We were measuring it wrong

Two reports this week letting us know that we should be measuring something better when it comes to earth system science in the tropics:

– We often aren’t surveying large enough areas to figure out how trees in the Amazon respond to and recover from big disturbances like storms and droughts.  If you do a landscape-level survey using a combo of field work and remote sensing, like these guys did (Chambers et al., 2013), you might find 9-17% more tree mortality than we would otherwise expect.

– We should have also been looking at the Indonesian peatland carbon getting exported in streams (Moore et al., 2013). This study shows that accounting for carbon that’s no longer locked up in peatland soils and is now being lost to the stream system could increase the total GHG emissions post-land use/land cover change from peatlands by more than 20%. Plus, that carbon being lost is crazy “ancient” and implies an inherent instability in the peatland carbon dynamics post-deforestation.  Bad news bears.

And the slow steady march of getting progressively better at answering questions (and at deciding which questions to ask) continues…

2013-02-03 10.31.29

Anta (tapir) sez: science!!!

Make graphs to answer questions: a great example of what ecologists do

So, reeeeaaaalllly big picture, what researchers do: ask questions, and then try to answer them. This week, one puzzle I was trying to solve was a great example of one tiny part of that process.

I’m here in Brazil in large part to measure nitrous oxide emissions, hoping to answer the question: how does land use affect the emissions regime and nitrogen cycle disruption of tropical soil? (Some more info here.)

To answer that question, I’m in charge of figuring out a protocol for how I’m going to measure nitrous oxide (N2O), when I’m going to measure it, how I’m going to process the data, eventually interpreting what the data tells me, and writing up the results. That, and several other projects, and I am hopefully deserving of a doctorate of philosophy (hey-o!).

N2O emissions are influenced by temperature, soil moisture, and the amount of available soil nitrogen, among other factors. When I’m designing an experimental protocol (i.e. when I’m going to sample and how), it’s bad news bears to vary many of those factors at once.

Last week, I turned to Paul and asked him the question, “so, do you think sampling in the morning is legit?” (Researchers: we’re less intense than you would think.) This post is about the quest to answer that question.

What might vary between the morning and afternoon that doesn’t vary day to day? Temperature. OK, how much does temperature vary throughout the day? And does “morning” (ehhhh, so imprecise!) do an ok job of estimating the average daily temperature?

We can answer that!

Check it out! Some data!


I mentioned last week that Paul and I put a thermocouple out into the forest and soy fields with the goal of taking some daily temperature data.

The graph above is the variation in temp in a soy field and in a shady forest over the course of several days, one of which was wicked rainy and one of which was wicked sunny. Let’s make the assumption that a rainy day and a sunny day can pretty much represents the rainy season here – not totally unreasonable.

I graphed the data (woot!) and calculated the daily mean soil temp in the two land uses (see the text just below the legend on the graph). You can see that the soy soil is, on average, warmer than the forest soil. No surprise there; it’s shady in the forest, and there’s leaf litter on the ground to help moderate temperature. (Side point that’s really cool – the soy soil also has much wider swings than the forest soil.)

Then I calculated the standard deviation of the daily temps in the two land uses, and the grey lines represent all the temps that fall BOTH within one standard deviation around the forest mean soil temp AND one standard deviation around the soybean mean soil temp. Where the graph falls between those lines, we’re within one standard deviation of the average daily soil temp of either soil – meaning we’re in the magical land of “representative daily temperature” for either land use. Meaning we’re good to go if I sample then. Huzzah!

What times of day are between the grey lines? Roughly, the morning hours (more specifically, between around 630 and 1130 am).

So, Paul, “do you think sampling in the morning is legit?” Answer: yes! Proceed, young grasshopper!

Answering that simple question took me a week: learning to use the thermocouple, setting the thermocouple up in the field, waiting a few days, removing the thermocouple from the field, downloading the data, putting the data in a graph and figuring out how to determine if the morning was ok, and, you know, deciding what colors to make the graph (fun!).

But…. problem solved! I won’t inadvertently mess up this project by sampling across a wide temperature gradient!


In conclusion: that’s what I do for a living, and get to put “tropical ecosystem ecologist” on my business card. Sweet deal! :)

Cool new Amazon studies: thinkin’ bout resilience

Three very cool new Amazon studies came out in quick succession this December and when viewed in combo they mostly remind me how all the raddest, baddest questions in Amazonia right now ask about how resilient the Amazon might be to environmental change.  The answer, based on these three articles: ummm, maybe not very resilient?

Exhibit A:  In 2005 there was a pretty epic drought during the Amazon’s dry season, especially in the southwest of the Amazon.  A group of remote sensing scientists used the backscatter from microwave data (gathered by satellites – coooollll!!!!) to look at how the drought affected the forest canopy (Saatchi et al. 2012).  They found that the forest hadn’t recovered from the ’05 drought by the time there was another major drought, in 2010.  Wah wah wah wahhhhhhh.

What’s super awesome about this approach is the microwave backscatter data they looked at only penetrates a few meters into the canopy, i.e. the backscatter coming back from bouncing off the Amazon that the satellites record only measures water information (leaf water content, etc.) about how the tops of trees are reacting to a big drought.  Hence, reduced uncertainty from things like how wet or dry the soil was.  It’s a really clever way of using remote sensing to look at how resilient these trees are to droughts (and particularly their productive canopy structure).  The future is now, folks.

Cool-ness of approach aside, the Saatchi et al. paper is a big reminder that drought regimes in the Amazon really, really matter for the state of the ecosystem.  As droughts get more common, if Amazonian trees in very drought-affected areas haven’t managed to recover by the time the next drought hits, that’s bad news bears for long-term ecosystem resilience.  Some trees become persistently drought-stressed, maybe what kinds of trees do well in the forest changes, the biodiversity in these forests might change over time, and we end up with an Amazonian plant community that’s different than the community we started with.  Good thing some of my friends are working on these exact relationships (Paulo Brando, whaaaaat!).

OK, Exhibit B: Another group of scientists (Rodrigues et al. 2012) checked out what happens to bacteria biodiversity in the soil of land that used to be forest but is now cattle pasture.  They found that while in a single soil sample soil microbe biodiversity went up, over a larger spatial scale (e.g. a big pasture field) the communities were much more related than they were over a similar landscape in Amazon rainforest – meaning that the system’s soil microbe biodiversity went down.  The authors call this “homogenization of microbial communities” and we have no idea if (a) the soil microbe community will bounce back if forest regrows on the deforested landscape or (b) how having homogenized soil microbe communities might matter for things like stabilizing soil carbon, promoting fertility in the soil, or fostering a biodiverse set of plants when/if regrowth does occur.  More evidence that the Amazon isn’t as resilient as we might hope.

Exhibit C’s counterpoint?  A third group of scientists (so many scientists!  we’re taking over!!!!) wanted to see whether a significant number of Amazonian tree species had been around back when the Earth was much warmer, e.g. during the Pliocene (2.6–5 Ma) and late-Miocene (8–10 Ma) geological epochs when it was hella warm (like, as warm as we’re projecting it might be in 2100) (Dick et al. 2012).  They found that 9 or 12 of the species they looked at were more than 8 million years old and so are likely good to go when it comes to warm temperatures.  It’s a controversial idea, though, since a lot of other studies have provided evidence that higher temperatures is going to be a problem for Amazon plant species.  The authors themselves note that just because these species could survive higher temps doesn’t mean that they’ll be able to survive higher temps in combination with forest fragmentation and changing rainfall patterns (like, for instance, the ’05 and ’10 droughts).  [Quick, un-data-substantiated plug: my hunch is that rainfall matters a lot more than mean annual temperature for a lot of these plant communities.  Questions to consider for a post doc.]

So what’s my take away?  (Other than the fact that tropical ecologists are a bunch of exciting and smart scientists?)  I suppose that the jury’s still out, but that we’re seeing more and more evidence that the very large-scale global changes being inflicted on the Amazon have a high likelihood of causing cascading, interrelated effects (lower canopy functioning, changes in soil habitats) that could lower plant community resilience to the point where ecosystems start to change in ways that we won’t be able to predict or manage.  What the magnitude of those changes might be is a totally open question.

PS.  Sorry no pictures in this one, guys!  I’m just super excited about these three studies.  Plus, I’m on my parents sofa in DC, so not a lot of plants to photograph.

PPS.  Check out the Mongabay reporting on all three of these articles.