Supplementary Material

Organic carbon hidden in urban ecosystems

J.L. Edmondson, Z.G. Davies, N. McHugh, K.J. Gaston, J.R. Leake

Soil sample preperation validation

Soil sampling methodology

Four replicate soil profiles were taken, from 0 – 100 cm depth, from a representative urban park in Leicester and a further four soil profiles were sampled from an arable field in the agricultural hinterland surrounding the study city. At each site soil profiles were taken from the four corners of a 1 x 1 m quadrat. Soils were sampled using a bulk density corer as detailed in the manuscript.

Soil sample preparation and analysis

It was not possible to split the soil cores in half lengthwise and still obtain reliable bulk density estimates for both halves as the distribution of stones would need to be identical for both halves. Thus, the eightsoil profiles were allocated alternately to one or other of the sample preparation treatments (the conventional method using a 2 mm sieve to remove stones, followed by oven drying at 105 oC and ball milling and analysis of the <2mm fraction; and our method which involved oven drying at 105 oC before ball milling, followed by 1mm sieving to remove stones, and analysis of the <1 mm fraction). In both cases the inorganic carbon was removed from the ball milled soil using HCl, and samples were analysed for organic carbon, using a CN analyser, following the methodology outlined in the manuscript.

Statistical analysis

The proportion of soil total weight removed by the 1 mm or 2 mm seive (%) and thestone-free (>2mm before ball milling, >1mm after ball milling) soilbulk density (g cm-3), OC concentration (mg g-1total soil weight), and soil OC density (mg cm-3) were all compared across the two methods using analysis of covariance (ANCOVA). The factors in the ANCOVA’s were method (1mm or 2 mm) and land-use (urban park or arable) and the covariate was soil depth. Per cent and concentration data were arcsine transformed prior to analysis.

Results

Descriptive statistics for the proportion of soil mass removed by the seiving process (% dry weight), soil fine earth bulk density (g cm-3), soil OC concentration (mg g-1of total soil weight), and soil OC density (mg cm-3), show no difference between the two soil sample preparation methodologies (Table 1). The mean soil OC concentrationdetermined across the soil profiles was 12.2 mg C g-1 ± 1.7 for the 1 mm methodology and 12.5 mg C g-1± 2.0 for the 2 mm methodology. The soil sample analysis methodology had no significant effect on the proportion of material removed by sieving, despite the differences in sieve sizes (Table 2). Furthermore there was no effect of sample analysis methodology on the soil fine earth bulk density (g cm-3), soil OC concentration (mg g-1) (Fig. 1), or soil OC density (mg cm-3) (Table 2).

Conclusion

In conclusion the two methods, as expected, are shown to be equivalent. The 1 mm method used for sample preparation and analysis did not cause a systematic over- or under-estimation of OC stocks when compared to the conventional 2 mm method. Furthermore the 1 mm method maximised the efficiency of sample preparation and analysis, and tended to give more accurate measurements of soil bulk density and OC concentration as reflected by slightly lower amounts of soil mass discarded and lower standard errors of the means.

Fig. 1: The relationship between soil OC concentration (mg g-1 total soil weight) and soil depth in arable and urban soils treated with the 1 mm sieve method (brown symbols and dashed line of best fit; y = 43.963(+2.313 S.E.)e - 0.031 (+ 0.002 S.E.)x) and 2 mm sieve method (green data points and a continuous line of best fit;y = 48.909(+2.378 S.E.)e - 0.035 (+ 0.002 S.E.)x).

Table 1: Analysis of covariance of the effect soil sample preparation method, land use and soil depth on the proportion of soil weight removed by seive, soil fine earth bulk density (<2mm and <1mm), soil OC concentration, and soil OC density.

1 mm method / 2 mm method
Mean / S.E. / Median / Range / Interquartile range / Mean / S.E. / Median / Range / Interquartile range
Proportion of soil weight removed by seive (%) / 1.8 / 0.6 / 0.6 / 0.0-15.1 / 2.3 / 3.3 / 1.3 / 0.4 / 0.0-35.0 / 2.2
Soil fine earth bulk density (g cm-3) / 1.4 / 0.1 / 1.5 / 0.9-1.7 / 0.4 / 1.3 / 0.1 / 1.3 / 0.9-1.7 / 0.5
Soil OC concentration (mg g-1total soil weight) / 12.2 / 1.7 / 6.5 / 2.5-35.9 / 15.2 / 12.5 / 2.0 / 6.2 / 2.9-39.1 / 15.4
Soil OC density (mg cm-3) / 14.7 / 1.9 / 9.1 / 4.1-33.6 / 16.8 / 15.2 / 2.3 / 9.0 / 3.6-37.3 / 23.7

Table 2: Analysis of covariance of the effect soil sample preparatin method, land use and soil depth on the proportion of soil weight removed by seive, soil fine earth bulk density, soil OC concentration, and soil OC density.

d.f. / F / P value
Proportion of soil weight removed by seive (%)
Factor: Method (1 mm or 2 mm) / 1,61 / 1.195 / 0.279
Factor: Land use (urban park or arable) / 1,61 / 1.125 / 0.293
Covariate: Soil depth / 1,61 / 3.472 / 0.067
Factor interaction: Method x land use / 1,61 / 0.173 / 0.679
Soil fine earth bulk density (g cm-3)
Factor: Method (1 mm or 2 mm) / 1,48 / 2.728 / 0.105
Factor: Land use (urban park or arable) / 1,48 / 3.972 / 0.052
Covariate: Soil depth / 1,48 / 96.418 / <0.001
Factor interaction: Method x land use / 1,48 / 0.018 / 0.894
Soil OC concentration (mg g-1)
Factor: Method (1 mm or 2 mm) / 1,63 / 0.005 / 0.942
Factor: Land use (urban park or arable) / 1,63 / 5.126 / 0.027
Covariate: Soil depth / 1,63 / 276.826 / <0.001
Factor interaction: method x land use / 1,63 / 0.002 / 0.888
Soil OC density (mg cm-3)
Factor: Method (1 mm or 2 mm) / 1,48 / 0.032 / 0.859
Factor: Land use (urban park or arable) / 1,48 / 1.430 / 0.238
Covariate: Soil depth / 1,48 / 187.039 / <0.001
Factor interaction: Method x land use / 1,48 / 0.047 / 0.816