Applications of the concept of phenotypic information
To conclude, several problems to which the concept of phenotypic information can be applied will be discussed.
The first concerns the life cycle, which is undoubtedly a sequence of causes and events dependent on gene activity. Genetic changes alone however, may be insufficient to explain the cycles' existence or creation. This may be seen by considering the following paradox similar to that of the 'chicken and egg'. It is that the cycle cannot be switched on unless gene products are available: but gene products cannot be produced unless the cycle is switched on. The paradox can be solved by assuming that the first synthesis of gene products occurred in response to an environmental stimulus and that this acquired phenotypic change was transmitted by the cycle and inherited by future generations. The phenotypic information may not now be identifiable experimentally because of lethal effects of cycle breakage.
A second phenomenon is the inheritance and evolution of asymmetrical phenotypic bias in animals. For most characters artificial selection is effective indicating the underlying presence of genetic variation.35 For asymmetrical bias this is not so; artificial selection is slow or ineffective and heritability low.36 Moreover there appear to be no instances where either the direction, or presence or absence of bias in an individual is determined by its own genotype.37 It is difficult in fact to imagine how a genetic change in a bilaterally symmetrical organism could induce bias since both sides must inherit the change and should react identically to it. A mechanism is needed to provide different positional information38 to the two sides. This can be achieved if bias is caused by a left-right gradient produced in the egg cytoplasm, or in the early stages of embryogenesis.37 Spatial anisotropy may also be required in the egg to provide positional information for the rostro-caudal and dorso-ventral plane but any simple mechanism of setting up gradients de novo may suffice for this. The problem with asymmetry is to consistently define a left or right bias.
A left-right egg gradient could be set up in two ways, as a result either of biased maternal gene action or influences, or molecular asymmetry. In the snail, Limnaea peragra, the former mechanism appears to operate; direction of shell coiling is influenced by maternal genotype39 perhaps relating to the asymmetrical arrangement of cells surrounding the oocyte.40 In this situation another paradox arises. This is that if the development of bias in one generation requires maternal phenotypic bias in the previous generation genetic change alone cannot lead to the origin of bias during evolution in a bilaterally symmetrical organism. Bias could however arise as a phenotypic response which initiates a cycle leading to the transmission of the bias in the form of phenotypic information to the next and future generations. This hypothesis leads to the prediction that non-lethal experimental interference with asymmetrical maternal influences during oogenesis might result in the heritable loss of bias.
A final problem again concerns maternal influences. There is evidence that the effects of an acquired phenotypic change in the mother can be transmitted further than one generation.41,42 An interesting example occurs in mammals where good maternal nutrition can cause high milk production and large offspring. These offspring may in turn produce more milk affecting progeny size in the second generation. It is assumed that such an effect will be diluted over generations; for this reason the phenomenon is not usually regarded as an example of extranuclear inheritance.1 In biometrical analysis the effect is seen as a result of maternal environment and becomes partitioned into the variance due to common environment.43 Transmission over several generations is interpreted as the inheritance of an acquired maternal environmental effect. This seems unsatisfactory since the agent of the effect is phenotypic not environmental. An alternative interpretation is that phenotypic information becomes stored in body size as a result of good nutrition and is then transmitted by a causal cycle linking body size to milk yield. This interpretation facilitates speculation that long term inheritance of acquired phenotypic effects may occur. Dilution of the effect can be interpreted as a trend in phenotypic information. Effects of this sort may also interact with cultural inheritance as when poor nutrition adversely affects the intelligence of offspring.
Acknowledgements
I thank Professor L. Van Valen for helpful critical comments on an earlier version of this manuscript.