There is no fixed sampling scheme, as it would depend on the question(s) you are studying. There are various systems by which you can stratify sampling into habitat categories. When I get to computer, I will attach classification of wetland and deep water habitats as one system of classification by Cowardin et al. Montgomery and Buffington have another classification of lotic habitat types for streams such as riffles, pools, glides or runs, step-pools, cascades. There are others that might be applied to help stratify stream categories, such as the Rosgen stream classification. Microbial and chemistry analysis may include a stratification of land use, along with geology.
Cowardin et al classification attached as one method to stratify units for sampling. I do not expect you would find a cook book approach, but reading about studies that have answered some of the questions about water quality and aquatic habitat, and consultation with the various types of aquatic specialists may be helpful to you as your proceed. Sometimes one might do a preliminary study before preceding. Often times, in lentic systems, measuring streamflow is very helpful to separate processes that occur during storms, baseflow, first storms after autumn leaf fall, variability of upstream land use activity, etc.