In this new post to his blog Chris Hartjes offers up a few of his own thoughts about testing APIs and some of the issues/concerns he's come across in his own testing (via Behat).

I have been asked a few times on Twitter for some advice on how to use testing tools in order to test APIs. It just so happens that I am writing a bunch of tests for an API at work so perhaps this is a very timely blog post. [...] When you are testing an API you main concern is your ability to reproduce the results of a specific API call. If your API is such that repeating a call several times will give you different results, you do have some challenges on your hands.

He talks about functional testing, the "moving parts" they involve, using only known data sources and the creation of integration tests. He also mentions the testing of your local application to ensure it's handling (or transforming) the results of the API calls into something useful.

In the end, it doesn't really matter what tools you are using to test either your own APIs or ones that your application depends on. Be clear in your mind about what it is you are actually trying to accomplish and don't get create tests that provide very little value to your application.