Conjugative transfer of R plasmids R64 and R64drd-11 has been compared in vitro and in vivo without selective pressure by antibiotics in a simplified experimental system; the ecosystem was the bowel of germfree chickens, with the host bacteria almost isogenic, and the plasmids differing only in their conjugative transfer frequency. The spread of repressed and derepressed (drd) R plasmids in recipient bacterial populations was very extensive. The repressed phenotype had only a transient effect during the first 4 h. The level of implantation of the donor bacterial population seems to be of minor importance. Only with a poor recipient (con strain) could the spread of R plasmids be reduced and a steady state with a predominantly sensitive bacterial population be established. It is suggested that this steady state results from an equilibrium between the frequencies of R plasmid transfer and loss.Conjugative transfer is a frequent property of enterobacterial plasmids in general and R plasmids in particular. This complex and multigenic mechanism has been observed in the laboratory (in vitro) and seems to exist also in nature, particularly in the gut of animals and humans (in vivo) (13).The intestinal ecosystem of monogastric animals is complex and open (37). Their procaryotic floras are extremely dense and include, among the major groups of microorganisms, strictly anaerobic species (10'0/g of feces) and Enterobacteriaceae (106 to 108/g of feces). The variety of species, mixing, and high bacterial density may be favorable conditions for conjugative transfer. But it is difficult to quantify in vivo transfer, and little is known about the major parameters controlling its occurrence.Epidemiological studies on antibiotic resistance in bacteria (7,9,12,14,28,39,48) and direct data obtained from contaminated humans (1,4,6,32,33,40,42,50) or animals (13,16,17,21,34,41,43,44) give evidence of an in vivo transfer with or without antibiotic selective pressure. A higher proportion of transconjugants is usually obtained in the presence of selective antibiotics (3, 4).Experiments involving germfree animals contaminated by a donor and a recipient population show the existence of an important plasmid transfer (22,23,31,35,36,47). This artificial system is a means of simplifying experimental models and allowing the comparison of experimental results to theoretical models such as those of Anderson and Lustbader (5) or Stewart and Levin (46). These authors have considered three important parameters: the growth rate, the conjugative transfer frequency, and the plasmid rate of loss through segregation. Their model is restrictive because the plasmids are not bacteriocinogenic and are independent of the chromosome, and because plasmid maintenance is not imposed by selective pressure.We have applied the same restrictive conditions to a qualitative and quantitative study of in vivo conjugative transfer in a simplified experimental model. The ecosystem was the germfree chicken gut. The contaminating strains used were derived from the same pare...