6. Variability, Instability, and Human Impact

Judging by the main pacemaker responsible for the temporal recurrence of monsoonal maxima and consequent sapropel formation, we don’t need to worry. It appears that the next catastrophe is some 11,000 years away! Although this is a very short period within the context of the geological depth of time, it is an eternity in human terms – especially when compared with sub-decadal election-driven political time scales. However, do we still feel so comfortable when we give the phenomenon some deeper thought?

The primary lesson to be learned from the repeated episodes of sapropel formation is that the Mediterranean circulation is not as robust at it might seem: some very unpleasant surprises might develop rapidly and irrevocably. Add to this the fact that, despite all the published hypotheses and models, we do not yet fully understand to what extent the present-day conditions in the basin may be considered stable, or not. In fact, the present-day conditions have already been found to show considerable sensitivity to interannual climatic extremes and anthropogenic perturbations, leading to very rapid large-scale changes in the temperature and salinity (and hence density) structure of the entire Mediterranean Sea.

Instability certainly seems to have been a characterisitic of the sapropel mode of circulation. Since 1993, targeted studies of sapropels in high resolution (continuous 0.5 or 1.0 cm sampling) have demonstrated serious variability in the climatic forcing and consequent ventilation conditions of the basin. Various sapropels, and especially the youngest one that formed between 9,500 and 6,000 years BP, have been found to contain periods of improved oxygenation that span several centuries

Besides these ‘interruptions’, the high-resolution studies have also highlighted several major spatial gradients and contrasts. The full spectrum of temporal and spatial variability considerably complicates the schematic account of sapropel formation given above, and has highlighted some essential processes and sensitivities involved.

A good example concerns an observation made already in the end 1980s to early 1990s, and which was confirmed by the recent studies: the upper depth limit for the youngest sapropel resided around 300 to 400 meters depth in the open eastern Mediterranean – marking the top of the anoxic conditions – while in the Aegean Sea it reached levels as shallow as 120 meters. From this, it was concluded that the intermediate water circulation, which ventilated the main basin down to 300/400 meters depth, failed to penetrate the Aegean Sea. The numerical model discussed in chapter 4 confirmed that the sapropel-time circulation of Adriatic Intermediate Water would be excluded from the Aegean Sea for dynamical reasons in conjunction with the southernmost Aegean’s complex topography. Apparently, the Aegean Sea is more susceptible to a ventilation collapse than the rest of the eastern Mediterranean, while the oxygen starvation is likely to reach much shallower levels in the water column. The latter enhances the potential threat to marine resource exploitation.

Another notable difference concerns the timing of the onset of anoxic conditions. In the open eastern Mediterranean and the Aegean Sea, these conditions started some 1,000 years before they developed in the Adriatic Sea. This is not totally unexpected, since the Adriatic’s continuing role in subsurface ventilation would favour a considerably better oxygen recharge of its subsurface waters than elsewhere. One could infer from this that the Adriatic is the ‘safest’ place with regards to development of anoxia. However, that conclusion would be unwarranted when directed towards the incidence of anoxic conditions in the Mediterranean today. Vast tracts of the sea floor on the extensive North Adriatic shelf today become anoxic in summer, due to the anthropogenically enhanced nutrient discharge from the Po river into a poorly-mixed warm shallow water mass.

The modern North Adriatic anoxia serve as a red flag: further research is needed to constrain whether the portrayed mechanism for basin-wide anoxia is unique, or whether there would be other ways to accomplish similar conditions – perhaps as a function of enormously enhanced nutrient concentrations. After all, a strongly increased biological oxygen demand (BOD) in subsurface waters could theoretically deplete all deep-sea oxygen even if some deep ventilation continued.

To evaluate possible alternative mechanisms, it is important that we use a variety of different sapropels from the recent geological past. To date, there has been great emphasis on the youngest sapropel that formed during the current interglacial period, and on the sapropel named ‘S5’, which was deposited at the height of the previous interglacial (about 125,000 years ago). Only a few very basic pilot studies have been undertaken that focussed on the two exceptional sapropels S6 (172,000 years BP) and S8 (217,000 years BP) that formed during high-amplitude precession minima at times when the world was gripped in a global glaciation state (ice age). However, having been deposited during times with fundamentally different (glacial) climate conditions from today, there is a good chance that these sapropels – if any – hold the key to a fundamentally different mechanism of sapropel formation. Unfortunately, there are so few quality cores through these glacial sapropels that we don’t even know to what depth their oxygen starvation extended in the basin, nor whether they developed at all in the Adriatic and Aegean Seas.

In summary, the development of a more thorough understanding of all of the basin’s major response modes to climate change requires that we identify and exploit the ‘natural experiments’ that have taken place during the recent geological past. Work on the intriguing glacial sapropels is badly needed to the same standard of the ongoing work on interglacial ones. These goals cannot be achieved without an intensive continuation of coring-expeditions in the Mediterranean.

* * *

With the advent of an almost standard high-resolution sampling and investigation approach through the Mediterranean sapropels, much attention has been focussed on the onset and ending of sapropel formation, and on the short-lived perturbations of the sapropel mode of circulation (note 13). These specific targets are selected because of their potential to elucidate: (a) the dominant processes involved in the basin’s response to rapid climatic/environmental change; and (b) the rapidity, magnitude, and duration of these responses. In addition, investigations of records covering the entire current interglacial period (called the ‘Holocene’ and spanning the last 11,000 years) provide insight into the recurrence of the various perturbations and responses through time. Viewing the sapropels within the context of  such longer-term developments will help in the development of better arguments for validation of model simulations.The ultimate aim is to build such a knowledge of the natural variability that we can make educated estimates of its most likely development into the near future. This would set the stage for a better appreciation of anthropogenic effects.

So what have we discovered to date about natural variability during the Holocene, including the sapropel deposited between 9,500 and 6,000 years ago? Most of it relates to cooling events and their impact on deep-water formation rates (note 14). In 1998, a first publication appeared that reported evidence of exceptional localised cooling over the Gulf of Lions (western Mediterranean) during times of intensive North Atlantic cold events within the last glacial cycle (the so-called ‘Heinrich events’). Also in later studies, only part of this cooling could be related to a drop in the temperature of Atlantic inflow. The most likely explanation for the fact that excess cooling anomalies were confined to the Gulf of Lions invokes increases in the frequency and/or intensity of Mistral-type polar air outbreaks through the Rhone Valley.

No cores of sufficient length have been recovered from the Adriatic and Aegean Seas to evaluate whether these basins were similarly affected by these extreme events. However, studies since 1997 have reported minor cooling events that appeared quasi-periodically throughout the western Mediterranean and in the Adriatic and Aegean Seas during the current interglacial period (Holocene). Recent work has related these minor cooling events to the climatic record from Greenland ice, confirming a link to high-latitude climatic shifts. It is now thought that these minor events were also related to winter-time intensifications in the frequency and/or intensity of cold outbreaks over the northern sectors of the Mediterranean. The events are estimated to have been characterised by 2 to 4ºC winter coolings in the Aegean and Adriatic Seas.
 

Figure 12. Schematic representation of the potassium ion record in the Greenland ice core from the GISP2 project, courtesy of American glaciologist Paul Mayewski. The dusty/cold phases in this record match up well with sea surface temperature drops (winter coolings) in the southern Aegean Sea (stylised representation). The stable oxygen isotope record of the same Aegean core clearly shows the ‘wet anomaly’ that defines the monsoonal maximum, as corroborated by the indicated interval characterised by a comprehensive lake in the now bone-dry Oyo depression of NW Sudan (the tapering end indicates the dessication phase). Click on thumbnail for full-sized jpeg image (or here for a pdf).
 

The most conspicuous Holocene event occurred around 8,000 years ago, from roughly 8,600 to 7,800 years BP. Coincident with its surface cooling in the Aegean and Adriatic Seas, an interruption is observed of the (sapropel) anoxic sea-floor conditions in those basins. The (partial) re-oxygenation allowed an immediate benthic foraminiferal repopulation, almost exclusively by opportunistic species that have the best life-strategy to rapidly invade a ‘sterilised’ environment when conditions improve. As the cooling came to an end, the sea-floor became anoxic once again. Conceptually, it was proposed that increased surface cooling would promote local ventilation in the Aegean and Adriatic basins, allowing some oxygenation at the sea floor and the consequent reinstatement of habitable conditions. However, this cooling took place within the general monsoonal maximum, and the ongoing reduction of excess evaporation meant that as the cooling came to an end, the deep ventilation remained freshwater inhibited. Inevitably, the anoxic conditions returned. A study in 2000 used the numerical model for sapropel-time circulation to investigate the consequences of enhanced cooling in the northern sectors. The results were astonishingly consistent with the observations and conceptual interpretation.

A similar cooling occurred around 6,000 to 5,500 years ago, i.e., around the time the monsoonal maximum had come to an end. Again, there is evidence for a rapid re-oxygenation at the sea floor, with a repopulation by opportunistic benthic species. This time, however, the benthic diversity continued to increase gradually, and opportunists gave way to more diverse and specialised faunas. Clearly, the ecosystem this time recovered from the anoxia in a way that was completely different from the benthic repopulation within the sapropel interruption around 8,000 years BP. When the cooling of about 6,000 years ago ended, the anoxic sapropel conditions did not reappear. The isotope data suggest that the monsoonal maximum influence on the Mediterranean had disappeared, and that excess evaporation had consequently recovered to levels comparable to those of today. We infer that, as the cooling came to an end, the salt-motor had gained sufficient momentum to keep deep ventilation going in similar ways to the present. The entire eastern Mediterranean basin became oxygenated again.

A third Holocene cooling event was centered on roughly 3,000 years BP. At that time, the Mediterranean was a very well ventilated basin, and we have no immediate ways to assess whether this ventilation was perhaps somewhat more vigorous during the cooling event.

The quality long cores available to date do not provide good records for the interval younger than about 1,500 years BP, since this still poorly compacted, water-logged, sedimentary top layer is commonly blown away by the pressure wave in front of the corer. Box-coring allows good coverage of this interval, but has traditionally been applied especially in areas with fairly low sediment accumulation rates, limitingv the possible temporal resolution. New efforts are now being guided to target recovery of the past 1,500 years in high-accumulation sites. There is a – testable – expectation that quality records of this interval from especially the Adriatic and Aegean regions will show a hitherto overlooked cool event dating from two to four centuries ago. This expectation derives from detailed comparison of the sequence of three main Holocene cooling events in the Mediterranean (independently identified also in the western Mediterranean using a different temperature proxy) with anomalies in the high-latitude climate record from Greenland ice cores. The ice-core ion series, an approximation of wind-blown dust loading, shows an excellent match of repeated cold/dusty episodes with the Mediterranean cooling events. The Greenland record is complete to the present day, and suggests that the most recent of the Holocene events occurred quite recently, peaking two to four centuries ago. This was the so-called ‘Little Ice Age’, the period that gave rise to the famous paintings of severe European winter scenes (eg. Winter Landscape by Breughel the Younger).

The suggestion from the Greenland record is that we are currently coming out of the Little Ice Age, or – in other words – that we are on a natural warming trend. The Holocene events seem to be separated by 2,000 to 3,000 years, suggesting that the next ‘warm peak’ in that cycle is to be expected 600 to 1,000 years from today. By analogy to what happened in the previous Holocene climate ‘cycles’, it should be expected that winter cooling in the northern sectors of the Mediterranean will become less effective over this time (note 15).

Less cooling means more chance for the second step of the basin’s deep water ventilation to become disrupted. Nobody yet knows how sensitive the basin’s deep circulation is, nor how it may respond. This is particularly alarming because the current natural warming trend might become accelerated by greenhouse warming effects. There is a clear need for sophisticated sensitivity studies to establish how much of a chance there is (or not) that the deep ventilation might collapse in reaction to the combined effects of a natural trend towards reduced winter cooling, and global anthropogenic greenhouse warming.

* * *

It will be obvious from the above that, in my opinion (subject to many unknowns about the sensitivity of the circulation), there is a realistic possibility that the combination of greenhouse warming with the gradual underlying natural trend could jeopardise the deep ventilation of the Mediterranean. How tangible is this, or is it just wild speculation in a scientist’s bid to justify his own existence? Or, in less hostile terms: now that both effects have been going on for considerable periods of time (about 1 century for the anthropogenic impact, and 3 to 4 centuries for the natural trend), can we actually see any responses in the basin? There is one straightforward and truthful answer to this question: YES!

In 1984, the French oceanographer Henri Lacombe and colleagues presented an initial summary of historical deep-water temperature readings for the western Mediterranean. They showed a clear hint of warming. More extensive databases have since been analysed, and the conclusion stands: a small but significant warming of the order of 0.1ºC is evident since 1900, throughout the western Mediterranean deep-water mass. The value may seem negligible, but integrated over the size of this deep-water reservoir, it represents a great amount of heat. The temperature increase was not evenly distributed over time, but shows an apparent acceleration around 1950. In fact, some three-quarters of the increase took place in the last 50 years.

There have been various attempts at explaining the deep-water temperature increase. It was found to most likely be related to some combination of general warming in the deep-water formation area (natural and/or anthropogenic), a possible climatic trend towards increased aridity over the Mediterranean, and damming of major rivers that drain into the (entire) Mediterranean basin. Warming influences speak for themselves, but how can aridity and damming have an effect? There has been a frenzy of damming and freshwater-diversion activity since the 1950s and 1960s in the Nile and the main Eurasian rivers, including those that indirectly affect the Mediterranean via the Black Sea. This freshwater diversion has noticably affected the Mediterranean’s freshwater budget, giving a 10% increase in excess evaporation. Increasing aridity would also enhance the excess evaporation. As a consequence, the basin’s general salinity has gone up. With increased salinity, the water needs to be cooled less for it to attain the same density as deep water formed before the freshwater diversion. Hence, the newly formed deep water would be somewhat more saline, and somewhat warmer. The damming-hypothesis has since been discarded as a complete explanation for the observed trend – direct warming in the deep-water formation area was found to play a dominant role, while shorter-term fluctuations in aridity also had notable effects. Although of secondary importance, however, the damming has a long-term impact that cannot be ignored. The adjustment triggered by the damming in the 1950-1960s will continue to affect the basin’s properties until at least 50 years from today.

What about changes in the eastern Mediterranean? The 1992 paper that elaborated the damming hypothesis discussed above noted an increase in the salinity of the Levantine Intermediate Water. This increase was strongly confined to the post-1950 interval, which formed the basis for the damming hypothesis. As mentioned before, the effect is likely to continue for at least some 50 years from today. However scientifically interesting such gradual adjustments in a fundamental property may be in their own right, they attracted relatively little attention. This changed when rapid reorganisations started to occur in the ‘accepted’ Mediterranean circulation state. One study, for example, indicated that the ‘traditional’ intermediate water formation may be changing (intermittently?) to deep water formation. An even more dramatic response to a change in climatic forcing, however, stole the limelight.

In 1996, a team led by the German oceanographer Wolfgang Roether presented their observations of a fundamental reorganisation of the deep ventilation state in the eastern Mediterranean, which took place in the late 1980s to early 1990s. Ever since oceanographic observations had been made, until that time, the Adriatic Sea had appeared as the primary source of Eastern Mediterranean Deep Water, while the Aegean Sea had been found to contribute in the 500-1,000 meters depth range. Then something changed. Between research cruises in 1987 and 1995, the Aegean had been transformed into a strong source of true deep waters. These were of higher salinity, and hence denser, than the resident, previously-formed, deep waters from the Adriatic.

It is thought that the circulation reorganisation was triggered by a climatic reduction of precipitation over the Aegean Sea in particular, and the easternmost Mediterranean in general, over the period 1988-1995. However, there also seems to have been an involvement of processes in the western Mediterranean which affected the salinity of the surface flow from the western into the eastern basin through the Strait of Sicily. Changes in the volume of this inflow could be related to the circulation change in the eastern basin, while changes in the properties of this inflow reflect fluctuations in the western basin, or even the Atlantic. To date, the exact forcing conditions and feedback mechanisms are unknown, and we also don’t know how long the reorganised circulation state will continue before (if) the system relapses into its old behaviour. In fact, it is not known whether a return to the previous state is to be expected at all, or whether the current state represents a new ‘stable’ configuration.

In summary, we have only a basic idea about what drove the historical changes from one circulation state to another, and we are left to guess whether the system will switch back or remain in the reorganised state. We lack fundamental insight into the full range of possible circulation responses, and into the fundamental sensitivities and feedback processes within the system. We don’t know whether the system’s internal variability itself is as strong as what has been observed, or whether an external perturbation is needed – and if so, how strong that perturbation has to be. These uncertainties should be considered within the geological context that the circulation is rather prone to catastrophic collapses. In addition, it needs to be taken into account that any natural climate trend will now – for the first time – be swamped by an anthropogenic disturbance of unknown nature and magnitude.

The conclusion must be that we cannot be complacent about the potential for very nasty responses in the basin, such as a partial or even full collapse of the deep circulation. This could be for several centuries or millennia, as was the case when the sapropels were deposited, or it could be for shorter periods of time, perhaps several years or decades. With the current knowledge, there simply is no way of judging such issues. However, the circumstantial evidence to date certainly appears to warrant the undertaking of a full-scale effort to quantitatively estimate the probability of a circulation collapse in the (near) future. This ‘risk-assessment exercise’ should include the scale (local, regional, basin-wide) and duration of such a potential event, as well as its likely implications for the Mediterranean ecosystem. On a more ambitious level, we might consider which properties could best be monitored continuously, and in which sensitive target area(s), to ensure some advance warning of impending circulation changes.

This is not scaremongering. It is simply the diagnosis of a serious potential threat, and an outline for a sensible course of action to understand the threat and minimise any potential impacts on society. All of the actions outlined are well within the grasp of a sound, structurally supported, trans-national scientific approach. Importantly, such work must transcend all disciplinary boundaries in the environmental sciences, spanning everything from (quantitative) studies of the climate and ocean states in the recent geological past to contemporary  oceanography, and from biology to numerical modelling.

* * *

Is that it, or do we need to take anything else into consideration – i.e., is man having any other significant impacts on the health of the Mediterranean? Unfortunately, the answer again is YES. We cannot forego a brief discussion of pollution, and especially the enormous anthropogenic enhancement of nutrient influxes into the sea. This release is a cause for concern within the context of development of anoxia. It artificially increases the biological oxygen demand (BOD), causing many near-coastal regions to experience serious seasonal oxygen starvation at the sea floor.

The problem is strongest, but not limited to the near-coastal zones. The modern Mediterranean in its entirety is affected. Its land-locked nature allows only limited exchange with the vast ‘open ocean reservoir’ through the narrow and shallow Strait of Gibraltar. Consequently, the Mediterranean steadily accumulates the high anthropogenic releases of dissolved nutrient salts and untreated sewage, and other major pollutants (hydrocarbons – petroleum products – heavy metals, etc.). The urgency is such that, in 1975, a consortium of Mediterranean nations composed the Mediterranean Action Plan, setting ten objectives for 1995 to clean up the sea and protect its environment. Unfortunately, much of the problem remains unresolved, and new initiatives have been started. It is outside the context of this book to dwell much further on the issue of pollution, but the following should be emphasised: with the potential for (partial) circulation collapse incompletely understood, it cannot be a good idea to dump excessive amounts of nutrients into the Mediterranean. The consequences for the deep ecosystem of any disruption of circulation – be it local, regional, or basin wide – would be hugely exacerbated by enhanced production and, consequently, subsurface BOD.
 

To Chapter 7