Modeling elementary chemical reactions in ocean ﬂuid dynamics simulations requires signiﬁcant computing resources, which can be diminished with model reduction techniques. Submesoscale ocean turbulence and biogeochemical reactions in the ocean occur on approximately the same time scale, 105 seconds. This similarity in time scales indicates a strong coupling between these two processes and the need to model the two together when performing large eddy simulations (LES). Existing biogeochemical models that are capable of accurately modeling ocean biogeochemistry are currently too large, from a computational standpoint, to couple with complex submesoscale turbulence models in LES. This highlights the need for a way to reduce the size of the biogeochemical models while still maintaining high model ﬁdelity. In this study, model reduction methods from the ﬁeld of combustion have successfully been used to reduce the size of a biogeochemical model for the ﬁrst time. Previously, researchers have used unstructured and unveriﬁed approaches of removing model components to reduce the size of biogeochemical models. This research involved developing a modiﬁed version of the directed relation graph with error propagation method and applying it to the 50-species Biogeochemical Flux model (BFM). For the reduction of the BFM, 24 di˙erent scenarios were considered, which resulted in four di˙erent reduced models with between 1 to 44 species remaining. Each reduced model is only valid for the scenario considered for the reduction. This work will allow for future investigation of the interactions between submesoscale ocean turbulence and biogeochemical processes.