Some background: Amanda, Ingrid, and Erin performed a very novel experiment in Churchill last summer (see their blog for more fun stuff). Amanda processed all the samples and is in the process of analyzing the data. Because of the design, it gets complicated very quickly, and so Amanda is honing her R-skills to visualize the results before we go p-value hunting. In the figure above you see the different sampling dates (the 5 big boxes), and the boxplots show dissimilarities between experimental pools, freshwater pools where we added salt to, with control saline pools (uneven numbers) and control freshwater pools (even numbers).
Now the fun stuff:
- On the first sampling date which was before the salt manipulation, no difference between the even and uneven comparisons, so far so good
- On the second and third sampling date, some clear differences between the dissimilarities with the salt and freshwater control pools: even numbers less dissimilar than the uneven numbers, so far so good
- But it is in the completely different direction BAM! Adding salt makes them more dissimilar to the saline control pools.
- This is consistent for all experimental treatments, and for 2 full weeks, so very likely not a spurious result.
- And then it disappears in the last 2 sampling periods
So the questions we have to answer:
- why in all pools, even some that were not manipulated?
- why does the effect disappear later on?
- what is the mechanism?
- and why opposite to our predictions?
This is what makes science so cool, completely unexpected results. Hopefully we will be able to tell you what the answer is in the not so near future. More to come.