The fact that the number of trade routes is limited by city, to me, this implies that the capital's routes should be used to boost outposts, and early cities, until those cities get their own Trade Depot. Then have the cities send convoys/vessels to the capital, and when you have quite a few routes available, send them between each other (works best if you have differences in the cities base yields).
My theoretical ideas, that I have NOT tested yet:
If you try to work your cities in a circle (capital in center, and six suburbs around it), then try to focus three of the suburbs on food, and three on production. Then, you can create a trade web, where each food center trades with each production center. This ensures large delta in each direction on every route, and maximizes the amount of 'free' (meaning not generated by tiles or people) yield you are getting.