A sea otter does not need to dive or flee for a drone flight to matter. It may lift its head, interrupt a rest or become more alert—small changes that are easy to miss through a live camera feed.

A field study published on 15 May 2026 gives those subtle reactions unusual weight. Researchers working at two sites on California’s Monterey Peninsula found that southern sea otters showed higher behavioural-response scores when a quadcopter was present than during a no-drone baseline. The scores generally rose as the aircraft descended. The finding is not a universal “safe altitude” for wildlife photography. It is evidence that distance, duration and context matter before an animal makes the dramatic movement a pilot might recognise as distress.

What the researchers actually tested

The study, published in Marine Mammal Science and catalogued by the US Geological Survey, covered 37 observation sessions at Cannery Row and Otter Point. Across the 2020 pilot work and the larger 2022 field season, the team observed 92 focal sea otters and recorded 3,529 maximum-behaviour observations.

Each session began with a 20-minute baseline period without a drone. Researchers then conducted three 20-minute flight trials, separated by battery-change pauses. A DJI Mavic 2 Pro or Mavic 3 Pro hovered above the focal animals while descending in 10-foot steps from 150 feet to 60 feet above sea level, then ascending through the same range. Experienced observers recorded behaviour once per minute on a seven-category scale running from resting to a full flush.

The design matters because the researchers did not simply decide in real time whether an otter “looked bothered.” They recorded defined behaviour codes and analysed the pattern afterwards. During the first flight trial, the median of each focal animal’s mean maximum-behaviour score was 2.53, compared with 1.33 during baseline observations. The statistical analysis found a measurable effect of drone presence, and modelled responses increased as the drone moved lower. Full methods, limitations and results are available in the open study.

Lower was worse, but altitude was not the only variable

The simple version is that closer flights produced stronger reactions. The more useful version is that the response varied with place, group size, pups, gull activity, flight sequence and the direction of drone movement.

At the less disturbed Otter Point site, behaviour generally became more reactive as the drone descended and eased as it moved away. At busier Cannery Row, responses could remain elevated while the aircraft ascended. Otters in larger groups tended to show smaller changes than isolated animals, but the authors explicitly warned against turning that pattern into a suggestion to target large rafts: doing so would expose more animals to the stimulus.

The researchers also suspended drone work for six weeks when nesting gulls became agitated and aggressive toward the aircraft. That operational decision is an important part of the result. A protocol that appears tolerable for one species can still disturb another species sharing the same airspace.

This is not a public safe-distance rule

The experiment used two closely related consumer quadcopters, two locations and a defined flight profile. Many focal animals were not individually marked, so researchers could not always know whether the same otter was observed on multiple days. Observers also could not be blinded to the drone’s activity because the pilot and observation team had to coordinate the aircraft over the focal animal.

Those limits do not erase the measured effect. They do prevent a responsible reader from declaring that 150 feet is automatically safe, that 60 feet is the only point of concern, or that the result transfers unchanged to seals, birds or terrestrial mammals. Drone size, sound, approach angle, background activity, breeding state, weather and previous exposure can all change the encounter.

There is another reason not to use visible flight as the sole test. A separate study of free-ranging American black bears found strong heart-rate responses to UAV flights even though movement changes were infrequent; the largest recorded increase was 123 beats per minute above the pre-flight baseline. That 2015 Current Biology study concerns a different species and cannot supply a sea-otter threshold. It does show why “the animal stayed there” is weak evidence of no effect.

Airspace permission and wildlife permission answer different questions

A flight can satisfy aviation rules and still create a wildlife problem. NOAA Fisheries’ current marine-life guidance says protected marine species must not be harassed and advises observers to keep a safe, respectful distance. It also says national drone guidance for marine mammals and sea turtles is still being developed, while regional, state and species-specific rules can differ.

Researchers face an additional layer. NOAA says scientific work directed at protected marine mammals or sea turtles below 400 feet requires appropriate permits and authorisations. California also restricts drone operations on wildlife-department lands and imposes specific aircraft limits over named protected areas, including the California Sea Otter Game Refuge. Those are US and California examples, not global rules. Pilots elsewhere need to check the wildlife, land-access and aviation authorities that apply to the exact site.

A better pre-flight question than “will it fly away?”

For recreational photographers and filmmakers, the practical lesson is not to memorise one number from a research protocol. It is to build a decision around uncertainty:

• Check the place and the species. A legal airspace map may not show wildlife refuge rules, local launch restrictions, seasonal closures or protected-species obligations.
• Avoid making the animal the flight target. Repeated hovering, vertical descent and efforts to provoke a look toward the camera turn observation into pressure.
• Use distance and time as protective margins. Higher and shorter operations reduced risk in the sea-otter study, but neither factor guarantees zero disturbance.
• Watch the whole scene. Nesting birds, pups and other animals may react before the subject framed in the camera does.
• Leave early. Increased vigilance, interrupted resting or feeding, coordinated movement, repeated head turns, aggression toward the aircraft or movement away are reasons to stop—not cues to follow for a better shot.
• Do not repeat a pass to “confirm” the reaction. The second test is another exposure, not a neutral measurement.

This approach also protects the quality of the footage. An image of wildlife behaving naturally is more truthful than a close-up created by changing the behaviour being documented.

Drones remain valuable conservation tools

The study is not an argument to remove drones from wildlife science. Aerial systems can reach difficult areas, support population surveys and help locate animals during an oil-spill response. The researchers’ aim was to make those uses less disruptive, not to deny their value.

That distinction matters. “Drones can disturb wildlife” and “drones can help wildlife research” can both be true. The responsible question is whether the information gained justifies the exposure, whether the work is permitted, and whether the flight plan uses the least intrusive method available.

The Monterey study makes one point especially hard to ignore: disturbance can begin before spectacle. Waiting for an animal to bolt is waiting for the clearest signal, not the earliest one.