The first thing you notice at ninety metres is how quiet the colour is. Red disappeared seventy metres ago, orange not long after, and what is left is a blue so complete it stops reading as a colour at all. Your torch beam puts the reef back for a moment — a shock of yellow, a fan of something impossibly pink — and then you move it and the wall returns to blue.
We were there because of an argument. For years the working assumption in [REGION] was that the mesophotic zone — roughly forty to a hundred and fifty metres, the twilight band below recreational diving — was a marginal habitat. Too dark, too cold, too food-poor to matter much. Nobody had really looked, which is a different thing from nobody having found anything.
Six weeks and a hundred and forty dives later, that assumption is difficult to hold. The wall between sixty and ninety metres carried more live coral cover than any shallow site we surveyed, and it carried it on a reef that had been through the same two bleaching events as the flats above.
What we actually did
The method was deliberately unglamorous. Closed-circuit rebreathers, fixed twenty-five metre transects at five depth intervals, photo-quadrats every two metres, and an ROV running the same lines afterwards so we could check what the divers had missed. Each dive bought us about eighteen minutes on the bottom and cost around two and a half hours of decompression.
- Five depth bands: 20 m, 40 m, 60 m, 75 m and 90 m, replicated across nine sites.
- Photo-quadrats scored blind by two analysts, with a third resolving disagreements.
- Temperature and light loggers left in place for a full annual cycle.
- Tissue samples from six coral genera present across the full depth range.
That last point is where the interesting question lives. If the same coral species occur from the shallows down to ninety metres, and the deep populations are healthier, are they simply surviving — or are they resupplying the reef above them? The deep reef refuge hypothesis has been around since the 1990s. It has been much easier to state than to test.
“We are not short of theories about how reefs recover. We are short of places where somebody has gone and looked at the same rock, at the same depth, for long enough to say.”
The parts that do not make the paper
Field seasons have a second, unpublished dataset: the human one. We lost nine days to weather. The compressor failed twice. One of our team dislocated a shoulder in a swell on day eleven and spent the rest of the trip running surface logistics with one arm and considerable grace.
Where this goes next
The genetic work is running now and will take most of a year. If the deep and shallow populations turn out to be exchanging larvae, the management implications are immediate: a protected area drawn at thirty metres is protecting the wrong part of the system. If they turn out to be isolated, that is arguably more urgent, because it means the deep reef is not a spare copy of anything.
Either way, we are going back. The loggers come up in March, and the wall will have had a year to tell us something we could not see in six weeks. Read more about how this fits the wider programme on the career timeline.
