By Ana Camila Gonzalez
“You can do math on excel?” I ask. I immediately imagine a face-palm response, but Dario, one of my advisors, is nice enough to hide it. I’ve collected tree core samples, I’ve prepared them and cross-dated them. Now what?
Oh, right. The Science.
I guess I never really understood there could be so much involved in answering a question. When I imagine the scientific method I’ve learned since the sixth grade, I somehow imagine a question that can be answered with a yes or no. If I let go of this apple, will it fall to the ground? Hypothesis: yes, it will. Experiment: yes, it does. Conclusion: yes, it will. To the credit of my high school science teachers, it’s not that they didn’t make it perfectly clear that the why and the how are just as important as the yes or the no. I just couldn’t imagine that you’d have to explain why the apple falls with four different figures: haven’t you seen an apple fall too?
Dario is helping me understand how to analyze the data from the black oak samples I have already been working with for some time now. I know these samples. Or at least I think I know these samples. I’m learning there’s more to know about them than I initially thought.
We’re analyzing the climate response, which proves to be exactly what it sounds like. We have recorded measurements of climate (precipitation records, temperature records) and a proxy for tree growth (our ring width measurements!) and by comparing those we can see how a tree population responds to a range of climactic conditions. Alright. I can do this. I’ve made graphs before.
“So we’re going to find correlations,” says Dario.
“Click on an empty cell.” I start to make a scatter plot; I think what we’re going to do is look at the slope of a line of best fit.
“So we’re going to see if the correlation is positive or negative?” I ask.
“Yes, but we also have to see if the correlations are significant.” Isn’t any correlation higher than a zero significant? They’re showing a relationship.
Dario continues, “Any correlation above a 0.2 or so is significant for the hundred years of ring width and climate that you have for this analysis.” I learn how to use the =correl function to compare the populations to temperature and I have to say I’m disappointed. I thought 0.2 sounded so low, but some of my data is showing a much lower correlation, and the data that is significant only ranges from about really close to 0.2 to 0.38 or so. I wanted to see a 0.5 correlation like I did between tree samples within a species as I was cross-dating. Comparing precipitation to ring width gives me slightly higher correlations, a few in the 0.3 range, but I’m still feeling underwhelmed.
“No, but it’s still significant! It matters!” Dario tells me to make a scatter plot comparing precipitation to ring-width measurements over time at both sites. At first it looks like a ball of yarn, but as I mask the plot out I can see why those 0.3 correlations are significant. I follow each curve, visually skateboarding up and down the peaks and valleys and noticing that I’m going up and down a lot of very similar hills as I do so. What’s most rewarding is looking for years I know are drought years (1966 and 1954 were big droughts) and seeing relatively low measures of precipitation and ring width during those years. I knew while I was cross-dating that those years were important when I saw how small the rings were, but now I can prove it. Like the apple falling, I can’t just say that because I see the rings are small those were dry years. I have to compare it to precipitation records, temperature records, and, dare I say it, the Palmer Drought Severity Index (I have to admit I don’t entirely understand the mechanics behind the index, but I understand that dryness is a composite of precipitation and temperature forcings).
Dario, over multiple days, teaches me a few more nuances of Excel and helps me understand the ARSTAN program and how we use it to make our ring-width measurements more effective as proxies for tree growth. He mentions this would all be easier if I knew how to use R. I make a mental note: learning R is the next step. If I thought that was scary, now I have to put this information on a poster. That real people will see. At a real conference.
Neil shows me a few poster examples, and the message is clear. Show your data instead of describing it in words. That also means I’ll have to explain my data by actually… talking… about it. Gulp. The North East Natural History Conference is next weekend, but I feel like I’m ready. I understand the why and how after analyzing my data. At least I understand it enough to give an answer better than yes or no.
Ana Camila Gonzalez is a first-year environmental science and creative writing student at Columbia University at the Tree Ring Laboratory of Lamont-Doherty Earth Observatory. She will be blogging on the process of tree-ring analysis, from field work to scientific presentations.