Design & Analysis of Ecological Experiments

Hello Everyone,

Even though it is currently Spring Break, I am subjecting myself to 8hrs a day of one of the most intense classes I have ever taken. However, I can say without exaggeration (or sarcasm) that this is a contender for best Spring Break yet. The class is “Design & Analysis of Ecological Experiments” and it is taught by Tony Underwood and Gee Chapman. If I can only figure out how to keep all of my brain synapses firing at this rate when class is finished, they might as well just hand me a PhD now.

If you know anything about me, you should know that statistics are one of my irrational fears, right up there with ladybugs and alligators/crocodiles. There is a book on my bookshelf called “Statistics for Terrified Biologists” (available in .pdf form at that link), I bluffed my way through my first undergraduate stats class, then failed and had to retake the second one (maybe I shouldn’t admit to that publicly), then barely passed the graduate stats class I took last semester from MY OWN ADVISER. Needless to say, my understanding of p-values are pretty solid, but I just assume you all know what you are doing in your results sections. Until now…

Something about this class is different. I didn’t know I could do anything for 8hrs a day! (Sorry Ken Sebens, UW, good thing I was solidly in the middle of that pay range huh? Kidding, I worked really hard as his research tech.) Anyway, I love theory, so the way Underwood explains every detail and why it is important, or not, is a really great way for me to learn. Today we went in depth into how to use fixed and random factors in multi-factor analyses. And while someone asked what one might call a “stupid question” I realized the proper design for my thesis, which I have been pretending I knew all along.

My experimental design has changed a little from what I explained in previous posts. Following my thesis proposal, my committee said “it’s fine that you chose your juvenile kelp settlement densities based on field observations, etc…but wouldn’t it be more interesting if you looked at survival of these juveniles at different densities as a way to tease out the actual self-thinning effects?” A mind-blowingly awesome suggestion. So now I am going to settle my microscopic sporophytes at 3 soon-to-be-known densities (small, medium, and high), these are my XS size classes, then braid the holdfasts of my S, M, L individuals into rope mats in small, medium, and high densities, and then outplant everything to my 3 sites. I couldn’t find any good density measurements for juvenile kelps in the field, in the literature. The closest thing I could find was Paul Dayton et al’s 1984 opus, but his methods section is infuriatingly vague and I cannot tease out anything remotely like individuals/standard area measurement. My adviser has several years of observations and thinks he has the data to help me figure some ballpark densities out, but I will believe it when I see it. Once I get a good estimate for low, medium, high densities for each size class, I will need to hit the ground running with my outplants as the naturally occurring individuals are beginning to recruit.

Anyway, here is what I figured out today which I am totally stoked about:

Size Class (fixed) = 4 factors

Density (fixed) = 3 factors

Location (random) = 3 factors

n = 5

Here is my model:

Xijk=M+Si+Dj+Lk+SDij+SLik+DLjk+SDLijk+e(ijk)

Ho: No Interactions, and no difference in mean survival between size classes, or densities

I would write out all of my mean squared terms, and I probably will on paper, but that might put us both to sleep now. Anyway, I am totally stoked on experimental design and my fabulous Australian teachers. If you or any of your students ever get the chance, please encourage them strongly to take this course. It is particularly useful for someone in my position who is just getting started with research and designing their own project, even someone with an intense hatred for stats. I also just purchased Underwood’s book “Experiments in Ecology: Their Logical Design and Interpretation Using Analysis of Variance“. They might just make a believer out of me yet. Can’t wait for the weekend to be over so I can go back to school!