From what I understand TRFLP is a easier method to conduct than DGGE, but both give similar results which are essentially simple profiles of the community based on which primer and digestion enzymes you choose. Think of it as a way to chose which samples to look into further with sequence methods.
DGGE can be very annoying; in theory it sounds good, but it can easily happen that you get 5 DNA bands, you isolate one specific band, and when you run it again you get another 2-3 extra bands, in different positions. It may be because of real different fragments, but also because the original band has lost 1-2 nucleotides. However, it can be a really pain in the ass.
Use a clone library approach by using both 16/18S and functional gene primers. This is a low cost but tedious method. Or else if you have funds then u can go for deep sequencing approach.
You can take the next generation sequencing approach [454, Illumina] for in depth coverage of phytoplankton assemblages from environmental samples. The dataset will be strong and subsequently you can undertake deep phylogeny. Resolution using DGGE may not be the best approach.