Detecting signatures of life on terrestrial and Martian rocks: contribution of microbial mats in the biogeomorphological responses of desiccated sediments

Detecting signatures of life in sedimentary rocks lies in the difficulty of distinguishing them from abiotic signals and interpreting their formational conditions, particularly when working on planetary systems that are different from Earth, such as Mars (Corenblit et al., 2023). Research in this field is booming, thanks to the development and deployment of detection tools either in orbit or on the surface. Mars is of great interest due to its early history comparable to Earth during the Noachian period > 3.7 Ga (Lapôtre, 2022). In addition, traces of favourable environmental conditions for the potential development of life have been found for this period, for example in Gale Crater (Rapin et al., 2023). Among the candidates for searching potential signatures of life, Microbially Induced Sedimentary Structures (MISS, Nora Noffke in 1996) have become a target. MISS are characteristic structures resulting from surface sediment disturbances induced by microbial mats (Schieber et al., 2007; Noffke, 2010). Their formational environments may correlate with early Mars conditions, and their terrestrial study is enriched by their representation in both fossil and modern records (Noffke 2015, 2021). The analogy between two planetary systems relates to the principle of abductive inference, which posits that similar (bio)geomorphological processes will result in similar (bio)geomorphological structures (Corenblit et al., 2019). Therefore, it is crucial to develop a clear conceptual framework for processing observations of modern and fossilized textures, forms, and patterns and for discussing the gradient of distinction between abiotic and biotic modalities (Davies et al., 2016).

Here, we focused on one type of MISS known as “mat cracks”, the biotic equivalent of abiotic structures “mud cracks” (Noffke, 2010). These are well-represented in the field in both fossil and modern records, and they are robustly repeatable under controlled laboratory conditions. They may correspond to ancient Martian environmental systems as attested by polygonal ridges in Gale Crater, which are characteristic of sustained wet/dry cycles (Rapin et al., 2023). The methodology is based on the visual distinction of biotic and abiotic classes of texture, form, and pattern using different visualisation methods such as photogrammetry and expert visual observations, statistical tools and classification with convolutional neural networks (CNNs). For an initial exploration of the mud crack variability, we set up an ex-situ experiment to produce mud cracks with three types of biofilms and three biomass levels according to variables observed in the field, and using 3D picture dataset of the resulting mud cracks. We have demonstrated significant differences between abiotic and biotic classes and between strain and biomass classes. CNN models outperformed the human-blinded classification by refining the diversity of criteria used and observations such as the textures of the sandy matrix. These significant distinctions and the finesse of the classification provided by artificial intelligence allow us to discuss the interest of the information gain in distinguishing potential textures, forms and patterns that are characteristic of MISS in the field where noise, alteration and erosion can be a problem in identifying the origin of signatures, particularly on Mars.