I have gotten REALLY into statistics lately, so much so that I am going to give a keynote speech in Taiwan in September as a sort of "call to arms" for marine biologists to use a really impressive set of analytical tools that most seem either not to know about or not to understand. We need to move beyond ANOVA (as an example) and EXPLAINING data to instead using what we already know to PREDICT what will happen to a particular coral or a particular reef.
I was recently inspired by a talk on partial least squares, which is (to my surprise) a GREAT tool for using transcriptome data to make predictions about animal behavior. Check out Peter Bartel's talk on this (Powerpoint slides here). Multiple regression is INappropriate for datasets in which tons of response variables are positively correlated......which is every coral dataset I have ever seen, especially those featuring gene expression!
I don't always read papers, but when I do, I make it a good one. Some of these have been sitting on my desktop for ages.....
Anderson et al. 2013 (ANOSIM vs. PERMANOVA)-if I read this correctly, why would anyone use ANOSIM when you can use PERMANOVA?
Cunning et al. 2018 (at long last, the Pocillopora damicornis genome)
d'Auriac et al. 2018 (cnidarian gene evolution)
Decelle et al. 2018 (oceanographic survey of Symbiodinium)
Johnston et al. 2017 (evolution of my favorite coral genus: Pocillopora)
Kornder et al. 2018 (climate change effects on coral calcification, and one of the authors, B. Riegl, is an all-around great guy.)