Whole-genome metabolic modeling can provide valuable insights to how species interact in microbial communities.  Here, I report on the use of phenotype simulations in the DOE Systems Biology Knowledgebase (KBase) environment, combined with experimental growth/no growth data, to improve individual species models being used for a hot lake bacterial community modeling project at Pacific Northwest National Laboratory. We compare the growth predictions to experimental evidence to provide guidance as to which missing reactions can be safely added to improve a species model or otherwise used for further investigation on model improvement, and show how an iterative improvement process could be employed.