I am still 'digesting' the methods and conclusions in this article.
With that said, I don't think that molecular weight (molecule 'size') is as important as whether the chemicals being skimmed have both oil loving (lipophilic) and water loving (hydrophilic) groups in their structure.
Compounds that have this characteristic are known as 'amphophyllic' compounds. Some practical examples are the active ingredients in soaps and detergents, also referred to as surfactants (although that is a broader term). Amphophyllic chemicals form bubbles easier than other type compounds. This makes for more foam, and easier capture in the collection cup.
http://en.wikipedia.org/wiki/Amphiphile">http://en.wikipedia.org/wiki/Amphiphile</a>
There are lots of bacteria and fungi in aquaria, which all have a role in molecular metabolic process. This should not be understated. Short chain fatty acids for example, are known to have crucial roles in higher organisms, through a symbiotic relationship with the microflora and microfauna living in association with them. That includes corals and fish.
The amount of organic carbon processed via these organisms was not mentioned. Organic carbon in the ocean, and certainly in our reefs, is a 'limiting factor'. One reason carbon dosing is successful, is that it removes the rate limit that exists. Carbon dosing adds TOC to the system.
Aside from removing 'organic carbon' (TOC) compounds, skimmers provide a crucial function in our reefs. They help substitute for the lack of waves. The gas exchange that occurs is not simply oxygenation, although that is certainly an imporatant function. The balance of CO2 is also very important, as is electron transport.
ORP, which is basically a measure of the potential for chemical reactions to occur, is greatly affected by skimming, and is based on 'electron availability'. The skimmer can greatly increase ORP, which is a direct indication of the overall health of the chemical system in our reefs.
I have been following the work these folks have done, and look forward to the next study. I would agree that more data points (replications) for each experiment are needed. Anything below 20 replications, starts to draw statistical variance into the equation (higher standard deviation).
Good read, thx.