Dear Evolutionists,

here are the responses from Dr. David Reznick.

David Reznick is a professor of Biology at UC Riverside. He is very interested in the evolution of life history traits, and has worked extensively with poecillid fish. In addition, he works with amphibians and reptiles. He has taught Comparative Vertebrate Anatomy, Evolution, and Population Biology. This year he received the award for "Outstanding Lecturer", which is the highest award given by the UCR academic senate. We are very lucky to have had his participation.

In addition to his thorough answers, he wrote "I probably tended to tell them more than they wanted to know, but part of it was deliberate to show some of the broader implications of the questions. I would be happy to field secondary questions of there are any. It was fun." Take him up on it—he's a great person and lots of fun to talk with!

His website is: http://cnas.ucr.edu/~bio/faculty/DaveR.html

Your experiment consisted of moving guppies from a high- to a low- predation site. Has there been an experiment done that would move the guppies from a low- to a high-predation site? If so, what were the results? If not, then would you hypothesize that the guppies from the low-predation site would evolve to be smaller and to reproduce at a younger age (characteristics of the high-predation populations)?

Yes, I have done such an experiment. The Aripo River has a barrier waterfall that stops the serious predators, but not guppies or some less dangerous predators. There are characins, pike cichlids, and Hoplias below the waterfall, but only Rivulus, Aequidens pulcher (a less piscivorous cichlid) and some non-predaceous species above the waterfall. In this case, I introduced pike cichlids over the waterfall. They extended their range to another waterfall around a half kilometer upstream. This means that I have a high predation control below the natural barrier waterfall, a low predation control above the secondary barrier waterfall, and an introduction site sandwiched inbetween. I report on some results in the following paper: Experimental Gerontology 32: 245-258 (1997). These are data based upon collections that were made four years after the predator introduction. As you predict, the introduction site fish are smaller and younger at maturity than those from the low predation control. They are larger and older at maturity than the high predation control. The controls differ significantly, but the introduction site does not differ significantly from either control - the sit right between the two. I have yet to evaluate mortality rates in these fish. It tested them again for genetic differences in life histories a few years ago and found that these initial differences had disappeared. At the same time, the range of the predators had contracted to only 100-200 meters of stream because they got washed out from the upper portions of the introduction site. It is possible that the life history differences disappeared because they were swamped out by gene flow from the upstream control (their values were the same as the upstream control in the second run of the experiment). These and other data have prompted me to think about the interplay between gene flow and local adaptation in this system, but I have yet to pursue this matter further. Consider this study a work in progress.

Besides this one river, I have found others which are high predation but lack guppies. In one case I have twice tried to introduce guppies from a neighboring low predation locality. It turns out that they go extinct rapidly. There are now five introductions of guppies from high to low predation sites and all five were very successful. I wondered why there was such a difference in the direction of the introduction. I deal with this in a paper that is in press in an edited volume on Evolutionary Conservation Biology. Helen Rodd and Len Nunney did population and genetic simulations of the alternative introductions and they show what common sense might otherwise have dictated. High predation guppies mature quickly and produce lots of babies. When you put them in a low predation environment they experience reduced mortality rates. Population models predict that the populations will always explode, and they do. Guppies from low predation environments are more slowly developing and produce few offspring. When you put them in a high predation environment they experience higher mortality rates. Models predict what we saw, which is that the populations will almost certainly go extinct. We finish that paper with a brief review of case studies of rapid evolution (e.g., industrial melanism, insecticide resistance, heavy metal tolerance, and others). It is almost always the case that rapid evolution is accompanied by unusual circumstances that either permit rapid population growth or that allow a lottery of repeated colonizations. For example, in the other three classes of study, the adaptation is to a novel environment that has reduced diversity (usually because an anthropogenic agent has wiped out a lot of the fauna or flora) but has source populations for repeated colonizations. The more general conclusion is that rapid evolution is more likely to be seen in specific circumstances. I would be happy to send a preprint of this paper is anyone is interested. I hope to work on this idea more in the future.

More answers later.

Sincerely, David

2. Scott Nichols, Rafael Oliveira, Malia Rivera, Chet Moritz ask:

It seems that the changes observed in the experimental guppie populations transplanted to a predator free upstream pool could be due to factors other than releases from predation. For example, If the upstream pool originally contained no guppies or predators, it is likely that other organisms are also absent. This could mean that there is less competition for food resources in the upstream pools. It seems that increased nutrition could also result in later maturation and increased size. Were you able to control for these factors? If not, on what basis are you asserting that the observed changes are due only to a release from predation?

Okay, I will try to sum up. I sent a warning via Tina that I write these replies off the top of my head inbetween all sorts of interruptions, so I hope that they make at least some sense.

There may be a short term difference between the introduction experiments and a typical low predation site. The introduction sites have the same basic ecology and faunal composition before the introduction, except that Rivulus appear to be more abundant than they are in a locality that also includes guppies (Frazer and Gilliam's work). The guppies are initially at a lower density also. We know that the guppy populations initially grow very rapidly, then stabilize. This means that the sort of selection that they are exposed to early along may be different from what they experience in the long term. Within around three years (the time interval might be shorter, but I did not think to check this when I set the experiments up so long ago) the conditions are very much like they would be in a typical low predation site.

Given all this, there still remains the possibility that the sort of selectin that guppies experience in low predation sites is more than just higher mortality. This would apply to all sites, not just the introduction experiments. Light and productivity levels are lower, population densities are higher, the age-structures of the populations are different (more old fish in low predation sites - They vs. high predation sites are like the age structure of Switzerland vs. Mexico). Some of these differences are independent of predators and some might be an indirect effect of predators. For example, higher densities and different size/age distributions are likely to be a consequence of lower predation.

I will go over your question again and make sure that I answered it. I certainly told you more than you asked for, but I think that the details are relevant.

Sincerely, David

The one thing that I forgot to emphasize in my response to question 2 is that the work that they read was intended to argue for predation as a likely factor in life history evolution, but not to argue that predation is the only factor. They are correct in noting that the movement of guppies to a new sites changes more than predation. It is important to think of the study in the context of the whole body of work, which established a very strong correlation between predation and life history traits that cut across ecological variation before the introductions were done.

More later.

sincerely, DAvid

(a few days later)

Actually, I agree that there are other possible explanations, but not the ones that you imagine. One of the themes of my work over the past 12 years has been to consider ecological correlates of predation and interactions between predation and ecology. I will give you a brief synopsis, but first will address the nutrition issue. It turns out that high levels of food availability cause earlier maturity, larger body size at maturity, faster growth rate, and higher fecundity. You saw later maturity, larger body size, and lower fecundity in the Nature paper, which is not consistent with food availability. I know from other results that guppies in such localities tend to be found at higher population densities and to have lower growth rates. It also turns out that these sites tend to have lower light levels and lower primary productivity. I describe these results in a paper that is currently in manuscript form. I hope to complete revisions in the next few weeks, but would be happy to send you a copy of what I have now if you are interested. I can send it as an attachment. Note that these data include this introduction site. The appropriate details are not contained in the paper, but we found that the introduction sites were ecologically indistinguishable from natural low predation site. Specifically, the low predation sites contained the same suite of invertebrates plus Rivulus that you would see at any low predation site. The one difference when guppies are absent is that the Rivulus tend to be far more abundant. This has been shown by Doug Frazer and Jim Gilliam. It turns out that guppies eat young Rivulus in the same way that Rivulus eat young guppies, so there could have been an initial competitive effect with Rivulus that is not typical of the usual low predation site. I have also compared mortality rates in the introduced guppies relative to the downstream control. The introduced population experiences lower mortality rates and has a mortality pattern that is consistent with what I see in other low predation localities. Some of these sorts of data appear in a paper that I published in Evolution in 1996. For some reason, I did not emphasize that the data were derived from the introduction experiments in that paper. I think that this was just an oversight on my part. These data do suggest that the differences inpredation are associated with differences in mortality rate which in turn is predicted to cause the life history to evolve as it has. This does not actually prove that the mortality rates caused the life histories to evolve.

I have to go to a meeting. I will add more detail later.

3. Jose, Mike German, Anne Marie France, Gautam, Maggie ask:

Could the changes seen in the guppies be the result of phenotypic plasticity, not actual new changes in the genome?

No. The details of what we did go by pretty quickly in the Science format, but the inference that there were genetic differences was based on the second generation of laboratory born fish. We collected adult females from natural populations. Female guppies retain sperm, often from multiple males, and will give birth to a number of litters when isolated in the lab.

These females were arrayed in a stratified, randomized block design in the lab. When their young were born, they were reared at a density of four to six per six liter aquaria and those aquaria were again arrayed in a stratified, randomized block design. When they matured, they were mated in a breeding design that retained the genetic variation of the initial sample, did not result in any inbreeding, and evenly represented all wild-caught females. The offspring of those matings, the grandchildren of wild-caught females, were reared until they were 25 days old, again at a constant density and in a stratified, randomized block design. We then selected two males and two females from each litter to be reared on high versus low food. From 25 days on, every fish received a quantified ration twice a day. Given multiple generations of a controlled environment, we interpret any differences that remain among populations as having a genetic basis. It has recently been shown that a mother can transmit environmental influences to offspring, but there is no known plasticity that will be transmitted across two generations in a constant environment. In other studies, I have hybridized females that produce large and small offspring and have shown that offspring size is a quantitative genetic trait. It is possible that there is some other mechanism that could account for differences among population persisting for two generations in a common environment, but I am not aware of any.

Another feature of this study that argues for a genetic basis of life history variation within a population is the two quantitative genetics assays. In this case, the fish were bred in a formal paternal half-sib design for two generations (described in one of the footnotes). These studies were the source of the tables of heritabilities and genetic correlations in the main part of the text. It was this approach that showed that the heritability of male age and size at maturity was so high, which yields the prediction for more rapid evolution, which is what we found. It is important to note that this sort of approach characterizes genetic vartiation within populations, while the approach above characterizes genetic differences among populations.

4. Patrick, Alan, Robert, Jan ask:

In this study it was found that the male guppies evolved more rapidly then the females. Has this been found to occur in other species as well?

Not that I am aware of, but that is probably more a matter of an absence of looking, rather than an absence of the phenomenon. There is only a small number of studies that have considered evolution in natural populations in this fashion. The only other formal experiment was done by Jonothan Losos (and co-authors) on Anolis lizards. This species is sexually dimorphic, but they do not report on differences in rate of change in males and females. Also, they do not demonstrate that there is a genetic basis to the changes.

Their unpublished work suggests that the differences are the product of phenotypic plasticity. The Galapagos finch work does deal with differences in selection on males, females and juveniles (I believe that a 1984 paper in Evolution by Price and Grant covers some of these results), but I cannot recall anything about differences in rate of change in males and females.

It is worth considering what would be required for sexes to evolve at different rates. It would have to be either that there are differences between the sexes in the heritability of a trait, generation time, or the coefficient of selection. It is easy to imagine that selection on males and females can be different, particularly when there is a strong sexual dimorphism, as in guppies. It is much more difficult to imagine that there could be a difference in the genetic basis of a trait in males and females, yet this is what appears to be the case in guppies. For example, size is a polygenic trait and I would have assumed at the outset that the genes that determine size in males and females would be pretty much the same. In human families, there seems to be an overlap in the effect of either tall mothers or fathers on the height of sons and daughters. The unusual feature of Poeciliid fishes is that the y-chromosome seems to contain a lot of important information on it, such as genes that control color patterns, age, and size at maturity. Such y-linked genetic variation for age and size has been demonstrated in swordtails and mollies. Mollies are in the same genus as guppies. So, this second mechanism for a male-female difference in rate of evolution is one that seems plausible in these fish and is strongly supported in guppies, but is not one that I would predict to be common in other organisms. The third possible mechanism is a difference in generation time. I have not seen a formal consideration of differences in generation time among sexes, but it seems reasonable to me that, if there are differences in the age at first reproduction in males and females and in the length of the reproductive lifespan, that there also could be a difference in generation time. For example, male guppies sometimes mature earlier, plus they have shorter lifespans than females. If they had a shorter generation time and equal coefficients of selection, then they could evolve more quickly in abosolut time.

So, the question is whether or not I would predict that such differences in the rate of evolution between the sexes might be common. I think that it will turn out to be uncommon under the mechanism that applies to guppies, which is differences between the sexes in the heritability of traits. I think that it could be very common under the mechanism of differences between the sexes in the coefficient of selection, particularly when there is a pronounced sexual dimorphism.

5. Jose, Mike German, Anne Marie France, Gautam, Maggie ask:

Could the increased rate of evolution of males be due to two factors: predation and sexual selection?

Yes, and I had predicted that this would be the cause, but the results don't yield strong support. The idea was that both females and the change in predation could "push" males in the same direction of delayed maturity at a larger size. If this were the cause, then our analysis would have demonstrated stronger coefficients of selection on males. It turns out that the differences in the coefficients of selection imply that there is slightly stronger selection on males than females in one experiment (the El Cedro River) but not the other (Aripo River). I think that it is possible that the difference between the two studies is an artifact of the time interval between the introduction and when I first assayed for genetic differences. In the El Cedro River, I looked after four years and found that the males had already evolved to the predicted endpoint. In the Aripo River, I first looked after 11 years, and found the same result as on the El Cedro. If you imagine that the fish on the ARipo evolved as quickly as on the El Cedro, then they would have reached the endpoint in less than or equal to four years, but we would have averaged the change over 11 years. This means that our methods would systematically underestimate the rate of evolution and the coefficient of selection in the Aripo. Given this sort of uncertainty, plus the fact that it is not possible for us to statistically compare the coefficients of selection for males versus females, I will never be able to say for certain what the intensity of selection was on males versus females.

Our results do show, unambiguously, that the differences in male and female rates of change are caused by differences in the genetic variation present in males and females.

6. Paul Moody and Hawkeye Sheen ask:

This may be a more complicated question than you had intended. Plasticity can arise in different ways. There is an environmental component to the phenotype for almost any quantitative (polygenic) trait that you care to measure, which means almost any aspect of the phenotype that you care to measure. For example, it is pretty universal for fish that are fed more to grow more quickly and mature at an earlier age and larger size, plus produce more babies. It is not surprising that if there is more food, then there is more production. This sort of plasticity will probably remain and probably does not constitute an adaptation. A second form of plasticity which you may have covered is adaptive phenotypic plasticity. For example, there are some classic cases of phenotypic dimorphisms that are caused by developmental switches. Curt Lively's work on barnacles is one of the clearest examples that I am aware of and it may be on your reading list. Such plasticity is expected to evolve in only restricted circumstances. One combination of circumstances is that the organism has limited dispersal abilities and that there are unambiguous cues about environmental quality. Such circumstances often apply to aquatic organisms and adaptations to predation, since predators can often be detected through chemical cues. Such plasticity also appears to require a tradeoff, so that one phenotype is best in one set of circumstances and an alternative is best in other circumstances. If such tradeoffs do not exist, then we might expect the evolution of one all-purpose phenotype.

Guppies satisfy some of these criteria, but I have yet to establish whether or not they have adaptive phenotypic plasticity. Whether or not they do will not necessarily be affected by their life history adaptations to predation, nor is there necessarily any constraint between the evolution of the optimal phenotype and the evolution of phenotypic plasticity. I suspect that my answer may have take a direction that you did not expect. Let me know if you have any new questions or if I can do a better job of answering this one.

7. Tonya Van Leuman, Tony Donoghue, Jonathan Woo, Elizabeth, Agrilla, and Danielle Lee ask:

Are the units of evolution described by Gould and Eldredge in the fossil record and the populations of guppies comparable units in their rates of morphological change?

There seem to be two different questions here. By units, Gould has recently argued that species are a plausible unit of selection, while I would argue that, in the case of guppies, individuals are almost certainly the unit of selection. From this perspective, we are talking about different things.

A second aspect of your question concerns rates of evolution. Gould and Eldredge originally argued, correctly, that Darwin predicted that evolution by natural selection would result in continuous change. The fossil record implies that the rate of change is discontinuous, with intervals of rapid change being interdispersed with intervals of little or no change. They argue that this pattern is contrary to what we would expect if natural selection is responsible for these events. My work, and work by others like the Grants on Galapagos finches, reveals rates of evolution by natural selection that are four to seven orders of magnitude faster than rates seen in the fossil record (ten thousand to ten million times faster). This means that there is a huge difference between the rates of change that can be perceived in real time and the rates of change that are necessary to account for punctuations in the fossil record. I would argue that this is a basis for saying that patterns of change in the fossil record are not a reasonable source of inference about the mechanisms that cause these changes. Natural selection is easily fast enough to account for all that we see and even the best fossil records lack the resolution necessary to evaluate the process as we know it can occur. Gould has argued that results like mine prove that natural selection cannot be responsible for evolution as seen in the fossil record because it is so much faster than the fossil record. In doing so, he has reversed the earlier argument by Gould and Eldredge, since they said that natural punctuations are too fast to be accounted for by natural selection, or at least too discontinuous to be consistent with Darwin's expectations. The key difference between Darwin, then Gould and Eldredge, and now, is that we now know that evolution by natural selection can be much faster than anyone might have predicted.

Our result raises some interesting questions. One is whether or not what I am studying is "important" evolution, meaning whether or not it can really go anywhere. Gould (and creationists) argue that it is just a fine tuning process that is limited in scope. Our record of domesticated animals proves that they are wrong. For example, the adult body sizes of domestic dog breeds ranges from 1.5 to over 80 kg. This range is larger than the range of body sizes among species in the family Canidae, but smaller than the range of body sizes among species in the order carnivora. This means that artificial selection, working in the same way and at the same rate as in my guppy study, has produced variation that ranks between the third and fourth steps in the seven step taxonomic hierarchy in only 10,000 years. This process combined with organisms capacity for change can certainly produce what we think of as macroevolution. Equally dramatic change is seen in the morphology of dogs and in a variety of other organisms.

A second question concerns the possible relationship between such change and speciation. Dr. Wake should be covering studies that document that speciation can result from natural selection played out over short time intervals. One recent example can be found in Heredity 82: 7-15 (1999). This study deals with incipient speciation in mosquitos adapted to the London Underground (subway system). The discontinuity caused by speciation combined with the potential rate of change is more than enough to account for macroevolution. There are other sorts of arguments that can be made that make it plausible for the process that I have studied being consistent with macroevolution. Such arguments do not prove that natural selection alone can account for all of evolution, but they can adequately disprove Gould and Eldredge's arguments. I hope to publish more on this point in the future.