Evolution to the edge of chaos: Multilevel selection and life history evolution in metapopulations

The history of life has been marked by transitions involving the association of lower level units into higher levels of organization-genes into chromosomes, cells into multicellular organisms, and individuals into social groups. With an emphasis on the transition between individuals and social groups, research in my lab seeks to address the causes and consequences of these evolutionary transitions. Areas of interest include (a) multilevel selection, (b) causes and consequences of social evolution, and (c) the ecology and evolution of metapopulations. We use a variety of research tools, including fieldwork in temperate and tropical areas, computer simulation, analytical modelling, and laboratory work employing cytological and molecular techniques. Central to our empirical studies and a source of inspiration for our theoretical work are the social spiders, a phylogenetically diverse set of species that have converged in evolving cooperative behaviour and highly subdivided population structures. Because colonies of these organisms constitute not only social groups, but also self-sustaining populations, they are ideally suited to address some fundamental and often controversial issues at the intersection of ecology and evolution.

Multilevel selection
Much of my life's work has involved the exploration of traits for which selection at multiple levels may be in conflict, such as sex ratios in subdivided populations and cooperation in social groups. My dissertation work at Harvard University (Ph.D. 1992) was a combination of empirical and theoretical research into one of the most hotly debated issues in evolutionary biology-whether selection at the level of groups can be an effective evolutionary force. At a time when the idea of group selection was viewed with hostility, I showed that social spiders represent a largely unequivocal natural example of this form of selection. By means of computer simulations, I showed that their highly female-biased sex ratios represent a compromise of selection acting within and among their more or less permanently isolated colony lineages. This and more recent work on the evolution of cooperation in single-generation groups of non-relatives demonstrate that there are general principles underlying multilevel selection processes that should apply to a broad range of population structures and to traits as varied as sex ratios, cooperation, pathogen virulence, and cancer. In a synthesis paper in preparation I show how these principles, derived from Darwin's four postulates of evolution by natural selection, can be used to test for the presence of selection at any level, either in our conceptual models or in our empirical systems. Philosophically, I argue that the levels of selection problem is not one of prediction, but one of causality: while models built from the bottom up or the top down can often be used interchangeably to correctly predict the expected evolutionary outcome, there should be only one answer to the question of whether one, two, or more levels are involved in yielding a particular equilibrium state.

Causes and consequences of social evolution
Based on the empirical finding that individual fitness often has a "humped" nonlinear relationship with colony size, I have developed a framework for the study of social evolution that uses the shape and magnitude of this function to make predictions about the origin, size, and dynamics of social groups. While subsuming complex ecological interactions in a simple three-parameter model of the relationship between average individual fitness and colony size, this approach brings to the study of sociality the theory and methods of nonlinear dynamics while representing a radically different, albeit complementary, approach to the existing paradigm -Hamilton's inclusive fitness framework- that has dominated the field for the last few decades. Three projects illustrate the value of this approach:

Cooperation and nonlinear dynamics.- I have shown, for instance, that the enhanced reproductive success that results from cooperation may allow the colonization of harsh or marginal environments in which solitary individuals would not be able to replace themselves. In contrast, in environments in which group living and cooperation allow access to plentiful resources, enhanced reproductive success may lead to a boom and bust pattern of group and population growth. The former situation is dramatically illustrated by the existence of eusocial mole rats in the extremely arid deserts of southern Africa; the latter, by the oscillations in colony size of some social spiders and the global population outbreaks and crashes of tent caterpillars, voles, migratory locusts, and tree-killing bark beetles.

Solving the freeloaders paradox. - Using agent-based simulation models I have extended this general framework to provide a solution to one of the enduring problems in social evolution-the maintenance of cooperation in the presence of freeloaders. I have found that even though freeloaders can benefit from cooperators when rare, they are selected against when common due to the reduced productivity of the groups they overburden with their presence. This effect allows the evolution and maintenance of cooperation under a wide range of parameter values, even when groups consist of non-relatives and cooperators suffer a relative fitness cost within their groups. These models provide the testable prediction that freeloader frequencies will oscillate through time and is consistent with the widespread occurrence of helping behaviour among nonrelatives in human and many animal societies.

Ecology, demography, and kinship in social evolution. - I have also used agent-based models to explore the interaction between ecology, demography, and kinship in social evolution. I have shown, for instance, that the decision to admit nonkin into groups has not only genetic, but also demographic consequences- while restricting groups to close kin facilitates the evolution of altruism, it may limit groups to suboptimal sizes; admitting nonkin, on the other hand, allows the formation of groups of any size, but limits the evolution of altruism. I have also shown that the random assignment of helping roles within groups breaks this trade-off, thus allowing the evolution of highly altruistic behaviours in groups of nonrelatives of any size.

Social spiders as model systems in ecology and evolution
A major contribution of my work has been to pioneer the use of social spiders as model systems for a variety of important and controversial issues in ecology and evolution, discovering in the process 6 of the 21 species known to the world. In addition to showing that their highly female biased sex ratios may represent the workings of interdemic (= intercolony) selection (see above), I have used them to explore the interplay between intrinsic and extrinsic factors in social evolution, the short and long term consequences of inbreeding, and, as proposed in this project, the interplay between local and global dynamics in metapopulations.

The ecology and biogeography of spider sociality. - Work from my lab in this area illustrates, as in perhaps no other social system to date, how sociality arises from the interaction between intrinsic features of organisms and the environments in which they live. We have shown, for instance, that social species in the genus Anelosimus, whose colonies contain from hundreds to tens of thousands of spiders, are restricted to the wet low to mid-elevation tropical areas of the New World, while the single-family groups of related subsocial species predominate at higher elevations and latitudes. We postulate that this pattern reflects an interaction between the dense three-dimensional webs characteristic of these spiders with two separate environmental gradients, one of insect size, which we have shown decreases with elevation and latitude, the other of the intensity of precipitation and abundance of potential ant predators, which we have shown exhibits the opposite trend.

The evolution of inbred social systems: short term benefits, ultimate costs. - One of the extraordinary features of spider sociality is its association with strong inbreeding as members of both sexes remain within their natal nests to mate generation after generation. My students and I have investigated the fitness consequences potentially associated with the origin of such inbred systems, including the costs and benefits of remaining in the natal group in social species, of dispersing in subsocial species, and of producing inbred offspring in ancestral-like outbred subsocial species made to artificially inbreed. These findings allow us to quantitatively show that the origin of inbred spider sociality was probably an easier transition than previously believed, thus explaining the repeated origins of this social system. Based on its spindly phylogenetic distribution, however, we also show that inbred spider sociality may in the long run constitute an evolutionary dead end.

Our research has been highlighted in Trends in Ecology and Evolution (1987, 1992), Science News (1999), Science Editor's Choice (2007), and Nature News online (2008). It has also been the subject of three TV documentaries (Scientific American Frontiers, USA 1999; The Desert Speaks, USA 1999; La Televisi&#243n, Ecuador 1997) and a radio interview (Canadian Broadcasting Corporation "Quirks and Quarks, Oct 2002). The models on cooperation and nonlinear dynamics were highlighted by James T. Costa in his recent book "The Other Insect Societies" (Harvard U Press, 2006), in a chapter entitled "Back to the Future." Our work on spider sociality has been transformative of the field. My 1997 review has been cited close to 150 times and in the more recent Lubin and Bilde's (2007) review, work from my lab is cited close to 100 times.