32
Lars Chittka and Adriana Briscoe
8 Selection Experiments
If we predict that animals will respond to different light habitats (or differently
colored objects in their diet) by adjusting their spectral sensitivity, we must test
this prediction by means of selection experiments. Such experiments are also
necessary to separate phenotypic from genotypic variance. Given this importance,
it is surprising that such studies are missing almost entirely in sensory ecology. To
be sure, such experiments are demanding, but they are so essential to the
reasoning of sensory ecology that we emphatically wish to encourage the taking
on of such studies!
We are aware of only a single selection experiment on spectral sensitivity in
guppies (Endler et a!., subm.) and this experiment showed significant heritability
for traits of the visual system. It also showed that animals may respond in multiple
ways to the same selective pressure. For example, if animals are selected for
higher sensitivity to long wavelengths, they may respond evolutionarily by
adjusting their spectral sensitivity, by increasing their overall sensitivity, or by
changing the relative strengths of postreceptor neuronal wiring (Endler et a!.,
subm.).
9 Fitness Tests
If we cannot show that a trait confers greater fitness to its bearer, then we cannot
claim that it is adaptive (Endler 1986). We must show, for example, that common
wild-type animals indeed produce more viable offspring than do those with
deviant traits, for example with a different set of color receptors. This is not
trivial. Quite often, we will find that performance of one phenotype is better than
another at a given task, but this may not have any impact on fitness. To return to
the human red receptor polymorphism above, would we predict that someone with
a red receptor with Amax=557 nm will be able to raise more children than someone
with a red receptor with "-max=552 nm, even if it turns out that the person with
Amax=557 nm is slightly better at detecting red fruit? Probably not. Even the
colorblind on the Island of Pingelap are able to detect and identify ripe fruits
(Sacks 1997), so in conditions which are not strongly limiting, even strong
deviations from the wild-type phenotype may not be selected against. In another
example, Caine and Mundy (2000) were recently able to show that trichromatic
marmosets were better at detecting orange fruit against a dappled foliage
background than were their dichromatic conspecifics. Nevertheless, dichromats
persist in the population.
Whether this occurs because of an unrecognized advantage of dichromats over
trichromats at a task not related to frugivory remains to be tested. Without
additional data, we also have to consider the possibility that the advantage of
trichromats at detecting fruit is so small under natural conditions that it is
irrelevant to fitness. The take-home message is that we need fitness tests in
sensory ecology.
Lars Chittka and Adriana Briscoe
8 Selection Experiments
If we predict that animals will respond to different light habitats (or differently
colored objects in their diet) by adjusting their spectral sensitivity, we must test
this prediction by means of selection experiments. Such experiments are also
necessary to separate phenotypic from genotypic variance. Given this importance,
it is surprising that such studies are missing almost entirely in sensory ecology. To
be sure, such experiments are demanding, but they are so essential to the
reasoning of sensory ecology that we emphatically wish to encourage the taking
on of such studies!
We are aware of only a single selection experiment on spectral sensitivity in
guppies (Endler et a!., subm.) and this experiment showed significant heritability
for traits of the visual system. It also showed that animals may respond in multiple
ways to the same selective pressure. For example, if animals are selected for
higher sensitivity to long wavelengths, they may respond evolutionarily by
adjusting their spectral sensitivity, by increasing their overall sensitivity, or by
changing the relative strengths of postreceptor neuronal wiring (Endler et a!.,
subm.).
9 Fitness Tests
If we cannot show that a trait confers greater fitness to its bearer, then we cannot
claim that it is adaptive (Endler 1986). We must show, for example, that common
wild-type animals indeed produce more viable offspring than do those with
deviant traits, for example with a different set of color receptors. This is not
trivial. Quite often, we will find that performance of one phenotype is better than
another at a given task, but this may not have any impact on fitness. To return to
the human red receptor polymorphism above, would we predict that someone with
a red receptor with Amax=557 nm will be able to raise more children than someone
with a red receptor with "-max=552 nm, even if it turns out that the person with
Amax=557 nm is slightly better at detecting red fruit? Probably not. Even the
colorblind on the Island of Pingelap are able to detect and identify ripe fruits
(Sacks 1997), so in conditions which are not strongly limiting, even strong
deviations from the wild-type phenotype may not be selected against. In another
example, Caine and Mundy (2000) were recently able to show that trichromatic
marmosets were better at detecting orange fruit against a dappled foliage
background than were their dichromatic conspecifics. Nevertheless, dichromats
persist in the population.
Whether this occurs because of an unrecognized advantage of dichromats over
trichromats at a task not related to frugivory remains to be tested. Without
additional data, we also have to consider the possibility that the advantage of
trichromats at detecting fruit is so small under natural conditions that it is
irrelevant to fitness. The take-home message is that we need fitness tests in
sensory ecology.