After completing the second round of fertilization last Monday we left the patch searching for a large iceberg in order to study the effects of its melt water on the surroundings. We had encountered extensive areas influenced by melt water further to the south, but these had been giant icebergs several or more kilometres long. We soon spotted one of the larger ones in our eddy on the radar screen and headed straight for it. The sea was calm and the sun was peeping out occasionally, so groups of rested scientists were sauntering around on deck, eager to view the iceberg. It turned out to be a compact chunk of ice: 60 m above water (measured with a sextant) and a few hundred metres in length and breadth. It was shaped like a squat sphinx, its flanks polished by the waves that surged up its sides. On its head were balanced huge chunks of ice like an oversized crown. It did not appear stable so we kept a prudent distance from it while taking CTD casts around it, looking for signs of melt water. Many of us were still watching it, fascinated by the way it changed shape and colour from pale blue to gleaming white, with many shades of grey in between. Suddenly a shout went up as pieces of the crown started tumbling down. Then before our startled eyes, the entire front started crumbling and collapsing like an avalanche, pushing a field of ice rubble into the sea. Big chunks toppled over and fell in the waves with huge splashes. Within minutes the iceberg was reduced to a vestige of its former towering presence. Although at a safe distance, we were stunned by the spectacle. One of us had caught the entire sequence on film so we could all witness later the fall of a stately iceberg.

The results of the iceberg survey were disappointing: there was no discernible signal of melt water at any of the corners we sampled indicating that the discharge rate, in relation to the rate of mixing around it, was negligible. Apparently these smaller icebergs leave behind only diffused signals that are accordingly less easy to map than the extensive fields of ice rubble ensuing from collapsed giant bergs. It is possible that iron released from the many smaller icebergs was responsible for the higher level of background chlorophyll in the eddy core as compared to the impoverished water further to the east and north. As these levels have been maintained since our arrival, the iron is either being continuously replenished by collapsing icebergs and dispersal of rubble, or the fertilized planktonic community is efficiently recycling the iron in a quasi-steady state in the surface layer. As we shall see later, the developments in our patch provide support for the latter explanation.

After the iceberg foray, we proceeded to the next out-station, giving our patch time to mix the new streaks of iron. In the meantime the drifting buoy around which we had re-fertilized, continued moving due northward, and we breathed a sigh of relief when, by mid-week, it was finally deflected to the northeast. That it was now disturbingly close to the band of swift currents sweeping around our eddy was demonstrated by its path and speed during the next few days. Since we were not sure whether the buoy had slipped out of the patch, as the first two buoys – the twins – had somehow accomplished, we decided to retrieve one of them, since it lay on our way to the patch, and use it to mark the site of the next in-station.

After a night spent mapping the new position of our patch along the north-eastern periphery of our eddy, we were troubled to find that it extended even further into the band of swift currents separating our eddy from the adjacent red one to the east, than the buoy did. We selected the most promising site in the patch, deployed the buoy we had recovered, which we now called 1a, and carried out the in-station while continuously correcting the ship’s position to keep up with the buoy, by now travelling southeast at the breathtaking speed of over 20 kilometres per day. Clearly the patch was moving with the band of strong currents we dubbed “the highway to hell” because it would drag the patch out of our blue eddy and drape it around the outer rim of the red eddy as a thin strip that would soon merge with its surroundings. We went about our work with a sense of impending doom. During the week we also collected and deployed neutrally buoyant sediment traps inside and outside the patch. The cups contained more material now than before, but the total amounts collected were small.

Despite the re-fertilization, chlorophyll concentrations over most of the patch stayed more or less constant at double the outside values. Dilution due to lateral mixing with surrounding, unfertilized water was not the reason, because, over most of its border, the patch was sharply demarcated: in sections crossing it during mapping at night time it appeared as a steep-sided plateau of dots extending for many miles in FRRF values recorded every half-minute. No doubt, mixing with surrounding water was taking place while the patch was gyrating within the eddy, but not fast enough to obscure continuous build-up in phytoplankton biomass reflected in the chlorophyll concentrations. Besides, a steady increase over the first weeks was observed in all previous experiments, so the plankton community in our patch was behaving in a fundamentally different manner, hence we were making new discoveries.

Admittedly, our observations were not unexpected because it has long been appreciated that planktonic communities fall into two broad categories: phytoplankton blooms and recycling or regenerating systems. Generally, blooms occur when nutrients are abundant and last from one to eight weeks depending on temperature and light supply. They are characterised by high chlorophyll levels that can fluctuate as different species replace each other until an essential nutrient becomes limiting to further growth. Blooms tend to be dominated by diatoms which require silicic acid, so their biomass build-up is terminated when this element reaches limiting concentrations, i.e. shells can no longer be made. This is generally accompanied by mass sinking and transport of organic matter to the deep water column and sea floor which represents a major source of food to the organisms living there.

In some regions both silicic acid and nitrogen are depleted simultaneously with thinner-shelled diatom species replacing the thicker-shelled ones as the critical resource dwindles. However, in many coastal and shelf regions, silicic acid is exhausted well before nitrate, the remainder of which is then utilized by non-diatom algae, in particular chalk algae and colonies of Phaeocystis of the haptophytes, or Ceratium of the dinoflagellates mentioned in previous reports. Although quite abundant, none of these groups “made it” in our bloom, the former presumably due to low temperatures and the latter two in all likelihood because they were kept in check by grazing pressure. So our bloom entered the next phase despite abundant nutrients. This is a crucial finding because it indicates that only diatoms regularly manage to make blooms in the Antarctic Circumpolar Current. We shall return to the implications of this important point later.

Recycling or regenerating systems maintain a steady state for months based on the balance between production and breakdown of organic carbon. Daily rates of chlorophyll-based primary production, expressed as the amount of organic material produced by photosynthesis under a square meter sea surface depend on the rate of supply of the limiting element from remineralisation of organic matter by respiration of all the non-photosynthesizing organisms collectively known as heterotrophs (as opposed to autotrophs) and ranging from bacteria to whales. Clearly our patch had entered a recycling state but what was surprising was that it appeared as stably balanced as the recycling systems around it, although the amount being recycled was at least twice as large. Since nutrients including iron were apparently not limiting further biomass build-up, it must be the rate of grazing which was imposing the upper level at which this system was operating. What we were not sure about was whether more grazers had accumulated in the patch (by migration or reproduction) or whether those present were eating twice as much as those outside. There was evidence for both scenarios.

First, a comparison of all net catches indicated that abundances of the large, non-reproducing  copepod (Calanus simillimus) tended to be higher inside the patch presumably because they were attracted to it. How such small animals manage to congregate within it will be dealt with later.  Second, the other, much smaller copepod (Oithona similis) was vigorously reproducing in the in-stations, although the numbers of larvae in the samples seemed to vary considerably, possibly due to predation by the large copepod. Third, at the last in-patch station we discovered a large stock of a planktonic snail – the pteropod Limacina – known to have very high reproduction rates. It is only a few millimetres in diameter but looks like the familiar garden snail. Its foot is modified into wing-like appendages with which it flaps though the water like a bat in slow motion, which has earned it the name of sea butterfly. Since they were present in low numbers in most catches throughout the region, this particular swarm, comprising both egg-bearing adults and juveniles, indicated that they had responded to the improved food supply but only in this previously un-sampled corner of the patch. Possibly, their numbers were kept in check by the roving swarms of predatory amphipods that seemed to be increasing in the region of the patch we had been sampling so far, but were less abundant here.

The other alternative, that the dominant, albeit non-reproducing copepods inside the patch were eating more, was supported by observations of freshly caught animals. They are quite transparent, so it is easy to see how much they have eaten. This can then be measured by counting the number of faecal pellets produced by freshly caught individuals maintained in a beaker. Indeed, distinctly fuller guts and higher pellet production rates of copepods from inside the patch were recorded, indicating that feeding rates of the animals outside were limited by the food supply. Clearly, the zooplankton was profiting from the iron fertilization.

At the end of the week we carried out a quick transect with CTD casts through the warmer red eddy, since we were close to it, to compare its physical structure with that of the colder blue eddy. There were no icebergs here and chlorophyll concentrations also lower. A brief glance at the plankton community of its closed core revealed that another species dominated the large copepods, Oithona was rare and the composition of its predatory zooplankton, comprising Chaetognaths (arrow worms) rather than amphipods, differed markedly; overall biomass was substantially lower. Unfortunately, we were not able to carry out a parallel experiment here, so were left guessing as to how this particular, silicon-limited system might have responded.

Our best wishes from a ship full of scientists, now fully adapted to life in the roaring forties, fervently hoping to gather more information from our garden before it is finally torn apart by the currents gnawing at its sides,

Wajih Naqvi and Victor Smetacek