Silson's computers, running programmes for weeks at a time, testing breeding programmes over many generations, have confirmed his findings. His work should also return to prominence the work of three professionals from the early 1900s, all with a world wide reputation at the time, who wrote books supporting inbreeding - Hagedoorn (1939 etc), Lush (1937) and Davenport (1907); and there are also encouraging reports from the same era of scattered experiments by breeders who followed Hagedoorn's ideas on their own farms. So there is practical evidence from the past backing Silson's model, and no evidence as yet that conflicts with it - because of conventional attitudes, few researchers have experimented with intermediate rates of inbreeding.
Silson, in the piece below, outlines his ideas and some of their implications. His additive gene model lends itself to disturbing social speculations: could it be, for instance, that the church through the ages has been ill-advised, from a genetics perspective, to discourage marriage between cousins? And were we genetically better off in the past when we lived within close-knit, partially isolated communities?
Textbooks demonstrate these by using the simple clearcut dominant and recessive examples that are seen with colour and pattern in plants and animals.
Since about 1910 it has been recognised that other qualities result from the total effect of many genes individually too small to be measured (see Note A below).
Many theories, often controversial, have been developed to explain the complex mixture of additive and non-additive effects seen in multiple gene systems.
Variable and/or non-linear activity by individual genes has been suggested and great emphasis placed on the concept that environment seriously affects gene activity.
About forty years ago, the author started working with large additive gene systems. These were modelled on paper and later by computer (Notes A & B).
Several decades of research have confirmed and extended the original discovery that wholly additive genes are able to produce most, if not all, of the non-additive anomalies found in real species. All of these anomalies have been shown to be due to simple mathematical interactions with no need for environmental effects or variable gene action.
Apparently many experts cannot believe that the solution is so simple. Nonetheless, no evidence has yet been seen that disproves the simple explanation.
An important anomaly is reversion to the mean which produces the often confusing effect known as heritability (Note C). For the same quality, heritability varies between populations as well as changing with time in the same population.
Natural populations show the anomalies of depression following inbreeding and heterosis (hybrid vigour) following crosses.
Inbreeding is conventionally treated as something to be avoided. The model, and evidence from real species, indicate that cousin matings are approximately optimum. Reduced inbreeding produces selection plateau (Note D).
Heritability has a high positive correlation with inbreeding.
The principles developed are also important for the study of evolution. The mathematics of large additive gene systems is such that there is an almost infinite number of possible gene combinations for any additive total.
In large populations, these combinations store immense amounts of cryptic genetic information. The population mean may be moved rapidly, in almost any direction, far beyond the most extreme individuals ever previously bred.
This result has major implications in understanding the fossil record with its sudden appearances of new species and the scarcity of intermediate fossils.
A single additive system may intersperse millions of generations with little obvious variation, with few or many short periods of rapid or minor change. It may produce numerous new species or vary its direction as a single line.
These ideas are also important in understanding social forces. The model shows that qualities are most easily optimised in small partially isolated populations. Such populations are common in many species including humans.
Extreme human mobility is very recent. Before 1900 the author's own name occurred in only two small areas.
In such circumstances gene flow would be low with most matings being between individuals in the same or closely adjacent populations.
Crosses sometimes show heterosis but any apparent superiority is deceptive. In the second generation genes from the two sources recombine in complex ways. Typically the progeny mean crashes: some progeny may show defects unknown in either parent population. Fewer will survive to reproduce.
If avoidance of crossing is due to genetic factors such genes will become more common because more progeny are surviving from the more closely related matings.
Initial restrictions may be as simple as different courtship signals or breeding periods.
Conversely considerable evidence is being found that individuals, even of the simpler species such as amoeba, have an inherent ability to evaluate the genetic similarity of other individuals; whether reared together or not.
In humans, actual experiments are difficult but much social evidence implies the existence of similar mechanisms. Murders, within families, rarely involve blood relations: cruelty is notably more common from step-parents.
At the other extreme studies have shown that married couples tend to be more similar, for a wide range of hidden as well as obvious characteristics, than random pairs of individuals from the same population. In those couples who divorce the similarities tend to be less than in those who don't.
Its reproductive aspect is clearly seen in communities where one or two foreign children have been adopted into local families. Generally accepted as infants they are suddenly rejected when they reach puberty.
It appears that society might more easily be improved if 'good' and 'bad' behaviour, within and between groups, were reconsidered in terms of inherent and unconscious control mechanisms and efforts made to minimise cultural exaggerations.
- Note A. The book discusses additive genes only and assumes additive values of '1' or '0' for each gene. A pair of genes could be '00', '01' or '11' giving additive totals of 0, 1 and 2.
With several pairs of genes the possible range of these additive totals, for 'n' pairs, may vary from 0 to 2n. It is theoretically important that all intermediate totals may be produced in many different ways.
- Note B. Each quality has its own optimum additive value that may change with time and place.
The author made the critical new assumption that selective pressures score ddifferences from optima rather than additive values directly. With an optimum of 6, totals of 5 and 7 would both score One and be indistinguishable.
- Note C. For each quality a population mean may be measured. From the total population the 'best' X% provide parents for a progeny generation. Both parents and progeny will have their own means.
Although expected from simple additive theory, the progeny mean does not equal the parent mean in either simulation or practice. It shows reversion towards the mean of the original population.
The percentage of the selection differential that is actually achieved is known as 'heritability'. Its value varies within the extremes of 0% and 100%.
- Note D. A plateau is when intense selection produces little or no change in a population mean.
- 'Additive Genes in Evolution and Selection' by Roy Silson is obtainable from Greenfield Publications, Mayflower House, Station Road, Tring, Herts, HP23 5QX, tel 044282 381, 1989, 312 pages, ISBN 1 871508 010, paperback, L10, or ISBN 1 871508 003, hardback, L20.