Appendix B. Details on the processing of water samples.
In this Appendix, we examine to what extent our results on nutrient concentrations in peatland water may have been influenced by the sampling procedures. We conclude that the observed trends in nutrient concentrations in peatland water were not affected by the sampling procedures.
In all pattern-localities, water samples were acidified to prevent microbial activity and chemical transformation of solutes (e.g., Wassen et al. 1990). The water samples from Scotland and Sweden were filtered before the acidification procedure (0.2 μm nylon filter, type SY25GN, mdi). The water samples from Siberia were not filtered. Instead, the unfiltered, acidified samples were later centrifuged in the laboratory. We performed both procedures for 24 samples from Scotland and Sweden to test whether the type of procedure (filtration or centrifugation) affected the nutrient concentrations of the water samples. The type of procedure did not affect total inorganic nitrogen and potassium concentration in the mire water (One-way ANOVAs, nitrogen: F1,46 = 1.26, P = 0.268; potassium: F1,46 = 0.173, P = 0.679). However, water samples undergoing centrifugation had higher phosphorus concentration (Mann-Whitney U test, phosphorus: U = 122, P < 0.001). For phosphorus, the increase due to the centrifugation procedure tended to be higher for hummocks as compared to hollows (255% increase in hummocks, 90% increase in hollows, two-tailed t test, t20 = 1.926, P = 0.070). If the Siberian water samples would have been similarly affected by the centrifugation procedure, we would have to reduce the phosphorous concentrations in ridges by 255% and the concentration in hollows by 90%. A t test for two populations suggests that this would still yield higher phosphorus concentrations in ridges as compared to hollows (two-tailed t test, t64 = 4.101, P < 0.001). Due to the reduced difference the value of the resource contrast would also decrease, by approximately 50%. Due to such a reduction, the resource contrast in water phosphorous concentration in Siberia (0.22) would be similar to the observed contrast in the Swedish pattern on flat ground (0.26). The result would still reveal the same trend in resource contrasts along the climatic gradient: the lowest and highest value of the resource contrast occur in Scotland, the Swedish and Siberian pattern-localities have intermediate values (see Fig. 3A in the main text).
However, it is unlikely that the centrifugation procedure has increased the phosphorus concentration in the Siberian water samples, probably because of the small amount of particulate matter in the samples. In Siberia, phosphorous concentrations were also measured in water samples during field sampling (before either acidification or centrifugation) using a colorimeter (Hach-Lange DS/890). The results from the colorimetric measurements yielded similar average concentrations as reported from the water samples analyzed in the laboratory; colorimeter results for average concentrations: PHUMMOCK = 0.31 mg P/L; PHOLLOW = 0.08 mg P/L; ICP-OES results for average concentrations: PHUMMOCK = 0.29 mg P/L; PHOLLOW = 0.10 mg P/L. Summarizing, it is unlikely that the different sampling procedure in Siberia as compared to that in Scotland and Sweden has influenced our results on nutrient concentrations in mire water.
Wassen, M. J., A. Barendregt, A. Palczynski, J. T. De Smidt, and H. De Mars. 1990. The relationship between fen vegetation gradients, groundwater flow and flooding in an undrained valley mire at Bierbza, Poland. Journal of Ecology 78:1106–1122.