Weathering rates of fine pebbles at the soil surface in Karkevagge, Swedish Lapland

In 1994, nylon mesh bags containing similar to 6.2-mm diameter fragments of either freshly crushed dolomite or granite were placed on meadow, solifluc tion meadow, birch, heath and alpine tundra vegetation that cover surfaces within Karkevagge, a glaciated arctic-alpine valley in Swedish Lapland. Eac h vegetation category contained three replicate sites comprising one bag of each rock type. In 1995, the study was extended to Dryas heath and willow (two sites only) surfaces. Percentage annual weight loss was used as a metr ic of chemical weathering rates. One site in each of the 1994 categories wa s measured in 1995. All sites were measured in 1999. Two dolomite and grani te standards stored in the laboratory and subjected to identical cleaning a nd weighing regimes to the field samples, exhibited no more than a 0.24% to tal weight loss after 5 years. The percentage weight loss in the subgroup e xamined after 1 year (dolomite 0.54%/year; granite 0.41%/year) significantl y exceeded the average annual loss for the subsequent 4 years in all cases (dolomite 0.40%/year; granite 0.07%/year). Ten sites with a 5-year record h ad a mean annual loss of 0.326 +/- 0.115%/year (two standard errors) for do lomite and 0.121 +/- 0.020%/year for granite. Assuming an exponential decay model, a 0.326%/year rate of loss produces a 'half-life' of approximately 212 years and a 0.121%/year loss rate, a half-life of approximately 570 yea rs. Kruskal-Wallis one-way ANOVA on rank-ordered data determined significan t differences in dolomite pebble mass loss between the Dryas heath and will ow plus meadow microenvironments, as well as between the birch forest and t he willow plus meadow microenvironments. The underpinning factor behind the se differences appears to be moisture differences. A similar test failed to determine statistically significant differences for the granite pebbles. D olomite appears to be a viable medium for determining mass loss rates over quite modest time periods providing that the number of samples is sufficien tly large. (C) 2001 Elsevier Science BN. All rights reserved.