need help. The days of single labs being able to do it all are long gone. Unless you're either a supermegalatron-sized lab or your research is irrelevant. Collaborations are where it's at. Build a network of friends, collaborators and even foes. Use them and let them use you. But not in a bad way. Be a great collaborator and you'll attract great collaborations. And keep an eye on things outside of your sub-sub-sub-field. You never know when you'll come across an awesome approach being applied in a different sub-field that can push forward your own.
Why cost-effective? Because that gives you...
The ability to change directions as necessary. Follow where the data is taking you. Start up the new projects (or new branches of existing projects) you're going to need for the 2-3 R01 (or equivalent) applications you're going to be submitting each year.*
I'm not advocating jumping on the latest "hot" science. I have a strong distaste for those that do that. I also dislike those who constantly change directions based on where the most funding is perceived to be (although to some extent this is something we all have to be cognizant of). But banging away at the same old, same old when the data is yelling "go that way bozo!" is just plain stoooooopid.
Learn new stuff. New techniques. You haven't picked up a new technique in the last year, either in-lab or via a collaboration? That's likely not good.
Darwin may not have been referring to research, but he was correct about evolving or disappearing.
* Assuming you already have funding. The number of submissions might need to be higher if you're not.