Environmental DNA and subsequent metabarcoding are useful tools for assessing marine fish biodiversity in a non-invasive manner. It is of particular importance to assess biodiversity in regions that are hard to access and thus less well studied. Thus, sampling and preservation methods tailored to the particular circumstances are required. Aquatic eDNA is often captured on filters made of different materials and pore sizes, followed by storage under divergent conditions and times scales. We examined the effects of filter type and storage time on DNA yield, fish diversity and community composition recovered form a marine ecosystem in Shark Bay, Western Australia, using two different filter types, two storage methods and two different fish-specific metabarcodes. Depending on filter type and storage method, our results showed that storage time decreased DNA yield and affected alpha- and beta-diversity estimates. CN filters stored in Longmire’s solution proved to be the best storage method. Glass-fibre filters stored in Longmire's solution led to a decrease in eDNA yield and alpha-diversity estimates with increasing storage time. Furthermore, the largest change in beta-diversity for each metabarcode was found for glass fibre filters regardless of storage method. Our results highlight the importance of considering storage time and interactions between storage medium and filter when analysing eDNA results, especially when storing samples for an extended period of time or comparing samples that have been stored for different periods of time.