Root colonization by introduced bacteria is an important step in the interaction of beneficial bacteria with the host plant. Investigators attempting to measure root colonization by bacteria must face several issues. A clear concept or definition of root colonization should be stated in each research summary, as several different definitions have been proposed. We consider true root colonists to be those bacteria that colonize roots in competitive conditions, i.e., natural field soils. Different methods of processing root samples are required if one is measuring external root colonization alone, internal colonization alone, or both. Given that most beneficial bacterial strains currently under investigation as root colonists are members of taxa naturally found in soils, a marking system is required to differentiate the introduced strain from members of the indigenous rhizosphere community. Spontaneous antibiotic resistance, immunological approaches, and foreign DNA sequences are among the marking systems that have been used and each has some possible advantages and disadvantages. More research is needed in the development and comparison of marking systems. The design of experiments to measure root colonization should take into account several statistical issues. One must decide what constitutes the sample unit for each replication of a given treatment, e.g., whole root systems or root segments. Consideration should also be given to how best to express the estimated population of root colonists (e.g., cfu/g fresh or dry weight root, cfu/cm root, or cfu/surface area root). Statistical analysis by standard analysis of variance tests should be used whenever possible to separate treatment means of colonization levels; however, one must determine that the underlying assumptions of these tests are correct for each experiment. Finally, in quantification of populations on roots, one will almost certainly encounter replications with no bacteria, i.e., zeros. There are several options for how to calculate treatment means when one or more replications is a zero, and the implications of these options are discussed.
