There is natural selection but there is also sexual selection which some regard as a category of natural selection. There is also artificial selection (by humans). The question is, what is a most appropriate for selection? What criteria does a force have to meet in order to be considered to have acted selectively?
My favourite definition
Selection is a fitness differential associated to a genetic variance among individuals in a population.
You might want to have a look at What does fitness really mean?.
What I like about this definition
I personally went with this definition because it is short and clear cut and directly highlight the importance of mathematical statistics in the evolutionary biology literature.
What I don't like about this definition
This definition has the issue that it may be hard to be fully understood to a layman reader. Also, I think my definition lack a mention of the phenotype as being causal to the fitness differential. While such mention is not necessary, I think it would clarify in most minds what selection is really about.
Now, let's consider the definitions that we typically get when we google the question
Natural selection is the differential survival and reproduction of individuals due to differences in phenotype
I am quite unhappy with this definition. While I like that it highlights the fact that fitness difference is caused via phenotypic difference, it completely lack the idea that for a change in allele frequency to happen, this variation must be correlated to genetic variation.
I would note that the rest of the introduction on the wikipedia article clarify my criticisms. I only attack the short version (one sentence) version of the definition.
the process by which forms of life having traits that better enable them to adapt to specific environmental pressures, as predators, changes in climate, or competition for food or mates, will tend to survive and reproduce in greater numbers than others of their kind, thus ensuring the perpetuation of those favorable traits in succeeding generations.
There are a few things I don't like about this definition
..traits that better enable them to adapt..
It sounds like they are talking about evolvability. I would just say
traits that are better adapted.
..too much emphasis on the role of the environment..
Also, I don't like that it is presented as if all selection pressures where environmental (predators, climate, competition for food, etc...) and completely ignore a very important source of selection which are simply related to the functioning of the body. A mutation that causes an individual to fail copying DNA anymore will obviously be lethal and under very strong purifying selection. Such the "outside-of-the-body environment" is important but it is definitely not the only thing.
It is kind of a common misconception about evolution that selection occurs when there is a climate change and not so necessarily in other case in most people's mind.
..ensuring the perpetuation of those favorable traits in succeeding generations..
Well.. it does not "ensure" that these traits will be kept. First because it only increases the probability of certains allele to be kept or to reach fixation, it does not ensure it. Also, because depending on the genetic framework, once the underlying QTL have reached a given frequency, depending on the actual genetic mechanism (epistasis, dominance and others), traits that are being express might differ quite a bit from those that have been selected for.
Overall, I think the definition lack any notion to genetics.
A process in nature in which organisms possessing certain genotypic characteristics that make them better adjusted to an environment tend to survive, reproduce, increase in number or frequency, and therefore, are able to transmit and perpetuate their essential genotypic qualities to succeeding generations
I like this definition. The only issues in my opinion are very similar to my above opinions
- too much emphasis on the environment as done before
- it seems that they define more the consequence of natural selection than what it actually really is.
What Google gave me when I entered 'what is natural selection?'
the process whereby organisms better adapted to their environment tend to survive and produce more offspring.
Again there is IMO, too much emphasis on the environment and no mention of thee actual mechanism (fitness - genetics correlation).
Some people might use the term "selection" to refer to the existence of phenotype - fitness correlation and the "response to selection" as being the effect of such correlation of the population genetics whenever there is a non-zero association between phenotype and genotype. This nomenclature would typically be inspired by the terminology used in the Breeder's equation (see How to interpret the breeders equation?).
However, none of the definitions I could find clearly suggested that selection was nothing but a phenotype - fitness correlation as all of them implied that selection has a long term impact on the population phenotype which such phenotype - fitness correlation would not have in absence of a phenotype - genotype correlation.
I am not sure a recipe count as a definition but Lewontin recipe is a good way to understand the what is natural selection
- Individuals in a population varies in terms of a given trait
- This trait has some heritability.
- The fitness varies (not necessarily linearly) as the trait varies.
The recipe hides the complexities behind the term heritability. You can learn about the math of heritability at Why is a heritability coefficient not an index of how “genetic” something is? .
How selection is "best defined" is clearly subjective. However, even though it is a key process in evolution, the exact delimitation of (natural) selection is not clearcut, or rather, authors have defined it in slightly different ways in relation to the evolutionary process (see below, especially quote at bottom from Wade & Kalisz). A paper worth reading as a starting point, if you really want to dig deeper into the definitions of evolution and natural selection is Godfrey-Smith (2007) "Conditions for Evolution by Natural Selection". There, he goes through several attempts of verbally defining "evolution by natural selection" (among them two version of Lewontin's recipe that Remi.b mentions in his answer).
Many of the definitions take a similar form as this modified version of Lewontin's recipe (the 1980 version, page 491 in paper):
A sufficient mechanism for evolution by natural selection is contained in three propositions:
(LI) There is variation in morphological, physiological, or behavioral traits among members of a species (the principle of variation).
(L2) The variation is in part heritable, so that individuals resemble their relations more than they resemble unrelated individuals and, in particular, offspring resemble their parents (the principle of heredity).
(L3) Different variants leave different numbers of offspring either in immediate or remote generations (the principle of differential fitness.
The aim in this paper is however not only to define "natural selection" but "evolution by natural selection" (which is a bigger question). So from the definition above (or others in the paper), the issue is on how to draw the line between "natural selection" (as a condition) and "evolution by natural selection" (as a process). If this is at all possible. To me, it is natural to draw the line between "selection of phenotypes" and "transmission of phenotypes to the next generation" (through genetic variance/heritability), which sort of mimics the breeders equation. So to me, it is natural to see "natural selection" as the result of a fitness-phenotype correlation. In some ways, this mimics an the 1970 version of Lewontin's recipe (page 490 in paper):
Different individuals in the population have different morphologies, physiologies, and behaviors (phenotypic variation).
Different phenotypes have different rates of survival and reproduction in different environments (differential fitness).
There is a correlation between parents and offspring in the contribution of each to future generations (fitness is heritable)
Here, point 1-2 corresponds to my view of "natural selection", while the 3rd point is necessary for "evolution by natural selection". Note however that this definition has other problems if taken on its own, mentioned in the paper.
A similar view is expressed in Haldane (1957):
Natural selection is a statement of the fact that the fictitious parental population differs significantly from the population from which it was drawn. For exemple, with regard to any specific metrical character it may differ as regards the mean, variance and other moments. A difference in means is called a selection differential (Lush,1954).
Selection may be genotypic or phenotypic. Phenotypic selection may or may not lead to genotypic selection.
Here, (natural) selection is clearly seperated from the possible subsequent effects of selection (evolution).
Even more clearly put (also highlighting that there are different views on this topic), from Wade & Kalisz (1990, Evolution):
When we measure phenotypic selection in natural populations, we compare the phenotypic distribution at one time with the phenotypic distribution at some later time(s) within the same generation. It is from this comparison that we drawn our inferences regarding the operation of viability and fecundity selection. We deliberately avoid examining changes across generations because, when the process of transmission changes the phenotypic distribution, we would confound selection and transmission. We consider the clear separation of these two processes to be conceptually and practically important to the study of selection although we acknowledge that other authors prefer to include inheritance as part of the definition of natural selection (e.g. Endler, 1986).