Active biomonitoring approaches are now recognized as relevant for monitoring water contamination and toxicity. Nevertheless, due to the confounding influence of variable and uncontrolled environmental conditions such as temperature, biological markers measured on transplanted individuals to assess water quality are difficult to interpret. The purpose of the present study is to propose a methodology for adapting a laboratory test of chronic sublethal toxicity based on the molting cycle of Gammarus fossarum to in situ assays. To this end, we 1) adapted the molt cycle temperature-dependent model developed in Part 1 (Chaumot et al. 2020, this issue) to the fluctuating temperatures measured in the field; 2) assessed the predictive power of our approach as a "reference value" from gammarids caged in 9 nonimpacted sites at different seasons; and 3) tested the relevance of our tool to interpret in situ reproductive bioassays from 5 upstream/downstream studies and a large-scale deployment in 12 sites. Our approach based on modeling the progress of gammarid molting cycle as a function of temperature appeared to be a relevant and robust tool for interpreting in situ observations in different environmental contexts in time and space. By avoiding using a "reference" or upstream situation as a baseline from which water quality could be assessed, this approach provides a real added value to water quality diagnosis in biomonitoring programs. Environ Toxicol Chem 2020;00:1-14. (c) 2019 SETAC