Abstract
Micro-organisms are ever more widely recognized as social. Many researchers, whose primary interest is the evolution of cooperation, have turned to microbes as the organisms of choice to test fundamental theories on the evolution of cooperative behaviour (e.g. Crespi, 2001; Brown & Johnstone, 2001; Ferriere et al., 2002; Turner & Chao, 2003; Velicer, 2003; Travisano & Velicer, 2004; Greig & Travisano, 2004; Griffin et al., 2004). The key problem for the evolution of cooperation is the fitness cost for the cooperating individual: cooperation between individual members of a group produces a public good that may benefit all group members, whether they cooperate or not, but only the cooperators pay the costs of producing the public good. Those who do not cooperate are called defectors, and those that gain an advantage from defection are called cheaters. Altruism is a behaviour that decreases the fitness of the altruistic individual while benefiting others. Cooperative behaviour in the presence of cheaters constitutes altruism towards the cheaters. The problem, therefore, is cheaters. Investment in cooperation reduces the fitness of the cooperating individual relative to the cheater, while the fitness of the group is increased relative to a group with a lower level of cooperation. In other words, we have a conflict of interest between individuals and the group. For recent more in-depth treatments, see Velicer (2003) and Travisano & Velicer (2004).
Mathematical models allow us to study the consequences of a coherent set of assumptions about the characteristics and composition of the system under study even when this system has complex dynamics and spatial structure. Using models, we can ask, for example, what are the minimum requirements for the evolution of cooperation? For several decades, models have shed light on the evolution of cooperation; for example, they have revealed the importance of repeated interactions and of spatial structure; see Sigmund (1994) for an introduction.
In biofilms or other complex dynamic microbial assemblages, bacteria live in crowded environments, where they interact with many neighbours in multiple positive and negative ways, much like our life in cities (Watnick & Kolter, 2000). Given that biofilms are multicellular, whether as a multicellular community or a multicellular organism, they are often seen as models for the development and evolution of multicellular organisms.
A report by Pfeiffer et al. (2001) was the first to identify a trade-off between growth yield and growth rate in heterotrophic organisms. They realized that trading an increase in growth yield for a decrease in growth rate is cooperative behaviour where the cooperation is passive and indirect, consisting in restraint from competition over the use of limiting, shared (external) resources. Furthermore, this study also highlighted a connection between the evolution of cooperation in resource use and the evolution of multicellularity (see the section on strategies below).
As an extension of this study, Pfeiffer & Bonhoeffer published an individual-based model of variants of the above cooperative strategy in 2003, while another individual-based model of the evolution of cooperation in biofilms had been submitted (Kreft, 2004), raising questions of comparability of the two models, which will appear similar to the non-specialist reader, yet differ in a number of basic assumptions.
Individual-based models are mathematical models of population dynamics without any specifications for population behaviour; rather, the characteristics of the higher level of organization, the population, result from the dynamics on the lower level of organization, the individual organisms: population characteristics emerge from the actions and interactions of the individuals with each other and the environment. While this is true for all bottom-up models, individual-based models are those bottom-up models that explicitly allow variability among individuals (DeAngelis & Gross, 1992). For example, an individual-based model of biofilms would not contain a single line of programming code describing biofilm structure or function. Rather it describes the properties of the bacteria (such as metabolism, growth, motility and quorum sensing), the system geometry (such as liquid flowing over a flat, inert substratum) and mass transport processes (such as diffusion and convection) and studies the consequences of these properties on biofilm formation.
The report by Kreft (2004) looks at the above yield versus rate trade-off from a more microbiological perspective, examining the evolution of altruism in biofilms, and the consequences this entails for biofilm structure and characteristics, pointing out the importance of purification steps for clusters of cooperating cells see the purification step section below.
The aim of this Comment is threefold. Firstly, we compare the above two models of the evolution of cooperation and cooperating groups, on the one hand the studies of Pfeiffer et al. (Pfeiffer et al., 2001; Pfeiffer & Bonhoeffer, 2002, 2003), and on the other the study of Kreft (2004). Secondly, the Comment wishes to emphasize the importance of conflicts of interest between the individual and higher levels of organization in biofilms and why this requires a cluster purification step. Thirdly, we hope that the Comment spawns a debate of the question of why biofilms have not evolved into multicellular organisms.
Comparison of models
Strategies: muscle cells versus Holophaga.
Both models provide the setting for the competition of two alternative survival strategies, which are based on a trade-off between growth rate and growth yield. One strategy is to grow fast at a low yield (rate strategy), the other is to grow slow at a high yield (yield strategy).
The trade-off, in turn, is based on irreversible thermodynamics which states that the rate of a process, such as microbial growth, is proportional to the thermodynamic driving force if the process is not too far from thermodynamic equilibrium. Consider the diffusion of a nutrient from a transient point source as a simple example. The rate of this process, the flux of nutrient, is proportional to the concentration gradient, which is the driving force in this case. Over time, the flux will decrease with decreasing force until equilibrium is reached. Other driving forces are temperature gradients for heat flow or gradients of chemical potential for chemical reactions (Westerhoff & van Dam, 1987).
In the studies by Pfeiffer et al., the two strategies reflect the switch from respiration to fermentation plus respiration as found in, for example, muscle cells and yeasts. These studies show that respiration, which results in higher yield but slower substrate turnover and growth rate, is in fact a group-beneficial trait because a high growth yield is equivalent to an economic utilization of the resource, which benefits all those sharing the (limiting) resource. Using fermentation in addition to respiration is a selfish trait, since it results in lower yield but higher substrate turnover and growth rate. They argue further that the evolutionary transition to multicellularity gives the newly formed multicellular organism an immediate advantage when the constituent cells use respiration, because the conflict of interest between the individuals (growth rate advantage) and the group (growth yield advantage) is diminished by aligning the interests of the individuals with the interest of the group. While many of the typical advantages of multicellularity became available only later, when further evolution had led to increasingly more sophisticated division of labour, forcing cells to cooperate in the use of common resources may have been the initial advantage of multicellularity.
The parameters of the rate versus yield trade-off used in the Kreft (2004) report were abstracted from growth data of the anaerobic bacterium Holophaga foetida which can double its growth rate at the cost of a halved yield, which is a moderate difference between the parameters of the high rate and the high yield strategy, compared to the difference in ATP yield between respiration (32 ATP per glucose) and fermentation (2 ATP per glucose) and the 100-fold higher maximal substrate consumption rate of fermentation (Pfeiffer & Bonhoeffer, 2003).