Abstract The ocean is experiencing unprecedented rapid change, and visually monitoring marine biota at the spatiotemporal scales needed for responsible stewardship is a formidable task. As baselines are sought by the research community, the volume and rate of this required data collection rapidly outpaces our abilities to process and analyze them. Recent advances in machine learning enables fast, sophisticated analysis of visual data, but have had limited success in the ocean due to lack of data standardization, insufficient formatting, and demand for large, labeled datasets. To address this need, we built FathomNet, an open-source image database that standardizes and aggregates expertly curated labeled data. FathomNet has been seeded with existing iconic and non-iconic imagery of marine animals, underwater equipment, debris, and other concepts, and allows for future contributions from distributed data sources. We demonstrate how FathomNet data can be used to train and deploy models on other institutional video to reduce annotation effort, and enable automated tracking of underwater concepts when integrated with robotic vehicles. As FathomNet continues to grow and incorporate more labeled data from the community, we can accelerate the processing of visual data to achieve a healthy and sustainable global ocean. 

/pdf/fathomnet-a-global-image-database-for-enabling-artificial-2zpt1cci.pdf

FathomNet: A global image database for enabling artificial intelligence in the ocean

https://besjournals.onlinelibrary.wiley.com/doi/pdf/10.1111/2041-210X.13617

A cost-effective video system for a rapid appraisal of deep-sea benthic habitats: The Azor drift-cam

Significance Symbolic marking is a hallmark of human cultures, but quantitative evidence for nonhuman animal cultures is comparatively limited. We show evidence that certain acoustic communication signals—“identity codas”—function as symbolic markers of cultural identity among Pacific Ocean sperm whale clans. Within clans, identity and nonidentity coda usage shows similar patterns of stability over geographic space, but between-clan similarity in identity coda usage decreases with increasing spatial overlap. The distribution of sperm whale clans—with multiple cultural groups that vary widely in ranging, are unevenly distributed, and frequently overlap along a latitudinal gradient—resembles that of human ethnolinguistic groups. These findings demonstrate how culture can structure and fortify segmentation of animal populations and reveal similarities between human and nonhuman cultures.

Evidence from sperm whale clans of symbolic marking in non-human cultures

The Salas y Gomez and Nazca ridges are underwater mountain chains that stretch across 2,900 km in the southeastern Pacific and are recognized for their high biodiversity value and unique ecological characteristics. Explorations of deep-water ecosystems have been limited in this region, and elsewhere globally. To characterize community composition of mesophotic and deep-sea demersal fauna at seamounts in the region, we conducted expeditions to Rapa Nui (RN) and Salas y Gomez (SyG) islands in 2011 and Desventuradas Islands in 2013. Remote autonomous baited-cameras were used to conduct stationary video surveys between 150-1,850 m at RN/SyG (N = 20) and 75-2,363 m at Desventuradas (N = 27). Individual organisms were identified to the lowest possible taxonomic level and relative abundance was quantified with the maximum number of individuals per frame. Deployments were attributed with associated environmental variables (temperature, salinity, dissolved oxygen, nitrate, silicate, phosphate, chlorophyll-a, seamount age, and bathymetric position index [BPI]). We identified 55 unique invertebrate taxa and 66 unique fish taxa. Faunal community structure was highly dissimilar between and within subregions both for invertebrate (p < 0.001) and fish taxa (p = 0.022). For fishes, dogfish sharks (Squalidae) accounted for the greatest dissimilarity between subregions (18.27%), with mean abundances of 2.26 ± 2.49 at Desventuradas, an order of magnitude greater than at RN/SyG (0.21 ± 0.54). Depth, seamount age, broad-scale BPI, and nitrate explained most of the variation in both invertebrate (R2 = 0.475) and fish (R2 = 0.419) assemblages. Slightly more than half the deployments at Desventuradas (N = 14) recorded vulnerable marine ecosystem taxa such as corals and sponges. Our study supports mounting evidence that the Salas y Gomez and Nazca ridges are areas of high biodiversity and high conservation value. While Chile and Peru have recently established or proposed marine protected areas in this region, the majority of these ridges lie outside of national jurisdictions and are under threat from overfishing, plastic pollution, climate change, and potential deep-sea mining. Given its intrinsic value, this region should be comprehensively protected using the best available conservation measures to ensure that the Salas y Gomez and Nazca ridges remain a globally unique biodiversity hotspot.

/pdf/deep-sea-biodiversity-at-the-extremes-of-the-salas-y-gomez-2l64vvdtmd.pdf

Deep-sea biodiversity at the extremes of the Salas y Gómez and Nazca ridges with implications for conservation.

Role of deep-sea equipment in promoting the forefront of studies on life in extreme environments.

There is a growing need for marine biodiversity baseline and monitoring data to assess ocean ecosystem health, especially in the deep sea, where data are notoriously sparse. Baited cameras are a biological observing method especially useful in the deep ocean to estimate relative abundances of scavenging fishes and invertebrates. The National Geographic Society Exploration Technology Lab developed an autonomous benthic lander platform with a baited camera system to conduct stationary video surveys of deep-sea megafauna. The first-generation landers were capable of sampling to full ocean depth, however the form factor, power requirements, and cost of the system limited deployment opportunities. Therefore, a miniaturized version (76 cm x 76 cm x 36 cm, 18 kg in air) was developed to provide a cost-effective method to observe ocean life to 6000 m depth. Here, we detail this next-generation deep-sea camera system, including the structural design, scientific payload, and the procedures for deployment. We provide an overview of NGS deep-sea camera system deployments over the past decade with a focus on the performance improvements of the next-generation system, which began field operations in 2017 and have performed 264 deployments. We present example imagery and discuss the strengths and limitations of the instrument in the context of existing complementary survey methods, and for use in down-stream data products. The key operational advantages of this new instrument are spatial flexibility and cost-efficiency. The instrument can be hand-deployed by a single operator from a small craft concurrent with other shipboard operations. The main limitation of the system is battery power, which allows for six hours of continuous recording, and takes up to eight hours to recharge between deployments. Like all baited-camera methods, this instrument is specialized to measure the relative abundance of mobile megafauna that are attracted to bait, which results in a stochastic snapshot of the species at the deployment location and time. The small size and ease of deployment of this next-generation camera system allows for increased sample replication on expeditions, and presents a path forward to advance cost-effective biological observing and sustained monitoring in the deep ocean.

/pdf/the-national-geographic-society-deep-sea-camera-system-a-low-3cn3s9u4yu.pdf

The National Geographic Society Deep-Sea Camera System: A Low-Cost Remote Video Survey Instrument to Advance Biodiversity Observation in the Deep Ocean

http://www.cell.com/article/S2589004222006642/pdf

Denley Delaney

Papers

The National Geographic Society Deep-Sea Camera System: A Low-Cost Remote Video Survey Instrument to Advance Biodiversity Observation in the Deep Ocean

Toward understanding the communication in sperm whales