Here is a short and trenchant little essay by philosopher Dan Dennett, addressed to graduate students, on how to choose a research topic of lasting value. His essay is aimed at philosophy students, so a few of his points don’t really apply to ecology (well, actually all of his points apply to a certain style of theoretical ecology…) But most of the essay is very relevant to every ecologist. Because ecological systems are so complicated, there’s an infinity of questions one could ask. And while I don’t think choice of question is the only thing that determines whether one’s research is of lasting value, I definitely believe that some questions are better than others. If you want to avoid jumping on the latest bandwagon, or choosing a topic that’s only of interest to you and your closest friends, it’s definitely worth a read.
Dennett suggests that one test of whether your topic is of broad interest and lasting value is whether you can get non-specialists, or bright undergraduates, to care about it. I like this suggestion, but I don’t think it’s anywhere close to a foolproof test (in fairness, Dennett doesn’t think it’s foolproof either). In my experience, it’s often difficult for even very smart non-specialists to distinguish the narrow, the esoteric, or the faddish from the truly important. That’s because even narrow, esoteric and faddish questions typically have some connection to big, important ones. A narrow, esoteric, or faddish question about, say, the consequences of species loss is still a question about the consequences of species loss, a very big issue about which one could ask important questions. When they hear that you are working on something that is somehow related (even tangentially) to a broad, important topic like species loss, non-specialists tend to conclude that the specific question you’re asking must therefore be of broad importance.
I think the best people for helping you judge whether your ecological question is of broad importance are ecologists who themselves have done important work on a broad range of problems, and who have been around long enough to see some fads come and go. Basically, if Bob Holt, or someone like him, thinks you’ve hit on a good question, chances are that you have.
Dynamic Ecology 
I think there are more considerations here than just what is important, however that’s defined. One of the main ones is: what skills and tools are you going to develop and come away with. That determines what problems you can and cannot tackle in the future. This is where you need help from your adviser or whoever you can get it from. Or the flailing and excessive pain method, that’s another way.
As for what exactly constitutes “important” that needs discussing. I don’t define it wrt theoretical advancement of the science. Doing so can easily lead to the sorts of troubles Dennett describes. Much healthier I think, to define it wrt what is important to larger societal needs and functions. Academia could do a lot better in that regard than it does.
Pingback: Advice: a compilation of all our advice posts, and a call for new advice topics | Dynamic Ecology