Patterns in Space and the Dynamics of Evolution

One of the central contributions of NECSI's research into evolution is the clarification of the important role that patterns of geographical distribution have on evolution.

Classical work in population biology typically assumes that geography doesn't matter, and treats systems as though everything happens in one place --- all organisms can, for example, access all resources and mate with any other organism. The real world, though, is spatially extended and traveling from one place to another takes time. There are many ways this changes the effect of evolution on populations. A few examples:

The gene-centered view has long held center stage as the dominant explanation of how evolution operates. According to this view we can understand evolutionary success by considering a fitness that can be assigned to each gene. However, taking space into account shows that the gene-centered view is insufficient, failing, for instance, to admit the possibility of self-sustaining regions of similar individuals. Even if the environment is homogeneous, the organisms in one region can be different from the organisms in another region. Such patterns of diversity across geographical space have a different dynamic behavior than the traditional population biology models---they depend on the pattern itself. This means that what is successful in evolution cannot be characterized by properties of genes. Moreover, unlike traditional models, the diversity in such models is much higher (because different locations have different organisms), consistent with what is found in nature. Furthermore, the ecological structure of the environment matters in a key way to the dynamics of evolution because the patterns of diversity are affected by them not just locally but across space.

Selection above the level of the individual has been thought unlikely or irrelevant for nearly fifty years. In spatial models, though, higher-level selection can very easily be an important evolutionary force, and give rise in a straightforward way to phenomena that are otherwise very tricky to explain. A canonical example is altruism.

When reproduction takes place locally, and children tend to inhabit the same regions their parents did, then a tendency to consume more resources than the environment can provide will cause problems a few generations down the line. The way individuals change the environment has effects that persist; the environment, as well as the genetic code, is in effect heritable; and actions taken now can have important consequences much later.

These considerations are not captured by individual- or gene-centered views of evolution, which take into account only the short-term benefits of exploitation and reproduction. In a spatial model the usual notion of reproductive "fitness" is not always a good predictor of long-term survival of descendants.

Altruism can be identified with behaviors that may not be optimal in the short term, but confer long-term success in a spatial environment. Selfish "cheaters" and "defectors" can enjoy immediate advantages over altruists, but if their behavior is more likely to lead to their eventual extinction, their success will be transient.

Social signaling can help to coordinate altruistic behavior. The strong advantage it can confer in the long term helps to explain why communication between members of the same species is so ubiquitous throughout nature.

Spatial models also give insight into issues like the distribution of biodiversity. Knowledge about where variation in a population exists, and how to estimate it based on a few samples, is crucial to inform decisions about subjects like how to set up nature preserves most effectively.