| Summary: | International audience Palaeoceanographic studies often rely on microfossil species abundance changes, with littleconsideration for species traits (e.g. size) that could be related to environmental changes. Wehypothesize that whole-assemblage and species-specific planktonic foraminifera (PF) testsize could be good predictors of environmental variables, and we test this using an EquatorialIndian Ocean (EIO) core-top sample set (62 viable samples). We use an automated imagingand sorting system (MiSo) to identify PF species, analyze morphology and quantifyfragmentation using machine learning techniques. Machine accuracy was confirmed bycomparisons with human classifiers. Data for 25 mean annual environmental parameterswere extracted from modern databases and, through Exploratory Factor Analysis andregression models, we investigate the potential of PF size, at the assemblage and specieslevel, for reconstructing oceanographic parameters in the Indian Ocean. Within our tropicaldataset, we find that SST is not a significant driver of assemblage size, although thermoclinedwelling species Globorotalia inflata and Globorotalia truncatulinoides show a significantrelationship with temperature. Our analyses indicate that deep carbonate ion concentrationand core depth may be important factors influencing PF size, especially in species that arelarge-sized or bear calcite crusts such as Globigerinoides conglobatus, Globorotaliamenardii, and Neogloboquadrina dutertrei. We propose that PF population size couldpotentially be useful to reconstruct bottom water carbonate concentrations and sea surfacetemperature. This approach will be tested on a new downcore record from the Arabian sea(ODP Site 722) during key Pleistocene glacial-interglacial transitions, where existing seasurface temperature and other paleo-reconstructions will allow meaningful comparisons.
|
|---|