% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@ARTICLE{DuarteGuardia:864364,
author = {Duarte-Guardia, Sandra and Peri, Pablo L. and Amelung, Wulf
and Sheil, Douglas and Laffan, Shawn W. and Borchard, Nils
and Bird, Michael I. and Dieleman, Wouter and Pepper, David
A. and Zutta, Brian and Jobbagy, Esteban and Silva, Lucas C.
R. and Bonser, Stephen P. and Berhongaray, Gonzalo and
Piñeiro, Gervasio and Martinez, Maria-Jose and Cowie,
Annette L. and Ladd, Brenton},
title = {{B}etter estimates of soil carbon from geographical data: a
revised global approach},
journal = {Mitigation and adaptation strategies for global change},
volume = {24},
number = {3},
issn = {1573-1596},
address = {Dordrecht [u.a.]},
publisher = {Springer Science + Business Media B.V},
reportid = {FZJ-2019-04159},
pages = {355 - 372},
year = {2019},
abstract = {Soils hold the largest pool of organic carbon (C) on Earth;
yet, soil organic carbon (SOC) reservoirs are not well
represented in climate change mitigation strategies because
our database for ecosystems where human impacts are minimal
is still fragmentary. Here, we provide a tool for generating
a global baseline of SOC stocks. We used partial least
square (PLS) regression and available geographic datasets
that describe SOC, climate, organisms, relief, parent
material and time. The accuracy of the model was determined
by the root mean square deviation (RMSD) of predicted SOC
against 100 independent measurements. The best predictors
were related to primary productivity, climate, topography,
biome classification, and soil type. The largest C stocks
for the top 1 m were found in boreal forests (254 ± 14.3 t
ha−1) and tundra (310 ± 15.3 t ha−1). Deserts had
the lowest C stocks (53.2 ± 6.3 t ha−1) and
statistically similar C stocks were found for temperate and
Mediterranean forests (142 - 221 t ha−1), tropical and
subtropical forests (94 - 143 t ha−1) and grasslands
(99-104 t ha−1). Solar radiation, evapotranspiration, and
annual mean temperature were negatively correlated with SOC,
whereas soil water content was positively correlated with
SOC. Our model explained $49\%$ of SOC variability, with
RMSD (0.68) representing approximately $14\%$ of observed C
stock variance, overestimating extremely low and
underestimating extremely high stocks, respectively. Our
baseline PLS predictions of SOC stocks can be used for
estimating the maximum amount of C that may be sequestered
in soils across biomes},
cin = {IBG-3},
ddc = {690},
cid = {I:(DE-Juel1)IBG-3-20101118},
pnm = {255 - Terrestrial Systems: From Observation to Prediction
(POF3-255)},
pid = {G:(DE-HGF)POF3-255},
typ = {PUB:(DE-HGF)16},
UT = {WOS:000456264900002},
doi = {10.1007/s11027-018-9815-y},
url = {https://juser.fz-juelich.de/record/864364},
}