I advise you to visit the website dedicated to the GBS method in Cornell University (http://www.igd.cornell.edu/index.cfm/page/GBS.htm). You will find most of the informations you want here. If you have any specific question, don't hesitate to contact Pr. Sharon Mitchell which is in charge of GBS.
For the enzymes, they usually use Apek1, which worked very well both in plants and insects for plenty of studies. But the enzymes you will use will depend on the objective of your study. For example, if you are doing population genomics, you want a moderate coverage but a good depth to detect heterozygotes. If your want to detect a specific locus linked to a pattern of adaptation (plant to specific condition, insect resistance, etc.), you want a very good coverage (to detect it) but also a good depth to detect potential SNP.
Again, I advise you to contact them, they are very welcoming and guide you step by step in your entire project.
If input amounts are limited, you might be interested in a method my lab has developed called nextRAD. It can use as little as 1 ng total genomic DNA. Feel free to contact me, or look at snpsaurus.com which offers it as a service.