CAN COMPRESSION ALGORITHMS HELP DIFFERENTIATE ABIOTIC STRUCTURES FROM BIOTIC STRUCTURES?

Differentiate biologic from abiologic structures based solely on morphology is problematic and constitutes a troubling factor in the search for life, whether on earth or elsewhere (e.g., Mars). Quantifying the information within a digital image using compression algorithms has the potential to help address the biogenicity of a sample by quantifying the amount of randomness in an image. We analyzed two structures that share morphologically similar features (radiating carbonate crystal fans) but have vastly different origins: 1) a putative abiotic crystal-fan stromatolite; and 2) dinosaur and extant bird eggshells. We produced digital images from a crystal-fan stromatolite and eggshell thin-sections magnified to 10X. We then digitally cut similar rectangular samples from the digital images (n=229), saved them as TIFF files, and compressed each file using gzip (a UNIX based lossless compression program). For each sample, the percentage loss was subtracted from the original file size thus generating a compression percentage. After, we compared compression percentages of the abiotic and biotic examples using a Student’s t-Test and Mann-Whitney one-way analysis of variance. We deduce, from compressing thin-section digital images, that although their morphologies are similar, abiotic samples compress less than biotic samples. At this scale of analysis, we interpret this to indicate that the abiotic samples record more random influence than the biotic samples. In addition to statistically comparing these abiotic and biotic examples, we tested whether or not our results were dependent on sample size. Samples were selected based on a random number generator and we compared the examples as stated above. Results indicate that regardless of sample size, abiotic samples compress less than biotic samples. However, compression variance trends seem to break down depending on sample size suggesting that 20 samples is the minimum samples needed to characterize compression standards. These results provide a promising, easily assessable and implemented technique that has the potential to assess the biologic origin of structures.