Siliceous and Delicious

Went straight to MIT for my 10:30 meeting with Zoe and Andrew. They were good and active listeners, and had a lot of both nice and constructive things to say, but at the end of the day it still left me dispirited. Firstly because I realized, as I have done oh so many times, over and over again, that I am the only one who can figure this out, and (almost) every time I ask for help, I realize that I have to answer the most difficult questions myself. And secondly because we ended up talking, inevitably, about the future, what I was going to do next, whether I wanted to a postdoc or not, and so on, a conversation that I try to avoid because always leaves me completely unsettled and usually depressed.

One useful piece of the conversation, though also source of my disappointment, was the point Zoe made that I needed to know what I wanted to say before I wrote the papers. What did you find? What are you trying to say? What question is this paper answering? These are the matters about which I’ve been plugging my ears yet Zoe is absolutely correct in asking these as prerequisites—and, sadly, there is probably justification in her incredulous eyebrow-raise when I mentioned that I wanted to get the first paper finished by the time I head to Denmark at the beginning of April. This is so much harder than I thought it would be.

Anyway, she was no help whatsoever on actually answering these questions for me, but she and Andrew did make a lot of other useful and helpful remarks, including these I’ve chosen to highlight as relevant and actionable:

1. If you divide morphospace occupancy (alpha volume) by the number of species for each time bin, is that a way of standardizing for sampling? [Note: maybe not, but it would give an interesting other measure of per-species morphospace occupancy, a sort of analogue of mean pairwise distance.]
2. Can you circumvent the sampling problem by plotting the morphospace occupancy of individual lists/sites for each time bin (i.e. alpha morphological diversity through time)? I think this is both interesting and helpful, and allows a discussion of the analogy of alpha and beta and gamma diversity with disparity through time, which I’m not sure has been done before (well, it probably has, but I don’t know about it).
3. What about plotting the number of realized character states through time? Wouldn’t that be a good measure? [Note: I think this is similar to the metric of ‘realized pairwise character combinations’ or whatever it’s called from the disparity literature.]
4. Zoe and Andrew both made a bit of a fuss about how my dataset isn’t independent of phylogeny because I used genera, which are defined morphologically, as the basis for my study. I’m not sure I understand why this is a problem, other than that there is sure a lot of intrageneric morphological variation that is not captured by my morphospace. The way around it of course would be to score individuals in samples (without necessarily even naming them or assigning them to a genus), and do this for a large number of samples through time, but of course that’s an utterly unrealistic effort and the mere thought of it makes me want to throw up.
5. Can you plot mean pairwise molecular distance through time? Using Neptune to time-resolve the pairwise distances from the Sorhannus tree. Would it look different from pairwise morphological distance?
6. Zoe also suggested what I’ve been wanting to do for a while now, but haven’t yet—i.e. looking directly at particular characters in the data set. When I pressed her to suggest ones that would be interesting to look at, she came up with a list that was (actually  reassuringly) similar to what I’ve had in mind: characters related to chain formation, chain formation characters resolved by silicification (i.e. are they using less silicified structures through time to achieve chain formation?), pore size through time, velum presence/absence through time (this relating to the viral/pathogen defense hypothesis), predation characters, sphericity (SA:V ratio, for both strength, nutrient uptake) and possibly labiate processes/raphes if anything meaningful could be gleaned about their homology from patterns seen in the morphospace through time.
7. With regard to my difficulty in thinking about the chain formation vs. predation characters, Zoe was encouraging—her take was that this is a real problem and rather than ignoring it I should simply write about how it is difficult to disentangle whether spines are for one or for the other, and just describe what my observations would mean for either hypothesis.
8. She also suggested going through my list of all characters and brainstorming creatively about what each one could mean functionally, keeping in mind what the main important factors are (which is another difficulty, but never mind—I wrote down “predators, sinking…” in my notes).
9. In the morphospace occupation density plot, she wanted to know what the morphotypes are that are disproportionately represented—a good question I should answer if I am going to put that plot in the paper.
10. Finally—what two things can you say about your data? This is the crux, really. I need to figure out what I can say about the data, and then shape my paper based on that.

Altogether a challenging, exhausting day. Some things gained, but much confidence lost. It just feels like there is an eternally long way left to go, and the path ahead just doesn’t seem to get shorter.