Arjun Chennu

Modern marine surveys generate tens of thousands of underwater images per expedition, yet many are unusable due to image quality deterioration. Manual quality screening is labour‐intensive, subjective and inconsistent across observers and teams, hindering reproducible science, creating large backlogs of unprocessed imagery, and limiting real‐time algorithmic and robotic applications. We present UNIQAT (UNderwater Image Quality Assessment Toolkit), an open‐source Python toolkit for automated underwater image quality assessment. The toolkit computes reference‐free quality metrics across nine categories of image characteristics relevant to underwater images, such as colour cast, turbidity and contrast loss. Metrics can be combined with user‐defined weights into task‐specific quality scores tailored to different ecological applications. High‐performance computing (HPC)‐aware multiprocessing and optional deep learning architectures for metric approximation support processing at scales matching modern survey data collection. We evaluated UNIQAT on 130,845 images from nine coral reef surveys conducted by the Australian Institute of Marine Science; every result reported in the main text was produced by the traditional metrics pipeline, with no deep learning model involved, and a separate training demonstration is reported in the . The traditional metrics pipeline, run as 48 CPU workers on a 64‐core cluster node, processed 7.9 4K‐images/s, with processing time scaling linearly with image count ( r ² = 0.98). The resulting quality scores captured expected patterns of image modulation based on the environmental context of the imaged scenes: All nine surveys showed negative depth–quality correlations ( r = −0.16 to −0.89), consistent with the depth‐dependent deterioration of image quality in underwater environments. Cross‐survey metric profiles were highly consistent (mean Spearman ρ = 0.99 across all 36 pairwise comparisons, minimum 0.96), confirming measurement stability across different dates, locations, camera systems and environmental conditions. Synthesis. UNIQAT can help deliver objective and tunable image quality metrics in a versatile toolkit that enables transparent reporting and supports unsupervised quality assessment for applications in marine monitoring and robotics.

The regulation of O₂ production by cyanobacteria is critical to understanding the co-evolution of oxygenic photosynthesis (OP) and Earth’s redox landscape. This includes their response to electron donors for competitive anoxygenic photosynthesis, like arsenic. In this work, we assessed the effect of arsenic cycling on photosynthetic activity in a modern cyanobacterial mat thriving beneath an arsenate-rich (∼9-10 µM) water column in the high-altitude central Andes, using biogeochemical and omics approaches. Microsensor measurements and hyperspectral imaging revealed two O₂-producing cyanobacterial layers. During the afternoon, OP ceased in the lower, Chl f-dominated layer. Ex-situ measurements revealed that the combination of high light and arsenic induced a prominent rise in local H₂O₂ concentration, which then coincided with the interruption of OP. Mat incubations suggested that after OP ceased, cyanobacteria transitioned to As(III)-driven anoxygenic photosynthesis using far-red light. Additional incubations and metatranscriptomics on in-situ samples reveal simultaneous As(V) reduction during the day, at sufficiently high rates to supply electron donors for anoxygenic photosynthesis. This study proposes that As(III) can serve as an alternative electron donor for cyanobacterial photosynthesis. It also reveals the crucial role of arsenic in moderating OP, a consequence of the interaction between arsenic and reactive oxygen species under high irradiance. Given the widespread abundance of arsenic in the Precambrian, we further discuss how this regulatory mechanism could have played an important role in the early evolution of OP.

Mangrove forests provide valuable ecosystem services to coastal communities across tropical and subtropical regions. Current anthropogenic stressors threaten these ecosystems and urge researchers to create improved monitoring methods for better environmental management. Recent efforts that have focused on automatically quantifying the above-ground biomass using image analysis have found some success on high resolution imagery of mangrove forests that have sparse vegetation. In this study, we focus on stands of mangrove forests with dense vegetation consisting of the endemic Pelliciera rhizophorae and the more widespread Rhizophora mangle mangrove species located in the remote Utría National Park in the Colombian Pacific coast. Our developed workflow used consumer-grade Unoccupied Aerial System (UAS) imagery of the mangrove forests, from which large orthophoto mosaics and digital surface models are built. We apply convolutional neural networks (CNNs) for instance segmentation to accurately delineate (33% instance average precision) individual tree canopies for the Pelliciera rhizophorae species. We also apply CNNs for semantic segmentation to accurately identify (97% precision and 87% recall) the area coverage of the Rhizophora mangle mangrove tree species as well as the area coverage of surrounding mud and water land-cover classes. We provide a novel algorithm for merging predicted instance segmentation tiles of trees to recover tree shapes and sizes in overlapping border regions of tiles. Using the automatically segmented ground areas we interpolate their height from the digital surface model to generate a digital elevation model, significantly reducing the effort for ground pixel selection. Finally, we calculate a canopy height model from the digital surface and elevation models and combine it with the inventory of Pelliciera rhizophorae trees to derive the height of each individual mangrove tree. The resulting inventory of a mangrove forest, with individual P. rhizophorae tree height information, as well as crown shape and size descriptions, enables the use of allometric equations to calculate important monitoring metrics, such as above-ground biomass and carbon stocks.

We present a new approach combining diffusive equilibrium in thin-film gels and spectrophotometric methods to determine the spatial distribution of arsenite, arsenate, and phosphate at submillimeter resolution. The method relies on the simultaneous deployment of three gel probes. Each retrieved gel is exposed to malachite green reagent gels differing in acidity and oxidant addition, leading to green coloration dependent on analyte speciation and concentration. Hyperspectral images of the gels enable mapping the three analytes in the 2.5-20 μM range. This method was applied in a contaminated brook in the Harz mountains, Germany, together with established mapping of dissolved iron. The use of two-dimensional (2D) gel probes was compared to traditional porewater extraction. The gels revealed banded porewater patterns on a mm-scale, which were undetectable using traditional methods. Small-scale correlation analyses of arsenic and iron microstructures in the gels suggested active iron-driven local redox cycling of arsenic. Overall, the results indicate continuous net release of arsenic from contaminant particles and deepen our understanding of arsenate transformation under anaerobic conditions. This study is the first fine-scale 2D characterization of arsenic speciation in porewater and represents a crucial step toward understanding the transfer and redox cycling of arsenic in contaminated sediment/soil ecosystems.

Coral reefs are the most biodiverse marine ecosystems, and host a wide range of taxonomic diversity in a complex spatial community structure. Existing coral reef survey methods struggle to accurately capture the taxonomic detail within the complex spatial structure of benthic communities. We propose a workflow to leverage underwater hyperspectral image transects and two machine learning algorithms to produce dense habitat maps of 1150 m² of reefs across the Curaçao coastline. Our multi‐method workflow labelled all 500+ million pixels with one of 43 classes at taxonomic family, genus or species level for corals, algae, sponges, or to substrate labels such as sediment, turf algae and cyanobacterial mats. With low annotation effort (only 2% of pixels) and no external data, our workflow enables accurate (Fbeta of 87%) survey‐scale mapping, with unprecedented thematic detail and with fine spatial resolution (2.5 cm/pixel). Our assessments of the composition and configuration of the benthic communities of 23 image transects showed high consistency. Digitizing the reef habitat and community structure enables validation and novel analysis of pattern and scale in coral reef ecology. Our dense habitat maps reveal the inadequacies of point sampling methods to accurately describe reef benthic communities.

Coral reefs are the most biodiverse marine ecosystems, and host a wide range of taxonomic diversity in a complex spatial habitat structure. Existing coral reef survey methods struggle to accurately capture the taxonomic detail within the complex spatial structure of benthic communities. We propose a workflow to leverage underwater hyperspectral transects and two machine learning algorithms to produce dense habitat maps of 1150 m ² of reefs across the Curaçao coastline. Our multi-method workflow labelled all 500+ million pixels with one of 43 classes at taxonomic family, genus or species level for corals, algae, sponges, or to substrate labels such as sediment, turf algae and cyanobacterial mats. With low annotation effort (2% pixels) and no external data, our workflow enables accurate (Fbeta 87%) survey-scale mapping, with unprecedented thematic and spatial detail. Our assessments of the composition and configuration of the benthic communities of 23 transect showed high consistency. Digitizing the reef habitat structure enables validation and novel analysis of pattern and scale in coral reef ecology. Our dense habitat maps reveal the inadequacies of point sampling methods to accurately describe reef benthic communities.

We developed and used a microsensor to measure fast (2O₂) on the polyp tissue of two scleractinian coral species (Stylophora pistillata and Pocillopora damicornis) under manipulations of illumination, photosynthesis, and feeding activity. Our real-time tracking of H₂O₂ concentrations on the coral tissue revealed rapid changes with peaks of up to 60 μM. We observed bursts of H₂O₂ release, lasting seconds to minutes, with rapid increase and decrease of surficial H₂O₂ levels at rates up to 15 μM s^–1. We found that the H₂O₂ levels on the polyp surface are enhanced by oxygenic photosynthesis and feeding, whereas H₂O₂ bursts occurred randomly, independently from photosynthesis. Feeding resulted in a threefold increase of baseline H₂O₂ levels and was accompanied by H₂O₂ bursts, suggesting that the coral host is the source of the bursts. Our study reveals that H₂O₂ levels at the surface of coral polyps are much higher and more dynamic than previously reported, and that bursts are a regular feature of the H₂O₂ dynamics in the coral holobiont.