ON 30 JUNE, 2013, a team of four – producer Miles Barton, cameraman Hugh Miller, camera assistant and safety diver Kat Brown and me, researcher and safety diver – set out in search of the Amami starry sky pufferfish (Torquigener albomaculosus). At this point, however, the pufferfish had yet to be named, and was barely known to science.
Almost a year earlier a marine-biologist friend had posted me a picture from a design journal showing a remarkably complex structure made from sand, and the small fish that was supposedly responsible.
If true, this would be the most complex structure known to be built by an animal anywhere on the planet.
After reading around the story I was convinced that it was for real, but it was going to take quite a lot to convince my colleagues. I had emailed our team, who were understandably sceptical, but at the same time intrigued.
The story had been published around April Fools’ Day, and it would be embarrassing to be taken in by a joke, so I spent some weeks trying to track down the scientists or the photographer.
It soon became apparent that the sand structures had first been observed off the island of Amami Oshima, which sits halfway between mainland Japan and Okinawa, by macro-photographer divers swimming between sections of coral reef.
They had not expected to see anything on the open sand patches, and had no idea what had made them.
One of those divers was 72-year-old Yoji Ookata. Japan’s version of David Doubilet, he had spent half a century diving and documenting Japan’s coast and seas.
Unfortunately Yoji didn’t speak much English, and my Japanese was non-existent, so I took a punt and paid for some emails to be translated back and forth to get the information I needed.
I had to persuade my colleagues that the story was perfect for the BBC Life Story television documentary series, which would look at the remarkable ways in which animals overcome the challenges of life.

YOJI HAD BEEN OUT with Japan’s natural history TV producer NHK the previous year, trying to document and identify what was responsible for the circles.
His team had pieced the story together through multiple short dives, and we got hold of a copy of the footage, which served as a very useful recce tool and showed the pufferfish at work.
We watched the video, talked to Yoji and started to hatch a plan for filming this little fish building its huge structure.
Our main worries were its depth and size. Everything Yoji and the footage told us was that it was going to need lighting, it was potentially deep, the fish were tiny and we’d need a lot of time in the water to pull it off, not to mention some luck, because breeding season was also cyclone season.
We planned for the worst but hoped for the best. Provided the fish was actually there, depth was going to be the key to making the sequence work. Anything below 21/22m would require us to start deco-diving, and we would need to stay down ideally for two to three hours. Deco wasn’t a problem as such, but would add extra complexity and logistical planning.
Also, the deeper we went the less light there would be for filming, and the more we would need artificial lighting.
With this in mind, Hugh had built a buoyant lighting rig that would be anchored to the seafloor, with six Keldan Video 8s to act as a “little ray of sunshine” should we need it.
We knew that once we found a male pufferfish building or tending a nest he would stay in a fairly similar place.
Once a nest was complete and the eggs had hatched he would start a new nest within 10m or so of the original, so all we had to do was find one and set up an underwater studio set with lights, jib (a small crane to move the camera around) and quadpod (a four-legged stand for under-slinging the camera and getting it close to the seafloor).
Finally, we constructed an A-frame to get the static top-shot that would reveal the structure’s full majesty.

WE HAD GIVEN OURSELVES 23 days to film this story, doing two dives a day with three of us in the water, as this would also provide our “making of” story for the Courtship programme.
We were diving rebreathers, so Hugh and Kat had their Revos and I had my Poseidon MKVI. We had calculated that we’d need 31,500 litres of oxygen (about five 200bar J-cylinders) and 220kg of Sofnolime scrubber. Alongside our camera kit we had close to a ton of equipment and supplies.
Yoji and Toyo had already been on location for a few days and had located an active nest. The scientific paper outlined seven days of construction with mating on the seventh night and then another seven days of incubation before hatching on the full moon. We had planned to arrive at the start of the cycle, but a cyclone had thrown the fish off track, so the timings were all out.
This turned out to be a running theme for the shoot. After the first nest was abandoned there was a day on which nothing happened, and the fish was nowhere to be seen. Like clockwork, however, he was back the next day and had started to construct a new nest.

THE GREAT THING about being able to stay under water for three-plus hours was that we were able to observe in a single dive what had taken multiple dives before.
It turns out that the pufferfish is a prolific builder. If required he can knock out a fairly impressive sandcastle in a few hours, but with more time comes more intricacy and successful mating, so ideally he would get a full cycle.
The seabed was of beige sand with few features and visibility was 20m or more but it was still rather disorientating, so we’d run a line and buoy out to where the male fish was constructing his sandcastle.
We had already lowered the quadpod, lighting rig and jib to the seabed a few days before to test it, and now we walked it out to the nest site.
It was easier to be a bit negative and take our fins off, walking the gear to the filming site 60m away rather than swimming, and it was quite a sight to see people wearing full-face masks with comms walking along the seabed like divers from some old sci-fi film.
Hugh and Kat assembled and tweaked the kit – with all the Allen keys and bolts it looked like IKEA furniture assembly but at 15m.
For three days we filmed the pufferfish constructing his circle at some speed. Dives were averaging three hours, and we’d get in two per day.
By the end of the third day, the middle of the circle was looking to be nearing completion. The fish had made fine wavy lines in the middle of the circle, the finishing touch before females would consider him a suitor.
At the end of the third day of filming, females had started to appear, and everything looked good for spawning the next morning. We were in the water shortly after dawn, but the circle looked completely different.
The wavy lines had gone, the male was no longer building and the structure was slightly degraded, yet the females were there and spawning!
We think the male mated with two or three females that day, and by 11am all the activity had dropped off. He was now tending the eggs and turning the substrate to oxygenate them.
It would be another seven days before he would start a new nest. We put this time aside for shooting some of the wide scenic shots and nice details that help to establish the location, as well as trying to get the super-big close-ups.
We had hoped to get three goes at filming the construction and courtship, but now we looked likely to get only two.

SO FAR WE HAD FILMED the behaviour of building the nest, cleaning the weeds, decorating the edges and grading the sand, but we hadn’t managed to film the big top-shot and reveal it in all its glory. The behaviour was amazing, but it was finally seeing the sandcastle from above that would amaze the audience.
Unfortunately, while waiting for the next building cycle we were also watching a cyclone develop and start to track in our direction.
It hit five days later, and with building wind and swell we retreated to the safety of shore, and the pufferfish retreated to the safety of the depths.
There seemed no point in us hanging around either. The swell meant that bottom currents and wave action were pushing all the sand around. Any attempt to build would have been in vein.
After the cyclone passed it took another few days for the swell to drop and visibility to improve. Yojo, Toyo and I made recce dives every day, but either couldn’t find any fish or else saw fish too deep and dark to match the existing film. All we could do was wait.
Time was running out, and it was apparent that we might not see this incredible spectacle ever again. We were leaving soon, and the decision was made to retrieve the kit and pack up ready to head home.
Having spent so much time focused on one small patch of sand, we hadn’t really looked around. Resigned to the fact that we wouldn’t see the fish again, we decided to go for a dive before recovering the kit.
Hugh and I swam off, and the first thing we found was an enormous conch shell that had been vacated some time before. It was the size of my head and resembled a morion, a Spanish conquistador’s helmet, and it was only right that we model it.
Putting it back after our fun, we continued looking around. Then Hugh stopped, looking into the distance, and as I settled alongside him I saw to my disbelief a small male starry sky pufferfish in the very early stages of building a new nest, with only two days to spawning.
It seemed inconceivable that he would be ready, but he seemed to be trying his hardest.
The topside team had been listening in to our conversation, so were aware that something exciting was happening.
Our dive to recover the kit had now turned into our last chance of getting that top-shot. Back at the boat, we explained that there was a fish at 20m building a nest, and that we needed to get back in.
Hugh and I headed back with the camera, but it was set up with a wide lens. The fish hadn’t done much work yet, so we decided that at lunchtime we would change the lens to get close-ups. We figured we had at least a day or two before the nest would be completed.
How wrong we were! The thing with newly discovered animals is that they never obey the scientific paper written about them.
When we got back, the male had finished the circle, all the furrows were ploughed and the wavy lines were complete. Females were lurking.
While we’d been away, the fish had done seven days’ work! We had missed the top-shot again.

WE CAME BACK the following morning to retrieve the kit, as we had a flight to catch and it was the end of the shoot. But it seemed that the male pufferfish hadn’t got lucky. Though the nest was degraded he was still there, and when we arrived at the nest he had started to retill the sand, tidying up and decorating it.
Hugh got the camera on the top of the A-frame, and we waited for the perfect conditions. We dived for more than 4.5 hours on that last dive, walking up and down the slope and managing our bottom time to extend our filming time. This was a classic example of “on the last day the team got what they came for”.
The pufferfish have been observed only on Amami Oshima island and in a couple of bays. I believe they may spend a lot of time in deeper, darker water, because they have very large eyes for their size, a common feature of deeper-water animals.
The fish are beige and very well camouflaged, so I think to find a mate the strategy is to come shallower where there is light, and construct a large “target”. But how they came to be able to build such an elaborate structure is still unknown.
Filming the starry sky pufferfish will go down as one of the best, if not the best, experience of my life so far, and I’d love to go back again.