![[Visit Client Website]](/img/gsa/banner.jpg)
Paper No. 123-11
Presentation Time: 4:00 PM
3D SHAPE EVOLUTION IN GRASS SILICA SHORT CELL PHYTOLITHS: TAXONOMIC PATTERNS IN BAMBUSOIDEAE AND OTHER LINEAGES
Grasses produce distinctive phytoliths which can be diagnostic at various taxonomic levels. Phytoliths are particularly important in paleobotanical work because they can help us reconstruct earth’s past vegetation and climate history. Phytoliths are complex in their 3D structure yet nearly all of the descriptive and classification work thus far has been based on 2D images and the use of categorical and subjective morphotypes. There is currently a lack of quantitative 3D data on phytolith morphology.
Using the Bamboo subfamily as a test case, we developed confocal microscopy and image processing techniques to create 3D surface representations of short cell phytoliths. We used geometric morphometric techniques to quantify surface shape and applied machine learning algorithms as a tool to identify the affinity of phytoliths at various taxonomic levels. Using these techniques we achieved a high degree of correct automatic classification of the phytoliths at sub-familial, tribal and genus level ranks.
The 3D objects produced during this study will contribute to a family wide digital reference collection for use by paleobotanists and grass taxonomists worldwide.