Beta Diversity
Ecologists like to use something called beta diversity to see how complex ecosystems compare to each other. This form of analysis is actually not something new as many of the mathematical models were developed in the 1950's and 60's. For better or worse, microbial ecologists have adapted these methods and put them on computational steroids. They are often the source of controversy and can easily be misused.
They tend to be so controversial because they are a form of data reduction and show an incomplete picture. However they offer a quick comparison between very complicated samples so we generally like to use them to quickly see if a question needs more interrogation. Just like any test you need to understand it's purpose and limitations.
Just remember, the closer the dots the more similar the communities are.
The figure below I showed previously and reduces all samples in each group into a single point.
From this we can tell that vermicompost is more similar to soil, precompost, and manure than to tea or compost. Tea and compost are also not similar to each other. I hope to get more into a tea vs vermicompost comparison but clearly we can here they are different. With such extreme difference in samples the bottom left cluster of samples is poorly differentiated.
Thawed vs Reshipped Samples
If you remember at the start of the project we had some problems with shipping. I decided to use cold packs instead of a DNA preservative because we didn't have time to test a preservative and there was some concern with sending people toxic substances.
Anyways, some samples thawed and were recollected to ensure the poor shipping conditions didn't change the community. Several of you said they didn't think it would make a difference because you normally didn't store it any special conditions. Still, scientists are controls freaks. Additionally, the samples that "thawed" had Mid West Lab data associated with them so we wanted to be sure the thawed samples did not change much.
Each dot represent an individual sample. In pink are the samples that thawed and turquoise are the samples that were successfully recollected. The numbers are the randomly assigned producer #'s from when we looked at the Mid West Lab data.
For the most part we see the pink and blue samples all clustering together. I think it's safe to say the thawing of samples had very little, if any effect. Clearly the samples by producer #19 and #2 are the most different. Interestingly producer #19 uses spent grain as bedding and #2 uses peat. These samples also had a lower amount of total sequences which could also be the cause of the difference.
If we take out producers 19 and 2 and "zoom" in even more we something pretty cool. Look at producer #7. First off it's great we have data for all triplicate samples, this really highlights the importance of replication. Secondly, look at that clustering; that is some really homogenous and consistent vermicompost. This also really highlights how little shipping affects the microbial communities.
We can also see producer 4 and 18 kind of creating their own clusters. Producer 18 actually submitted vermicompost from two different systems which is likely why we see a broader spread of their samples.
That's all I have for now. I really hope these plots are not too confusing... because I'm going to be throwing more at you soon. Next up we'll be looking at combining these plots with questionnaire data and physical and chemical data.
Any questions? Please ask!
Comments