Evidence of biogenic alteration in sub-oceanic basalt glass: complexity image analysis, elemental abundance distributions, and Bayesian probabilistic classification

Abstract

Microbial activity in subsurface basalts produces characteristic alteration in the volcanic glass readily observed in petrographic microphotographs 1 . Basalts containing volcanic glass cored from three different sites beneath the ocean floor along a mid-ocean ridge exhibiting such alteration was determined to also exhibit co-registered regions of elemental abundance alteration. 2 Further investigation one of these sites presenting examples of all three states (unaltered glass and glass altered to produce clays by abiotic and biotic processes) now reveals that lossless compression analysis 3 of the RGB photomicrographs detects significant differences between altered and unaltered basalt glass. The technique can distinguish images of all three groups without knowledge of the opinion of the expert human observer. Electron microprobe elemental abundance measurements were obtained from sites in all three regions. Principal Component Analysis using eleven elements as input (Na 11 , Mg 12 , Al 13 , Si 14 , P 15 , Cl 17 , K 19 , Ca 20 , Ti 22 , Mn 25 , and Fe 26 ) extracted two factors accounting for more than 75% of the data variance. When these same elements were used as inputs to a Hierarchical Cluster Analysis algorithm to provide unsupervised sample classification all samples clustered as predicted both by human visual observation and complexity analysis prediction. Finally, the PCA components extracted were used as inputs to a stochastic, non-linear and Artificial Neural Network to produce a Bayesian probability of correct classification. The technique provides a quantitative methodology for discrimination of biotic and abiotic alteration of sub-ocean basalt glass using both structural and chemical information.