How to taxonomically structure comparisons

For a recent grant, we proposed to measure aspects of the nitrogen and water economy of 30 species at Konza. The novelty of the proposed research was in measuring both water- and N-related traits for a wide variety of species, and then test how well they explain the abundance of the species in a native grassland.

One point that came up was how to frame the research. Part of our framing was that the results should help us understand the evolution of plant strategies and selection forces on species. Reviewers seemed to disagree.

One reviewer said, “The problem here is that because of the close evolutionary relationships of many of the selected plants, traits and responses will be co-correlated through evolutionary relationship and will therefore give an inflated estimate of independence.”

Another said, “It seems to me that the work in this project will yield much in the way of an understanding of the influence of resource availability in the evolution of land plants, since gaining such insights would really require a more extensive phylogenetic and perhaps phylobiogeographic sort of approach.”

This is something I still do not understand. How many species does one have to measure to be able to infer selection pressures and evolutionary tradeoffs? Ironically, we had initially proposed to measure 100 species, but were encouraged to measured fewer species. 30 is not enough? Shouldn’t 2 well-contrasted species be sufficient to provide some inference? Most of the initial work on C4 photosynthesis compared 4 species. Granted the work is still being refined, but isn’t 30 a good start? Also, why would 30 species be enough to test ecological processes, but not evolutionary?

I think the standards here have less to do with the science, than the scientist.

The current review is immaterial—the panel summarized that “The placement of this research in evolutionary context was undeveloped but will not affect the quality and novelty of the project outcome.” Yet, the gap in our scientific process is clear in the lack of anabolic comments being paired with the catabolic ones. Experimental designs to test for evolutionary patterns seem to require I-know-it-when-I-see-it tests. Constructively, we need some resolution on standardizing designs. I’ve pushed before for a standard species set, but we also need resolution on some key questions outside of any standardized set.

If there is one question I'd like to see answered, it's "how many species need to be measured and how should they be related?"

I don't expect one answer to this, but if we are serious about wanting to understand the evolution of ecological traits, we have to make the bar visible, rather than always place it just above our leaps.

The model species set

By restricting our own freedom, we gain collective power. It's a tenet of larger society, but also scientific society. 

For some the restriction is in the form of Arabidopsis. Zea for others. Populus, Lotus, Medicago...the list of model organisms that are used to answer fundamental questions about the genetics of plants goes on.

But what about the evolution of plants? To a degree, we can compare the genomes of model organisms to hint at some of the broader evolutionary patterns. But evolutionary patterns are generally derived by comparison with multiple members of a single clade. If one wanted to understand the evolutionary patterns of grass, we couldn't just look at a single model organism. We would need to look at a model set of species.

What would a model species set for grasses look like? It would have to be large enough to cover the major clades (~10), but restricted enough that researchers could measure standardized metrics on every species. Probably about 100 species. For grasses, they should come from different continents, span multiple origins of C3 and C4, and cover a wide range of environmental tolerances. Seed should be readily accessible. Most likely seed sets would have to be collected by a central agency for distribution to willing researchers. A central database would be needed to store all the data for other researchers to use.

Once that happened, an individual researcher that was interested in the cold tolerance of grasses could grow up all 100 species, measure their cold tolerance, and then examine the evolutionary patterns of cold tolerance. The next researcher that wanted to examine stomatal density could do the same, and then would be able to compare it with cold tolerance. Root anatomy, mycorrhizal dependence, genome size, carbonic anhydrase activity, flowering phenology, drought tolerance...the database would build. Each time we would learn more about multivariate trait selection in ways that no one lab could do.

Why doesn't this exist? Hard to say. Part of it is probably some small group just deciding which 100 species to use. Would it be perfect and cover all the potential evolutionary questions? No, but there are researchers that are asking these questions anyways, so they might as well be using the same species. Plus, there always could be a second species set identified to fill the gaps in the first for a second round of measurement.

Why not just keep a database and let researchers work on whatever species they felt best allowed them to examine specific ecological and evolutionary contrasts? Never enough overlap. Brassicaceae has 3700 species and even the Arabidopsis genus has 9 species. But everyone works on thaliana even if other crucifers might be better to answer some questions.

Once the scientific community agrees to encourage the restriction of freedom of inquiry into plant evolution a little more, a large amount of collective power will be realized. How long should it take? A few informed individuals who are not afraid to make political sausage would need to be in the same room for about 2 days. How long will it take to get people in a room for 2 days? Hopefully within a year or two.

Do I have to phylogenetically correct my grocery list?

The figure of Westoby et al. (1995) that summarizes their view of the tension between phylogeny and ecology in understanding trait relationships.

For some, a simple grocery list can pose a dilemma. Just yesterday, I went to the store with 21 items to buy. Others would look at my list and suggest I only bought 11 items. Fresh peas and frozen peas shouldn't really be counted as different items--they were both peas. Cauliflower, broccoli, and collard greens are the same species. Mustard part of the same genus as the previous three. Hot dogs and pork chops both from pigs (I hope). So although 21 items went into the cart, one could phylogenetically correct my list and arrive at the conclusion that I only bought 11 unique items.

It might seem silly to phylogenetically correct one's grocery list, but how to consider both phylogenetic and ecological data when examining species relationships lays bare the same fundamental tension as describing my last trip to the grocer.

In 1995, Westoby, Leishman, and Lord published a forum piece, “On misinterpreting the 'phylogenetic correction'”. The genesis for the forum piece came during the review process of a paper on seed mass in plants. Most likely, during the review of that paper, differences in opinions between reviewers and authors were laid bare. In the original paper, the authors showed that tall plants had large seeds. The reviewers likely insisted that the relationships between plant height and seed size could be due to phylogenetic relationship. The authors disagreed. Differences in opinions became forums, which by ecology standards unleashed a bit of a storm.

The questions associated with the topic of how to match ecological and phylogenetic data are ripe, but “phylogenetic correction” essentially adjusts relationships by weighting closely related species less than distantly related species. The fundamental differences of opinion pin whether closely related species hold similar traits because of phylogenetic constraint or ecological constraint. Closely related species might have similar traits because there has been little time for radiation, or because they are under similar ecological selection pressure. Distantly related species might have different traits because initial trait differences have long been conserved due to fundamental difficulties associated with character displacement or because they have been under the same ecological pressures for a long time.

The issues of how to identify adaptations or evolutionarily beneficial relationships cannot be covered here, but these fundamental issues have never been resolved, near as I can tell. The current détente that seems to exist is to examine ahistorical and “phylogenetically corrected” relationships among traits and hope that the patterns are the same. When setting to test relationships among species, choose congeneric species pairs from distantly related genera and hope the patterns work out consistently.

It’s currently an uneasy impasse. Both sides recognize that correlation does not necessarily imply causation. But outside of hoping that the evolutionary and ecological patterns parallel, there is still no resolution to the question of how to compare the traits of species.

I do know that if I want to shrink my grocery list, I'll start by not buying both cauliflower and broccoli rather than phylogenetically downweighting closely related taxa on my list.

The evolution of grasses: phylogeography of C4 photosynthesis

The temperature niches of grasses of the world overlaid onto their phylogenetic relationships.

The two great datasets in biology are the tree of life and the global biogeographic distributions. The first describes the phylogenetic relationships among organisms. The second describes their distributions on our planet. In a rich and well-nuanced paper, Edwards and Smith have brought the two together to shed light on one of the most fundamental questions regarding the evolutionary ecology of plants, namely the origin of C4 photosynthesis. The authors first use an expanded grass phylogeny to describe the origins of C4 photosynthesis in more detail than has done before. They then determine the current distribution of the grass species to determine the climates they occupy.

With regard to the evolution of C4 photosynthesis, the authors conclude that shifts from C3 to C4 photosynthesis did not involve shifts to warmer macroclimates, but instead to drier macroclimates. This results comes as a bit of a surprise--it is less clear that C4 photosynthesis is a response to low water availability as much as high temperatures. Their next logical step is a bit of a leap--namely that these modern geographic differences can be associated with habitat shifts in the past.

As important as the insights into the phylogeography of C4 photosynthesis is that the evolution of cold-tolerance in grasses is more difficult evolutionarily. Cold-tolerance apparently evolved vary early on in the grass radiation and has not been repeated to the degree that C4 photosynthesis has.

In all, this hardly seems like the last word on the topic. The biogeographic data needs to be improved, climatic ranges rather than centers will likely be used, and the grass phylogeny is still relatively unresolved. Also, we still have little understanding of why C4 photosynthesis would benefit plants in dry environments. That said, there is a lot of insight for many types of researchers and a solid step in understanding the strategies of plants to resource scarcity.

Edwards, E. J. and S. A. Smith. 2010. Phylogenetic analyses reveal the shady history of C4 grasses. Proceedings of the National Academy of Sciences 107:2532-2537.

What to grow after the tree of life?

(crocuses at the Missouri Botanical Garden)

I had a fun conversation with Peter Raven of the Missouri Botanical Garden the other day. In essence, the topic that we discussed is what happens after the tree of life has been constructed? Today, thousands of researchers are working with a primary goal to reconstruct the phylogenetic relationships among organisms. But what happens after the tree is largely reconstructed? Yes, the tree of life will never entirely be reconstructed. One can argue that we can work on finer and finer branches, for example, trying to understand relationships among populations. 

Yet, I can't help but look at the tree and begin to ask why. Not why we are reconstructing it, but why it is shaped the way it is. What ecological forces might have been behind so many of the radiations? For example, we know about 10% of the world's flora either use the C4 or CAM photosynthetic pathways. Broadly speaking, these pathways are an evolutionary response to low atmospheric CO2 concentrations. These species are not just adapted to low CO2, nor have the other species not been influenced by low CO2, but what about the other 90% of the world's species? How many are a result of selection for success in shade? Or competition for nutrients? Or the ability to tolerate drought? Moreover, can we begin to compare the regions of the world and better understand the biogeography of species? Sure, we can see who is where, but do we know why? 

Evolutionary ecology is more than microevolutionary examination of processes that affect selection. There are broad questions about the evolution of the world's that cannot be answered by small-scale manipulations. One of the hurdles we need to get over is bias against what before have been called screening experiments. Hundreds of species are grown in common conditions and their traits measured to better understand the fundamental differences among species. These experiments are not trivial, but it seems clear that if we are going to answer these broad questions about the evolution of the world's flora, we need to screen more species. Not tens of species, but tens of thousands. Imagine a single greenhouse with all 10,000 of the world's grass species growing side by side. As far as I know, the largest ecological screening project measuring ecological traits of plant species included 640 species. After that, few experiments measured a hundred.

So what do we grow after the tree of life? If you ask me, it would be a garden.