Siliceous and Delicious

Wednesday

The reality of everyday “working” life hit like fist to the nose this afternoon. The ease and relaxation won over ten days of gradual forgetting evaporated within an hour as the suspended stresses came rushing back. First, the FIB-SEM training session fell through because it turned out that, as I should probably have known (but didn’t), analyzing samples in this machine requires them to be sputter-coated with metal (to prevent them from both charging and flying off the stub under vacuum). This requires both time and formal training on the machine, neither of which I had in the few minutes before the training began.

Then came a <sarcasm> lovely </sarcasm> email from Christian, the head of the PlanktonTech group (who, as you might recall, are presently paying my salary and tuition). He has hired a group of engineers who are going to make CAD models of plankton skeletons and wanted to know if I could send him high-quality 3D images of radiolarians of different degrees of silicification. You can imagine the shame and embarrassment I felt when I had to reply that I had no such 3D images of radiolarians, and that in fact I had not obtained any sort of 3D images (or indeed any images at all) of any kind. It was not a pleasant email to write. I have been wasting their money completely—in the year and a half since I’ve been on their payroll, I have provided them with exactly nothing. It’s a rotten feeling.

Other emails threw more yule logs of anxiety onto the rapidly catching fire of despair. I got a reminder of the lecture I agreed to give at Mass Bay Community College. Back in August. It’s now December. I haven’t so much as thought about what I’ll talk about, let alone put together a presentation. Good-bye, weekend…

Worked for a good long while on a detailed reply to Zoe’s email. Then had to dash off to Charles’ good-bye bash.

Thursday

Spent about an hour faffing around with OmniFocus and OmniPlan, trying to sort out projects and tasks into somewhat manageable chunks and get up do date with the tasks floating around in my head. Then took a glance at the Mesozoic rad catalog and was not overwhelmed with joy. The SEMs don’t generally show broken specimens or specimens porous enough to determine wall thickness. The resolution is not particularly high (probably 150 dpi or less). The best I could hope to do is a very rough estimate of surface “porosity”, which is hardly an all-encompassing measure of silicification. Meh. Is it worth doing that? I don’t know.

The morphologies of the things are also really quite variable. One major issue is that it seems lots of taxa have heavily silicified spines coming off them, in many cases it looks like most of the silica is in those spines. So trying to get at silica use from porosity would be pointless in those cases. Then there’s lots of three- or four-armed club morphologies that just don’t lend themselves to the sort of simple geometrical modeling I used in the MSc project.

Blergh. Once again, my response is mostly overwhelming disgust at how pathetic and daft this whole project is. I know I need to get over that and get it done. But really I don’t see any point or value in what I’ve set out to do here. Maybe I need to just axe the literature-based preliminary study and press ahead with getting the samples ordered for the full-fledged study. Not that that’s any more pointful. Yech! Decided to give up at 11:30 and start thinking about the lecture at Mass Bay instead—first step is to read through the textbook chapter the students will have read in preparation for the lecture. Maybe that’ll give me some idea for what on Earth to talk to them about.

Briefly delved back into the world of R in the afternoon. Spent most of the allotted time getting back into it, having not thought about it at all for two weeks. Did manage to convince myself that I’ve succeeded in the basic data culling to an acceptable degree. His database has 66,941 occurrences, mine has 68,195. At least I have more and not fewer now (phew!), that’s better than where we were before. What’s even better is that his database has 1,024 unique species, and mine now has 1,055. Which is close enough.

Tried to learn a little more about how to plot data in R, with some moderate success. Here’s the distribution of the data through time, as it stands:

It looks like the pattern is the same whether you look at individual occurrences or the number of lists, suggesting that the lists are (on average) the same length over time. Here’s another way of looking at that:

This linear relationship means the average list length is more or less the same through time. That’s interesting. If diversity is increasing in a particular site, I would expect a more recent list to be longer than an older one, and so to have a greater than proportional increase in occurrences for a given increase in number of lists. Hmm! Let’s just look at list length through time, then.

Yup. Should have been obvious from the other plot, but indeed it looks like the average list length isn’t changing appreciably over time. This is intriguing, because it suggests that the gain in diversity over time (if it is real) is due entirely to a change in beta diversity, not alpha diversity. That is, the diversity at any one place was always about the same (between eight and sixteen species), but while those species were the same in place A and place B early on, those species are completely different in A and B in the more recent time bins.

The alternative explanation, of course, is that nothing changed, and it’s merely the number of different places that have been sampled that has changed through time. This seems like a good place to stop for the day. The department “holiday party” (why not just call it a Christmas party? Who’s really going to be offended by that?) is on, and I should probably show my face. I hope I can grit my teeth enough to mask my growing fed-uppedness with being here.

Friday

Huzzah, the Day of Fry is here! Had to make a trip to the International Office this morning to get my visa documents renewed, then traipsed along to get a coffee with Andy, who told tales of the illustrious science folk he dined with at the conference in Zurich from which he just returned.

Spent the remainder of the morning trying to find papers that might give me a clue on what fluorescent compound to order in order to start preparing samples for the confocal scope. No luck. Just the usual frustration and overwhelming tidal wave of irrelevant information. Was happy that I had booked a squash court before lunch and took off to take out the anger on a squishy ball.

Lunchtime was taken up (and exceeded) by an excruciatingly dull seminar on reactive oxygen species produced in freshwater environments, a quasi-mandatory talk taking the place of lab meeting this week. Deeply yawn-inducing. I am struck by how extremely dorky scientists are, and how deeply awful at presenting their work. In their defense, it might just be that it’s not at all interesting, and that nothing can be done to polish up turd-like content into an attractive and captivating presentation, but using hierarchical bulleted lists and slides filled with (improperly formatted) chemical equations just can’t be the way to go. The saddest part is that I seem to be the only one in the room to whom this seems sad, pitiful, and frankly pathetic. Everyone else appears more or less in their element.

Spent the few moments of motivation I had left in the afternoon trying to put together a few ideas for the Mass Bay talk. Didn’t really get anywhere with it. My throat’s starting to feel sore so I’m going to run to CVS to see if I can somehow halt the inevitable…

Monday

Reached the definitively lowest point of the semester thus far. Spent Friday evening and Saturday feeling sick in bed. Felt better on Sunday but found it hard to motivate to work on the Mass Bay talk. When I finally did sit down, and sat for hours, I drew nothing but blanks. I couldn’t figure out what to say, or how to say it. My mental capacities have been numbed by an overwhelming sense of dread, of disgust at my own lack of progress, and bitter distaste of the whole endeavor of science. This morning I had to go to the dentist, and by the time I was at work I realized there was just no way I could put together an hour-long lecture by tomorrow morning, regardless of my mental state, but certainly not in the pits of despair and frustration I find myself today. I finally decided to send an email and back out of the commitment, a really lame thing to do less than twenty hours ahead of time, but I felt it was my only choice. Today I feel that I have really failed myself on all fronts. In all honesty I don’t know how I can go on doing this for another two years or more. I am bashing my head against the wall and am accomplishing nothing. I have had many times in my life where I have felt depressed, frustrated, and afraid. But never have I been this deeply miserable about what I spend my days doing.

A couple of hours later, I started toying gently with R. Thought I’d see what ever came of the job I ran on the server—stripping out the extraneous text from the species name column in the database—which I never followed up on since it had eventually worked locally on my laptop. Turns out the process never completed on the server: the output file was not written. I specified an error stream, but that file was empty; perhaps the cluster terminated the process before it was finished. Weird. Anyway, it’s all moot now because I’ve decided that the file I have is good enough to proceed…

Tuesday

Have been continuing with the step-through of Rabosky’s R code, with reasonable success. I am beginning to understand better what he’s done (somewhat), although there’s a lot of code remaining to be commented. If nothing else, I’m continuing to learn how things work in R. And I’m having a reasonable amount of fun with it, which is about all I can hope for after the last couple of days.

After deciding to leave the input dataset as it was, I thought I’d look at what the actual raw diversity plot looked like for that dataset (this is something I believe Rabosky never did—in his paper, he only showed what I think must be the previously published diversity curve, which applies the range-through correction), and I got surprisingly high diversity values:

As you can see the diversity is close to 200 at 30 Ma, and tops out at about 400 in the recent. In comparison, here’s the figure from Rabosky’s paper:

Ignore the shaded grey area and look at the black line (scale is to the left)—diversity is around 100 at 30 Ma, and tops out at just over 160. Whoa, Nelly! How am I getting about twice the diversity in my data set relative to the published values? Perhaps this is a function of bin size, as I am looking at 2-myr time bins. Let’s see what it looks like with 1-myr bins:

OK, so it’s a little lower now. This suggests there’s quite a bit of turnover in those recent time bins—i.e. there are similar numbers of species in the 1-2 Ma bin and the 1-0 Ma bin, but what those species are is quite a bit different. Thus if you lump those two bins together you get a 2-0 Ma bin that has a higher diversity than either of the bins by themselves. Still, the shocking fact remains: the overall diversity I get is about twice what it is in the figure from Rabosky’s paper. The last possibility is that I’ve goofed something up in my dataset, so let’s try this with Rabosky’s dataset and his routines. Here’s what that looks like:

Well, that’s pretty much indistinguishable from mine, and it certainly doesn’t match the plot in his paper. The range-through method can only increase some of the y-values on this plot, but it will never decrease them. This leaves only one possibility in my mind (barring major cock-ups along the way): Rabosky copy-pasted (or otherwise took) his graph from the previously published literature. When that graph was made, the Neptune database must have had significantly fewer entries in it (or, perhaps more likely, had been heavily culled by Cinzia Spencer-Cervato, who published the paper the graph is most likely derived from). This culling was subjective (as far as I can tell from reading that paper), and so who knows what sort of “bad” data has been removed from her analysis that was used in Rabosky’s work without further concern.