soil carbon storage controlled by interactions between ... · carbon, the stability of organic...

1

The role of geochemistry-climate interactions on soil carbon storage

Sebastian Doetterl, Antoine Stevens, Johan Six, Roel Merckx, Kristof Van Oost, Manuel Casanova

Pinto, Angélica Casanova-Katny, Cristina Muñoz, Mathieu Boudin, Erick Zagal Venegas, Pascal

Boeckx

Supplemented information

SI 1 Supplementary Discussion

1.1 Interpretation of the partial correlation analysis

The variable importance assessment on the best predicting models (Tables 1, S5) shows very clearly

that precipitation, as a predictor variable, plays a minor role when compared to geochemical predictors

when building prediction models for SOC response variables.

However, this does not mean that the importance of precipitation as a driving factor for SOC dynamics

can be dismissed. But from our data we conclude that geochemistry has higher information content.

Partial correlations support the conclusions of this analysis, showing that the correlations of SOC

response variables with precipitation drop to zero when the correlation is controlled for the

geochemical variables (Figures 2 & S2). One reason for this is that precipitation does not include

information on the geochemical features of the parent material. On the other hand, precipitation has a

similar effect on the correlation of Si and BS with SOC variables (extended data figure S2, extended

data table S8). This is due to their high dependency on precipitation, as precipitation is the main driver

for the chemical weathering, and hence loss and alteration of minerals (see for example 37).

In summary, the data indicate a strong interrelation between geochemical soil properties and climatic

conditions along the investigated transect, predominantly through precipitation patterns and not

through temperature patterns. The control of temperature on geochemical predictors was not

significant along the investigated transect. However, temperature does have an influence on the

Soil carbon storage controlled by interactions between geochemistry and climate

1

The role of geochemistry-climate interactions on soil carbon storage

Sebastian Doetterl, Antoine Stevens, Johan Six, Roel Merckx, Kristof Van Oost, Manuel Casanova

Pinto, Angélica Casanova-Katny, Cristina Muñoz, Mathieu Boudin, Erick Zagal Venegas, Pascal

Boeckx

Supplemented information

SI 1 Supplementary Discussion

1.1 Interpretation of the partial correlation analysis

The variable importance assessment on the best predicting models (Tables 1, S5) shows very clearly

that precipitation, as a predictor variable, plays a minor role when compared to geochemical predictors

when building prediction models for SOC response variables.

However, this does not mean that the importance of precipitation as a driving factor for SOC dynamics

can be dismissed. But from our data we conclude that geochemistry has higher information content.

Partial correlations support the conclusions of this analysis, showing that the correlations of SOC

response variables with precipitation drop to zero when the correlation is controlled for the

geochemical variables (Figures 2 & S2). One reason for this is that precipitation does not include

information on the geochemical features of the parent material. On the other hand, precipitation has a

similar effect on the correlation of Si and BS with SOC variables (extended data figure S2, extended

data table S8). This is due to their high dependency on precipitation, as precipitation is the main driver

for the chemical weathering, and hence loss and alteration of minerals (see for example 37).

In summary, the data indicate a strong interrelation between geochemical soil properties and climatic

conditions along the investigated transect, predominantly through precipitation patterns and not

through temperature patterns. The control of temperature on geochemical predictors was not

significant along the investigated transect. However, temperature does have an influence on the

SUPPLEMENTARY INFORMATIONDOI: 10.1038/NGEO2516

NATURE GEOSCIENCE | www.nature.com/naturegeoscience 1

© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved© 2015 Macmillan Publishers Limited. All rights reserved

http://dx.doi.org/10.1038/ngeo2516

2

availability of water for biochemical reactions and is non-linearly correlated to SOC storage (See

interactive figure). Although precipitation has a strong influence on the correlation between SOC

response variables and geochemical predictors, some relationships are unaffected by climatic

conditions (i.e. Al or clay content) which are predominately related to the mineralogy of the parent

material and its weathering status.

Indirect and direct control on soil carbon storage

We defined a predictor as a direct control if a change in the predictor triggers a direct response of

the concerning SOC variable. A predictor is defined as an indirect control if change in the predictor

triggers change in another factor that in turn influences the investigated SOC variable directly. For

example, increased precipitation leads to higher soil reactivity and net primary productivity (NPP),

hence stimulating the association of carbon with minerals resulting in higher SOC stocks and lower

specific respiration rates (Figure 1).

Naturally, precipitation affects our geochemical indices as a control on chemical weathering and also

indirectly through influencing plant (NPP) and microbial activity, which could induce biological

driven mineral weathering, producing biogenic silica to stabilize C, etc. All variables in our model

have to be seen as proxies for the key environmental conditions controlling SOC dynamics. However,

these proxies are either directly related to the actual processes controlling SOC dynamics, and hence

more exact (geochemical variables), or further removed from these processes and rather integrative

across a wide range of factors (climate variables).

1.2 Caveats and other influences on soil carbon stabilization

In this section we discuss further factors controlling SOC stabilization in soils: this discussion covers

the main elements of our current knowledge on the factors influencing SOC stability in soils, i.e. (I)


3

Temperature sensitivity of different carbon pools, (II) physical and geochemical protection

mechanisms of SOC in soils, (III) molecular structure and recalcitrance of molecules to

decomposition, (IV) potential sources for SOC priming or nutrient limitation in soils, (V) microbial

community structure and enzyme activity as well as (VI) environmental controls (soil temperature,

soil moisture and oxygen limitation).

Chemically, one of the most important factors for stabilization of carbon in soils is availability of

mineral surfaces for sorptive protection of SOC38. Among the factors that determine the availability

of mineral surfaces for sorptive stabilization are soil texture and soil mineralogy, which control the

chemical interactions that SOC forms with the mineral phase10. Reactive soil minerals and availability

of sorptive surfaces also contribute towards the formation of soil aggregates that provide physical

stabilization of SOC, by allowing the SOC to become encapsulated inside aggregate spaces where the

diffusion of oxygen, water, and enzymes needed for breakdown of organic matter is limited39. Clay-

sized minerals tend to be the most important mineral constituents of organo-mineral associations given

their high specific surface area and hydroxylated reactive surfaces40. Reactivity of silicate minerals

with SOC depends on their type (expandable 2:1 versus non-expandable 2:1 or 1:1 phyllosilicate clay

minerals) and size (i.e. specific surface area). However, there is an increasing number of studies that

show that the relationship between surface area/clay content and organic matter accumulation is not

as straight-forward as previously assumed. Vogel et al.41 showed that carbon is preferentially

stabilized in certain hot-spot zones where stabilized C is already present, and that only a limited

portion of the clay-sized surfaces contribute to SOC sequestration. This is in line with studies on

grassland showing the limited importance of clay content for explaining C accumulation42. In andic

soils, the accumulation of reactive Fe-, Mn-, and Al-oxy-hydroxides rather than silt or clay content

and climate conditions, was identified as the most important factor that controls organic matter (OM)

levels43,44.


4

In addition, certain C compounds in soil, such as fire-altered C (also known as black carbon (BC) or

pyrogenic C (PyC)), lignin, and lipids are assumed to be ‘inherently recalcitrant’ and remain stable in

soils over centuries to millennia45,46. However, recent studies indicate that this may only hold true for

short-term changes (< 10 yrs), while environmental variables of SOC decomposition become more

important on longer time scales. Carbon quality can further influence decomposition rates as fresh,

undecomposed material can be an essential source of energy for decomposers to volatize more

recalcitrant organic matter, an effect commonly described as priming47. In the absence of fresh organic

carbon, the stability of organic carbon might be maintained. Similarly, soil organic matter turnover is

affected by the size and the diversity of microbial populations, which is in return influenced by the

supply of energy-rich litter compounds48. Hence, carbon stability in soils depends on a complex

interplay of bio-geochemical factors, constrained by environmental conditions3,49.

The environmental controls on stability of SOC19 ,50,51,52 include availability of optimal temperature

and moisture conditions for C sequestration through plants photosynthesis and C release through

decomposition by microorganisms. For example, studies have shown that soil humidity53,54,

aeration55,56 and soil temperature3 are key controls on microbial activity and hence SOC stability.

Physical inaccessibility as dictated by aggregation, and spatial separation of decomposers and SOC in

soils19,57,58 are further environmental controls on SOC persistence. Finally, it is likely that each of

these variables interact in a complex manner affecting SOC stability49,59. For example, organo-mineral

associations can, in addition, be physically separated from decomposers or placed in a low oxygen,

water saturated environment. Similarly, the temperature sensitivity of C decay is dependent on its

chemical properties. More biogeochemically recalcitrant organic matter has greater temperature

sensitivity than more labile material59.


5

1.3 Interpreting the role of clay for stabilizing SOC.

Despite the well-investigated relationship found between higher clay content19, or short range ordered

minerals60 and higher SOC stocks and slower turnover, clay content did not emerge as an important

correlate in our analysis for SOC storage (Table S5). First, our dataset does not span the full range of

values needed to identify clay as a strong driver for SOC stocks. Studies illustrating the importance

of clay content are usually spanning 0-80% clay19. In our study, except for one observation, clay

content is between 8-24%.

Finally, by integrating geochemical soil properties such as Si and soil texture into the same analysis,

clay content will not pop-up as an important predictor if its explanatory power is covered already by

Si or another variable, which contains more information than the clay content alone. Furthermore,

Percival et al.42 could show that in New Zealand grassland soils, clay content relates only poorly to

long-term soil organic C accumulation, while other geochemical characteristics had a much higher

explanatory power. In addition, recent studies61 indicate that the type of clay minerals (expandable vs.

non-expandable clay minerals) and the presence of very reactive Al and Fe-hydroxides are more

important parameters to explain correlations of SOC with minerals than clay content by itself.

1.4 The time-scale of the soil mineral reactivity response to climatic changes and its effect on SOC

dynamics

The alteration of minerals can act very rapidly in areas where the degree of former weathering is low

and both water and heat are sufficiently available to fuel (bio)chemical weathering15,62 of minerals.

Hence, weathering rates can increase sharply under certain conditions, bringing them very close to

those of C stabilization. Empirical data from chronosequence studies63,64 and current soil weathering

models65 suggest exponential weathering curves, meaning that less weathered material at the surface

will change very rapidly in the beginning and this rates decrease with time or soil depth. Examples of


6

an increase in weathering rates due to global warming are arctic soils or forefields of retreating

glaciers, where organic matter is stabilized by association with minerals. A series of studies on recent

warming include arctic regions66 or areas of glacial retreat67 showed that a less weathered soil,

experiencing suddenly higher chemical weathering can respond very rapidly (annual to decadal

timescales) in stabilizing C with minerals. The mineralogy in areas with low amounts of available

water and/or low temperatures will then react particularly strongly to climatic changes. For example,

in arctic soils temperature increase can lead to higher chemical reactivity of these soils, which might

enhance the stabilization of SOC with the mineral phase and outbalance to some extent higher

decomposition rates66. In soils from hot arid areas, C dynamics will most likely not be affected by a

temperature increase, but higher amounts of precipitation might stimulate the weathering cycle and

the biologic activity. In tropical areas, reactive minerals have been washed out and altered to less

reactive forms due to millions of years of weathering. Hence, temperature or precipitation changes

will likely not lead to significant changes in mineral reactivity. In humid temperate zones, where

mineral weathering is not as advanced as in the tropics due to rejuvenation of the soil landscape during

glacial cycles, the weathering cycle might be stimulated by higher temperatures and precipitation.

However, the effect of this on SOC dynamics cannot be fully detangled from NPP increases, as

conditions for higher chemical weathering are also conditions where plant life can thrive68,69. SOC

stocks in permafrost soils, for example, restructure very rapidly by losing unprotected POM after

thawing, but at the same time increase C stabilization with minerals and experience higher NPP after

warming takes place66. Furthermore, a system with low NPP can still have high SOC stocks if

conditions for stabilization are favorable in soils and vice versa. Support for this claim is shown in

Figure 1: Areas with presumably high NPP are at the same time areas where most of the C is stabilized

with minerals, as soil reactivity/weathering is high under these conditions. Potential specific

respiration rates (normalized for SOC content) deliver further support for our claim showing that the


7

potential specific respiration rate in areas with mineral stabilized C is very low, indicating high

stability and, hence, tendency to accumulate. The opposite can be observed for the profiles in the hot

arid and cold arid climate zones.

Some of our sites in climatically extreme conditions might face rapid changes in SOC stocks if

chemical weathering rates increase, given the high amount of primary minerals and the low amount

of weathering they have been confronted with in the past. The existing research on SOC dynamics in

these areas suggests that C stabilization mechanisms like aggregation and association of C with clay

particles are likely to respond very rapidly to increased C input and chemical weathering.

1.5 Distribution of SOC fractions

Similar to SOCStock and SPR, geochemical variables are of great importance for predicting SOC

fractions (Table S5) and resulted generally in good predictions (R2 = 0.39 - 0.79), while precipitation

or temperature hold little prediction power (Table S4; Extended Figure 2). In contrast to the main C

response variables described above, soil texture (silt and clay content) was identified as an important

predictor for the CPOM (silt negatively correlated to CPOM) and the s+c associated fraction (clay

positively correlated to s+c). However, soil texture was not identified as an important variable for the

aggregate associated C fraction. One possible explanation could be that aggregate C consist of both

particulate organic matter and silt and clay19 associated C. As a consequence, the formation of

aggregates is dependent on high C input rates and a suitable mineralogical set-up in a chemically

highly reactive system. This is supported by our observations on the spatial distribution of the three

investigated fractions: silt and clay associated C does not show a distinct latitudinal pattern and is

likely the fraction with the biggest dependency on geochemical factors. However, it can contribute

significantly to total SOCStock (up to 47%), both in areas where CPOM or aggregate associated C are

high. Similar to the SOCMicrobial and SOCMineral pools, the CPOM and aggregate-associated C fractions


8

follow opposing spatial trends. The CPOM associated C fraction is largest (up to 57% of SOCStock) in

climatically extreme regions (hot or cold arid) with low chemical weathering rates, small SOCStock

and presumably small C input rates. In contrast, the microaggregate associated C fractions is largest

(up to 79% of SOCStock) where chemical weathering rates and SOCStock are high, with presumably

high C input rates to soils (warm-humid) (see Table S1; S3 and interactive graph). In conclusion,

similar as for the investigated main SOC response variables, the distribution of SOC fractions reflects

the complex interplay of climatic and geochemical factors, (Figure S2, Table S8) resulting in a

complex spatial pattern. In conclusion, changes in some of the identified controls might lead to a

strong differential response of these fractions, but not necessarily affect SOC storage, respiration or

turnover in the same way.

SI 2 Supplementary references

37. Lincoln, N., Chadwick, O. & Vitousek, P. Indicators of soil fertility and opportunities for precontact

agriculture in Kona, Hawai'i. Ecosphere 5, 1-20 (2014).

38. Kleber, M. et al. Mineral-Organic Associations: Formation, Properties, and Relevance in Soil

Environments (Elsevier Ltd, Oxford, 2014).

39. Berhe, A.A. Decomposition of organic substrates at eroding vs. depositional landform positions. Plant

Soil 350, 261-280 (2012).

40. Totsche, K.U. et al. Biogeochemical interfaces in soil: The interdisciplinary challenge for soil science.

J. Plant Nutr. Soil Sci. 173, 88-90 (2010).

41. Vogel, C. et al. Submicron structures provide preferential spots for carbon and nitrogen sequestration

in soils. Nat. Commun. 5, doi: 10.1038/ncomms3947 (2014).

42. Percival, H., Parfitt, R. & Scott, N. Factors controlling soil carbon levels in New Zealand grasslands:

is clay content important? Soil Sci. Soc. Am. J. 64, 1623-1630 (2000).

43. Matus, F., et al. Relationship between extractable Al and organic C in volcanic soils of Chile.

Geoderma 148, 180–188 (2008).

44. Matus, F., Rumpel, C., Neculman, R., Panichini, M. & Mora, M.L. Soil carbon storage and stabilisation

in andic soils: A review. Catena 120, 102–110 (2014).

45. Kleber, M. et al. Old and stable soil organic matter is not necessarily chemically recalcitrant:

implications for modeling concepts and temperature sensitivity. Glob. Planet. Chang. 17, 1097–1107

(2011).

46. Singh, N., Abiven, S., Torn, M.S. & Schmidt, M.W.I. Fire-derived organic carbon in soil turns over on

a centennial scale. Biogeosciences 9, 2847-2857 (2012).


9

47. Fontaine, S. et al. Stability of organic carbon in deep soil layers controlled by fresh carbon supply.

Nature 450, 277-280 (2007).

48. Fontaine, S. & Barrot, S. Size and functional diversity of microbe populations control plant persistence and

long-term soil carbon accumulation. Ecol. Lett. 8, 1075-1087 (2005).

49. Schmidt, M.W.I. et al. Persistence of soil organic matter as an ecosystem property. Nature 478, 49-56

(2011).

50. Salomé, C., Nunan, N., Pouteau, V., Lerch, T.Z. & Chenu, C. Carbon dynamics in topsoil and in subsoil

may be controlled by different regulatory mechanisms. Glob. Planet. Chang. 16, 416-426 (2010).

51. Six, J., Elliott, E.T., Paustian, K. & Doran, J.W. Aggregation and soil organic matter accumulation in

cultivated and native grassland soils. Soil Sci. Soc. Am. J. 62, 1367-1377 (1998).

52. McHale, G., Newton, M.I. & Shirtcliffe, N.J. Water-repellent soil and its relationship to granularity,

surface roughness and hydrophobicity, A materials science view. Eur. J. Soil Sci. 56, 445-452 (2005).

53. Thomsen, I.K., Schjønning, P., Olesen, J.E. & Christensen, B.T. C and N turnover in structurally intact

soils of different texture. Soil Biol. Biochem. 35, 765-774 (2003).

54. Arya, L.M., Leij, F.J., Shous, P.J. & Van Genuchten, M.T. Relationship between the hydraulic

conductivity function and the particle-size distribution. Soil Sci. Soc. Am. J. 63, 1063-1070 (1999).

55. Zhang, L., Zeng, G. & Tong, C. A review on the effects of biogenic elements and biological factors on

wetland soil carbon mineralization. Acta Ecol. Sin. 31, 5387-5395 (2011).

56. Holden, P.A. & Fierer, N. Microbial processes in the vadose zone. Vadose Zone J. 4, 1-21 (2005).

57. Rumpel, C. & Koegel-Knabner, I. Deep soil organic matter-a key but poorly understood component of

terrestrial C cycle. Plant Soil 338, 143-158 (2011).

58. Lehmann, J., Kinyangi, J. & Solomon, D. Organic matter stabilization in soil microaggregates,

implications from spatial heterogeneity of organic carbon contents and carbon forms. Biogeochemistry

85, 45-47 (2007).

59. Craine, J. M., Fierer, N. & McLauchlan, K.K. Widespread coupling between the rate and temperature

sensitivity of organic matter decay. Nat. Geosci. 3, 854-857 (2010).

60. Giardina, C. P., Litton, C. M., Crow, S. E. & Asner, G. P. Warming-related increases in soil CO2 effux

are explained by increased below-ground carbon flux. Nat. Clim. Chang. 4, 822-827 (2014).

61. Lawrence, C. Harden, J., Xu, X., Schulz, M.S. & Trumbore, S.E. Long-term controls on soil organic

carbon with depth and time: A case study from the Cowlitz River Chronosequence, WA USA.

Geoderma 247-248, 73-87 (2015).

62. Song, Z., Müller, K. & Wang, H. Biogeochemical silicon cycle and carbon sequestration in agricultural

ecosystems. Earth-Sci. Rev. 139, 268-278 (2014).

63. Cornelis, J.T. et al. Silicon isotopes record dissolution and re-precipitation of pedogenic clay minerals

in a podzolic soil chronosequence. Geoderma 235-236, 19-29 (2014).

64. Alexandrovskiy, A.L. Rates of soil-forming processes in three main models of pedogenesis. Rev. Mex.

Cienc. Geol. 24, 283-292 (2007).

65. Finke, P. & Hutson, J. Modelling soil genesis in calcaerous löss. Geoderma 145, 462-479 (2008).

66. Sistla, S.A. et al. Long-term warming restructures Arctic tundra without changing net soil carbon

storage. Nature 497, 615-618 (2013).

67. Duemig, A., Smittenberg, R. & Koegel-Knabner, I. Concurrent evolution of organic and mineral

components during initial soil development after retreat of the Damma glacier, Switzerland. Geoderma,

163, 83-94 (2011).


10

68. Elmendorf, S.C. et al. Plot-scale evidence of tundra vegetation change and links to recent summer

warming. Nat. Clim. Chang. 2, 453-457 (2012).

69. Pearson, R.G. et al. Shifts in Arctic vegetation and associated feedbacks under climate change. Nat.

Clim. Chang. 3, 673-677 (2013).


11

Extended data

SI 3 Extended data tables

Table S1. Basic characteristics of selected and sampled sites.

(- = no data available, STR = soil temperature regime, SMR= soil moisture regime, MAP= mean annual precipitation, MAT= mean annual temperature [Period: 1950-

2000]).

Site Local name Soil Series Chile Region WGS1984 X WGS1984 Y Elevation STR SMR MAP MAT

(ddd,ddd) (ddd,ddd) m asl mm yr-1 °C Soil Taxonomy (USDA) WRB (ISRIC)

1 Las Cardas Tambillo IV -71.25071 -30.20157 242 Thermic Aridic 98 15.2 Cambidic Haplodurid Petric Durisols

2 Los Vilos Los Vilos IV -71.50008 -31.81284 95 Isothermic Aridic 208 16.4 Torric Psamment Arenosol (Eutric)

3 Los Andes Calle Larga V -70.52162 -32.87609 768 Thermic Xeric 349 14.6 Typic Argixeroll Luvic Kastanozem / Luvic

Chernozem

4 Alhué Pudahuel MR -71.14047 -33.99911 194 Thermic Xeric 450 16.6 Vitrandic Durixeroll Duric Kastanozem / Duric

Chernozem

5 Maipú Cuesta Barriga MR -70.83242 -33.47780 489 Thermic Xeric 401 13.8 Typic Haploxerolls Kastanozem/Chernozem

6 San Antonio Bochinche V -71.61243 -33.33901 159 Isothermic Ustic 580 15.1 Typic Haploxeroll Kastanozem/Chernozem

7 Matanzas Matanzas VI -71.87564 -33.96865 124 Isothermic Ustic 568 16.9 Oxic Haplustoll Chernozem / Kastanozem /

Phaeozem

8 Chillán Santa Bárbara VIII -71.69721 -36.45816 452 Thermic Xeric 1321 12.0 Typic Haploxerand Andosol

9 Arauco Carampangue VIII -73.26686 -37.25344 13 Isomesic Udic 1431 13.4 Fluvaquentic Cambisol (Dystric)

10 Puerto Saavedra Peule IX -73.38968 -38.77412 3 Isomesic Udic 1205 13.0 Typic Endoaquept Gleysol

11 Panguipulli Choshuenco XIV -72.11120 -39.85941 263 Isomesic Udic 2108 11.0 Andic Dystrudept Cambisol (Dystric)

12 Corral Hueicoya XIV -73.41283 -39.93245 6 Isomesic Udic 2174 11.6 Typic Haplohumult Acrisol (Humic)

13 Purranque Corte Alto X -73.15404 -40.90340 118 Isomesic Udic 1456 10.9 Typic Hapludand Andosol

14 Chiloé Island Pachabrán X -73.82429 -42.42186 204 Isomesic Perudic 2233 9.9 Histic Duraquand Petroduric Histic Andosol

15 Chiloé Island Aituí X -73.61712 -43.05791 30 Isomesic Perudic 2232 10.8 Hydric Fulvudand Fulvic Andosol

16 La Junta La Junta XI -72.39653 -43.96393 43 Isomesic Perudic 2308 10.3 Thaptic Hapludand Thaptic Andosol

17 Chacabuco Aisén XI -72.81482 -45.48736 33 Mesic Udic 2120 6.7 ~Andic Dystrudept Cambisol (Dystric)

18 Coyhaique Simpson XI -72.90757 -45.78639 462 Mesic Udic 1524 3.2 Andic Dystrudept Cambisol (Dystric)

19 Puerto Sánchez Murta XI -72.61276 -46.56823 330 Mesic Udic 1048 7.0 Typic Hapludand Andosol

20 Puerto Natales Ultima Esperanza XII -72.16488 -51.80696 67 Isomesic Udic 394 6.5 ~Mollisol Kastanozem

21 Punta Arenas Agua Fresca XII -70.98860 -53.43267 58 Cryic Udic 620 6.2 ~Entisol Leptosol

22 Porvenir Santa Olga XII -70.36106 -53.31478 59 Cryic Udic 483 6.3 ~Entisol Leptosol

23 Admiralty Bay King George Island - -58.46611 -62.16196 52 Cryic Ustic 797 -2.9 Gelisol Cryosol

24 Byers Peninsula Livingstone Island - -61.08332 -62.65007 5 Cryic Udic 648 -2.3 Gelisol Cryosol


12

Table S2. Description of dominant and secondary vegetation for all sampling sites.

Values for biomass dry weight are taken from Ruiz13 for Chile (Sites 1-22) and from Barcikowski et al.14 for the

Antarctic Peninsula (Sites 23-24).

Site Local name Biomass

Dry weight kg

ha-1 yr-1

Vegetation

Dominant + (secondary)

1 Las Cardas 300-600

Woody plants : Colliguaya odorífera, Trichocereus chilensis, Fluorensia thurifera, Acacia

caven Grasses: Erodium cicutarium, Adesmia tenella, Vulpia dertonensis, Plantago hispidula,

Erodium moschatum, Stipa lacchnophylla, Pectocarya dimorpha,

2 Los Vilos 400- 800 Woody plants: Bahia ambrosioides, Puya chilensis, Oxalis gigantean, Fuchsia lysioides. Grasses: Nassella chilensis, Piptochaetium stipoides, Dichondra repens, Trifolium

megalanthum

3 Los Andes 800-2000 Woody plants: Acacia Caven

Grasses: Plantago firma, Adesmia angustifolia, Adesmia tenella , Medicago polymorpha , Plantago hispidula, Aira caryophillea , Avena barbata

4 Maipú-Alhué 1000-2500 Woody plants: Acacia caven

Erodium moschatum , Avena barbata, Bromus hordeasus, Vulpia dertonensis, Raphanus sativus + Chaetanthera chilensis, Aira caryophyllea

5 Maipú 1000-2500 Woody plants: Acacia caven, Baccharis linearis

Grasses: Trisetobromus hirtus, Erodium moschatum, Avena barbata, Bromus hordeasus,

Vulpia dertonensis, Raphanus sativus, Chaetanthera chilensis and Aira caryophyllea )

6 San Antonio 1200-3500 Woody plants: Baccharis linearis

Grasses: Lolium multiflorum, Medicago polymorpha, Bromus hordeasus, Leontodon

taraxacoides, Nasella chilensis

7 Matanzas 1500-3500 Grasses: Lolium multiflorum, Briza máxima, Briza minor, Medicago polymorpha, Trifolium subterraneum, Hordeum murinum

8 Chillán 2000-5000 Grasses: Stipa neesiana, Piptochaetium stipoides, Lolium multiflorum, Briza máxima Bromus

hordeasus Hordeum murinum , Plantago lanceolata, Medicago polymorpha

9 Arauco 3000-6000 Grasses: Holcus lanatus, Trifolium repens, Agrostis tenuis, , Lotus uliginosus, Taraxacum

officinalis, Cynosurus echinatus

10 Puerto Saavedra 3000-6500 Trifolium repens, Holcus lanatus, Hypochoeris radicata, Plantago lanceolada,

11 Panguipulli 4000-7000 Holcus lanatus, Taraxacum officinale, Deschampsia Antarctica and Leucanthemum vulgare

12 Corral 5000-8000 Bromus valdivianus, Trifolium repens, Lotus uliginosus, Lolium perenne .Taraxacum officinale

13 Purranque 4000-6000 Grasses: Dactylis glomerata, Paspalum dilatatum, Trifolium repens

14 Chiloé Island 6000-8000 Grasses: Lotus uliginosus, Holcus lanatus, Hypochoeris radicata, Poa pratensis,

Leucanthemum vulgare.

15 Chiloé Island 6000-8000 Grasses: Trifolium repens, Dactylis glomerata, Agrostis sp., Taraxacum officinale.

16 La Junta 6000-8000 Grasses: Holcus lanatus, Lotus uliginosus s, Hypochoeris radicata, Plantago lanceolada +

Agrostis tenuis

17 Chacabuco ≈4000 Grasses: Lotus uliginosus, Holcus lanatus, Deschampsia berteroanum, Hypochoeris radicata, Dactylis glomerata, Arrhenatherum elatius,

18 Coyhaique 2000-5000 Grasses: Dactylis glomerata, Trifolium repens, Holcus lanatus, Agrostis tenuis, Acaena

pinnatifida, Geranium magellanicum

19 Puerto Sánchez 3000-5000 Grasses: Arrhenatherum elatius, Dactylis glomerata Geum magellanicum, Plantago lanceolata, Acaena pinnatifida, Anemone multifida, Rumex

acetosella 20 Puerto Natales 500-1000 Grasses: Poa pratensis, Acaena magellanica

Holcus lanatus, Deschampsia flexuosa, Agrostis sp., Festuca , Hypochoeris incana (Rumex

crispus + Achillea millefolium+ Dactylis glomerata), Agrostis capillaris 21 Punta Arenas 350-850 Grasses: Festuca pallescens, Hieracium aurantiacum, Trifolium repens, Agrostis spp.,

Festuca arundicnacea, Berberis buxifolia, Deschampsia Antarctica

22 Porvenir 350-850 Grasses: Chiliotrichioum diffussum, Baccharis magellanicum, Festuca pallescens, Poa

pratensis, Trifolium repens, Dactylis glomerata, Taraxacum officinale

23 Admiralty Bay, King George Island

500-2000 Grasses: Deschampsia antarctica + (Moss Herb Tundra)

24 Byers Peninsula,

Livingstone Island

500-2000 Grasses: Deschampsia antarctica + (Polytrichastrum alpinum)


13

Table S3. Soil mineral and organic carbon parameters.

(CECpot = Potential cation exchange capacity; BStotal = Total base saturation of CECpot; TEC = Total elemental content; TRB = Total reserve in base cations; BD = Bulk

density; MD = Mineral density; SOC% = Soil organic carbon concentration; SOCStock = Soil organic carbon stock; Respiration = Specific potential CO2-C respiration

(SPR); SOCmineral = Mineral associated SOC; SOCmicrobial = Microbially available SOC; CPOM = Coarse particulate organic matter; s+c = Free silt and clay associated

C; m = (Micro-)aggregate associated C).

Site CECpot BStotal Texture TEC Al/Si TRB pHKCl BD MD SOC% SOCStock Respiration SOCMicrobial SOCMineral CPOM s+c m

2000-50 50-2 ˂2 µm Si Al Fe Mn P

0-10cm depth (SPR)

cmolc kg-1 % Vol % g kg-1 cmolc kg-1 g cm-3 g cm-3 g C kg-1 kg m-2 µg C-CO2 g-1 SOC h-1 % SOCStock (± rel. SD)

1 11.6 89 35 41 24 254 90 72 3.2 0.9 0.35 366 5.8 1.7 2.7 6.6±2.3 1.1±0.4 25.4±3.3 47±32 58±0.5 42±0.3 40±0.2 18±0.1

2 4.4 94 78 15 7 317 74 25 0.6 0.6 0.23 331 5.4 1.4 2.7 7.4±1.4 1.0±0.2 26.5±3.9 41±15 58±4.5 42±2.7 14±9.6 43±5.6

3 28.7 120 20 57 23 243 92 56 1.3 0.9 0.38 422 5.9 1.1 2.8 18.4±3.1 2.0±0.3 17.9±2.6 21±19 75±1.2 25±1.2 38±2.8 37±4.0

4 11.4 95 46 36 18 259 90 53 1.0 0.9 0.35 465 6.0 1.1 3.0 14.3±2.5 1.6±0.3 23.1±1.8 60±16 67±0.8 33±3.2 23±1.4 44±1.3

5 14.4 147 25 52 23 240 89 70 1.7 1.5 0.37 423 5.4 1.7 2.8 21.7±1 3.6±0.2 20.2±3.2 41±5 95±0.2 5±0.2 47±2.4 48±2.6

6 9.1 60 41 45 16 359 51 19 0.7 0.4 0.14 160 4.9 1.3 2.6 12.9±2.4 1.7±0.3 25.9±5.7 52±15 86±4.3 14±3.6 44±9.6 42±6.0

7 21.7 90 16 65 19 265 63 78 1.7 0.9 0.24 270 5.0 1.1 2.9 32.0±1.1 3.4±0.1 12.1±0.6 ≈1 64±2.2 36±3.2 16±1.6 48±1.6

8 33.9 15 21 66 13 203 101 58 1.8 2.2 0.49 197 5.0 0.8 2.4 60.4±4.8 5.0±0.4 3.8±0.4 8±7 97±2.4 3±2.3 46±1.2 51±1.1

9 16.4 32 16 68 16 287 84 32 0.5 0.7 0.29 185 4.1 1.6 2.6 26.7±7.4 4.3±1.2 12.2±3 28±24 64±0.4 3±0.8 46±6.1 51±7.3

10 28.2 58 10 67 23 249 90 35 0.4 1.1 0.36 278 4.5 1.1 2.4 43.9±1.7 4.9±0.2 15±2.3 89±3 96±0.6 36±0.5 23±0.3 41±0.9

11 25.7 14 61 31 7 221 80 50 1.1 1.7 0.36 429 4.5 0.8 2.6 74.7±9.3 5.8±0.7 6.8±0.6 5±10 70±2.4 4±1.5 38±1.2 59±6.4

12 19.4 7 38 57 6 265 74 35 0.5 0.9 0.28 248 3.9 0.8 2.6 57.0±1.6 4.6±0.1 9±1.2 45±2 76±2.9 30±9.3 13±1.8 57±9.3

13 46.6 30 15 45 40 179 103 70 2.0 2.2 0.57 128 4.5 0.8 2.5 77.3±2.4 6.3±0.2 7.7±3.1 5±3 90±4.0 10±3.6 38±1.0 52±6.5

14 46 11 15 62 23 200 60 48 0.4 1.2 0.30 114 4.2 0.6 1.7 123.4±9.8 7.3±0.6 7.8±0.4 4±7 89±1.8 11±1.6 37±0.2 52±1.8

15 50.4 20 32 60 8 174 48 38 0.5 2.2 0.27 117 4.6 0.6 1.7 148.3±8.5 8.2±0.5 6±2.3 ≈1 98±0.4 2±1.4 19±1.4 79±6.4

16 41.5 49 24 64 12 198 52 31 0.9 2.8 0.27 199 4.9 0.5 2.1 127.0±4.5 6.7±0.2 6.7±1.4 4±3 91±1.0 9±0.9 43±0.5 48±1.3

17 35.3 11 36 56 8 179 63 45 0.7 2.6 0.35 260 4.5 0.5 2.0 116.9±6.7 5.6±0.3 5.6±1 ≈1 88±0.6 12±9.6 11±2.5 77±9.2

18 27.4 40 24 52 19 227 85 59 1.2 2.5 0.38 445 5.3 0.7 2.6 43.4±6.5 3.2±0.5 9±1.7 28±13 91±3.0 9±2.8 43±1.3 48±1.6

19 38.6 45 41 45 14 184 99 58 1.9 3.2 0.54 301 5.3 0.6 2.4 74.4±13.5 4.7±0.9 7.6±1.8 8±16 65±5.9 35±1.1 11±1.9 54±1.9

20 27.6 88 43 45 12 273 63 28 1.2 1.4 0.23 247 5.1 0.7 2.6 70.2±5.5 5.0±0.4 18.7±1.4 11±7 69±1.0 31±4.9 10±1.7 59±1.1

21 38.3 93 41 39 20 259 59 25 1.6 1.5 0.23 223 5.5 0.7 2.5 94.9±5.4 6.7±0.4 9.1±1.8 39±5 94±0.1 6±0.1 30±1.7 64±1.8

22 29.3 72 28 51 21 282 62 28 0.9 1.5 0.22 213 4.6 0.8 2.5 67.1±2.2 5.3±0.2 14.8±1.9 36±3 94±2.7 6±2.5 24±1.2 70±1.3

23 35.2 87 45 37 19 229 98 61 1.1 3.3 0.43 555 4.2 1.7 2.4 13.2±1.3 2.3±0.2 10.7±2 19±24 43±2.3 57±1.1 14±6.6 30±4.6

24 3.2 158 81 11 8 273 75 48 0.8 1.4 0.27 438 3.8 1.7 2.5 7.3±3.5 1.2±0.6 27.5±2.7 28±18 54±0.5 46±0.3 13±2.7 40±2.9


14

Table S4. RMSE and R2 of cross-validation for the different applied modeling strategies.

(LARS = Linear least angle regression; BSS = Linear best subset; CUBIST = non linear model tree; for SOC response

variable abbreviations see caption of Table S3).

Unit RMSE Rcv2 Dependent Model

g C kg-1

11.54 0.95

SOC%

LARS

11.24 0.96 BSS

17.91 0.87 CUBIST

kg SOC m2

1.30 0.75

SOCStock

LARS

1.22 0.71 BSS

1.21 0.77 CUBIST

µg CO2-C g-1 SOC h-1

5.00 0.76

Respiration (SPR)

LARS

3.13 0.86 BSS

4.99 0.74 CUBIST

% of SOCStock

15 0.63

SOCMineral

LARS

13 0.69 BSS

17 0.44 CUBIST

% of SOCStock

21 0.41

SOCMicrobial

LARS

18 0.54 BSS

20 0.39 CUBIST

% of SOCStock

13 0.54

CPOM

LARS

11 0.59 BSS

14 0.43 CUBIST

% of SOCStock

11 0.58

s+c

LARS

10 0.65 BSS

12 0.50 CUBIST

% of SOCStock

12 0.48

m

LARS

8 0.79 BSS

14 0.47 CUBIST


15

Table S5. Relative contributions of geo-climatic variables in predicting SOC response variables.

Standardized coefficients indicating their importance in the models. R2 and RMSE are computed by cross-validation of

the LARS (Least Angle regression) and BSS (Best Subset) models (see methods for details) (MAP = Mean annual

precipitation; MAT = Mean annual temperature; Clay & Silt = Soil clay & silt content; TRB = Total reserve in “base”

cations; BS = Base saturation of potential cation exchange capacity; Si, Al, Fe, K, P, Mn = soil total content of tested

elements; pHKCl = soil pH measured in KCl solution; blanks = not selected; for SOC response variable abbreviations see

caption of Table S3).

Least Angle Linear Regression (LARS)

MAP MAT Clay Silt Si BS Al TRB pHKCl Fe K P Mn R2cv RMSEcv RMSE Unit

SOCMineral 7.61 0.63 15

% of

total SOCStock

SOCMicrobial 101 0.41 21

CPOM -15 0.54 13

s+c 17 5 1 0.58 11

m 5 -0.05 -4 -16 0.48 12

Best Subset Selection Linear Regression (BSS)

MAP MAT Clay Silt Si BS Al TRB pHKCl Fe K P Mn R2cv RMSEcv RMSE Unit

SOCMineral 0.54 -0.33 0.69 13

% of

total SOCStock

SOCMicrobial 0.32 0.72 0.54 18

CPOM -0.54 0.33 0.59 11

s+c 0.57 0.37 0.65 10

m -0.71 -0.30 -0.61 0.79 8


16

Table S6. Correlations (Pearson’s r) between predictors and SOC response variables.

SOC% SOCStock Respiration (SPR) SOCMicrobial SOCMineral CPOM m s+c

MAP 0.74 0.72 -0.77 -0.42 0.40 -0.40 0.59 -0.14

MAT -0.15 -0.10 0.21 0.25 0.18 -0.18 -0.11 0.34

Clay -0.13 0.02 0.02 0.18 0.25 -0.25 -0.31 0.65

Silt 0.40 0.50 -0.56 0.08 0.63 -0.63 0.27 0.49

Sand -0.27 -0.42 0.47 -0.13 -0.63 0.63 -0.09 -0.69

Si -0.67 -0.62 0.75 0.62 -0.41 0.42 -0.37 -0.12

BS -0.62 -0.63 0.75 0.28 -0.34 0.34 -0.48 0.10

Al -0.49 -0.36 0.01 0.00 -0.26 0.26 -0.51 0.23

Al/Si ratio 0.02 0.09 -0.40 -0.34 0.03 -0.04 -0.16 0.21

TRB -0.62 -0.66 0.36 0.06 -0.53 0.53 -0.51 -0.11

pHKCl -0.26 -0.35 0.33 0.02 0.02 -0.01 -0.25 0.29

Fe -0.24 -0.21 -0.10 -0.14 -0.02 0.02 -0.48 0.50

K -0.68 -0.67 0.70 0.42 -0.49 0.49 -0.65 0.09

P 0.46 0.38 -0.64 -0.52 0.11 -0.12 0.25 -0.13

Mn -0.21 -0.19 0.07 -0.02 -0.06 0.06 -0.44 0.41

For variable abbreviations see captions of Tables S3 and S5.


17

Table S7. Correlations (Pearson’s r) between predictors for SOC response variables.

MAP MAT Clay Silt Sand Si BS Al Al/Si ratio TRB pHKCl Ca Fe K Mn P

MAP 1 -0.12 -0.26 0.43 -0.25 -0.61 -0.81 -0.24 0.15 -0.40 -0.58 -0.24 -0.12 -0.61 -0.42 0.43

MAT -0.12 1 0.13 0.31 -0.29 0.24 -0.08 -0.01 -0.13 -0.27 0.43 -0.34 0.04 0.33 0.12 -0.60

Clay -0.26 0.13 1 0.21 -0.59 -0.12 0.14 0.39 0.39 -0.1 0.22 -0.12 0.44 0.02 0.44 -0.06

Silt 0.43 0.31 0.21 1 -0.91 -0.32 -0.48 -0.09 0.09 -0.5 -0.09 -0.43 0.06 -0.28 -0.1 0.07

Sand -0.25 -0.29 -0.59 -0.91 1 0.33 0.34 -0.1 -0.25 0.43 -0.02 0.38 -0.25 0.23 -0.11 -0.06

Si -0.61 0.24 -0.12 -0.32 0.33 1 0.44 -0.22 -0.67 0.1 0.1 -0.04 -0.44 0.56 -0.18 -0.77

BS -0.81 -0.08 0.14 -0.48 0.34 0.44 1 0.11 -0.21 0.55 0.41 0.43 0.16 0.49 0.23 -0.28

Al -0.24 -0.01 0.39 -0.09 -0.1 -0.22 0.11 1 0.83 0.49 0.19 0.46 0.6 0.18 0.44 0.21

Al/Si ratio 0.15 -0.13 0.39 0.09 -0.25 -0.67 -0.21 0.83 1 0.21 0.07 0.28 0.65 -0.23 0.45 0.59

TRB -0.40 -0.27 -0.10 -0.50 0.43 0.10 0.55 0.49 0.21 1 0.29 0.94 0.41 0.34 0.17 0.13

pHKCl -0.58 0.43 0.22 -0.09 -0.02 0.10 0.41 0.19 0.07 0.29 1 0.23 0.21 0.35 0.53 -0.13

Ca -0.24 -0.34 -0.12 -0.43 0.38 -0.04 0.43 0.46 0.28 0.94 0.23 1 0.35 0.06 0.02 0.30

Fe -0.12 0.04 0.44 0.06 -0.25 -0.44 0.16 0.6 0.65 0.41 0.21 0.35 1 0.11 0.65 0.28

K -0.61 0.33 0.02 -0.28 0.23 0.56 0.49 0.18 -0.23 0.34 0.35 0.06 0.11 1 0.38 -0.52

Mn -0.42 0.12 0.44 -0.1 -0.11 -0.18 0.23 0.44 0.45 0.17 0.53 0.02 0.65 0.38 1 0.14

P 0.43 -0.6 -0.06 0.07 -0.06 -0.77 -0.28 0.21 0.59 0.13 -0.13 0.30 0.28 -0.52 0.14 1

For variable abbreviations see captions of Tables S3 and S5.


18

Table S8. Partial correlations between SOC response and selected geochemical variables (GCV),

controlled for temperature (MAT) and precipitation (MAP).

SPR SOCStock SOC%

GCV Zero-

order MAP MAT Climate GCV

Zero-


Zero-

order MAP MAT Climate

Clay 0.02 -0.30 -0.01 -0.32 Clay 0.02 0.31 0.04 0.32 Clay -0.13 0.10 -0.11 0.11

Silt -0.56 * -0.40 -0.67 * -0.52 * Silt 0.50 * 0.30 0.56 * 0.34 Silt 0.40 0.13 0.47 * 0.19

Si 0.75 * 0.55 * 0.73 * 0.53 * Si -0.62 * -0.33 -0.62 * -0.34 Si -0.67 * -0.41 -0.66 * -0.4

BS 0.75 * 0.33 0.78 * 0.41 BS -0.63 * -0.12 -0.65 * -0.13 BS -0.62 * -0.04 -0.64 * -0.07

Al 0.01 -0.28 0.01 -0.28 Al -0.36 -0.27 -0.36 -0.27 Al -0.49 * -0.48 * -0.5 * -0.49 *

TRB 0.36 0.09 0.44 0.16 TRB -0.66 * -0.57 * -0.71 * -0.62 * TRB -0.62 * -0.51 * -0.69 * -0.58 *

pHKCL 0.33 -0.22 0.27 -0.33 pHKCL -0.35 0.12 -0.34 0.14 pHKCL -0.26 0.31 -0.22 0.39

Fe -0.10 -0.30 -0.11 -0.31 Fe -0.21 -0.17 -0.2 -0.17 Fe -0.24 -0.22 -0.24 -0.22

K 0.7 * 0.47 * 0.69 * 0.44 K -0.67 * -0.42 -0.67 * -0.44 K -0.68 * -0.43 -0.67 * -0.42

P -0.64 * -0.53 * -0.66 * -0.54 * P 0.38 0.11 0.40 0.13 P 0.46 * 0.23 0.47 * 0.23

Mn 0.07 -0.43 0.05 -0.45 Mn -0.19 0.17 -0.18 0.17 Mn -0.21 0.16 -0.20 0.17

SOCMineral SOCMicrobial

GCV Zero-


Zero-


Clay 0.25 0.40 0.23 0.39 Clay 0.18 0.08 0.15 0.06

Silt 0.63 * 0.55 * 0.61 * 0.51 * Silt 0.08 0.31 0.00 0.25

Si -0.41 * -0.23 -0.48 * -0.30 Si 0.62 * 0.51 * 0.6 * 0.49 *

BS -0.34 -0.02 -0.33 0.06 BS 0.28 -0.1 0.32 -0.04

Al -0.26 -0.19 -0.27 -0.18 Al 0.00 -0.11 0.00 -0.11

TRB -0.53 * -0.44 -0.51 * -0.39 TRB 0.06 -0.13 0.14 -0.06

pHKCL 0.02 0.33 -0.07 0.25 pHKCL 0.02 -0.3 -0.1 -0.45

Fe -0.02 0.03 -0.03 0.02 Fe -0.14 -0.21 -0.15 -0.22

K -0.49 * -0.34 -0.59 * -0.46 K 0.42 * 0.23 0.37 0.17

P 0.11 -0.07 0.28 0.11 P -0.52 * -0.42 -0.48 * -0.37

Mn -0.06 0.13 -0.08 0.11 Mn -0.02 -0.24 -0.06 -0.26

CPOM m s+c

GCV Zero-


Zero-


Zero-


Clay -0.25 -0.4 -0.23 -0.39 Clay -0.31 -0.21 -0.30 -0.20 Clay 0.65 * 0.64 * 0.65 * 0.65 *

Silt -0.63 * -0.55 * -0.62 * -0.51 * Silt 0.27 0.03 0.32 0.05 Silt 0.49 * 0.61 * 0.43 0.55 *

Si 0.42 * 0.24 0.48 * 0.31 Si -0.37 -0.02 -0.36 -0.01 Si -0.12 -0.25 -0.22 -0.35

BS 0.34 0.03 0.33 -0.06 BS -0.48 * -0.01 -0.50 * -0.03 BS 0.10 -0.02 0.14 0.09

Al 0.26 0.18 0.26 0.18 Al -0.51 * -0.47 * -0.52 * -0.48 * Al 0.23 0.21 0.25 0.24

TRB 0.53 * 0.44 0.51 * 0.39 TRB -0.51 * -0.37 -0.56 * -0.41 TRB -0.11 -0.18 -0.02 -0.07

pHKCL -0.01 -0.33 0.07 -0.25 pHKCL -0.25 0.13 -0.23 0.17 pHKCL 0.29 0.26 0.16 0.13

Fe 0.02 -0.03 0.03 -0.03 Fe -0.48 * -0.51 * -0.48 * -0.51 * Fe 0.50 * 0.49 * 0.51 * 0.51 *

K 0.49 * 0.33 0.59 * 0.46 K -0.65 * -0.45 -0.65 * -0.46 K 0.09 0.02 -0.02 -0.10

P -0.12 0.07 -0.28 -0.11 P 0.25 0 0.24 -0.03 P -0.13 -0.08 0.09 0.16

Mn 0.06 -0.14 0.08 -0.12 Mn -0.44 * -0.27 -0.43 -0.26 Mn 0.41 * 0.39 0.39 0.39

Differences between zero-order and partial correlations indicate the level of dependency of a geochemical predictor to

climatic predictors: MAP = Mean annual precipitation, MAT = Mean annual temperature or the combined effect of both

indicated as “Climate”); significance of the correlations (*) is evaluated at the 0.05 level. For variable abbreviations see

captions of Tables S3 and S5.


19

SI 4 Extended data figures

Figure S1. SOC fractionation scheme. Applied SOC fractionation scheme following 31 and its

interpretation in terms of functional SOC pools and protection mechanisms.


20

Figure S2. Partial correlations between SOC response and climatic variables. Change of correlation

between SOC response and climatic variables controlled for geochemical variables separately and

all combined (column “Geochem”). Difference between zero-order and partial correlations indicate

the level of dependency of a given predictor and the SOC response. Color and displayed numbers

indicate the strength and sign of the correlation. (No change in color between controlled variable

and zero-order = no dependency; decrease/increase of color intensity = loss of / gain of correlation).

Significance of the correlations (*) is evaluated at the 0.05 level. Abbreviations are explained in

caption of Tables S3 and S5.


21

Figure S3. Geographical location of the study plots in relation to mean annual temperature and

precipitation. Annual temperature and precipitation are averages of the period 1950-2000 taken from

the worldclim dataset24 for Chile and from the Bellingshausen and Esperanza research stations for

the Antarctic Peninsula.


soil carbon storage controlled by interactions between ... · carbon, the stability of organic...

Documents