In a post from several years ago, Russ Poldrack said pretty much everything I have to say about why you would write academic papers in LaTeX. Like Russ, I also sometimes have trouble convincing collaborators to work with me in TeX. But I've found a surprising number of journals will accept TeX source. Elsevier, for all its faults, has an in-house TeX style that looks pretty nice. The only journal that has consistently asked for manuscripts to be reformatted in Word has been Perspectives in Psychological Science.
Here are just a few notes on my TeX workflow:
I still haven't found much in the way of track changes-style functionality aside from texdiff, which is better than nothing but occasionally barfs on more complex changes.
One of the best parts of writing APA formatted papers in TeX is using apa.cls (though now there's apa6 as well). The only issue from my perspective is that apa.cls typesets figures correctly according to APA style: at the end of the manuscript. But APA style is completely silly (as some journals have noted) : putting figures at the end makes reading and reviewing papers with many figures into a nightmare of flipping or scrolling. So apa2.cls is my minor modification that leaves figures in situ.
For bibtex, apacite is great, but it includes issue numbers for journals; apacite2.cls doesn't do this.
Very minor modifications, but hopefully useful for someone else besides me.