Genetically engineered microorganisms (GEMs) released into the environ
ment may persist and spread, depending on their features and condition
s encountered. In streams, the extent of dispersion depends largely on
cycles of attachment to, and detachment from, biofilms, because distr
ibution of microorganisms is limited only by stream flow and settling
rates, and because biofilms are the primary generator of bacterial cel
ls. To simulate dissemination of introduced bacteria, multiple antibio
tic-resistant bacteria (Chryseobacterium (Flavobacterium) indologenes)
were introduced into microcosms containing water, sediments, and leav
es. Marked bacteria reached greatest abundances in sediments, and cont
ributions of bacteria from sediments to other habitats was relatively
low. Bacterial attachment and detachment occurred rapidly, but the abi
lity of marked bacteria to successfully exploit receiving habitats was
comparatively low. Current speed influenced bacterial dis semination.
A mechanistic model, using mortality and attachment/detachment rates,
determined experimentally, was developed to predict bacterial exchang
es in nature. The model was predictive of experimental results when on
ly 5% of bacteria in sediments were available for detachment. Based on
model results, an introduced bacterial strain, with mortality rates c
omparable to those of the model strain, is predicted to maintain highe
st abundances in sediments. However, within a month, abundance was pre
dicted to be reduced by 98%; long-term persistence is possible if thes
e low population sizes can be sustained.