Caenorhabditis elegans has frequently been isolated from decaying plants and fungi, suggesting that plant-pathogenic and fungal-pathogenic bacteria form a major component of its natural diet. As a result, the bacteria and the worms may have experienced significant selection, either in the bacteria to reduce predation, or in the worms to increase their feeding efficiency. Plant-pathogenic bacteria have been much studied because of their agricultural and economic importance. We have examined a large panel of such bacteria for possible pathogenic or other effects on the growth of C. elegans. Over 80 plant or fungal pathovars of Pseudomonas, Xanthomonas, Erwinia and Acinetobacter species have been examined so far, by growing worms (wild type and various immune-compromised mutants) on lawns of each bacterial type. As food sources for C. elegans, the strains can be grouped into four broad categories: first: growth significantly better than on E. coli OP50; second: growth comparable to that on OP50; third: growth significantly worse than on OP50; fourth: little or no growth, with toxic or pathogenic effects. Strains in the first category demonstrate that OP50 is not an optimal food source for C. elegans, despite its convenience in a laboratory context. Bacterial preference assays reveal complex behavior in C. elegans with respect to taxis towards good and bad food sources. A variety of deleterious effects have been observed for strains in the fourth category, some of which have been examined in detail. The NZI7 strain of P. syringae is both repellent to C. elegans and fails to support growth of worms, although non-growing L1/L2 larvae can continue to swim through a lawn of bacteria for days. We have screened a transposon-insertion library of NZI7 for derived strains that can be eaten by C. elegans, and have isolated and characterized a number of palatable single-gene insertion mutants, which are now good food sources. The P. fluorescens strain Pf-5 has severely toxic effects on C. elegans, probably acting by more than one mechanism. This strain produces a diffusible factor which can kill eggs, as well as having other deleterious effects. Toxicity of this and other strains is not correlated with cyanide production, which is responsible for worm killing by P. aeruginosa PAO1 (Gallagher and Manoil, 2001). Another candidate for Pf-5 toxicity is the
mcf2 gene, homologous to the insecticidal mcf gene of Photorhabdus, but a deletion mutant of Pf-5
mcf2 was unaltered in toxicity.