DISCUSSION

Extraordinary organic matter contents and severe dysoxic to anoxic conditions are characteristic of the sediments of the DMB. This allows live staining of foraminifers with Rose Bengal even several months after death (Murray and Bowser 2000). Therefore, our staining indicates living specimens only with regard to this time span. However, the four stained species recovered repeatedly from surface sediments and deep infaunal habitats (Table 2, Table 3; Figure 4) are the main contributors to the thanathocoenoses. Virgulinella fragilis undoubtedly thrives under the harsh conditions of the DMB. It was recovered living with protruded reticulopodia from surface sediments, occurs Rose Bengal stained only at stations ≤ 0.3 ml l^-1 O₂, and was not recovered outside the DMB (Leiter 2008). For Nonionella stella, Fursenkoina fusiformis, and Bolivina pacifica one might suggest lateral advection from more oxygenated habitats, and a limited time span of their subsequent survival. All three species occur stained at deeper stations seaward of the DMB, and thus could be transported by the prevailing landward bottom currents. In this case, it seems improbable that none of the many other unstained species that were recovered (Figure 3, Figure 5) displays a similar type of advection and staining of decaying protoplas10 cm³ unusually rare occurrences. More sophisticated live staining techniques may perhaps provide a better resolution of ratio of living and dead stained individuals (Bernhard et al. 2006a). However, increased numbers of dead stained individuals does not automatically mean that lateral advection of living species into the DMB has occurred. With regard to the severe environmental fluctuations and gas outbreaks, high mortality rates must be considered a pulsed attendant for established populations as well. As a result, we consider the syntopic occurrence of V. fragilis, N. stella, F. fusiformis, and B. pacifica as characteristic for the oxygen depleted DMB. Moreover, we conclude that our thanathocoenoses firmly transmit the ecological basics derived from the stained taxa (Murray and Alve 1999). The standing stock of most populations ranges at <1 individual 10 cm^-3, sometimes even <0.1 ind. 10 cm^-3. Staining with CellTracker Green would most probably result in an even smaller standing stock (Bernhard et al. 2006a). Such populations are nearly impossible to recover from common split fractions of sample sizes equivalent to 10 cm^-3 or less. Much larger sample sizes would have to be scrutinized in order to recover the tiny, but widespread populations. It is likely that the absence of foraminifera considered for other extreme environments is due to sample size, rather than the actual absence of foraminifera.

Taxa recovered exclusively as unstained tests show increased abundances towards the outer sedimentary edges of the DMB. They indicate sedimentological advection and/or unsuccessful migrational attempts. The 'rich' benthic foraminiferal fauna reported on previously from the mud belt was either restricted to the outmost seaward edge (Bremner 1983), or presented as a total assemblage with questionable origin (Dale and McMillan 1998).

The deep infaunal occurrence of four stained foraminiferan taxa probably represents the most significant result of our study. Several centimeter-long annelids of uncertain affinity were recovered during cruise Ahab 5 at the crater gasdome 2' (Altenbach and Struck 2006). Several polychaetes thrive under microxic conditions on the northernmost Namibian shelf (Zettler et al. 2009), and worm feces are considered abundant for the Namibian mud belt (Dale and McMillan 1998). This suggests that passive transport of stained tests downcore may have occurred through bioturbation. On the other hand, worm borrows strongly affect lateral migration of foraminifera and their colonisation behaviour in deeper sediment layers (Thomsen and Altenbach 1993). Core 199 was carefully selected, because it did not show indications of bioturbation or sediment replacement from ascending gas bubbles. The number of stained and unstained tests is lowest within the top 5 cm, and increases in below-positioned layers. An erratic transport downcore would probably lead to a decline in test numbers with increasing sediment depth. Stained taxa occur in distinct depth layers, accompanied by varying numbers of preserved unstained tests (Figure 4). This coincidence perhaps is a result of the imprint of various microbial and biogeochemical processes. Depth defined layers typically occur in areas in which certain reduction rates of sulfate and nitrate are present within the uppermost 30 cm of the Namibian shelf sediments (Brüchert et al. 2003). The sulfate reduction rates of the DMB were modelled to rapidly increase above 15 cm sediment depth and sharply decrease in the deeper sediment column (Dale et al. 2009). This depth layer coincides with the last appearance of F. fusiformis and the first occurrence of N. stella downcore (Figure 4).

Teratological tests of Virgulinella fragilis were commonly observed in the year 2000 (cruise M48) in the size fraction of >250µm. Three years later (cruise M57/3) no teratological tests were recovered from this size fraction, neither stained nor unstained. Numerous gas blow outs of hydrogen sulfide and methane were recorded within this time span (Weeks et al. 2004). These short-term perturbations of pH and redox conditions in the bottom and pore water are known to trigger fusions of embryonic stages (Stouff et al. 1999), and subsequently lead to the corrosion of foraminiferal tests (Keir 1980). The taphonomic loss of the teratological tests likely occurred rapidly and resulted in complete test dissolution, because no V. fragilis advected from the DMB was recovered in 2003, neither on the outer shelf nor downslope (Leiter 2008). Offspring of the populations of V. fragilis was detected in size fraction 150-250 µm, with relatively high numbers of teratological tests. None of these specimens, however, had matured, and all tests were slightly to heavily corroded, in transfer to dissolution. The extremely thin wall of V. fragilis (approximately 1 µm, according to Grindell and Collen 1976) might offer an explanation for these observations. If post mortem dissolution of calcitic foraminiferal tests can be considered a logarithmic function of wall thickness (Keir 1980), the exceptionally thin test walls will rapidly dissolve after death, if they remain in carbonate-aggressive sites. The lateral transport over long distances can only occur if empty tests are advected towards more normal marine hydrochemical conditions. As a result, environmental ranges deducted from sites that only produced unstained tests of V. fragilis are most likely exaggerated towards more normal marine conditions (Grindell and Collen 1976, Bhatia and Kumar 1976, Nigam and Setty 1982, Takata et al. 2003).

For palaeoecological investigations, the genus Virgulinella provides a very interesting extant species because of it's environmental thresholds within dysoxic to reducing conditions. All other taxa recovered stained from the DMB can also persist under normal marine conditions, as well as all other agglutinated, miliolid, or rotaliid taxa that have been recorded to date from anoxic environments (Bernhard and Sen Gupta 1999; for an exceptional allogromiid see Bernhard et al. 2006b). The first modern record of the genus Virgulinella can be found in Todd and Brönnimann (1957), including a reference to a collection of extant tests at the San Marcos University (Lima/Peru). These authors assigned their material to Virgulinella pertusa and regarded the taxa Virgulinella gunteri and Virgulinella miocenica as 'very similiar or possibly identical' with this species. The first record of stained specimens was provided by Seibold (1975), including a note on modern virgulinellids recovered from the NW African shelf (Lutze in Seibold (1975)). Seibold (1975) noted the problem of assigning a Cenozoic fossil form to a modern taxon, and placed the specimens under Virgulinella cf. gunteri. The specimens illustrated in this paper, however, reveal a morphology that appears to be intermediate between Virgulinella fragilis and the Pliocene Virgulinella lunata. One year later, the only modern species V. fragilis Grindell and Collen 1976 was described, with V. lunata named as the most similar fossil taxon. Nevertheless, Nigam and Setty (1982) reported on the variability of the modern taxon. These authors described modern species assignable to both V. guntheri and V. pertusa. This is consistent with our observations. We found considerable variability with regard to the size of the sutural openings, the interposed arches, and the chamber elongation. Grindell and Collen (1976) have included all these morphological variations in the differential diagnosis for V. fragilis. As a result, it appears that the extreme fragility represents the sole specific diagnostic feature of V. fragilis. The description of the type material indicates that the wall of the later chambers is approximately 1 µm. This stands in contrast to the more robust specimen recovered from the Cenozoic (Haman 1977, Haman et al. 1993). Several authors did not consider fossil taxa in their evaluation of modern virgulinellids for various reasons, including fewer sutural bridges in comparison to the fossil taxa (Seibold 1975), presumed problems related to reworking (Zobel 1973), or more general considerations (Bhatia and Kumar 1976). However, several authors did not recognize considerable morphological differences between the fossil taxa and their modern virgulinellids (Todd and Brönnimann 1957, Sellier de Civrieux 1977, Setty and Nigam 1980, Diz and Frances 2008). Some fossil taxa reported from surface sediments of the eastern Pacific ocean may fall within the intraspecific pattern of V. fragilis (Hayward et al. 1999). We see the need to more precisely define the patterns of variability for the Cenozoic and modern species, until synonymy or separation can be handled precisely. A detailed discussion on the structure and taxonomy of the genus Virgulinella proposes to elevate this taxon to superfamily status and to remove it from the buliminids (definition in Revets 1991, rebuttal in Haman et al. 1993). Regardless of these taxonomic problems, the question remains as to whether a relatively homogenous test morphology implies corresponding ecological layouts for extant and fossil species.

It has been proposed that upwelling and dysoxic conditions existed at the Namibian shelf since the Miocene (Baturin 2002a). During the Pliocene-Pleistocene, reducing conditions amplified towards conditions comparable to the modern DMB (Baturin 2002b). This indicates that the spread of the genus Virgulinella off south-western Africa could have taken place under severe environmental conditions since the Miocene. One Miocene taxon first appeared at the African west coast (Haman et al. 1993).

However, even the earliest records of the genus Virgulinella from the middle to upper Oligocene come from areas that are believed to represent anoxic and sulfidic environments. For example, Loxostomum chalkophilum Hagn is described for a pyritic fill-out of a foraminiferal test interior. Already Hagn discussed a possible relationship between this Rupelian stone core and Virgulinella pertusa. But in the absence of apertural structures, he preferred to select a more primitive genus for the typus description (Hagn and Hoelzl 1952). Our study of the holotype specimen (BSPG Prot. 1952-8) revealed a chamber arrangement and sutural bridges that are in fact quite similar to that seen in the genus Virgulinella. Hagn's discussion is exhaustive. Mineralization and test dissolution must have occurred under sulfidic and carbonate aggressive hydrochemical gradients. In the Upper Oligocene, Virgulinella ex gr. pertusa is recovered from several outcrops of the Eastern Paratethys; oxygen depleted or sulfidic palaeoenvironments are evident from sedimentological and geochemical data for all these sites (Stolyarov 2001).

In conclusion, our results add support to the hypothesis that Virgulinella fragilis represents a characteristic species for environments in which sulfate reduction occurs (Bernhard 2003). We also conclude that, wherever fossil or modern virgulinellids are recovered, it is reasonable to assume temporal or stagnant anoxia for the source stratum or adjacent areas. The co-occurring Nonionella stella and Fursenkoina fusiformis may expand this supposition towards foraminiferal denitrification at redox boundary conditions (additional taxa discussed in Risgaard-Petersen et al. 2006, Høgslund et al. 2008). Increasing numbers of other benthic foraminiferal taxa indicate increasing lateral advection, and increasing distance from oxygen depleted environments.