SIR,

Seasonal variations in insoluble particle concentrations with large spring peaks have been observed by Reference HammerHammer (1977) and Reference ThompsonThompson (1977) in Greenland ice cores, and the phenomenon has been used for ice-core dating. The origin of the dust in the peaks is, however, still unknown. Reference Gayley and RamGayley and Ram (1984) have found diatoms, mainly of fresh-water origin, in a section of ice core from Crête, central Greenland (lat. 71°N., long. 37°W.; Fig. 1). We have now measured the time variation of the diatom concentration in a 2 year section of this ice and found that diatom abundances also exhibit seasonal changes with a spring maximum that coincides with the dust maximum. In this letter, we would like to suggest the possibility that diatoms could be used as tracers for the source of dust in ice cores.

The ice we studied was a 2 year section of 200 year old ice from Cr�te. The ice was divided into ten samples and, as described previously (Reference Gayley and RamGayley and Ram, 1984), diatoms were recovered from each sample by filtration of ice melt water through a 13 mm diameter Nuclepore membrane filter with pore-size diameter of 0.08 ¼m for each of the samples. The typical mass of water filtered per sample was 30 g. Using a Scanning Electron Microscope (SEM), we determined the concentration of diatoms and diatom fragments whose largest linear dimension was greater than or equal to 10 ¼m.

All of the diatoms that could be clearly identified were of fresh-water origin. Genera observed include Achnanthes. Amphora, Eunotia, Fragilaria, Melosira, Navicula, Nitzschia, Pinnularia, and Stephanodiscus. Many of the specimens present were fragmented, and could only be identified to genus. Complete specimens included both species commonly found in soils and other aerophytic communities (e.g. Navicula mulica var. cohnii; Fig. 2a), and species which grow only in plankton communities (e.g. Melosira italica, Fig. 2c; M. granulala, Fig. 2d). Some planktonic species (e.g. Stephanodiscus niagarae; Fig. 2b) were surprisingly well preserved.

Fig.1. Map of Greenland showing the location of Crête.

Figure 3 shows how the concentration of large insoluble particles and diatoms, and diatom fragments varied with time in the 2 year period, 1783–84. Both large particles and diatom concentrations exhibit simultaneous spring peaks in each of 2 years. Diatom remains in the smaller 1783 peak consist entirely of species derived from soil or aerophytic communities, plus a few badly fragmented and corroded specimens possibly derived from fresh-water diatomites. The larger 1784 peak contains, additionally, relatively large numbers of complete and well-preserved valves of planktonic species. The most probable sources of these specimens are shallow, productive lakes in semi-arid regions which undergo large periodic fluctuations in water level. Aeolian transport of fresh-water planktonic diatoms derived from such lakes in sub-Saharan Africa via the “Harmattan haze” has long been known (Reference KolbeKolbe, 1957).

The two conditions that are essential for atmospheric transport of planktonic diatoms from source regions are: (a) lowering of lake levels to the point where diatomaceous sediments are exposed, and (b) wind velocities high enough to entrain and transport particles as large as whole diatom valves. The presence of specimens of planktonic species in Greenland ice from a given year may thus provide a signal of both aridity and particularly active aeolian transport. Possible source areas that satisfy these conditions are sub-Saharan Africa, south-central Asia, and south-western North America.

The diatoms found in our present samples do not allow an unequivocal determination of source area. Most species which can be firmly identified have a wide geographic distribution. The presence of Stephanodiscus niagarae, a species particularly abundant in North America, supports the semi-arid south-western United States as a possible source. The work of Reference Jackson, Gillette, Danielson, Blifford, Bryson and SyersJackson and others (1973) suggests the possibility that dust storms originating in this region in the spring are carried in an ascending flow along a north-easterly path that intersects Greenland before descending and heading south over the Atlantic Ocean. We note, however, that the species found in our samples are very similar to those illustrated by Reference KolbeKolbe (1957), which are believed to be derived from Africa.

Fig.2. SEM micrographs of diatoms found in the 1784 spring abundance maximum, (a) Navicula mutica var. cohnii, a species commonly found in soils and other aerophytic habitats. (b) Stephanodiscus niagarae. (c) Melosira italica, and (d) M. granulata. The gap between the two horizontal lines at the bottom of the pictures is 0.5 ¼m wide.

It is hard to see how diatoms and particles could arrive in such well co-ordinated peaks over such a short seasonal time-scale ( Fig.3) unless they came from the same general region under the influence of the same general wind patterns. This idea is supported by the observation that the ratio of diatoms to insoluble particles in the two spring peaks is equal within the experimental uncertainty. This suggests that study of diatoms in Greenland ice could eventually help elucidate the origin of the dust in the spring peaks.

Since diatom searches in ice cannot be automated, they are very time-consuming and our preliminary study covered only 2 years of ice core. Because of the possible importance of our findings, it is, nevertheless, essential that more extensive measurements be carried out to verify that dust and diatom concentrations co-vary in time over longer sections of ice core covering more years of snow accumulation.

Fig.3. (a) Concentration of diatoms and diatom fragments in the Crête ice core as a function of time of deposition. Only diatoms whose largest linear dimension is greater than or equal to 10 μm were counted. Since each sample represents a fraction of a year, the number of diatoms in each is not large and in some cases it is zero. Note that since snowfall is not constant throughout the year, one cannot treat the time-scale as linear.

(b) Concentration of insoluble microparticles in the Cr�te ice core as a function of time of deposition. In our work (Reference Gayley and RamGayley and Ram, �985), we measured the size distribution of insoluble microparticles in the radius range 0.05–1.31 ¼m. Only those particles in our largest radius range, 0.38–1.31 um, were included in this study, since they are closer in size to the diatoms. The qualitative features of the curve remain unchanged when all particles are used rather than the larger ones. The dashed line indicates a region where particle measurements were not made (the particle distribution on the filler was not uniform), although the diatom content was determined.

Acknowledgements. Part of this work was supported by the National Science Foundation, Division of Polar Programs, under grant No. DPP-8619198. We thank L.H. Burckle for his assistance in the early stages of the work and for incisive comments. We also thank CC Langway, jr for making available to us the ice used in this study, and J.R. Petit for directing our attention to the critical paper of Reference Jackson, Gillette, Danielson, Blifford, Bryson and SyersJackson and others (1973), and for helpful comments. Excellent technical assistance was provided by E. Cockayne.