Down the rabbit hole

Well it has been a while hasn't it? I started writing up a chapter of the dissertation and each paragraph has required extensive citation mining and additional research, and the data analysis hit a few snags, including a 12 hour day troubleshooting a single line of code in R, etc. I can see from this concrete process how people end up taking 11 years to get a PhD or whatever. The potential to just fall endlessly through the abyss of prior research, methods, controversy, new methods, classic papers, and the vast amount of information available now, especially on the internet, is enormous. Terrible sentence, but you get the idea. It's also a funny kind of writing, because even statements of the obvious usually have to have at least one citation. "Cacti are slow growing long-lived perennial angiosperms known only from the so-called New World, except for one species of Rhipsalis that apparently is native to Africa (Obvious and Obvious 1986, Boring et al. 1990, Redundant and Soporific 2004). 

Anyway, I am in the final phase now of adding all of the in-line citations and tidying up the lit cited, adding the figures and indexing them all, etc. I think I'll finally be able to send it to my committee chair tomorrow. 

This is a cool figure that I have to tweak a little that is a representation of 500 stochastic projections over the next 100 years for the species I am studying, given a higher probability of extended drought versus bonanza years of precipitation. 

The "total" axis is 1/10th of the total population, and the "Year" axis is in increments of 20 years at a time. You can see the vast majority of the random projections are below 100 individual plants. The stochasticity is cool to see too though, with the exact same parameters resulting in less likely but much more favorable scenarios. 

The figure below represents 100 "random walks" calculating "quasi-extinction" probability over time. Quasi-extinction is a time at which the species goes below the minimum population size it would need in order to persist, which in this case I arbitrarily estimated to be 100 individuals. "G Weighted" is a reference to the most robust subpopulation. A lot of the projections I am running do not bode well for the species. As you can see, at best this set of parameters gives even the most beneficial habitat only a 20% chance a century from now. This is a model developed with climate change effects added, such as increased max and min temps annually and reduced winter precipitation, which is the strongest predictor for establishment of new plants. 

Most of the analysis that is cross validated and tested iteratively all points to Mammillaria halei being more vulnerable that I had thought it would be. I'll be interested to see what the next chapter, on species distribution modeling, reveals. 

I am also investigating why it is that a lot of these extinction risk modeling techniques seem biased toward dire predictions. I think it's something a well versed statistician could explain to me. I bet it has to do with the sensitivity of these models to very small changes in initial conditions, and the high variance of small data sets. I know in my bones that Mammillaria halei is more resilient than my models are showing, but I can't find a quantitative justification for changing the model parameters. I am going to try generalized linear models of the count data using a Poisson family regression, or negative binomial if it's a better fit, and then generating a pseudo-count data set of, maybe, 100 years of counts, and then try the modeling using that, just to see if that process flattens out some of the noise. I think a great deal of the apparent fate of the species is a result of error structure in the data, rather than real threats. 

In other news, I am chronically unhappy the majority of the time, without regard to most efforts to ameliorate the condition. The work is a balm by comparison. I'm glad to be going to a few meetings a week, meditating, plugging away at being on the planet, one day at a time. But I find myself thinking about an escape more often than I would prefer. I had dinner with an old recovery friend last night who is visiting Phoenix from Santa Fe, and of course the first thing out of his mouth when I asked how he was doing was "I have prostate cancer and it's spread to my hip bones so the doc gives me about 2 years. I'm okay with it, I just don't want to suffer." He then related a story where, a couple days after the diagnosis, he was walking late at night in Albuquerque and a car drove by and slowed and a kid pointed a gun out the window at him. He stopped and stood there and the car slowly drove on, the kids inside laughing. He said the first thought that crossed his mind was "go ahead, shoot. Shoot!" He found himself extremely angry when they drove away. It was a good conversation as we could compare notes on a lot of these impulses and perspectives. We then went to a speaker meeting that was okay, although in the light of day today I suddenly have deep, deep reservations about it. Maybe I'll write about that later. 

Right now, it's a lovely soft day out, cool and raining, a soaking winter type rain. I look forward to going back to sleep later. 

The cacti aren't too sure.