Early Online Release

Oceanography in the Age of Intelligent Robots and a Changing Climate By Chris Scholin

Early Online Release | Oceanography

all too familiar fashion, it did not take long to learn that human

activities are a source of ocean soundscape pollution that can be

injurious to marine wildlife (e.g.,  Hildebrand, 2009). Although

the notion of an ocean soundscape is ancient, and its use in ocean

studies has long been the subject of intensive research and devel­

opment, we continue to make remarkable discoveries by simply

listening with increasingly sophisticated means for doing so.

Today, detailed observations of the comings and goings of

marine animals is greatly enhanced by soundscape analysis

(e.g., Oestreich et al., 2022, 2024; Ryan et al., 2022, 2025). The com­

bination of passive and active acoustic observations has proven

useful in investigating predator foraging behaviors and the ecology

of fear (e.g., Benoit-Bird et al., 2019; Urmy and Benoit-Bird, 2021).

“Listening with light” by way of using fiber-optic cables as vibra­

tion sensors—a technique known as distributed acoustic sens­

ing (DAS)—is the latest evolution in the ongoing push to broaden

access to and analysis of the ocean soundscape (Saw et al., 2025).

By combining fleets of ASVs and AUVs equipped with acoustic,

imaging, and water sampling payloads, a new perspective on the

movements of animals traversing the environment in response to

ever changing ocean conditions is emerging, including by tracking

the traces of “genetic soup” shed in their wakes (e.g., Zhang et al.,

2021b; Figure 7).

The use of organisms’ DNA and RNA (and other methods) to

reveal what species are present and how they are responding to

their environments has advanced in concert with the develop­

ment and application of ocean imaging and acoustics. The tools

and techniques employed have storied pasts and spring from the

creativity and insights of many investigators over decades. What

has come to be known as “ecogenomics” is deeply rooted in sub­

cellular biological studies and molecular analytical methods for

detecting and decoding the very essence of life itself. Just as under­

water imaging and soundscape analysis grew from industrial uses

and for purposes unrelated to ocean ecology, molecular biology,

nucleic acid sequencing, bioinformatics, and other methods unre­

lated to marine science were adopted for ocean applications, for­

ever altering the course of modern marine biology. Microbial ecol­

ogists arguably led the way (e.g., Pace, 1985; Karl, 2014).

In a surprising twist, Ficetola et al. (2008) discovered that DNA

shed by frogs could be detected in the environment in which

they lived even when you could not see the animals themselves,

sparking an environmental DNA (eDNA) forensics revolution

(e.g., Kelly et al., 2014; Stoeckle et al., 2024). The analysis of eDNA

offers a noninvasive method for assessing biodiversity and track­

ing animal movements by collecting samples of water and sequenc­

ing the recovered material, enabling simultaneous detection of

marine organisms across multiple trophic levels (e.g., Chavez et al.,

2021b). As eDNA analysis has evolved, our eyes have been opened

to the notion of “genetic dark matter” that is recoverable from the

environment but has no described source or, in some cases, no

well-characterized function (e.g., Venter et al., 2004; Roux et al.,

2015; Delmont et al., 2022). Analysis of the sea’s genetic soup tells

us that there is a great deal of marine life and genetic capacity that

has not yet been characterized.

Just as machine learning and artificial intelligence have played

a huge role in analyzing and reacting to ocean imagery and sound,

they are likewise fueling the analysis of eDNA to synthesize an

integrated picture of a complex web of life. Although the detec­

tion and real-time analysis of imagery, sound, and other bulk water

properties are now commonly employed to guide autonomous

platforms during targeted field observations, devices that enable

in situ, “hands off,” real-time analysis of eDNA and other cellu­

lar metabolites are still very much in their infancy (e.g., Scholin

et al., 2017). With a few notable examples (e.g., Truelove et al.,

2019; Peter Thielen et al., Johns Hopkins University, pers. comm.,

2025), marine eDNA surveys rest largely on the acquisition, pres­

ervation, and return of samples for shoreside analysis (Yamahara

et  al., 2019; Zhang et  al., 2021a; Truelove et  al., 2022; Preston

et  al., 2023). Despite the progress, scaling up the use of robots

that enable integrated optical, acoustic, and “omic” characteriza­

tion of the sea presents a very significant technological challenge

when compared to using profiling floats to conduct global scale

biogeochemical observations.

TO THE SEAFLOOR

Descending to the seafloor, whether using a crewed submers­

ible or an ROV, has been likened to being dropped into a pitch-

black room and using only a flashlight to see what lies ahead.

Remarkable discoveries have been made by picking dive sites that

are known to offer different types of terrain that might lead to find­

ing something novel. The discovery of the “octopus garden” near

the base of Davidson Seamount offers an excellent, recent exam­

ple of using ship-acquired bathymetry to guide an exploratory

ROV dive that serendipitously uncovered something remark­

able (King and Brown, 2019). No doubt that method works, but

the area that can be covered is limited, and for the most part, you

have no detailed map to lead the way. AUVs are changing that

FIGURE 7. A fleet of long-range autonomous underwater vehicles (AUVs;

Hobson et al., 2012) fitted with different imaging, water sampling, and eDNA

collection payloads are lined up alongside a Liquid Robotics Wave Glider, all

readied for deployment in Monterey Bay. The fleet of vehicles allows coor­

dinated observations for extended periods to provide a multifaceted view of

dynamic ecosystem processes. After Zhang et al. (2021a,b). Susan von Thun

© 2017 MBARI