Exploring a spectroscopic signature for templated biomineralization in modern eukaryotes

Session: New Voices in Geobiology

Presenting Author:

Authors:

Abstract:

Biomineralization is the process of biologically mediated precipitation of minerals onto a macromolecular template. Mineralized structures are critical for the structural support of complex body plans which allow organisms to explore new ecological niches. Biomineralization is found in prokaryotes and eukaryotes and has emerged multiple times independently throughout Life’s evolutionary history. During diagenesis biominerals often experience recrystallization and/or dissolution, leaving behind only the fossilization products of the organic template. Consequently, the challenge of identifying biomineralization in diagenetically altered fossils has obscured the early evolution of biomineralization. To address this issue, we have developed a spectroscopic signature of templated biomineralization applicable to all modern eukaryotes: previous work has identified specific functional groups that are robustly associated with templated biomineralization in modern animals and demonstrated the survival of such informative heterogeneities in deep time. We collected in situ Raman spectral data from n=40 modern samples of algae, protozoans, and multicellular animals with known biomineralization status using a Horiba LabRam 800 with a 532nm excitation (10% laser power at sample surface).  All spectra were baselined, normalized, and denoised in Labspec 5.0, Spectragryph, and Quasar. Outputs were collated into a taxon-character matrix including variables that capture the biomineralization status of a given taxon. Data were subjected to multivariate statistics (Principal Component Analysis [PCA], Canonical Correspondence Analysis [CCA], Linear Discriminant Analysis [LDA]) in PAST to explore to which extent macromolecular template features allow for the distinction of biomineralized and non-biomineralized modern eukaryote samples. PCA separates the clusters of biomineralized and non-biomineralized samples.  In the PCA distinct groupings of non-biomineralized, carbonate-biomineralized, and silica-biomineralized samples emerge. A PCA biplot of the contributing chemical vectors identifies key discriminant features which yielded an importance ranking of chemical template features informing biomineralization; we subsampled these bands and subjected the resulting data set to LDA. For the training data set, LDA yields a posterior predictive accuracy of 90.2% for templated biomineralization.  This prominent signal for templated biomineralization in spectral fingerprints of modern eukaryotes is our starting point for the exploration of preservation kinetics of this signal in deep time. A time-integrative biomineralization signature has the potential to illuminate the early evolution of hard parts as the key innovation required for the emergence of complex body plans.