Paenibacillus polymyxa belongs to the group of plant growth promoting rhizobacteria (PGPR). Activities associated with P. polymyxa-treatment of plants in earlier experiments include, e.g., nitrogen fixation, soil phosphorus solubilization, production of antibiotics, auxin, chitinase, and hydrolytic enzymes, as well as promotion of increased soil porosity. My thesis work showed that, in stationary phase, P. polymyxa released the plant hormone cytokinin isopentenyladenine, in concentrations of about 1.5 nM.

In a gnotobiotic system with Arabidopsis thaliana as a model plant, it was shown that P. polymyxa-inoculation protects plants; challenge by either the pathogen Erwinia carotovora (biotic stress) or induction of drought (abiotic stress) showed that pre-inoculated plants were significantly more resistant than control plants. By RNA-differential display on RNA from P. polymyxa-treated or control plants, changes in gene expression were tested. One mRNA, encoding ERD15 (drought stress-responsive gene) showed a strong inoculation-dependent increase in abundance. In addition, several biotic stress-related genes were also activated by P. polymyxa.

Antagonism towards the fungal pathogens Phytophthora palmivora and Pythium aphanidermatum was studied. P. polymyxa counteracted the colonization of zoospores of both oomycetes on A. thaliana roots, and survival rates of plants treated with P. polymyxa were much higher when challenged by P. aphanidermatum.

Using a green fluorescent protein-tagged isolate of P. polymyxa, colonization of A. thaliana roots was investigated. Two main conclusions can be drawn. Firstly, the bacterium enters the root tissue (but not leaves) and is abundantly present in intercellular spaces. Secondly, the root becomes severely damaged, indicating that – under some conditions – P. polymyxa is a "deleterious bacterium", and in others it promotes growth. Based on work presented in my thesis, I argue that a balance between the activities of a PGPR, the genetic background and physiological state of a plant, and the environmental conditions employed in test systems, ultimately determines the resulting effect.