On the journey back from the neighbouring red eddy (the warm one) which we had visited over the previous weekend, we anxiously watched the path of the 2 buoys in the patch: buoy # 4, around which we had re-fertilized a week ago, and buoy # 1a, which was one of the twins we had re-deployed at the last in-station, when the patch was at the north-eastern periphery of the eddy. The other twin, buoy # 2, around which we had carried out the first fertilization, had disengaged itself from the patch weeks ago and was now closer to the eddy core than the patch. All three buoys had been moving south-eastward along parallel arcs over the weekend with buoy # 1a along the outermost track. With furrowed brows and gloomy looks we gathered round the computers in the user room, updating each new position of the buoys, read from the computer with internet access, onto the laptop which displayed their tracks relative to the borders of the eddy’s current location. The ship’s track and the underway measurements of temperature and salinity were on a third computer screen and the FRRF values appeared on a fourth. Then suddenly, our faces lit up as the new positions indicated that buoy # 4 was turning slightly south-west, i.e. veering out of the band of strong currents encircling the blue and red eddies – the “highway to hell” – and back into our eddy’s influence. There was hope that our patch would be retained. When the FRRF values shot up after we crossed its path, indicating that the patch we had just entered was on the other side and within the eddy, we smiled and shook hands.

We again spent the night mapping the position of the patch and found to our relief that buoy # 1A had left it, so, when some hours later the unfaithful buoy again veered to the south, we were no longer concerned and in the following days watched it “go to hell” around the periphery of the red one, to be followed a few days later by buoy # 4. After mapping the patch, we carried out our in-station at the site where chlorophyll concentrations were highest. There were only few pteropods (sea butterflies) here, unlike the swarm we had encountered during the last in-station of the previous week. Clearly there was spatial heterogeneity within the elongated patch, so we decided to assess its extent with a Scanfish survey crisscrossing it, followed by two transects – one north-south and the other east-west - with short stations comprising CTD dips that sampled only the upper 200 m and a single multinet tow from 200 m depth to the surface.

The Scanfish is a sleek instrument, carrying the same sensors as the CTD, which undulates from surface to 200 m depth while towed behind the ship. We had brought it along to map the patch but the many growlers (small pieces of icebergs) had deterred us from using it previously, except for one survey just after the first fertilization. However, this time there were no icebergs around, so after completing the in-station, we started the Scanfish survey followed by the south to north transect. A short, sharp storm had been forecasted and indeed it arrived almost suddenly before we reached the northern boundary of the patch; luckily, the last station indicated that we were close to it. Although wind speeds were predicted to be very high, their duration would be short, so there was no need to leave the eddy. Polarstern was buffeted by winds of force 10 (over 100 km hour) for 8 hours with gusts reaching to force 11 but she is a sturdy ship and, with her nose in the wind proceeding forward at slow speed, we weathered the storm during a night when nobody got much sleep. The wind subsided by morning so we returned to the patch and, after the obligate re-mapping, marked the site with the highest chlorophyll concentrations with our remaining buoy # 5 and completed the east-west transect when wave heights had declined enough to resume station work. Another long in-station was taken over the weekend close to buoy # 5 which had not moved much since deployment. As usual, free-floating sediment traps were deployed and recovered inside and outside the patch at intervals of a few days in between the stations.

The results of the transects showed that our patch was indeed fairly homogeneous within its boundaries, which were sharply delineated to the south and along its sides, but was trailing a long tail of diluted water to the north. The pteropod swarm was not encountered again but the VPN images (the camera system attached to the CTD) showed that there were about twice as many copepods (Calanus simillimus) per square metre inside the patch compared to outside. Even more striking were the high densities of their main predators – the amphipod Themisto gaudichaudii – we had found at 3 previous in-stations: their numbers ranged from 30 to 138 specimens per square metre. Imagine that many 2 – 3 cm large beetles living off tens of thousands of 2 mm long aphids (the equivalent of C. simillimus densities), themselves feeding on the plants on a square metre of garden and you have an impression of the biomass of larger zooplankton maintained by our patch. At our latitude the amount of sunlight available to the plants is the same as in a cloudy August in northern Europe, so, although the plankton are distributed over a deep water column, the analogy is justified.

We found several species of amphipods in the RMT tows but Themisto was by far the most abundant. They are tough, active animals, equipped with a range of grappling, sharply hooked legs in the front and paddle-like legs on their abdomen with which they scurry about in the buckets in which the net catches are emptied. It is a voracious carnivore that evidently feeds on other large zooplankton such as salps, chaetognaths (arrow worms) but also the local euphausiids (cousins of the better known krill of the south). Small groups even attacked the tiny fishes (myctophids) in the net catches, reducing them to skeletons in tens of minutes. In feeding experiments on board they also captured and ate copepods. They are visual predators with large compound lenses on top of the head (the translucent “caps” in the picture) indicating that they hunt by looking for prey above them, silhouetted against the weak light of the night sky. The black spots on the sides of the head in the picture are actually the retinas below the lenses. In the virtual absence of fishes (we caught very few mesopelagic fishes in the night net tows), amphipods are the only visual predators on plankton in this stretch of ocean, so the amazing transparency of their potential prey - from copepods to salps and chaetognaths – is witness to their predation pressure and their acute eye sight.

Amphipod numbers increased within the patch in the weeks following fertilization and by the middle of the experiment there were ten times more inside it than outside. They probably roam about in swarms and presumably entered our patch from the sides and stayed within it because of the higher copepod (C. simillimus) density. Previous experiments have also found higher copepod densities within iron-fertilized patches compared to outside. Since copepods, unlike amphipods and krill, are too small to swim into the patch horizontally, they are believed to congregate within it by adjusting their patterns of daily vertical migration. They feed on plankton in the surface layer only during the night and descend to a depth of around 100 m (demonstrated by the VPN images) where they spend the day, presumably invisible to visual predators. Their light sensors (they do not have lenses, so cannot see images) enable them to determine the appropriate depth, which depends on the depth of light penetration, which in turn depends on the amount of light-absorbing particles, particularly phytoplankton, in the surface layer. If surface and deeper layers move at different speeds then copepods spending the day at deeper depths, i.e. under a more transparent surface layer will be transported away from it. If they encounter a surface layer with more plankton, they will stay higher up during the day and hence add to the population already there. At this stage we cannot judge whether the VPN data are representative but will have to wait until the copepods in the net catches are counted and the relative movements of surface and deeper water layers in the patch have been analysed. So we cannot yet say what attracted the amphipods to the patch.

Our blue eddy is located on the western flank of an undersea mountain range known as the Mid-Atlantic Ridge which extends through the middle of the Atlantic Ocean from about our latitude in the South to Iceland in the North. The sea floor under our eddy rises from 4200 m to only 3200 m depths from its southern to its north-eastern boundaries. Since we did not know what to expect from our bloom, we had taken cores of the surface sediment with a Multicorer from a shallow (3,360 m) and a deep (4,183 m) site during the first week of the experiment, in order to find out whether the previous bloom had left behind a layer of fluff on the sea floor. Although we expected the sediment composition to be different, we were surprised by its magnitude: at the shallow site the sediments were largely calcium carbonate (coccoliths and foram shells) with almost no diatom shells but at the deeper site, which was below the carbonate dissolution depth, the sediments consisted mainly of diatom shells as elsewhere under the deep ACC. We also found many small black stones, both of basaltic origin (magma) and encrusted nodules, lying on the surface at both sites. There were hardly any signs of freshly sedimented diatoms (fluff) at either of the sites. During this week we took our third Multicorer sample from 3758 m depth. The sediment cores were quite similar to those from the previous deep station.

We are now nearing the end of this arduous cruise but our spirits are high because we are enjoying what we are doing. One hears a lot of laughter in the labs and corridors, even at the evening meetings where the different groups present their results “hot off the instruments”. The presentations are applauded enthusiastically as we are intrigued by the processes apparently going on in the eddy, both inside and outside the patch. The initial disappointment when our bloom stopped increasing biomass has given way to the dawning of new insights on how iron-limited systems dominated by zooplankton recycle iron. Ours is a truly interdisciplinary endeavour without a hint of rivalry, and cooperation has tightened as we have come to know each other better and appreciate the individual contributions to unravelling the story of the LOHAFEX patch. The crew of Polarstern are as superb as ever.

Our only companions are the majestic wandering albatrosses, effortlessly soaring over the tossing waves, that have come to accept us as part of the seascape of the eddy. Possibly, they consider us a peculiar looking iceberg. The several other bird species regularly around us are less conspicuous and ignore our presence. We have been visited during stations by a few lonely seals or penguins, generally juveniles who appeared to be lost. They were excited by our presence and evinced eagerness to come on board. At each visit, word quickly spreads and everybody goes rushing out from their labs with their cameras, striving to get as many pictures as possible. We were sorry that we could not offer them hospitality. That whales did not make an appearance (except for 2 sei whales early on in the experiment, and one or two others glimpsed in the distance) is the only disappointment harboured by most of us.

With our very best wishes from the Roaring Forties,

Wajih Naqvi and Victor Smetacek