« April 2009 | Main | June 2009 »

May 30, 2009

Whom do cheating bacteria cheat: host plants or other bacteria?

Bacteria known as pseudomonads produce and release chemicals (defensive toxins) that protect plants from fungi that would otherwise attack their roots. In return, the roots release various organic compounds that serve as food for the bacteria.

The "in return" part has always bothered me. Each root system is associated with millions of bacteria. In a 2003 paper (Cooperation in the rhizosphere and the "free rider" problem. Ecology 84, 838-845), we pointed out that this system is a potential tragedy of the commons. Mutant bacteria that don't make root-protecting chemicals free up resources for their own reproduction, so we might expect them to out-compete more-beneficial strains. If these "cheaters" become common enough, the host plant might be killed by fungi, but that would hurt the beneficial strains around that root system just as much as it hurt the cheaters. We suggested that the bacteria make these toxic chemicals to protect themselves, with protection of roots as a side effect. Research by others, including some of the authors of this week's paper, has provided data consistent with this hypothesis. For example, toxins made by pseudomonads protect them from predators.

More recently (in Annual Review of Ecology, Evolution, and Systematics), Toby Kiers and I suggested that cooperation between microbes and plants is better understood as cooperation among microbes. For example, by providing their host plant with the nitrogen it needs to grow, rhizobia (root-nodule bacteria) help all the other rhizobia infecting the same plant.

This week's paper shows that defensive toxin production by pseudomonads is similar. Toxin production by pseudomonads may benefit the plant and may benefit individual cells, but it also benefits other pseudomonads nearby. "Predators promote defence of rhizosphere bacterial populations by selective feeding on non-toxic cheaters" was published in The ISME Journal by Alexandre Jousset and colleagues in Germany and Switzerland. These pseudomonads produce various toxins, especially when there are many of them in close proximity. This dependence of toxin production on bacterial density is an example of quorum sensing. Mutants "defective" in quorum sensing have been shown to grow (i.e., reproduce by dividing) faster, because they save the cost of toxin production. Can these "cheaters" free-load on defensive toxin producers nearby, essentially hiding behind their chemical defenses?

To find out, the researchers gave the two genotypes (toxin-producers and nonproducers) genes for two different fluorescent proteins, so that they could be told apart under a microscope (or on culture plates; see photo). They then mixed the two strain together in different proportions (10:1, 1:1, or 1:10). In their most interesting experiment, these mixtures were applied to the roots of rice plants. They then added predators that eat bacteria (amoebae or nematodes) to some of the roots.

When they started with mostly cheaters, the frequency of cheaters decreased. When they started with mostly cooperators (defensive-toxin makers), cheaters became more common. This kind of result is known as negative-frequency-dependent selection. Under nematode predation, they estimated that the two strains would have equal fitness when the frequency of the cheaters was 20%.

The frequency-dependent selection appeared to result from the interaction of two factors. The predators preferred to eat nontoxic bacteria. By itself, this effect might be strong enough to eliminate the growth advantage of the cheaters, perhaps eliminating them from the population. But when nontoxic mutants were rare, they were protected by toxin produced by others.

One curious result was that they also found negative frequency dependence (although the equilibrium frequency of cheaters was greater) even when no predators were added. Previously, some of these authors have suggested that the toxins may have a role in competition, as well as in protection from predation.

A role in competition might also help explain another thing I wondered about. If toxins protect individual pseudomonad cells from predation (because predators preferentially eat nontoxic individuals), why is toxin production linked to quorum sensing, so that little toxin is produced by isolated individuals? Maybe they use the toxins to kill competitors or to chase them away. Or maybe predators are attracted to groups of bacteria, so that isolated bacteria have less need for protection against predation.

If cheating pseudomonads have an advantage only when they are relatively rare, then maybe we don't need to worry that production of pseudomonad toxins (which protect roots from fungal pathogens) will disappear over the course of evolution. Instead, we should expect some mixture of protective and nonprotective pseudomonads under most conditions. On the other hand, if selection for toxin production depends on how well these toxins protect the pseudomonads from predators, then there is no guarantee that the toxins will also protect plants from every new fungal strain that evolves.
Photo from Alexandre Jousset shows two strains marked with different fluorescent markers, growing together.

May 23, 2009

Livescribe SmartPen Review and solution to "unable to access your database folders" problem

My wife bought me a Livescribe Smartpen for my birthday. It's an amazing device, but I can't recommend it at this point. First, the positive: as advertised, it records handwritten text (using special notebooks), displays the text on a computer, and recognizes hand-printed text well enough to search through stored pages for keywords. It can also record sound. It doesn't work with Windows 2000, so I switched to Windows XP, something I haven't had to do for any other program. But I thought my planned uses justified the switch:

1) lab notebooks. I often need to refer to something I wrote months or years ago. With Livescribe, it should be possible to find it quickly.
2) taking notes in seminars. The audio recording is good enough I can can just write keywords and make sketches of graphs, knowing that I can refer back to the audio for details I didn't get written down.

But, after only a few days of use, the program suddenly informed me that it was:

"Unable to access your database folders. Please contact customer support."

Reinstalling didn't help. OK, I'll contact customer service and report here (and Amazon.com, etc.) how they respond.

Apparently other people have had the same problem.

Update: I got a reasonably prompt generic ("what operating systems are you using?", etc.) response but no actual help so far. I was able to install on a different computer, but worry that the same problem could arise there at some random time in the future, meaning I would lose any files that weren't still on my pen.

Update2: After a few days, "customer service" sent me another generic request for information. Maybe they figure repeatedly asking for more information and just hoping users will solve the problem themselves can be outsourced, whereas they would have to hire someone competent to actually figure out what was wrong. The reason I suspect this is that they didn't seem to do anything with the information they asked for the first time, and they didn't answer a very specific question I asked when I sent them the first bunch of information they asked for, namely, whether it might help to copy the (hidden) MyLivescribe directory from another computer where I'd gotten it to work. Anyway, I went ahead and tried it and that seems to have fixed the problem. So I guess backing up the MyLivescribe directory periodically would be a good idea. There's apparently some way to back stuff up on the company web site, but what happens to the data when the company goes out of business?

I agree with comments on their website: until there's a way to send Livescribe files to colleagues without going through the Livescribe website, they will lose millions of potential customers. For example, this system would be great for notebooks used to document patentable inventions, but nobody working on a patentable invention is going to trust their notes to an outside company.

May 22, 2009

Oxytocin and the genetics of altruism

Where to publish a paper on the genetics of altruism? In an open-access journal, of course! One day after publishing the fossil primate paper that's creating so much excitement -- it's a great fossil, but too old to tell us anything about our recent ancestors, shared with other apes, or the less-recent ones shared with monkeys -- PLoS One published "The Oxytocin Receptor (OXTR) Contributes to Prosocial Fund Allocations in the Dictator Game and the Social Value Orientations Task", by Salomon Israel and colleagues. Like all papers in open-access journals, the full text is available on-line.

These researchers measured altruism in 200 students, based on how each chose to divide a pool of money with another unknown individual. Their hypothesis, based on various past studies, was that the hormone oxytocin is important for social interactions in general and for human altruism in particular. For example, Zak and colleagues showed that sniffing oxytocin made people offer a more generous split when the recipient had the chance to retaliate for a low offer (the "Ultimatum Game"), although not when there was no chance to retaliate, as in the Dictator Game used in the current study.

The researchers tested for statistically significant relations between and different variants of the oxytocin receptor gene, which codes for the protein that responds to this hormone signal in the brain, and "prosocial responses" (generosity) in the Dictator Game and a more-complex version, the SVO. Interestingly, none of the genetic differences they looked at were in the protein-coding part of the gene (orange). Most were in an intron, which would be transcribed from DNA into messenger RNA but then cut out before the mRNA is translated into protein. So I assume these genetic differences could affect how much oxytocin receptor protein is made where and when, but not the structure of the protein itself.

Generosity depended most strongly on whether a DNA base shortly after the protein-coding region (rs1042778) was a guanine (G) or a thymine (T). The difference was statistically significant, but also fairly small (5.1 vs. 5.5). So choosing a mate (or a business partner) based on this genetic difference might be premature. I see two possible explanations for these results. Maybe even a large difference in the amount of receptor protein has only a small effect on generosity. Alternatively, maybe this specific difference (and other genetic differences studied, some of which also had significant effects on generosity) has only a small effect on the amount of receptor protein made. In that case, there could be other genetic differences (perhaps not included in the study group) that would have a larger effect.

May 8, 2009

"If evolution is true, why are there still chimps?"

I once heard PZ reply to this popular creationist question by pointing out that, although many Minnesotans are descended from Norwegians, there are still Norwegians. This isn't really a good analogy, however, because Minnesotans and Norwegians aren't separate species. We know this because they can interbreed, producing healthy children. At the end of this post I suggest a better answer, indirectly inspired by this week's paper.

Two of evolutionary biology's central questions are: how do species change over generations? and how does one species split into two? We have many detailed examples of small evolutionary changes occurring over days (in bacteria) or years (in animals and plants), so one would have to be very close-minded to deny major evolutionary change over millions of years. But major evolutionary change is not enough, by itself, to split one species into two. One subpopulation within a species must change, while the rest of the species either stays the same or changes in different ways. This divergence cannot happen if the two subpopulations continue to interbreed at high rates. In other words, speciation requires some reproductive isolation.

Often, reproductive isolation is a byproduct of geography. After a few individuals (or a pregnant female) cross a mountain range or are blown from the mainland to an island, they no longer interbreed with their ancestral population. Over many generations, random genetic drift or nonrandom natural selection can change the isolated population enough that they can no longer produce healthy offspring with the original population, even if they come back into contact.

Sometimes speciation can occur without a major geographic barrier, but reproductive isolation is still required. This week's paper shows that this has happened and is still happening in Europe.

"A continuum of genetic divergence from sympatric host races to species in the pea aphid complex", by Jean Peccoud and others, was just published online in the Proceedings of the National Academy of Science.
Photo by Jean Peccoud

The authors collected 1090 pea aphids feeding on 19 different species of wild and cultivated legume plants across Europe. When they developed a family tree for these aphids, based on their DNA, they found that they fell into 11 distinct groups. The most closely related aphids were those feeding on the same host species, rather than those from the same geographic area.

There were 11 distinct "biotypes", but are these all separate species? In other words, how complete was reproductive isolation? About 9% of aphids were found on the wrong host for their genotype. These "migrants" could, in theory, have mated with individuals there. But was there enough interbreeding to prevent speciation?

To find out, they looked for hybrids, whose DNA showed they resulted from mating between different biotypes. Three of the biotypes are apparently separate species, because no hybrids between them and other biotypes were found. The other 8 biotypes did produce hybrid offspring with other biotypes, at least occasionally. Genetic differences between biotypes were greater, but not much greater, for those identified as separate species.

Given the rarity of hybridization among some of the other biotypes, will some of these also evolve into separate species? This could happen if differences among host plants result in tradeoffs, such that aphids that grow well on one host grow poorly on others. In that case, hybrids may do poorly on either host and there would be strong selection to feed and mate only on the preferred host.

Here's my answer to the creationist question:
"Have you ever had sex with a chimp? No? Neither has anyone else, for a long time. That's why there are still chimps."

May 1, 2009

Sibling rivalry in plants

This week I will discuss two papers, both dealing with plants and competition, in the context of genetic relatedness that might be expected to moderate competition:
"Growing with siblings: a common ground for cooperation or for fiercer competition among plants?" by Ruben Milla and colleagues (Proceedings of the Royal Society), and
"Do plant parts compete for resources? An evolutionary viewpoint" by Victor Sadras and me (New Phytologist).

Earlier I discussed a paper by Susan Dudley and Amanda File showing that some plants grow less root when interacting with related than with unrelated neighbors. Spending less resources on roots could have freed resources for more seed production, but they didn't measure that. Now Milla and colleagues have.

They grow three lupine plants per pot, using either three seeds from the same plant, three seeds from different plants in the same area, or three seeds from different parts of Spain, and measured various aspects of plant growth and reproduction. In contrast to what I might have expected from Dudley and File's work, plants surrounded by siblings produced no more seeds than plants surrounded by strangers. In fact, one of their measures showed significantly more seed production from plants growing with plants from other regions.

They suggest two possible explanations. First, there was some tendency for plants to grow taller when growing with close kin, perhaps because they all germinated at the same time and thereby triggered an "arms race" to get above each other. The resulting over-investment in stem could leave less resources for seed production. Their other explanation is almost the opposite. What if closely related plants invest less in root, as Dudley and File found, and (under the conditions of Milla's experiment) this resulted in too little root for optimal uptake of water and nutrients?

When wild plants are grown in pots in a greenhouse, they may not allocate resources optimally, nor respond normally to environmental cues, including cues about the relatedness of their neighbors. But if hypothetical cooperation among closely related plants is weak enough to be undermined (even reversed) by growth conditions, the tendency to cooperate can't be very strong.

I discussed a paper by Victor Sadras in one of my first posts in This Week in Evolution, so I was intrigued when he invited me to collaborate on a paper reviewing the idea of "competition" among parts of the same plant. We argue that mechanisms that look like within-plant competition often act to maximize overall plant reproduction. A branch shaded by another branch may die, but this is more like suicide than murder. We know this because the same degree of shading isn't lethal when the whole tree is shaded equally. When only one branch is shaded, however, it can increase the frequency of its genes in the next generation by sending its nitrogen to better-lit branches, where the photosynthesis rate per unit nitrogen is greater. Seeds produced on those branches carry the same genes as those that the shaded branch could have produced itself. Selfish genes lead to unselfish branches.

Competition among seeds on the same plant is a different story. These seeds may have different fathers, whose pollen contained competing versions of various genes. Gene variants that help a seed take more than its share of resources from the mother plant will tend to increase over generations, unless countered. But mother plants have various counter-measures that tend to equalize resources among seeds. (This contrasts with birds that can only bring enough food to feed one chick. They may lay two eggs, but then let the stronger chick kill the weaker.)

We suggested that natural selection for equalizing resources among seeds has often set limits on how much seeds can grow, even when conditions turn out to be unusually favorable during seed-fill. This tradeoff may have been worth it for genetically diverse wild plants. In modern agriculture, however, whole fields may be almost identical, genetically. We might therefore be able to eliminate some of these ancestral seed-balancing mechanisms, letting seeds grow more when conditions are good.

Such tradeoffs between past natural selection and present human goals are a major theme of my forthcoming book, "Darwinian Agriculture: where does Nature's wisdom lie?"