If the sensory bias is "co-opted into a new function" then it is an example of exaptation. It is perhaps notable that exaptations may be much more common than previously thought (Barve & Wagner 2013, Nature doi: 10.1038/nature12301).

Perhaps your example for exaptation is not particularly good because original and new function and context are quite close to each other. If the original use of your sensory system is distinguishing poisonous from suitable food, preparing dishes in the course of mating rituals might be a better example for exaptation...

I am attaching a paper on sensory biases co-opted into a mating context that you may find interesting.

about exaptation and questions about this concept, it is possible to read a interesting paper in Trends in Ecology and Evolution : Exapting exaptation (Greger Larson, Philip A. Stephens, Jamshid J. Tehrani, Robert H. Layton).

I agree with Jon that the exaptation framework is not the most useful. Using "co-option" without assigning adaptive value to a trait walks a safer line. Many co-opted traits may be initially non-adaptive in a new context (as in mating and sensory exploitation), and then later shaped by selection to have adaptive value. Avoiding the use of exaptation thus allows for many scenarios to account for the eventual adaptive use of a trait in a novel context.

I have to admit I have serious problems with the concept of "exaptation", or at least that people can publish in Nature saying that "exaptation" is more common than previously thought (see Corrie Schoeman's comment). In my view EVERY adaptation deserves also the name "exaptation" when analysed on all possible levels. Because none of the adaptations was originally "designed" for a function, because there is no designer creating adaptations. This is not a new discovery, but the foundation of darwinian evolutionary theory itself. Every adaptation starts with random variation of existing structures, if structures are changing and fulfil now a new function (because there is a natural selection pressure towards another optimum), then this is an adaptation as well as an "exaptation" (because it always derives from a former structure, which often has a different function, and before that there was another structure and so forth). As you can see the word "exaptation" doesn´t give us any new insight because it basically describes the darwinian mechanism; new structures evolve from existing structures.

A last word on the example with the evolution of the feathers: People trying to distribute the (quite unnecessary) word "exaptation", including Barve and Wagner 2013 (the Nature paper), claim that the feathers are one example. They were first used for thermo-regulation and later in evolution for flight. Yes, this is a very clear example how natural selection works it fits very well in Darwins original theory, no need to say "this is not adaptation". However, people say now, the feathers originally did not evolve for flight, therefore we call it "exaptation". Interestingly, in the feather-example it all gets blurry now: Adaptation/Exaptation for what? The feathers are even in this definition still adaptations, for thermo-regulation, and would be exaptations for flight. Really would they? What is with the 150 million years that passed since Archaeopteryx, aren´t the feathers of modern birds very well adapted to flight? So as you can see the wording you are using depend only on the scale you are looking at. Modern birds clearly have feathers that are adapted to flight, feathers may have originally evolved because of a selection pressure to keep the body temperature constant, however, also feathers didn´t fall from the sky they evolved out of other structures, with different functions as well.

Bottom line, I would really hope that evolutionary scientists don´t waste to mcuh time talking about a purely semantic concept, which does not give us any new insights or explanations.

A sensory bias that is exploited for the purposes of its utility in sexual selection, could then become an exaptation. If it is exploited by another species for the purposes of enhancing that other species fitness it could become a mutualism, commensalism....which don't hurt the focal species fitness....if the interaction reduces its fitness (e.g. predation, parasitism) then that bias becomes the basis for manipulation. Finally, if that sensory bias is inadvertantly triggered by cues no intentionally sent by a 'sender' (see communication theory) then it becomes the basis for an evolutionary trap.

I find Lukas's argument rather interesting. However, in this perspective even adaptation is a semantic concept. It depends on the function(s) we interpret, observing a specific phenotypic trait. I don't find the term exaptation useless, since it simply highlights aspects of the adaptive evolution of that trait, with the aim of describing its evolutionary history. However, I completely agree (and I find this interesting), that ANY trait or trait complex is a bundle of adaptive and exaptive functions, exactly because 1) it comes from somewhere (it "abvolves", Begon would say) and 2) it explores adaptive landscapes through populations and individuals.

In this context, I find the concept of Baldwinian or facilitated evolution particularly inspiring (e.g., Kirshner & Gerhart (2005) The plausibility of life: resolving Darwin’s dilemma. Vail-Ballou Press). This theory predicts the presence in natural populations of subsets of individuals that "explore" the adaptive landscape under suboptimal fitness conditions (e.g., in different time and space). Everything is exaptation, in this view, since every trait can work in a range of conditions. In a striclty neodarwinian world, in order to adapt to novel conditions, population should "wait" for mutation or other evolutionary forces to create or bias genetic variance. However, if a stressed or "poorly functional" heritable phenotype does survive at relatively low frequencies (e.g. due to lower competition levels in marginal habitats or suboptimal conditions, or due to only slightly negative or even neutral effects), it exposes descendants to selective conditions that might eventually select for better adapted variants, provided by neodarwinian evolutionary forces.

According to this model, organisms would always explore "marginal" adaptive landscape in suboptimal conditions, and since they "come from" somewhere, their past evolutionary history (which sets the scope of their somatic adaptation) will determine which portion of this landscape can be explored. Therefore, somatic (physiological) adaptation provides the opportunity for genetic adaptation. I would add, that even physiological adaptation (more or less durable changes at organismal or lower levels) is not needed, since as we said, each somatic trait is inherently working in a range of conditions, with variable efficiency.

Lukas Landler has said it all! Every adaptation is an exaptation. No need for the word.

John Alcock

I have to agree with all of those arguing 'exaptation' is not a particularly useful term. My first reaction upon reading Gould and Vrba (1982. Exaptation a missing term in the science of form. Paleobiology 8:4-15) was the same as Alcock's, every adaptation is initially an exaptation. That said, I do think people have described at least two distinct origins of sensory bias. First is the co-option of a pre-existing bias for something else such as you describe for foods with a certain smell (the example I now best is red-coloration for highly desireable fruits and red coloration in male guppies). The other is a bias that exists with no obvious precursor. The preference for males with swordtails in species of mollies that do not have swordtails is usually presented in this way. This would not be considered co-option since the bias is not taken from another function. I'm not sure if any such examples exist, however. Fish in the family Poeciliadae (which includes mollies) often have a streak of color along the ventral edge of the tail fin. This looks very much like a sword (an elongation of the ventral most spine beyond the normal margin of the tail fin) and is probably the origin of the preference. I suspect the sword itself is merely a cheap way to exaggerate this character and one that appears to have very little cost to the males (see Oufiero et al. 2012. Physiological and Biochemical Zoology 85:684). Are there any good examples of sensory bias where the bias is not for something that already exists?

I'm always curious why these debates are quickly dismissed as "issues of semantics" as if this is a disadvantage. The essence of Gould and Vrba's paper was (1) to explore the meaning of terms related to the historical genesis of traits and (2) introduce terminology that would clarify such meaning. Thus, yes, this debate is entirely an issue of semantics, and an important one at that. That said, I have a few disagreements with contributors thus far.

One element that inevitably is undervalued in these discussions--yet is absolutely critical to them--is the term "aptation." This speaks to Lukas' statement that "Every adaptation starts with random variation of existing structures...". It's again important to remember that Gould and Vrba chose their language very carefully. If Gould and Vrba had omitted the term "aptation" (or if we continue to ignore it), then this statement would be perfect. Rather, to be conservative about what we view as an adaptation (i.e. before studying a trait's selective impact), this statement should begin with "Every aptation starts with random variation..." It is absolutely imperative that we begin with no preconceived notions--or, at the very least, reasonable a priori assumptions--about how traits, or suites of interacting traits, impact survival and fecundity. If we simply assume that all traits are adaptations, we do two things: (1) we come dangerously close to falling into the adaptationist abyss, and (2) we become automatically canalized into thinking that all adaptations are exaptations.

This brings me to contributors (including Prof. Alcock) who dismiss the utility of the term "exaptation" simply because of the statement "all adaptations are exaptations." While this is unarguable, the converse is not. All exaptations are not adaptations. Certainly, as originally defined by Gould and Vrba, they are "fit for their current role." But these traits did not arise as a consequence of the same selective agents that they currently experience. Unless we are prepared to ignore the type of selection outright, both "adaptation" and "exaptation" are required terms, whether they have both been adopted by practitioners or not. In other terms, all ducks are birds, but not all birds are ducks. Both terms enjoy unparalleled utility.

Towards my first point, Gould and Vrba's paper was about understanding historical genesis of traits. Though this is a weak example that does no justice to either Gould's writings or his lectures, I'll still use it to reiterate a point made by others (and partly to get the thoughts straight in my own head). As an example (inspired by Losos' work), consider a population of lizards in which predator-driven natural selection has selected for long limb length. That population--or a subset thereof--is swept away to another island where the long-limb length confers a selective advantage for accessing a novel food supply. Long limb length is an adaptation in the original population (i.e. being positively selected by a predator) but an exaptation (or co-optation) in the founder population. We use these terms to describe historical genesis of the traits at a given point in time; this was the goal of Gould and Vrba's paper. If one wants to conservatively and completely describe the genesis and consequences of a traits over time (as evolutionary biologists do as part of their job), the terms aptation, adaptation, and exaptation are all absolutely necessary.

I'll end with one more statement. One reason for such staunch opposition to the term "exaptation" is most likely the way in which eco-evolutionary feedbacks have been pigeon-holed into the ultimate-proximate dichotomy. The useful distinction between adaptation and exaptation contributes directly to our understanding of the genesis of traits, which, in turn, adds to our understanding of those traits' control and consequences (within an individual's lifetime). We need an integrative framework that allows us to quantify genesis, control, and consequences of traits rather than debating whether the trait should be considered at the ultimate or proximate scale. (Can of worms hereby opened...)

Let us try and distinguish the two scenarios below in the context of a population of bacteria experiencing an antibiotic in the environment -

1. The bacteria, through lateral gene transfer, acquired genetic material that allowed them to survive and break the antibiotic down.

2. The antibiotic was structurally similar to another antibiotic for which the bacteria already have resistance machinery. The bacteria exploited the structural similarity to break down the new antibiotic.

Would example 1 qualify as an adaptation and example 2 as an exaptation?

Jon,

I like your answer. It is true that a mutation would have sufficed in case of Example 1. I just wanted it to be a big change! :)

I like your analysis of Example 2 as well as the argument of what would constitute a nontrivial example of exaptation. And I agree with your last paragraph that in some cases semantics are very important and differentiate how someone thinks about the issue. However, I still don't think that bacteria using the mechanism to break antibiotic A to break down antibiotic B is entirely trivial. Though it does sound trivial, I don't think it is. Because

1. The co-option of the ability to degrade antibiotic A allowed them to survive when facing a new antibiotic B, a non-existent requirement.

2. In the long-term and if exposed regularly, the bacteria would fine tune the ability to break down antibiotic B. This may later on allow them breakdown antibiotic C, which may be kind of similar to B, but no so similar to A anymore. (We routinely re-purpose antibiotics with minor tweaks).

I understand that point 2 here sounds like a bit of handwaving without substance. But I think it is a plausible scenario especially because a lot of changes are gradual as described above. This gradual reworking can lead to large long-term effects and that is why I don't think it as a trivial example of exaptation.

Evolution is a historical science.

We can only observe the effects of evolutionary forces, after they act on allelic frequencies, and phenotypes.

In this view, what about this definition?

Adaptation = a clade's inheritable phenotypic trait with a 'long' evolutionary history of selection by a measured (observed) selective factor, relative to the evolutionary history of its mrca

Exaptation = a clade's inheritable phenotypic trait with a 'short' evolutionary history of selection by a measured (observed) selective factor, relative to the evolutionary history of its mrca

The advantage of such a simple definition is that in order to operate this distinction, we don't need to know what is the exact mechanism that alters the relative reproductive fitness of the variants, through the effects of phenotype-environment interactions.

In the previous example, we don't need to discern whether the biochemical mechanism that protects the bacterium from antibiotic B is either exactly the same mechanism that protects it from antibiotic A ... or it is something completely different, that evolved as a 'by-product' of the evolution of resistance to A.

It is sufficient to say: 'the bacterial population never encountered the antibiotic B during its previous evolutionary history'.

Note that the focus is on the monophyletic nature of the evolutionary unit. A population, a species or a group of species have a measurable evolutionary history. A trait which is 'young' relative to the age of the mrca of the clade would be an exaptation.

This would equally apply both to biological (i.e. reproductive isolates) and evolutionary species (i.e. clones).