Research Critique The Simulated Evolution of Biochemical Guilds
Research Critique: The Simulated Evolution of Biochemical Guilds: Reconciling Gaia Theory and Natural Selection K. Downing & P. Zvirinsky, 2000 Presenter: Joanne Lee Date: 30 th August, 2004
Talk Outline Neo-Darwinism vs. Gaia Theory Daisyworld Guild Model Simulation Results Critique of Guild Model Conclusion 2
Question: How did giraffes get their long necks? Inheritability of Acquired Characteristics: The giraffes stretched their necks, and so their children and subsequent generations were born with long necks. 3
Question: How did giraffes get their long necks? Survival of the fittest: Giraffes born with long necks had a better chance of survival than those born with short necks, and so had an increased reproduction rate. Over time, the giraffe population became longnecked. 4
Neo-Darwinism Main ideas: Survival of the fittest: individual selection, not group selection Combines Darwin’s views with genetics Neo-Darwinism is the most widely taught and accepted view on evolution 5
Gaia Theory Organisms both affect and regulate their environment. – James Lovelock 6
Observation Local biotic mechanisms regulate global chemical concentrations N: P ratio in oceans is identical to N: P ratio in algae and zooplankton There exist efficient recycling pathways for poorly-supplied nutrients High cycling ratios for carbon, nitrogen and phosphorus support far more biomass than what external fluxes alone can support 7
Carbon Cycle Carbohydrates Photosynthesising plants Carbon dioxide 8 Herbivores, detritivores
Neo-Darwinism vs. Gaia Theory How do recycling networks and chemical regulation emerge? Neo-Darwinists accuse Gaia theory of: Group selection Teleology Adaptationist wing of Neo-Darwinism states that organisms adapt to their environment, while Gaia Theory claims that organisms adapt but also influence their environment 9
Daisyworld Simple differential-equation model to refute Gaia theory criticisms Simulation of two species of daisies living on a planet Same preferred temperature of 22. 5 C Black daisies have a low albedo, creating warmer local temperatures White daisies have a high albedo, creating cooler local temperatures 10
Scenario Daisyworld is subject to levels of increasing temperature At low temperatures, black daisies proliferate As the temperature increases, white daisies take over Inevitably, temperature becomes too hot and no daisies survive Observation: For a limited range of temperature inputs, daisies are able to keep the temperature at 22. 5 C Conclusion: Simple local interactions among the biota can have global regulatory consequences 11
Criticisms of Daisyworld Small genotype space: doesn’t show relationship between evolution and regulation What if Daisyworld was extended to include genotypes for temperature preference? At any point in the simulation, the population comprises daisies that: prefer the current temperature; prefer a higher temperature and have a low albedo; or prefer a lower temperature and have a high albedo Simulations show that daisies will simply adapt to the rising temperature, rather than regulate it 12
Guild Model Objective: To simulate the emergence of nutrient recycling networks and chemical ratio regulation using natural selection mechanisms Key element borrowed from Daisyworld: Organisms are able to create local buffers against the environment 13
Guild Model Biochemical Guild: Organisms that have the same nutrient inputs and outputs Organisms cannot consume and produce the same chemical 14
At the Environmental Level Nutrients N 1…Nn Input fluxes 1 … n Output fluxes Environment chemicals (internal amount) 1 15 … n
At the Genome Level en. Zyme genes: Zk = 1 means that organism produces an enzyme to free Nk from the detritus Chemical genes 1 … n Fin percentage of each nutrient consumed (input) Fout percentage of each nutrient produced (output) 16
Organism characteristics Rf Rm X ksat 17 : base feeding rate : metabolism rate : biomass : satisfaction
Satisfaction An organism’s satisfaction is based the deviation of its local perception of the relative fractions of the environmental nutrients from a user-defined optimal ratio An individual’s input and output fractions are taken into account when computing the effective nutrient fractions that it experiences The closer the ratio is to the user-defined optimal ratio, the higher the satisfaction 18
Local Chemical Ratio 19
Feeding and Metabolism Afeed = X 0. 75 * rf * S Example: X 0. 75 = 900, rf = 0. 01, S = 1, then Afeed = 9. The organism attempts to consume 9 units of nutrients, in the proportion specified by its input alleles. The input nutrients are immediately converted into biomass Ametab = X 0. 75 (rm + nz * costz) 20
Death and Decay An organism dies if it cannot access sufficient input nutrients Mortality rate is dependent on population density The biomass of a dead organism is partitioned into the detritus in direct proportion to its input nutrients An organism feeds on detritus only if there are no available nutrients left to feed on, and if it produces a nutrient-specific enzyme to free the nutrient from the detritus 21
Reproduction is permitted only if the population has not reached its carrying capacity Reproduction through replication: an organism splits into two when it has doubled its biomass Mutations may occur during replication A percentage of organisms are randomly selected for conjugation (chromosomal crossover) 22
Global Measures of System Performance An efficiently recycling ecosystem is where: The outputs of one guild are consumed by another guild The detritus of one guild is freed by the enzymes of another guild and immediately consumed These processes prevent chemical loss from the environment and increase the biomass 23
Global Measures of System Performance: Cycling Ratio The amount of nutrients consumed over the amount available from the input flux The higher the ratio, the more selfsufficient the environment is 24
Ideal Free Distribution (IFD) IFD error compares the ratio of the available nutrients (environment and detritus) against the average input nutrient ratio of the biota. Essentially, IFD measures how well the biota has adapted to its environment. The biota has completely adapted when IFD = 0 25
Guild Simulation in 1 D Initial population size: 100 Max population size: 2000 Number of generations: 800 Timesteps per generation: 50 Mutation rate / individual: 0. 5 Conjugation fraction / generation: 0. 2 Number of nutrients: 4 Initial biomass units: 20 26
Guild Simulation in 1 D Homogenous population of 100 individuals: All individuals produce N 1 All individuals consume N 2 No individuals produce enzymes Nutrient inputs: [20, 20, 20] [5, 10, 25, 50] [50, 25, 10, 5] (Generations 1 – 400) (Generations 401 – 600) (Generations 601 – 800) Goal environmental chemical ratios: [0. 4, 0. 3, 0. 2, 0. 1] 27
Population Size 28
Population Size Initially every individual consumes N 2, but there is not enough N 2 to support the whole population. Population size drops to below 50 at startup. Due to mutation, some individuals can now consume a nutrient other than N 2. With an alternative nutrient to feed on, the population starts increasing after 100 generations. 29
Diversity 30
Diversity At startup, all individuals produce N 1 and consume N 2 Over 300 generations, the production and consumption of the 4 nutrients converge to an equal proportion 31
Enzyme Production Increasing enzyme production in the first 100 generations is followed by decreased enzyme production in generations 101 - 300 There is insufficient detritus to support the growing number of decomposers, and so the metabolic cost of producing enzymes does not pay off 32
Increase in N 1 -only Consumers After 300 generations, an N 1 -only consumer emerges Because all individuals produced N 1 at startup, there is an abundant amount of N 1 in the environment Conditions for N 1 -only consumers are ideal, and so the population of N 1 -only consumers multiplies rapidly 33
Population Boom Increased diversity, but constant biomass Advent of N 1 -only consumer allows conversion of N 1 into biomass The output nutrients of the N 1 -only consumers supply other organisms with nutrients – this triggers a population boom as organisms feed and multiply. Population size is now over 900 Throughput of the recycling networks increase. Cycling ratios increase 34
Population Limit Reached When the population exceeds 900, the system reaches a new steady-state limit, which can only be increased by changes in the external nutrient fluxes At this density, competition for nutrients is fierce. Enzyme production is an advantage, allowing individuals to tap into the nutrients stored in the detritus. Increased detritus feeding increases the cycling ratio 35
Emergence of global nutrient-ratio control Prior to the population-size and recycling booms, N 1 made up a large fraction of the environmental nutrients. After generation 300, the input diversity is diverse enough to ensure the consumption of most available nutrients, rather than having them left untouched in the environment. Recycling loops primed by N 1 consumption then facilitate a biomass increase 36
Cycling Ratio 37
Extreme Control Problems Input flux: [5, 10, 25, 50] Optimal ratio: [1/18, 10/18, 5/18, 2/18] Control is only feasible when biotic diversity reduces the dominance of any one nutrient. After this, the chemicals partition into values close to the desired ratios 38
Guild Model in 2 D: Implementation in SWARM Agents move around a 2 D grid, eat nutrients and produce other nutrients as metabolic waste Additional vision and metabolic genes Gene mutations occur throughout a lifetime, but phenotypic results are manifest only in the next generation Findings are consistent with the simulations in the 1 D environment 39
Simulation Results Emergence of nutrient recycling networks Set of nutrients + vast number of organisms resource competition emergence of many biochemical guilds nutrient recycling networks Emergence of global regulation of chemical ratios Under-consumed nutrients + new consumers population explosion increase in cycling ratios high transfer fluxes between guilds control of global chemical ratios, via their cumulative production and consumption. Coordinated behaviour is not due to group selection or teleology. It can be explained by individual-based natural selection 40
Significance of Findings Previous models such as Daisyworld support the compatibility of Gaia and natural selection, but they exhibit a certain hard-wiredness The Guild Model showed that global regulation can also emerge from the aggregate metabolism of a community 41
Critique of Guild Model In the real world, recycling networks refer to when the same nutrient cycles through the network (albeit in different forms). An organism cannot feed on a nutrient and then output an arbitrary nutrient as waste Recall the Law of Conservation of Matter: Matter cannot be created or destroyed Limited genotype space: what if biota adapted to the current chemical ratios, rather than trying to change it? 42
Conclusion Guild Model supports the view that the emergence of nutrient recycling networks and regulation of chemical ratios are consequences of natural selection Needs to strengthen its argument by revising its chemical model and the issue of evolving preferences 43
References http: //alife. tuke. sk/projekty/mag_html/guild/ guild. html http: //neuronai. tuke. sk/~zvirinsk/projects. htm http: //neuron. tuke. sk/~zvirinsk/thesis/index. html http: //www. idi. ntnu. no/grupper/ai/eval/guild/ guild. html http: //userpage. chemie. fuberlin. de/~steffen/bcc/111. html http: //www. alife. org/links. html 44
- Slides: 44