[Paper Review] Diversity Evolution

Russell K. Standish|arXiv (Cornell University)|Oct 15, 2002

Evolutionary Game Theory and Cooperation18 references9 citations

TL;DR

This paper investigates unbounded diversity evolution in artificial ecosystems using the Eco Lab model, demonstrating that specific migration rate tuning induces exponential diversity growth—revealing a critical 'resonance' effect. It identifies specialization and biogeographic provincialism as mechanisms enabling unbounded creative evolution, challenging artificial life systems to replicate biosphere-level complexity.

ABSTRACT

Bedau has developed a general set of evolutionary statistics that quantify the adaptive component of evolutionary processes. On the basis of these measures, he has proposed a set of 4 classes of evolutionary system. All artificial life sytems so far looked at fall into the first 3 classes, whereas the biosphere, and possibly the human economy belongs to the 4th class. The challenge to the artificial life community is to identify exactly what is difference between these natural evolutionary systems, and existing artificial life systems. At ALife VII, I presented a study using an artificial evolutionary ecology called \EcoLab. Bedau's statistics captured the qualitative behaviour of the model. \EcoLab{} exhibited behaviour from the first 3 classes, but not class 4, which is characterised by unbounded growth in diversity. \EcoLab{} exhibits a critical surface given by an inverse relationship between connectivity and diversity, above which the model cannot tarry long. Thus in order to get unbounded diversity increase, there needs to be a corresponding connectivity reducing (or food web pruning) process. This paper reexamines this question in light of two possible processes that reduce ecosystem connectivity: a tendency for specialisation and increase in biogeographic zones through continental drift.

Motivation & Objective

• To investigate mechanisms enabling unbounded, creative evolutionary dynamics in artificial ecosystems, contrasting them with the biosphere’s apparent open-ended diversity increase.
• To test whether artificial evolutionary systems can achieve the 4th class of evolution—unbounded creative evolution—identified by Bedau as characteristic of the biosphere and possibly the human economy.
• To examine the role of ecological processes such as specialization and biogeographic provincialism in facilitating sustained, exponential diversity growth in artificial ecosystems.
• To determine whether artificial systems can self-organize toward a critical surface balancing speciation and extinction, enabling maximum complexity for given diversity levels.

Proposed method

• Employed the Eco Lab artificial evolutionary ecology model with spatially structured populations, mutation, migration, and Lotka-Volterra-type population dynamics.
• Applied Bedau’s evolutionary statistics to quantify adaptive activity and creative evolution, distinguishing bounded from unbounded diversity growth.
• Used a neutral shadow model to isolate adaptive from neutral evolutionary components, enabling measurement of true evolutionary creativity.
• Systematically swept the maximum migration rate (|γ|∞) across a range of values to observe its effect on diversity dynamics over 18 million timesteps.
• Conducted parameter scaling and timing variations (e.g., migration every 100 vs. 1000 timesteps) to test for numerical artifacts and validate the resonance phenomenon.
• Analyzed the system’s behavior under different grid sizes (2×2 and 3×3) to assess robustness of observed effects across spatial configurations.

Experimental results

Research questions

• RQ1 Can artificial evolutionary systems achieve unbounded, creative evolution similar to the biosphere, as defined by Bedau’s 4th class of evolutionary systems?
• RQ2 What ecological mechanisms—such as specialization or biogeographic provincialism—can drive sustained, exponential growth in diversity within artificial ecosystems?
• RQ3 Does a critical surface exist in artificial ecosystems where speciation balances extinction, and how does it constrain maximum complexity for a given diversity?
• RQ4 Is the observed exponential diversity growth at a specific migration rate a numerical artifact or a genuine dynamical resonance in the system?

Key findings

• A distinct resonance effect was observed at a migration rate of approximately 1×10⁻⁵, where diversity exhibited exponential growth, independent of random number seed or initial conditions.
• The resonance persisted across different grid sizes (2×2 and 3×3), indicating it is not an artifact of spatial discretization or system size.
• Scaling model parameters by 0.1 (altering timescale) did not shift the resonance location, suggesting the effect is intrinsic to the system’s dynamics rather than a timescale-dependent artifact.
• Changing the migration frequency from every 100 to every 1000 timesteps shifted the resonance by one order of magnitude, ruling out certain software or algorithmic faults.
• The system did not exhibit the expected power-law dependence of diversity on migration rate (D ∝ A^c), but instead showed a nearly constant c across most migration rates, indicating non-monotonic and complex response.
• The results suggest that connectivity-reducing processes—such as specialization and biogeographic isolation—can enable unbounded creative evolution by allowing sustained speciation without immediate extinction.