Summary: We used light and scanning electron microscope analyses to quantify morphometric features (valve length, width, stria density, lineola density and valve curvature) from the observation of valves representing Seminavis pusilla. Cluster analysis based on Gaussian mixture models and the expectation-maximization algorithm was used for delineating two species, Seminavis pusilla sensu stricto and Seminavis lata (Krammer) Rioual comb. et stat. nov. By comparison with S.pusilla, S. lata is characterized by wider valves and lower stria density. The two species have also markedly different ecology. S.pusilla is most abundant in the most saline lakes of the dataset, while S. lata is most abundant in the less saline lakes. Our results indicate that combining the two species into S.pusilla sensu lato would lead to a loss of ecological information and a decrease of the performance of transfer functions developed for quantitative reconstruction of past salinity from fossil diatom assemblages in sediment cores.
Bibliographical notePublisher Copyright:
© 2014 Japanese Society of Phycology.
- Badain Jaran Desert
- Expectation-Maximization algorithm
- Model-based clustering
- Morphometric analysis
- Seminavis lata comb. et stat. nov.
- Seminavis pusilla