Ecological modeling of the temperature dependence of cryptic species of planktonic Foraminifera in the Southern Hemisphere

• Ecological modeling of cryptic species of four planktonic Foraminifera
• Simulation of core-top counts dataset in Southern Ocean
• Calibration of three transfer functions: IKTF, WA-PLS and MAT
• Significant decrease of SST estimates error rate in transfer functions

Cryptic genetic species of planktonic Foraminifera often exhibit narrower biogeographic distributions and ecological preferences than the respective morphospecies. In theory, it should therefore be possible to improve the resolution of the paleoceanographic reconstructions based on sediment assemblages of these species. Here, we use observational data on the latitudinal distribution of 11 genotypes of Globigerina bulloides (including a newly described genotype), Orbulina universa, Truncorotalia truncatulinoides and Globoconella inflata in plankton tows from south Indian Ocean to model the relationship between their abundance and the sea surface temperature (SST). We then use this model to assess the potential benefit of knowing the ecological preferences of cryptic species on assemblage-based transfer functions. In doing so, we first apply this model to a database of assemblage counts in 1334 surface sediment samples from the Southern Hemisphere and simulate the expected abundances of individual genotypes in sediment samples. This simulated dataset is used to calibrate three different transfer functions: the Imbrie and Kipp Method, Weighted Averaging Partial Least Squares and Modern Analog Technique. Trials show that such simulated splitting of morphospecies into their respective genotypes leads to a substantial (7 to 25%) overall reduction of the error rates of SST estimates in the calibration dataset. The degree of error rate reduction is sensitive to the increase of taxonomic richness in the simulated assemblages induced by the co-occurrence of genotypes of the same morphospecies. Although the studied species occur across a broad SST range, the largest reduction of error rate by the transfer functions is obtained within the 4 °C to 12 °C SST range, where most of the studied genotypes are found. Our results show that all genotypes are not expected to improve the accuracy of transfer functions at the same level: while integrating cryptic diversity in G. bulloides, T. truncatulinoides and G. inflata clearly improve assemblage-based SST reconstructions, genotypes of O. universa have a negligible effect.