Dramatic changes in Earth’s climate, especially noticeable in the Arctic, require prompt actions and, thus, a better understanding of the processes involved in those changes. Rapidly disappearing sea ice is a vital part of the planet’s climate, both affecting the system and being subject to climatic feedbacks. However, to fully understand the effects of sea ice and to incorporate them into climate models, a longer record of sea ice is needed. Proxies for sea ice conditions are the only way to achieve this goal. In this dissertation, I explore the limitations and viability of a proxy based on fossilized diatom communities in marine sediments and present the first quantitative diatom-based proxy for sea ice concentration in Beringia.

Possible sources of alteration must be tested to ensure that diatom communities in marine sediments are a reliable archive and can be used as a proxy. I explore the connection between phytoplankton and taphonomic assemblages, proving sediment records to be representative of phytoplankton and not depleted in sea ice diatoms due to preferential dissolution. I also confirm that diatoms respond to sea ice conditions and that this response is detectable in sediment assemblages. One such diatom that responds to extensive sea ice is Sinerima marigela, a new, monotypic genus and species described for the first time in this dissertation. Based on a training set of 104 surface sediment samples from the Bering and Chukchi seas and satellite-derived sea ice concentration, I constructed a proxy for sea ice in Beringia using the Generalized Additive Model fitted to relative abundances of five diatom species: Neodenticula seminae, Paralia sulcata, Fragilariopsis cylindrus, Fragilariopsis reginae-jahniae, and Sinerima marigela. This proxy is available as a web-based application at http://seaiceproxy.geol.iastate.edu. It doesn’t require full diatom counts and scrupulous taxonomic work because the model only uses five easily identifiable species. Users can upload their counts to the R Shiny app and download sea ice reconstructions based on the model. The proxy was then applied to two sediment cores, Healy 0202 51JPC and IODP Site U1345, from the Bering Sea, that produced quantitative reconstructions in agreement with previous qualitative estimates.