Figure 1 Spatial distributions of coiling ratios and isotopic differences.   Full legend   High resolution image and legend (104k)

The general modern hydrography of the Nordic seas is reflected in the stable oxygen isotope values of both N. pachyderma (sin.) and N. pachyderma (dex.)²¹, since the ¹⁸O incorporated into test calcite is a function of temperature and salinity²². The ¹⁸O values of N. pachyderma (sin.) and N. pachyderma (dex.) are largely identical under polar conditions (Fig. 1b) in the western region, where relative abundances of N. pachyderma (dex.) are low (Fig. 1a). But in the east and south of the Nordic seas, where the percentage of N. pachyderma (dex.) increases rapidly under more-temperate conditions (Fig. 1a), a significant isotopic difference between left- and right-coiling N. pachyderma of up to about 1 in ¹⁸O is observed (Fig. 1b).

Molecular analysis of the small subunit (SSU) ribosomal RNA sequences has previously shown that the N. pachyderma morphospecies consists of a complex of distinct genotypes⁷, and that the coiling direction in the Southern Hemisphere is not a phenotypic response to temperature, but reflects genetic divergence and the probable existence of two separate species⁷. In this study (Fig. 2), we have found that throughout the Nordic seas, all left-coiling N. pachyderma are genotypically [Type I (sin.)]. This is not the case for N. pachyderma (dex.). In the region of the Norwegian Sea and Denmark Strait south of 64° N (Fig. 2), all N. pachyderma (dex.) are genotypically [Type I (dex.)] and are genetically highly distinct from the left-coiling variant⁷. Under polar conditions north of the latitude 70° N where the coiling ratios are confined to less than 5% (Fig. 1a) and the ¹⁸O values of N. pachyderma (sin.) and N. pachyderma (dex.) are largely the same (Fig. 1b), genotyping reveals that the left- and right-coiling variants of N. pachyderma are genotypically identical. Both are [Type I (sin.)]. Clearly, N. pachyderma [Type I (dex.)] is thriving in the relatively warm North Atlantic waters, and is adapted to a totally different hydrographic regime than left- and right-coiling [Type I (sin.)], which is dominant in polar waters. Such genetic and environmental evidence indicates that these two genetic types should be considered different species. As might be expected, there is a mixing zone between the right- and left-coiling provinces (70° N to 64° N) where the dextral morphotypes can be either N. pachyderma [Type I (dex.)] or [Type I (sin.)].

Figure 2 Geographical distribution of genotypes of left-coiling N. pachyderma [Type I (sin.)], the right-coiling form of N. pachyderma [Type I (sin.)], and right-coiling N. pachyderma [Type I (dex.)].   Full legend   High resolution image and legend (66k)

The genetic evidence for N. pachyderma shows that the morphological distinction of coiling directions is not sufficient to distinguish between genotypes. Since fossil specimens do not contain DNA for genetic analysis, we need an indicator proxy to distinguish between the two species of right-coiling N. pachyderma. From the presented sediment data in combination with the genetic evidence it can be assumed that in regions where the percentage of N. pachyderma (dex.) does not exceed a 'threshold' value of 5% in coiling direction, both N. pachyderma coiling variants will be identical, genotypically [Type I (sin.)], with a correspondingly identical isotopic composition. In regions where the percentage of N. pachyderma (dex.) exceeds this 'threshold' value, N. pachyderma (dex.) is expected to have a progressively different isotopic composition from N. pachyderma (sin.) as the N. pachyderma [Type I (dex.)] genotype increasingly dominates the assemblage.

In order to further test our conclusions from the modern data set, we investigated two sediment cores representing the warmer and colder regions of the Norwegian Sea (see Fig. 1 for locations). Samples from the glacial sections of the two cores show a low relative abundance (1–5%) of N. pachyderma (dex.) (Fig. 3). Since the early Holocene (9–10 kyr ago), dextral coiling ratios are higher, with levels of 5–10% in core PS1243 and 40–60% in core M23323 due to climate warming at the end of the deglaciation. These faunal differences reflect the basic water-mass temperature gradient between the two sites. A similar picture is recognized in the stable oxygen isotope records of N. pachyderma (sin.) and N. pachyderma (dex.), which both depict the overall climatic change since the Last Glacial Maximum (LGM) (Fig. 3). However, while ¹⁸O in both species are nearly the same during the glacial sections, when dextral coiling ratios are extremely low, a systematic offset of about 0.5 is noted since about 10 kyr ago when dextral coiling ratios increase, accompanied by a decrease in the relative abundance of N. pachyderma (sin.) (Fig. 3). In light of the genotype evidence for the Nordic seas, this Holocene offset in ¹⁸O has to be interpreted as an alternation in the species of N. pachyderma (dex.) from [Type I (sin.)] during glacial times to a [Type I (dex.)] domination in the interglacial interval. The isotopic offset of about 0.5 diminishes in core PS1243 since about 4 kyr ago, and corresponds to a period of increasing abundance of N. pachyderma (sin.) and a coeval abundance decrease of N. pachyderma (dex.) fairly close to the threshold value of 5% coiling ratio. This would result in a greater contribution of the right-coiling [Type I (sin.)] genotype to the ¹⁸O signal during this time in core PS1243, and explains the reduced offset in ¹⁸O between N. pachyderma (sin.) and N. pachyderma (dex.) compared to that of core M23323. Altogether this strongly supports our conclusion that a species-dependent isotopic shift in N. pachyderma (dex.) ¹⁸O signal can be inferred when relative abundances of N. pachyderma (dex.) cross a specific threshold value of 5%.

Figure 3 Time series showing climate proxies of two sediment cores versus depth and versus calendar years.   Full legend   High resolution image and legend (87k)

The ¹⁸O isotopic difference of 0.5 between the right-coiling forms of N. pachyderma [Type I (sin.)] and N. pachyderma [Type I (dex.)] in the Holocene translates into a temperature or salinity difference of approximately 2 °C or 2 practical salinity units, respectively. This isotopic difference is an additional increment, superimposed upon the global LGM/Holocene isotopic signature. For all future palaeoceanographic ¹⁸O isotope studies carried out on N. pachyderma (dex.) at low percentage coiling ratios in the Northern Hemisphere, it will be of considerable importance to account for this derived species dependent isotopic shift of up to 0.5. Interestingly, while the ¹⁸O values reveal such a marked difference during the Holocene, there is no systematic offset in ¹³C between N. pachyderma (sin.) and N. pachyderma (dex.) over the entire time interval (Fig. 3). Since ¹³C values of foraminifers are thought to record specific environmental conditions^{23, 24}, the Holocene ¹⁸O offset may represent a genotype-specific vital effect. However, we have shown that N. pachyderma [Type I (sin.)] and N. pachyderma [Type I (dex.)] are clearly adapted to different hydrographic regimes, and it remains to be determined whether the species-dependent ¹⁸O offset represents annual, seasonal or water-depth hydrological differences, a vital effect or a combination of them. Although our findings add yet another aspect to the interpretation of ¹⁸O values of planktonic foraminifers in the Northern Hemisphere, by using the indicator proxy of the coiling ratio, the correct interpretation can be deduced. Clearly the relevance of coiling ratios for isotope records of N. pachyderma (dex.) in other palaeoceanographically important regions needs to be addressed.

There is accumulating evidence that several morphospecies do not represent genetically continuous species over their entire environmental adaptive range, as newly discovered potential 'cryptic species' seem to have different environmental preferences^{6-8, 25}. Isotopic differences may also be associated with other genotypes of morphospecies commonly used in palaeoceanographic reconstructions^{26, 27}. An understanding of the nature of these adaptations will considerably enhance other palaeoceanographic and palaeoclimate proxies if the 'cryptic species' can be distinguished in the fossil record, as was possible for this study.

Methods
Foraminiferal counts In down-core sediment samples about 300 individuals of the size fraction >125 µm have been counted, while in surface sediment samples individuals >150 µm have been counted10. Comparison of counting results of these different size classes in the Nordic seas has shown that there is no significant difference in relative faunal abundances and thereby also coiling ratios²⁸. In the N. pachyderma (sin.) dominated polar regions, typical coiling ratios of N. pachyderma (dex.) are between 1% and 5 1% from counts of 300 individuals¹⁰. The upper value of 5% coiling ratio has been adopted as the threshold value for the start of the influx of N. pachyderma [Type I (dex.)] into the N. pachyderma [Type I (sin.)] fossil assemblage. This threshold conforms to the upper values observed in the polar-region core tops (Fig.1a), the glacial down-core sediments (Fig. 3), and is also approximately the threshold observed in core PS1243 at which the isotopic offset is observed to decrease in the later Holocene (Fig.3).

Isotope analysis of calcite shells About 20–30 individuals were picked from the 125–250 µm size fraction, and analysed at the Leibniz Laboratory of Kiel University using a fully automated Kiel Device plus MAT 251 isotope mass spectrometer system. Results are reported in the usual -notation relative to PDB, and have an accuracy of 0.07 and 0.04 for ¹⁸O and ¹³C, respectively.

Sampling localities Foraminifera were collected on board RV Polarstern (ARKXV/1 + 2); specimens were obtained using multinets (upper 500 m) at 75° N (from 13° W to 13° E), and from a mixture of multinet and surface pumped water samples in the Nordic seas between 80° N and 60° N.

Isolation and sequencing of SSU gene fragments DNA extraction, amplification by polymerase chain reaction (PCR) and automated sequencing of either a 500-base-pair (bp) or a 1,000-bp region of the terminal 3' end of the foraminiferal SSU rRNA gene was as described previously^{7, 8}. The gene fragments for N. pachyderma (dex.) were directly sequenced. Some limited degree of ambiguity was detected in the gene repeats for N. pachyderma (sin.), and consensus sequences were obtained by cloning using a pCR 2.1 TOPO TA cloning kit (Invitrogen) and then sequencing using universal primers.