Transcript Evolution-Natural and Artificial John Maynard Smith

Evolution-Natural and Artificial
John Maynard Smith
Interdisciplinary Program in Cognitive Science
Lee, Jung-Woo
June, 7, 1999
99-06-07
1
Introduction
• Simulation of Evolution on a Computer
– Is it an efficient way of answering practical questions?
– Can it contribute to our understanding of evolution?
• Three general question
– Has there been time for natural selection to produce the
complex creatures we see around us?
– What is the nature of the adaptive landscape?
– What features of the genetic system are necessary and
what contingent?
99-06-07
2
Time for natural Selection(1/2)
• Has there been time for natural selection to
produce the complex creatures we see around us?
– The genome of higher animals and plants contains 108
to 109 base pairs of informative DNA.
– This is enough, together with the environment and the
laws of physics and chemistry, to specify the structure
of the adult organism.
– But, there has not been time for selection to generate
that much information.
99-06-07
3
Time for natural Selection(2/2)
• To get a handle on the question
– Suppose that we start with a set of random DNA
sequence, and try, by selection, to produce a unique
optimal sequence.
– If in each generation we are allowed to weed out half
the sequences, we can specify one base in two
generations, or 109 bases in 2 x 109 generation.
– The time available since the origin of life is
approximately 4 x 109 years, and during most of that
time most organisms got through many generations a
year
99-06-07
4
Nature of Adaptive Landscape(1/4)
• What is the nature of the adaptive landscape?
• Fisher’s view
– evolution by natural selection is a hill-climbing process
that can happen only on a smooth landscape.
– If mutational steps are small, and populations large and
random-mating, it is hard to escape this conclusion.
– But, the fitness landscape is determined not only by the
physical environment but also by other species, its
competitors, predators, prey, which can also evolve.
99-06-07
5
Nature of Adaptive Landscape(2/4)
• Wright’s view
– If a species is divided into a large number of small and partially
isolated populations, a small population may occasionally jump
across a local valley, purely by chance. Once a new adaptive peak
has been reached, the new type can selectively replace the old.
– First difficulty :
• The valley must be small (ab->AB)
• The mechanism will not help us to cross large valleys.
– Second difficulty :
99-06-07
• If recombination is common, new genotype, AB, is
immediately destroyed by recombination
• If the ability to make such transition has been important in
evolution, it is hard to understand why sex is so widespread.
6
Nature of Adaptive Landscape(4/4)
• My suspicion
– Evolution has depended on local hill-climbing, and
long-term entrapment on local peaks has been avoided
because the landscape is always changing.
– This does not mean that there are not peaks that cannot
be reached.
• The wheel may be an unreachable peak, but animals have
evolved effective alternatives.
99-06-07
7
Features of the Genetic System(1/2)
• Two features necessary for any genetic system that
is to support adaptive evolution
– The system should be digital and
– should not permit the ‘inheritance of acquired
characters’.
99-06-07
8
Features of the Genetic System(2/2)
• The need for a digital system
– arises from the familiar difficulty of maintaining
information in a system able to vary continuously.
• The reason for avoiding ‘Lamarckian’ inheritance
– Most changes induced in organisms by the impact of
the environment are non-adaptive : they are the effects
of injury, disease, and ageing.
– A genetic system equipped with a mechanism of
reverse translation would lead to deterioration, not to
adaptation.
99-06-07
9