I suggest the NGS to determinate microbial diversity. Thought it is a little expensive, it is efficient and more accurate than DGGE or Clone Library Construction
DGGE is difficult, CLC is laborious and actually expensive. NGS by pyrosequencing method is most popular but include bias. Read whole metagenome and sorting rRNA gene sequence is bias free but require extensive informatic works. I thingk, NGS by pyrosequencing is still best in accuracy, price, time, etc. * chunlab.com
Above the experts advice is right. NGS is better than CLC and DGGE.
Based on you have to choose from these two methods, so:
CLC is more accurate than DGGE, if microbiology diversity is high. But CLC is laborious.
You could apply them together.
First, you can see the whole change by DGGE, if you'd like to see more accurate differences, then, you could apply CLC after filtering DGGE results. This could reduce clone libraries you need to construct.