In the early 1800s, David Ricardo developed a theory of comparative advantage as an explanation for the origins of trade. And this explanation has substantial power, particularly in a pre-industrial world. Assume, for example, that England is suited to produce wool, while Portugal is suited to produce wine. If each nation specializes, then total consumption in the world, and in each nation, is expanded. Interestingly, this is still true if one nation is better at producing both commodities: even the less productive nation benefits from specialization and trade.
In a world with industrial production based on division of labor, however, comparative advantage based on weather and soil conditions becomes secondary. Ricardo himself recognized this in his broader discussion of trade, as Meoqui points out. The reason is that division of labor produces a cost advantage where none existed before--an advantage based simply on specialization. Consequently, even in a world without comparative advantage, division of labor would create incentives for specialization and exchange.
This is from Michael Munger, "Division of Labor," in David R. Henderson, ed., The Concise Encyclopedia of Economics.
Mike persuaded me that the Encyclopedia was missing a lot by focusing on comparative advantage as if it was the main reason for the productivity of division of labor. That's why I commissioned this entry. Read his whole piece and I expect that many of you will be as persuaded by his argument as I am.