Host Control of Symbiont Transmission: the Separation of Symbionts into Germ and Soma

—Obligate, vertically transmitted symbionts occur in many species. Hosts often have elaborate developmental processes and specialized organs to control the reproduction and transmission of their symbionts. Control mechanisms divide the symbionts into reproductive, germline lineages and nonreproductive, somatic lineages. This germ-soma distinction favors reduced competition among symbionts and fewer virulent effects on the host. Observations suggest a repeated evolutionary trend toward host control of symbionts and division of symbionts into germ and soma. Theory predicts that host control evolves only in response to particular mechanisms of symbiont competition, for example, when symbionts disrupt host development. Many symbionts have a beneficial effect on their host. For example, leeches and lice cannot survive on their blood diet without bacterial symbionts, many sea organisms have complex organs to house luminescent bacteria, and insects that feed on cellulose or plant sap rely on a variety of bacterial partners (Buchner 1965). Although the symbionts provide benefits to the host, those benefits may be partly offset by the disruptive, virulent effects of competition among symbionts. Hosts often appear to control competition by limiting the reproductive opportunities of the symbionts. I argue that these host-imposed limits separate the symbionts into transmissible, germline lineages and nonreproductive, somatic lineages. I develop the argument in five parts. First, I summarize the theory of hostsymbiont conflict. The theory explains why symbiont competition tends to disrupt the host and therefore why the hosts gain from controlling the reproductive opportunities of the symbionts. I relate theories about symbiont competition to Buss's (1987) argument that competition among cell lineages explains the origin of the germ-soma distinction in metazoans. In the second section, I describe examples of host-symbiont natural history. This section is taken up with biological details of symbiont transmission and host development. The observations support my claim that symbiont lineages are often separated into germ and soma. I develop new theory about host-symbiont conflict in the third section. One interesting conclusion is that the average fitness of a host population generally increases by imposing germ-soma differentiation on the symbionts, but host al*E-mail: [email protected]. Am. Nat. 1996. Vol. 148, pp. 1113-1124. © 1996 by The University of Chicago. 0003-0147/96/4806-0009$02.00. All rights reserved. 1114 THE AMERICAN NATURALIST leles that cause such traits do not necessarily increase in frequency. This conclusion is similar to a result from the theory of genomic conflict: mean fitness of hosts generally increases by imposing uniparental inheritance of cytoplasmic symbionts, but host alleles that prevent cytoplasmic mixing spread only when they are associated with an immediate reduction in cytoplasmic virulence (Hoekstra 1987 and Frank 1996a discuss uniparental inheritance of symbionts). This theory focuses attention on host characters that, by controlling symbionts, cause an immediate increase in host fitness. I argue that limiting symbiont disruption of host development may be particularly important. In the fourth section, I summarize observations showing that symbionts often disrupt host development. In the final section, I consider the types of comparative and experimental data that could be gathered to learn more about host-symbiont interactions and the evolutionary processes that favor differentiation of germ and soma. THEORETICAL BACKGROUND The evolution of within-host competition among symbionts is an example of the general problems of kin selection and group selection (Hamilton 1972; Bremermann and Pickering 1983; Frank 1996b). Each genotype gains within the group by outcompeting its neighbors but loses as the overall efficiency of the group declines (virulence increases). The tension is resolved according to the distribution of genetic variance. Relatively more variance within groups of symbionts favors more intense competition, whereas relatively more variance among groups favors greater cooperation within groups and lower virulence. The problem can be described equivalently in terms of kin selection. The coefficient of relatedness is the ratio of variance among groups to total variance in the population. As the relative variance among groups rises, relatedness and competition increase among neighboring symbionts. Any problem of within-group competition among genotypes can be viewed as a problem of the evolution of virulence. For example, Buss (1987) used a group selection argument to suggest that cell lineages within a metazoan compete in ways that lower individual fitness. In this case an "individual" is really a group of cell lineages. A lineage gains a reproductive benefit relative to its neighbors by increasing its probability of contributing to the germ tissue. The intensity of competition and the potential for reducing individual fitness depend on the genetic variation among cell lineages. One way to control renegade cell lineages is with "policing" traits that enforce a germ-soma split early in development. This split prevents reproductive bias among lineages during subsequent development. Once the potential for bias has been restricted, a cell lineage can improve its own fitness only by increasing the fitness of the individual. Maynard Smith (1988) agreed with Buss's (1987) logic about the potential for cell lineage competition, but he argued that metazoans solved their problems of cell lineage competition by passing through a single-celled stage in each generation. When an individual develops from a single cell, all variation among subsequent cell lineages must arise by de novo mutation. In Maynard Smith's view, SYMBIONT GERM AND SOMA 1115 such mutations must be sufficiently rare that the genetic relatedness among cells is essentially perfect. Thus, the soma sacrifice reproduction as a natural, altruistic act in favor of their genetically identical germline neighbors. Buss recognized the importance of de novo mutations within an individual but argued that these would be sufficiently common to favor significant cell lineage competition and policing. Cell lineages of symbionts face a similar tension between competition within hosts and the overall success of their host. However, two additional factors influence symbionts when compared with metazoan cell lineages. First, when symbionts are transmitted vertically, from parent to offspring of the host, hundreds or thousands of symbiont cells typically infect each offspring. This large founding population contrasts with the single-celled bottleneck typical of a newborn metazoan. Thus, a host develops with a potentially diverse set of symbionts. Second, symbionts may infect a host horizontally, from another host individual or from the environment. Such mixing of symbiont lineages greatly decreases relatedness within hosts and favors within-host competition. Selection will always favor the symbionts to transmit partly by horizontal routes to avoid competition against relatives within the host (the dispersal effect of Hamilton and May 1977, applied to symbionts in Frank 1994). In summary, metazoans appear to maintain cooperation among cell lineages by passing through a single-celled bottleneck in each generation. By contrast, symbiont lineages have greater potential for diversity within hosts and require other mechanisms of control. This leads to my argument that hosts often impose a germ-soma split on their vertically transmitted symbiotic lineages. This split prevents reproductive bias and reduces opportunities for virulent competition among