Fitness. Of the many concepts of evolution, this is perhaps one of the more widely misunderstood. It comes from the unfortunate slogan written by Herbert Spencer and urged on Darwin by Wallace and others: survival of the fittest. People think it means the strongest, or the most aggressive, and that it means evolutionary theory is a tautology. We'll look at these in a bit. but first, what does it really mean in evolutionary biology?
Fitness is a property of a competing variant in a population. It means that X, whatever it might be biologically, is increasing in its frequency in a population faster than its competing variants. X can be a gene, or a trait, or even an entire organism's form and functionality.
What fitness isn't, is something absolute. There's no universal measure of fitness that applies to all organisms. Every different kind of biological variant is only fit or not compared to the other variants in its population, whether that's a local population, or a gene pool, or even a smaller group like a kin group (an extended family). Fitness is relative, literally and figuratively.
Fitness has little directly to do with violent behaviour, or strength. A behaviour can reduce fitness if the damage done in a fight or the energetic cost of being strong makes the organism less likely to survive in hard or combative times. Whether or not violence is fit depends on the nature of the organism, its populational neighbors, the times and climes, and so on. Criticisms such as that of philosopher David Stove
Usually, fitness is ascribed by biologists, particularly geneticists, to forms of genes, called alleles (from the Greek, of course, for "others"). Although the term was coined in the context of genes, by R. A. Fisher in the 1930 The Genetical Theory of Natural Selection, so far as I can tell, it doesn't have to apply just to genes, and Fisher himself sometimes spoke as if it referred to a variety of organism or the progeny of organisms not genes.
Fisher made what strikes me as an interesting point about fitness:
It will be noticed that the fundamental theorem [of natural selection] ... bears some remarkable resemblances to the second law of thermodynamics. Both are properties of populations, or aggregates, true irrespective of the units which compose them; both are statistical laws; each requires the constant increase of a measurable quantity, in the one case the entropy of a physical system and in the other the fitness, measured by m, of a biological population. [Genetical Theory, 1958 edition, p39]
The variable m he mentions is the "Malthusian parameter", or the coefficient of selection in a given environment. And note: m must increase! Populations must get more fit (he allows they may reach a limit due to a lack of variance for selection to act upon, and stall). But what is fitness? Is it a physical property? No, he says, for "although measured by a uniform method, [fitness] is qualitatively different for every different organism" [loc. cit.]. In other words, a virus and a plant it infects can have equal fitness, but there is clearly little shared physically by them that contributes to their respective fitness.
So, we know what fitness isn't. It isn't absolute, and it isn't a physical property. We'll get back to this in a bit, but we want to know what fitness is. I can hear you grinding your teeth - tell us what it is. OK, simply put,
Fitness is the rate of increase of X in a population P.
Well, almost. Things can increase in their frequency for reasons other than because they have high fitness. So we have to add a caveat: in a population of sufficient size that small scale sampling effects don't cause drift (a concept that I bet Larry Moran can define better than I can), and in which the same ecological pressures obtain on all members. Neither of these are realistic assumptions. All real populations, if they are not artificially restricted to a lab bench, are of finite size, have patchy environments (consider this: there have to be some members of a population at the fringes, and that causes different selection pressures to apply)., and have variable population densities across the range of the population. Some live in suburbs, some live in urban townhouses, and some live in rural shacks, as it were.
So fitness is something of an abstraction. It's what would have occurred if none of these other factors intervened. Fitness is in my view an abstract property of the models of population genetics. It basically means the "reproductive value" (Fisher's preferred term, and metaphor being investment practices. Evolution is capitalism in his mind!) in progeny over many generations. Any organism that has many surviving descendants after at least three generations is fitter than one that doesn't.
The same thing applies whether you are talking about genes, traits or organisms - if the copy of the gene in one generation has many descendant copies in a later generation, and a competing allele doesn't, then it is fitter than that competitor.
There are two ways to interpret fitness: one is as the fitness of an individual. This need not be an individual organism; it might be an individual gene or an individual trait. Imagine that an organism has a trait like longer legs (suppose it's an impala). That individual trait will allow its bearer to escape predation, and this will lead to it being able survive long enough to have more progeny than a shorter legged individual, and so we could say that the long legged individual has a higher fitness than the shorter legged individual.
But individuals are also subjected to accidents. The long legged individual might be in an area subject to fire, and die for reasons that have nothing to do with the trait. So typically fitness is not ascribed to individuals, but to populations. The fitness of a trait is the fitness of a type in a population of other types, in a given environment.
Now I would like to add some personal comments, of a philosophical nature. They are not necessarily the received view of biologists or even philosophers. If you aren't interested, stop reading now and go for a walk.
Fitness has some philosophical accounts - one is the propensity definition, in which the fitness of a gene/organism is its propensity to survive and reproduce. Another is the frequency definition (basically the frequentist statistics of gene/organism variants).
The frequentist account leads to tautology - a fitter allele/organism variant is just that which has more progeny after some number of generations. It's true by definition, and hence doesn't explain why [but remains true for all that].
But the propensity account is at best merely a sketch, since the principle of natural selection (NS) fails to give any account of the propensity itself. In my view, NS is not an explanation of changes in the frequencies of variants in a population; it's an explanation sketch. In short, it's like a syllogistic form - you must plug in true values to get an actual explanation. NS is not a mechanism, it's a form of dynamic adduced to explain particular cases.
Sober noted that fitness was a "supervenient property" in his 1984 book The Nature of Selection - the same fitness can be realised by many different physical systems - a virus can have the same fitness as a flowering plant. But if any two physical systems - say members of the same flower species - are physically identical (in the relevant respects, and there's the rub) they should have the same fitness in the same environment. But what are these "relevant respects"? The answer that satisfies me is that they are those physical and biological properties of the organisms that are identical with respect to a fitness account, that is, which satisfy the criteria for inclusion in a population genetics model. Hence, in my view, fitness is an abstraction, not a property of realworld systems. Particular fitnesses, say of the impala's legs, are real enough. But fitness is not in itself a real thing, but a feature of explanations.
This explains why some people think that selection is a tautology. The property of being fitter than something else is purely abstract, something that "exists" only in a model. The property of being a fitter impala in a particular herd in a given environment, though, is a real property. It isn't something that exists only in the heads of evolutionary biologists, but is a complex set of biological features that means the case fits neatly into the standard selectionist model, and given that it does, we can make some very non-tautological inferences about what will happen, and why. Plugging in real facts means our model ceases to be merely formal and fitness ceases to be a variable in that model. The particular case has real propensities (the impala's longer legs make it outrun predators). Fitness is not a propensity. That fitness is.
Enough philosophy for now.
See also Richard Dawkins "An Agony in Five Fits" in his Extended Phenotype: The Gene As the Unit of Selection.
- Log in to post comments
Someone once said that it should more properly be called "survival of the fitters", that is, those who find a fit in a particular niche. Sorta takes the 'red in tooth and claw' flavor out of it.
First, good article. I think you're right that fitness is an often misunderstood concept and that providing a clear description of it is very helpful.
Second, I just wanted to share a thought that came to me while reading the article. The way in which you were describing fitness made me think of Platonic solids. Fitness as a thing is an ideal and what we see and talk about are the messy, less-than-perfect realizations of the ideal.
I think people often get hung up on the ideal and what they think that implies for humans (and other organisms) that they forget that the real thing is quite so clean cut.
Just my $0.02
"...that they forget that the real thing is quite so clean cut."
Or rather not quite so clean cut being what I meant to say.
Presumably, you have to measure fitness over some reasonable number of generations. Since pumping out a huge number of progeny isn't always a good strategy, I assume we don't want to say that a mutation that favors short-term fecundity is automtically fitter than a mutation that makes it more likely that the next generation will survive and reproduce without increasing the number of offspring. It depends. On the other hand, since in the long run we're all dead, fitness loses its meaning if you look at things on too long a scale--even for species and traits that don't go extinct, fitness must certainly fluctuate along with the ever-changing environment.
Do you have any suggestions about how to handle the time-scale issue, assuming that you agree there is one and I'm not missing something obvious?
John, thanks for taking a stab at fitness. I take issue with a few of you points here
Jim: measuring fitness usually occurs over, as I said, at least three generations, although mostly the measures assume that many generations prior to that have contributed to the current fitness value, with the environment held constant, ceteris paribus, caveat emptor, etc.
The reason why, contra Matt's response, that I think that fitness is a rate of frequency change, is that it provides a momentum view of fitness - it's the rate right now at which the frequency of that X is changing, irrespective of what happened in the past or will happen in the future. This is how I get over the generation issue.
John -
Your presentation seems to suggest that the notion of 'fitness' can only be applied when the relative frequencies of the traits in question are undergoing change. I think that is incorrect, since equally fit variants produce stasis in relative frequencies.
This issue of stasis in frequencies is one of the main reasons why "rate" based definitions don't seem to work. Is phenotype X without fitness if its rate of transmission is equal to other phenotypes within the population?
I think we all agree that rate is important- but there is more to fitness that rate alone...
I agree that "fitness" is often misunderstood. It is deceptively simple, but difficult to grasp. Further, there is some disagreement on precisely how to define it. As if that was not enough, there are a number of complexities that may be introduced beyond the basic concept. I certainly laud you for taking up the task of clarifying this term for the non-expert.
You did a good job of dispelling the very common misconception that higher fitness necessarily means physically stronger or more aggressive, and this is probably the most important point to make in this context. You also note that fitness is only "usually" defined in terms of genetic differences among individuals (I do not favour that definition, but it is not incorrect).
However, fitness can be absolute, it can be a property of individuals, and there are universal measures of fitness that apply to all organisms (even without universal agreement on which one to use).
That fitness can be absolute is implied by any calculation of relative fitness. Relative fitness is the average fitness of individuals with particular alleles (or traits) compared to the average fitness of other individuals in that population. The average fitness of individuals is the sum of individual fitnesses divided by the number of individuals.
A simple definition of fitness is 'the number of viable offspring produced per unit time.' This is absolute, a property of individuals, and applies to all organisms. Is fitness really that simple? No, but trying to address the complexities will not help the average reader to understand it any better. Fundamentally, fitness is all about the reproductive success of individuals. From there, we can address the average fitnesses for different groups of individuals, and the relative fitnesses of such groups. We could then move on to fitness over multiple generations, inclusive fitness, components of fitness through life histories, and other issues.
There is an excellent discussion of these issues in Douglas Futuyma's Evolutionary Biology, Third Edition (1993), p. 349.
I didn't raise the issue earlier, but it seems excessively verificationist (or maybe operationist) to identify fitness with rate of change in relative frequency (or any other measure of reproductive success). Plunging headlong into metaphysical waters, I would argue that the relative fitness of variants is what is appealed to _explain_ rates of change (or other such measures).
Fitness is the rate of increase of X in a population P.
Well, almost. Things can increase in their frequency for reasons other than because they have high fitness. So we have to add a caveat: in a population of sufficient size that small scale sampling effects don't cause drift (a concept that I bet Larry Moran can define better than I can), and in which the same ecological pressures obtain on all members.
Very nice discussion. I will be posting on the difference between RS and fitness a bit later on (in a week or two?)
One way to manage this issue to which you refer here is to count as fitness the effect attributable to a property of the thing (allele, etc.) being measured. Stochastic variation is not attributed to such a property.
It is a good article, and I think I agree with the main conclusions. But I have some difficulties with the particulars.
I agree that the model perspective is fruitful - it is the one I prefer. But I don't see how a model definition can be tautological. True, "Hooke's law describes linear-elastic materials, and those materials that obeys Hooke's law are linear-elastic". Similarly, "Evolutionary models describes fitnesses of properties X in a population, and the rates of increase of properties X described by evolutionary models are fitnesses". Descriptions of models may seem tautological at a glance.
But underlying the model is the real world, and the observed material expansion or X increase is lifted up into the model. No circularity here what I can see, in an account looking at both model and its application.
Perhaps that is what is alluded to here: "Plugging in real facts means our model ceases to be merely formal and fitness ceases to be a variable in that model." But IMO the model is still formal until tested, and the model rate is still a variable even though it was instantiated under test or application.
And it is likely that part of the post is confusing me. "The property of being a fitter impala in a particular herd in a given environment, though, is a real property." But it was, rightly I think, considered a property of properties over populations earlier. Isn't here the reason behind an individuals fitness conflated with the population measure?
Btw, Fisher's analogy strikes me too. Entropy is another elusive property that is captured with several measures and models. The most fundamental definition, as the number of the possible microscopic configurations of the system, could be analogous to the different individual causes which makes the rate of increase of property X.
Also, some synchronized populations generation discreteness is then analogous to classical Carnot cycle discreteness. No one complains about total thermodynamical entropy change being observed after cycling through the later.
"No one complains about total thermodynamical entropy change being observed after cycling through the later."
Actually, better than that, since IIRC Otto engines are now known to be chaotic regards combustion and resulting pressure. So here one must observe a couple of cycles to get the averages.
Oh, now I get the "formal model" part - keeping the formal expression, of course. Right, and the model with substituted values are an individual model. But I still think the tautology part is confused considering that models are applied when used. (Ironic, that looks like another tautology. :-)
John - I think equating fitness with the rate of change is almost right. I would equate it with the expected rate of change (in the mathematical sense). This gets around drift, and also matches the mathematical descriptions.
I would also argue that fitness has to be defined over a generation. It's dependent on the environment, which will change over time, so a long-term fitness would be the aggregated effect of lots of fitnesses. Calculating it would involve writing down too many integral signs...
Finally (for now) I think the way out of the tautology is to view fitness as an intermediate step in the full description of a system. Fitness is the result of a lot of processes, which impact reproduction and survival. It's these processes that make up fitness, and are what evolve. Fitness is just the currency used to measure their effects. If you look at Fisher's Genetical Theory of NS, you'll see that he gets this, defining fitness in this way. But somewhere since then the link has been lost from the way it is taught.
Bob
Okay, I have been cogitating on all your comments (see, I do listen!*) for a couple of days now about rates. I think I will stick to it so long as the following points are made clear:
The rate is the rate at a given moment. "Expected" rate of change in the future is really just extrapolation from the present. It is the present rate that defines the present fitness. Consider this: a case where two variants reach equilibrium in a population (say, because they are density dependent in their fitness). The rate of change when the novel variant is introduced is very much higher than when they reach equilibrium, and so, I would say, its fitness is higher. But once it reaches fixation, its fitness relative to the variant it displaced is effectively 0 (although it may be fitter relative to later variants).
Bob, I agree with you (and you said it so much more clearly than I): the tautology lies in the fact that we haven't yet finished doing the explaining. Fisher's use of m indicates that it is a variable not a property, but one used in "qualitatively different" cases, each of which have their own biological explanation.
Ian, this gives an answer to the notion of (mathematical) absolute fitness. That in a particular time interval there are a particular number of copies of X is true. That X has some fitness in all environments and conditions is false (which is what I meant in that case). The rate of change is different when averaged over one, or many individuals and generations. It's a problem of differential calculus that those of you who can do this stuff will have no trouble dealing with.
* I listen, but I don't always get it.
Hi John, thanks for the response.
I do not understand what you mean by a "mathematical" absolute fitness. Absolute fitness is expressed as a number, of course, but it is no more mathematical than running speed or body length. An individual with any given allele (or trait) may have different absolute fitnesses in different environments, just as they may have different running speeds and different body lengths. In fact, the individual might have a different absolute fitness in the same environment just by chance (this is getting into genetic drift), depending on how one defines it. Here are a few ways of thinking about "fitness":
We can ascribe an absolute fitness to an individual, in which case it is just the reproductive rate of that individual. This is not usually of interest except as part of the calculations of other types of fitness. The individual might be �lucky,' or they may have adaptive traits, but this measure of fitness does not tell us which.
We can ascribe an average absolute fitness to a population, in which case it is the average reproductive rate of the individuals in that population. This is important in itself in population ecology, and is part of the calculations of other types of fitness.
We can ascribe an absolute fitness to a trait in a given population, in which case it is the average reproductive rate of the individuals with that trait in that population. We are moving away from seeing fitness as an individual trait, but it still is findamentally that.
We can ascribe a relative fitness to a trait in a given population, in which case it is the absolute fitness of the trait divided by the average absolute fitness of the population. This is the fitness that relates directly to selection.
We could also ascribe a relative fitness to an individual, in which case it is the absolute fitness of that individual divided by the average absolute fitness of the population. This is not usually calculated.
Returning to the relative fitness of a trait, the selection that this imples is what some call phenotypic selection. Specifically, this is selection that does not necessarily result in any evolution since variance in the trait in question does not necessarily have a genetic basis. We can define selection (and fitness) in another way, leading to:
We can ascribe an absolute fitness to an allele in a given population, in which case it is the average reproductive rate of the allele in that population.
and
We can ascribe a relative fitness to an allele in a given population, in which case it is the absolute fitness of the allele divided by the average absolute fitness of all alleles in the population.
The latter seems to be what you are getting at. Note that the relative fitness of an allele (or trait) is undefined if there are no other alleles (or traits) in the population, it is zero only if the allele (or trait) is completely lost from the population, and it is one if the allele (or trait) has the same average absolute fitness as the rest of the population.
These do not explicitly consider environment, as this is subsumed under "population." That is, an actual population is living in a particular environment. In theoretical terms, we would add "in a particular environment" to each of these measures of fitness.
Perhaps it would be useful to quote a paragraph from the Futuyma that I refered to earlier:
"For our purposes, we will define natural selection as any consistent difference in fitness (i.e., survival and reproduction) among phenotypically different biological entities. The entities may be individual genes (which must have some phenotypically variable property if they differ consistently in fitness), groups of genes, individual organisms, populations, or taxa such as species. (Although we have adopted a phenotypic definition, we wil almost always discuss the fitness of phenotypes that are inherited to at least some degree, because selection has no evolutionary effect unless there is inheritance.)"
Ian
Okay. How about "fitness landscape" now?
By the way, may I have your permission to quote (with citation) your extended definition in a paper that prof. Philip Fellman and I are drafting right now, on measuring entropy of evolution by natural selection, where the fitness function is very important?
... concepts are not physical properties ... Voltage is not a physical property, nor a propensity. But _this voltage_ is. Mass is not a physical property ... after all, two things that have nothing in common except their mass have only that in common ...
The 'is it a rate' issue: just make it rate + 1.
I think you're right about fitness being relative (one doesn't have to outrun the grizzly bear, only the slowest member of one's group) but I think defining it as a rate makes it unnecessarily complicated. You could measure fitness by looking at rates, but fitness itself seems to be more of a qualitative state, proportional to an individual's liklihood of survival in a given environment. I think Spencer had it right.
The trouble with likelihood is that it is either Bayesian, in which case it is an epistemic property, or it is frequency based, in which case it is post hoc and not predictive. The propensity account is designed to deal with this problem by removing likelihood, but it doesn't give anything more than a tautology.
The reason I went with a rate account is that rates are crucial for fitness measures (there are well known equations that deal with discrete and continuous traits), and the fitness at any time in a population in the environment just is the instantaneous rate of increase of heritable traits in a population, and the overall or average fitness of the population is the sum of those fitness values in a static population size, or the sum in conjunction with the rate of population increase (or decrease). So rates give you the fitness (ceteris paribus stochastic effects).
Spencer's version, however, did imply strength and violence. All we owe to him is the word.
Rick Michod's book "Darwinian Dynamics" goes through all the major definitions of fitness, examining their properties and shortcomings (his book even has an appendix giving 28 different proposed definitions of fitness). It is the best treatment of fitness from a mathematical and philosophical perspective in a single volume (at least in my view). Given the back-n-forths on this post, some folks may find this source useful.
Both Andreas Wagner and Richard Lewontin make arguments as to why fitness is not a rigorous scientific concept.
I thought Lewontin had some powerful insights in a 2003 Santa Fe Institute Paper:
Santa Fe 2003, Lewontin on 4 Complication in Evolution
Lewontin seems to argue "fitness" is only something superflous to statistical frequencies in the first place. If one wishes to describe biology in terms of function rather than selective value (as Wagner does somewhat) this is a stealth design approach to describing biology, as function transcends concepts of reproductive fitness.
While I appreciate Lewontin's definition (and agree mostly with it, see my comments about tautology), this is not the post in which to debate design, Salvador. If you try to run it here, I'll block it.
That said, there is an interesting ambiguity in the pronouncements about design in biology by the biologists (as opposed to the IDevotees). I believe talk of design in biology is a holdover from the natural theology tradition in British natural history started by Harvey and Ray, and to which Darwin was heir (being able to almost recite the Evidences by memory, as he himself noted).
The problem with fitness is that it is an abstract property due to the model applied to various cases. It only becomes a concrete property in each particular case, in which eventuality we can dispense with fitness per se and talk about the physical reasons why a particular type might increase in frequency, such as the ability to mimic a poisonous species, or to detoxify a food source, or to run faster.
Functional talk is an artifact of the interests of the researcher. Functions, in my idiosyncratic view, are not properties or facts about the non-cognitive world. Identifying something as a "function" is, essentially, saying "that's interesting [to me]!"
Once we take this step, we no longer feel the need to project our own teleological dispositions onto the world.
You (John Wilkins) defined fitness in your first post (22 January 2007):
"Fitness is the rate of increase of X in a population P."
You also pointed out that genetic evolution also occurs via sampling error ("genetic drift"). You and others discussed the "physiological fitness" of individuals and "expected evolutionary fitness" in addition to "realized evolutionary fitness" of individuals (the actual reproductive success of individuals) in real populations.
Charles Darwin did a nice job describing the "struggle for existence" (ecological) causes of selection and the hereditary (genetic) consequences of selection in the first edition of the Origin of Species. Darwin drew attention to the fact that he used "the term Struggle for Existence in a large and metaphorical sense, including dependence of one being on another, and including (which is more important) not only the life of the individual, but success in leaving progeny." (1859, p. 62). Darwin then went on to summarize the ecological causes of selection by stating that "as more individuals are produced than can possibly survive, there must in every case be a struggle for existence, either one individual with another of the same species, or with the indivdiuals of distinct speces, or with the physical conditions of life. (1859, p 63).
Scientists all too often have only measured differences in survival when studying the effect of selection on evolution. They have rarely taken differences in fertility and generation length into account. Fisher understood this when he proposed using m (the intrinsic rate of increase) to measure evolutionary change. The life history approach to measuring selection involves using age specific survival and age specific fertility to calculate m.
Adaptive evolution by selection is a biological process with ecological causes and genetic consequences. Accurately measuring the genetic outcome of selection is important. That is why you (John Wilkins)pinted out that measuring fitness should involve more than just one generation (21 January post). This is important because both parental investment and selective mating affects the survival and reproductive success of those offspring that do survive.
Most of the confusion over evolutionary fitness seems to occur when individuals want to use "physiological fitness" or "expected evolutionary fitness" or just differences in survival rather than a more comprehensive and thus more scientifically accurate measure of realized evolutionary fitness of X in some population P.
P.S--Creationists talk about designs and evolutionary biologists talk about adaptations for survival and reproductive success.
My philosophy(of Science) teacher Otto Lappi said it well: "Not fit like bodybuilder, but fit like a glove"
I thought the paper by Lewontin was fairly significant, and I think it has not been sufficiently discussed. I have little intention of debating design here, but thought Lewontin's research into the matter was relevant and compelling. No need to worry about blocking my posts, I don't intend to visit too often if at all. However, thank you for commenting on Lewontin's article. I've been trying to solicit reaction to his ideas.
Lewontin is always worth reading on these matters, and here I fully agree with him.
The article, for others, is available here.
I think fitness is a mathematical variable in a series of equations. Explaining why that variable is instantiated in terms of this or that property of actual organisms depends on whether the equations fairly model the situation under explanation. The variable is not a property of the world, but of the models.
So if the models of fitness fail to properly describe the case, the solution is to add equations or remove them until it does fit. Some assumptions might include there being an n-ary relation between organisms and the abiotic resources that doesn't nicely map onto a two locus-two allele model. Or we might introduce stochastic effects such as drift or nearly neutral Markovian processes. Once we have a model or set of models that fairly describes the observed dynamics, and makes fairly accurate predictions of future behaviour (and retrodictions of past behaviour), we can say we have a representation of the phenomenon, and use that to guide research.
This is part of the usual explanatory dynamic of science.
Looney:
Um, unless one subsumes it in the model as a constraint of course, or conversely incorporate mechanics into a mixed model with EM. More caveats. :-)
And so, one last (?) note into the fray:
There are many quantities that can be used to measure fitness, btw.
"A simple definition of fitness is 'the number of viable offspring produced per unit time.' This is absolute, a property of individuals, and applies to all organisms. [...]
Fundamentally, fitness is all about the reproductive success of individuals. From there, we can address the average fitnesses for different groups of individuals, and the relative fitnesses of such groups. We could then move on to fitness over multiple generations, inclusive fitness, components of fitness through life histories, and other issues.
There is an excellent discussion of these issues in Douglas Futuyma's Evolutionary Biology, Third Edition (1993), p. 349."
( http://scienceblogs.com/evolvingthoughts/2007/01/fitness.php#comment-32… )
Frak! Please disregard previous comment, posted in the wrong browser tab.
wocderful topic
i have a question. what is one to do when it is not obvious what to count as a generation? im thinking of examples such as aspen, where it is not obvious to distinguish between reproduction and growth. i wonder if fitness concepts tend to assume an overly simply view of what an individual organism is. applying fitness to alleles instead does not really help, because alleles are only counted once per organism. what do you think?