New paper: predicting climate from leaf venation networks

cr 2

If you travel from the alpine zone to the lowland tropics, more than just the climate will change. Species composition follows changing environments: there is a reason why a Cyathea tree fern doesn’t live in Colorado but does live in Costa Rica. (Note of course that many millions of years ago, when the earth was warmer, rain forest could be found even in what is now Antarctica!).

What is it about a tree fern that prevents it from living in a cold environment? A central idea in modern ecology is that measuring properties that reflect an individual’s function, performance (and ultimately fitness) can explain this pattern. My main hypothesis has been that leaf venation networks are the key ‘trait’ to measure. Here’s why. Plant growth requires water loss through transpiration in the leaves, and transpiration requires water supply, which is provided by the leaf veins. If there is a preferred growth rate and water loss rate set by the environment, then only certain leaf venation networks should be viable in each environment.


I just published a paper that tests this idea. I had noticed early on that different vein networks were associated with colder and warmer, or wetter and drier environments, as you see above. To see how general these patterns were, I went in search of leaves from a broad range of climates.

First, Dr. Brad Boyle led a botanical expedition on the western slope of Costa Rica. We started in lowland moist forest, as you see below. Access to one field site required hiking past a beautiful beach, and it was very tempting to ditch our equipment and sweaty clothes in favor of a late-afternoon swim.

cr 1

We collected plants all the way up to cloud forest and páramo vegetation, across an elevation gradient of more than 3000 meters.

cr 3

Second, I made collections in the Colorado Rocky Mountains. Here, sites ranged from high desert to subalpine meadow.


My favorite sites were in aspen forest, where Neill Prohaska helped me with tree climbing (aspen bark is very slippery).

co (1)

We found an intriguing pattern – vein density (length of veins per unit area) decreases with elevation, meaning that lowland sites have higher resource fluxes than montane sites. But the slope of the line wasn’t constant: the relationship depended on whether we were in the tropics or not.


I decided to try to explain this pattern based on plant physiology: perhaps the slope difference reflects differences in climate, i.e. growing season temperature, carbon dioxide availability, or intensity of solar radiation. We developed a mathematical model that uses vein density to quantitatively predict these climate variables, based on the idea I wrote about at the beginning of this piece: there is an optimal physiology for a given environment, so that the water supply in a leaf matches the water supply in the environment.

I was surprised by how well it all worked. Below is an observed-predicted plot: we compare observed values of climate to values predicted from measurements of vein density, such that a perfect fit falls on the diagonal.


You can see that we sometimes over-predict or under-predict – but in all cases, the model explains nearly all the variation in the empirical data, with trends in the correct direction. I think this means that the model is capturing some important aspects of reality. The paper’s main contribution is showing that leaf veins provide an important explanation for why species are only found in certain climates around the world.

You can read the paper in New Phytologist (link, or free PDF reprint). It’s been more than three years between conceiving the project, doing the fieldwork, doing the math, and seeing this published. Science isn’t always fast. We certainly had setbacks, and I’ll close by showing you one of them. Visualization of leaf veins requires some chemical work in the lab, and we had a very exciting hotplate malfunction one day. Fortunately everyone’s eyes and fingers survived the accident, and our dataset is just a little bit smaller as a result!


9 Comments Add yours

  1. monomiao says:

    Would you like some leaves from wetland trees? I am a bit curious… 🙂

    1. bblonder says:

      I don’t know if I have time to analyze a new dataset – but do you want to collaborate? 🙂

      1. monomiao says:

        Hi Ben, I would be happy to try some leaves if it won’t take much time. Now we are in the good season to collect leaves. 🙂
        I might write an email to you later for details.

  2. Greg Blonder says:

    the hotplate seems more dangerous than a rattlesnake…

  3. Do you think this could be used to predict past climates from venation patterns on fossil leaves?

    1. Nevermind, just looked at the actual paper, and the answer is right there in the abstract! Cool!

      1. bblonder says:

        Yes, we are trying to do this on another dataset right now – the paper is in review and I will hopefully be able to share more soon!

  4. monomiao says:

    By the way, how did that accident happen? Is the hotplate too old or overheated? Good nobody was hurt….

Leave a Reply