\makeatletter
\@ifundefined{HCode}
{\documentclass[screen,CRGEOS,Unicode,biblatex,published]{cedram}
\addbibresource{crgeos20250938.bib}
\newenvironment{Table}{\begin{table}}{\end{table}}
\newenvironment{noXML}{}{}
\let\citep\parencite
\let\citet\textcite
\def\xcitealp#1#2{\citeauthor{#1}, \citelink{#1}{#2}}
\newcommand*{\citelink}[2]{\hyperlink{cite.\therefsection @#1}{#2}}
\def\defcitealias#1#2{}
\let\citetalias\textcite
\let\citepalias\parencite
\def\thead{\noalign{\relax}\hline}
\def\endthead{\noalign{\relax}\hline}
\def\tabnote#1{\vskip4pt\parbox{.95\linewidth}{#1}}
\def\botline{\\\hline}
\def\dollar{\$}
\def\tbody{\noalign{\relax}}
\def\tsup#1{$^{{#1}}$}
\def\tsub#1{$_{{#1}}$}
\def\og{\guillemotleft}
\def\fg{\guillemotright}
\let\permil\textperthousand
\RequirePackage{etoolbox}
\usepackage{upgreek}
\usepackage{wasysym}
\usepackage{pifont}
\def\jobid{crgeos20250938}
%\graphicspath{{/tmp/\jobid_figs/web/}}
\graphicspath{{./figures/}}
\newcounter{runlevel}
\usepackage{multirow} 
\def\nrow#1{\@tempcnta #1\relax%
\advance\@tempcnta by 1\relax%
\xdef\lenrow{\the\@tempcnta}}
\def\morerows#1#2{\nrow{#1}\multirow{\lenrow}{*}{#2}}
\let\MakeYrStrItalic\relax
\def\refinput#1{}
\def\back#1{}
\csdef{Seqnsplit}{\\}
\newcommand*{\hyperlinkcite}[1]{\hyper@link{cite}{cite.#1}}
\newenvironment{inftab}{}{}
\DOI{10.5802/crgeos.332}
\datereceived{2025-11-21}
\daterevised{2026-02-09}
\dateaccepted{2026-03-16}
\ItHasTeXPublished
}
{
\PassOptionsToPackage{authoryear}{natbib}
\documentclass[crgeos]{article}
\usepackage[T1]{fontenc} 
\usepackage{pifont}
\def\CDRdoi{10.5802/crgeos.332}
\def\href#1#2{\url[#1]{#2}}
 \def\citelink#1#2{\citeyear{#1}}
 \def\xcitealp#1#2{\citealp{#1}}
\makeatletter
\def\CDRsupplementaryTwotypes#1#2{}
 \def\selectlanguage#1{} 
 \def\hyperlinkcite#1#2{\citeyear{#1}}
\let\newline\break
}
\makeatother

\dateposted{2026-05-05}
\begin{document}

\begin{noXML}

\editornote{Article submitted by invitation}
\alteditornote{Article soumis sur invitation}

\CDRsetmeta{articletype}{review}

\TopicFR{\'Ecosyst\`emes terrestres et aquatiques}
\TopicEN{Terrestrial and aquatic ecosystems}

\title{Advancing our understanding of the dynamics of silicon at the
soil-plant scale since Bartoli's work (1980s)}

\alttitle{Avanc\'{e}es dans la compr\'{e}hension de la dynamique du
silicium \`{a} l'\'{e}chelle sol-plante depuis les travaux de Bartoli
(ann\'{e}es 1980)}

\author{\firstname{Jean Dominique} \lastname{Meunier}\CDRorcid{0000-0002-0017-5877}}
\address{CNRS, Aix Marseille University, IRD, INRAE, CEREGE,
Aix-en-Provence, France}
\email{meunier@cerege.fr} 

\keywords{\kwd{Silicon}\kwd{Biogeochemical
cycles}\kwd{Phytoliths}\kwd{Plant available silica}\kwd{Human
activities}}

\altkeywords{\kwd{Silicium}\kwd{Cycles
biog\'{e}ochimiques}\kwd{Phytolithes}\kwd{Silice disponible pour les
plantes}\kwd{Activit\'{e}s humaines}}

\begin{abstract} 
Over the past thirty years, the silicon cycle at the scale of
continental ecosystems has become an important topic within the
scientific community. Since the 1990s, this research has been driven
primarily by growing awareness of the need to quantify the impact of
anthropogenic activities on the environment, as well as by advances in
understanding the beneficial role of silicon in cultivated plants and
its application in agricultural practices. This review highlights the
major developments in the study of the silicon cycle at the
soil-plant scale since the pioneering work of Bartoli, particularly his
cycle model (Bartoli, F. \textit{Environ. Biogeochem. Ecol. Bull.}
\textbf{35} (1983), pp. 469--476).

Phytoliths are now recognized as key players in the silicon cycle at
both local and global scales. They represent a significant source of
dissolved silicon (DSi), especially in highly weathered soils. Human
activities such as deforestation, grazing, agriculture, and urban land
use have been shown to disrupt the Si cycle. The concept of
plant-available silicon indicator (PAS) has been developed and is now
used alongside other routine soil parameters (e.g., pH and grain size)
to identify the drivers of DSi.

In the future, greater attention should be given to the determination
and the dynamics of the available Si pools in plants, the
quantification of the effects of biological factors other than plants,
and the impact of atmospheric silicon inputs. The main factors
influencing the terrestrial biogeochemical cycle of silicon should be
integrated into models to predict the evolution of silicon in soils.
\looseness=1
\end{abstract}

\begin{altabstract} 
Au cours des trente derni\`{e}res ann\'{e}es, le cycle du silicium
\`{a} l'\'{e}chelle des \'{e}cosyst\`{e}mes continentaux est devenu un
sujet majeur au sein de la communaut\'{e} scientifique. Depuis les
ann\'{e}es 1990, ces recherches ont \'{e}t\'{e} principalement
motiv\'{e}es par une prise de conscience croissante de la
n\'{e}cessit\'{e} de quantifier l'impact des activit\'{e}s anthropiques
sur l'environnement, ainsi que par les progr\`{e}s dans la
compr\'{e}hension du r\^{o}le b\'{e}n\'{e}fique du silicium chez les
plantes cultiv\'{e}es et de ses applications en agriculture. Cette
revue met en lumi\`{e}re les principales avanc\'{e}es dans l'\'{e}tude
du cycle du silicium \`{a} l'\'{e}chelle sol-plante depuis les travaux
pionniers de Bartoli, en particulier son mod\`{e}le de cycle (Bartoli,
F. \textit{Environ. Biogeochem. Ecol. Bull.} \textbf{35} (1983), pp.
469--476).

Les phytolithes sont de plus en plus reconnus comme des contributeurs
significatifs au cycle du silicium, tant \`{a} l'\'{e}chelle locale que
mondiale. Ils repr\'{e}sentent une source majeure de silicium dissous
(DSi), notamment dans les sols fortement alt\'{e}r\'{e}s. Les
activit\'{e}s humaines, telles que la d\'{e}forestation, le
p\^{a}turage, l'agriculture et l'urbanisation, ont \'{e}t\'{e}
identifi\'{e}es comme perturbant le cycle du Si. Le concept
d'indicateur de silicium disponible pour les plantes (PAS) a
\'{e}t\'{e} d\'{e}velopp\'{e} et est d\'{e}sormais utilis\'{e} aux
c\^{o}t\'{e}s d'autres param\`{e}tres routiniers du sol (comme le pH ou
la granulom\'{e}trie) pour identifier les facteurs influen\c{c}ant le
DSi.

\`{A} l'avenir, une attention particuli\`{e}re devrait \^{e}tre
accord\'{e}e \`{a} la d\'{e}termination et \`{a} la dynamique des
r\'{e}serves de Si disponibles dans les plantes, \`{a} la
quantification des effets des facteurs biologiques autres que les
plantes, ainsi qu'\`{a} l'impact des apports atmosph\'{e}riques de
silicium. Les principaux facteurs influen\c{c}ant le cycle
biog\'{e}ochimique terrestre du silicium devraient \^{e}tre
int\'{e}gr\'{e}s dans des mod\`{e}les afin de pr\'{e}dire
l'\'{e}volution du silicium dans les sols.
\end{altabstract} 

%\input{CR-pagedemetas}
\makeatletter
\def\cdr@preabstracthook{\par\medskip}
\makeatother

\maketitle

\twocolumngrid

\end{noXML}

\defcitealias{Conleyetal2008}{ibid.}
\defcitealias{Caubetetal2020}{ibid.}

\section{Introduction}\label{sec1}

After oxygen, silicon is the second most abundant element on Earth's
surface, and its importance in global geochemical cycles is
considerable. Its geological cycle is closely intertwined with that of
other elements. This is particularly true for carbon through the
supergene weathering of silicate minerals, which leads to the capture
of atmospheric carbon in the form of carbonates that are then stored at
the bottom of the ocean.

At the oceanic scale, silicon is absorbed by microorganisms such as
diatoms, which play a major role in biogeochemical cycles. On land, the
terrestrial biogeochemical cycle of silicon remained largely
understudied until the early 1990s. The pioneering work of Bartoli
\citep[][and references herein]{Bartoli1983} proposed a steady-state
cycle model with four reservoirs (Figure~\ref{fig1}): soil water, 
living biomass, dead biomass, and soil minerals. This system also
included atmospheric inputs and outputs through dissolved silicon and
particles exported to rivers and \mbox{groundwater.} Measurements of certain
fluxes, such as \mbox{litterfall,} allowed \citet{Bartoli1983} to demonstrate
that silicon recycling by vegetation could regulate the silicon cycle
in ecosystems, such as the deciduous forest of Housseras (eastern
France). In other ecosystems, such as the coniferous forest of
Biffontaine (also in eastern France), vegetation recycles very little
silicon, and the silicon found in drainage water mainly originates from
the dissolution of soil minerals.

\begin{figure}
\includegraphics{fig01}
\caption{\label{fig1}The terrestrial biogeochemical cycle of Si
according to \citet{Bartoli1983}.}
\vspace*{-2pt}
\end{figure}

In plants, silicon is present in the form of amorphous silica
particles, which range in size from a few micrometers to several tens
of micrometers. These particles are called phytoliths. They have been
known since the 19th century \citep{Piperno2006}, but a WOS (Web of
Science) analysis shows that the number of publications with the
keyword ``phytolith'' has increased significantly since the early
1990s, thus showing a momentum of interest in the scientific community
(total counts ${=}$\ 3053; Figure~\ref{fig2}). Although a great number of
publications concern the application of phytoliths in the archaeology
and paleoenvironmental disciplines, approximately 30\% (883) are
counted if the keyword ``soil science'' is added to
``phytolith''. Additionally, a greater number of publications
appear with the terms ``silicon'' and ``agriculture'' (8011;
Figure~\ref{fig2}b), reflecting the growing interest in Si in the
fields of biology and agronomy.

\begin{figure}
\includegraphics{fig02}
\caption{\label{fig2}Counts of article citations from Web of Science
(up to June 2025); (a) using the keyword ``phytoliths''; 
(b)~using the keywords ``silicon'' and ``agriculture''.}
\end{figure}

The objective of this review is to present the progress made in our
understanding of the silicon cycle in ecosystems since 
\citeauthor{Bartoli1983}'s (\citeyear{Bartoli1983}) model was
introduced and to highlight the significance of these findings for
understanding the global silicon cycle and its potential implications
for agricultural practices.

\section{Drivers of Si mobility}\label{sec2}
\subsection{Phytoliths as major players}\label{ssec21} 

At the core of \citeauthor{Bartoli1983}'s model 
(\citeyear{Bartoli1983}) is soil water, which contains dissolved
silicon (DSi)  (Figure~\ref{fig1}). DSi results from the chemical
weathering of silicates, a process driven by water and carbon dioxide
that releases silicon into the soil solution, as well as into rivers
and oceans. The concentration of DSi depends on the intrinsic
properties of minerals (such as solubility) and environmental
parameters (e.g., pH or temperature). In 1993, \citet{Lucasetal1993}
changed the strictly abiotic view of soil formation in tropical
latitudes, which considered a highly leached surface horizon rich in
gibbsite, a more siliceous horizon below, and at the base, a smectitic
accumulation horizon. The presence of kaolinite above the gibbsitic
horizon could not be explained by this abiotic model; a model involving
a Si contribution from plants that recombines with Al to form kaolinite
then appeared more coherent. Other studies followed, showing the
importance of plants in weathering balances. \citet{Alexandreetal1997}
analyzed the phytoliths of a lateritic soil profile in the equatorial
forest of the Congo and used \citeauthor{Bartoli1983}'s model 
(\citeyear{Bartoli1983}). They revealed two important points: the Si
cycle was mainly controlled by plants, which reinforced the studies of
Lucas and Bartoli, and some of the phytoliths (approximately 10\%) were
not rapidly dissolved and thus constituted a stable pool of amorphous
silica in the soil. \citet{Meunieretal1999} analyzed a soil rich in
amorphous silica located on the volcanic island of R\'{e}union (Indian
Ocean). They demonstrated that this 15 cm thick layer was composed of
Poaceae phytoliths. These phytoliths accumulated because of the
convergence of 3 factors: soil developed on volcanic ash rich in Si,
Si-accumulating plants (bamboo), and, finally, acidic soil pH, which is
unfavorable for the rapid solubilization of amorphous silica. The
Bartoli model was thus enriched with a new pool with stable phytoliths
in soils.

However, very few studies on the physicochemical properties of
phytoliths were available before the mid-1990s. It was therefore
assumed that the phytoliths had the same properties as the other
amorphous silica phases did. This hypothesis was the subject of the
work of \citet{Fraysseetal2006,Fraysseetal2009}, who analyzed the
solubility and dissolution rate of phytoliths in the laboratory. Their
results confirmed that the solubility of phytoliths was identical to
that of other forms of amorphous silica but that their dissolution rate
was lower than that of vitreous silica. The dissolution rates increase
with increasing pH and thus explain why phytoliths accumulated on
R\'{e}union \citep{Meunieretal1999}. \citet{Fraysseetal2009} reported
that, beyond pH 4, the dissolution rates of phytoliths were higher than
those of aluminosilicates, which explained why phytoliths can control
the Si cycle in \citeauthor{Bartoli1983}'s model 
(\citeyear{Bartoli1983}), despite their low proportion in soils
(generally less than~1~wt\%).\looseness=1

By measuring germanium and silicon in stream waters in Hawaii, 
\citet{Derryetal2005} provided new evidence of the importance of
phytoliths. Their results revealed that stream waters were distributed
along a hyperbolic curve between a pole characterized by low dissolved
Si (DSi) values and high Ge/Si values (clays) and a phytolith pole
characterized by high DSi values and low Ge/Si ratios. They showed that
contribution of phytoliths predominates in the stream waters. At the
global scale, \citet{Conley2002} evaluated the role of plant recycling
by comparing the annual production of amorphous silica (ASi) in the
form of phytoliths to that of diatoms in the ocean. The results
revealed that the annual production of silica from phytoliths was
between 60 and  200 Tmol${\cdot}$yr$^{-1}$ and of the same order of
magnitude as that of silica fixed in the oceanic biogeochemical cycle
(240~Tmol${\cdot}$yr$^{-1}$). Thus, for approximately thirty years,
phytoliths have become major players in the Si cycle at the local and
global scale.

\subsection{Biological factors other than plants}\label{ssec22}

Several studies have provided new insights into the role of living
organisms other than plants 
\citep[see review in][]{deTombeuretal2021}. 
\citet{Sommeretal2013} quantified the
pool of testate amoebae in a German forest. Although the Si pool of
testate amoebae tests was small, approximately 100 times less than that
of phytoliths, the authors calculated an annual production of 17~kg of
Si ha$^{-1}{\cdot}$yr$^{-1}$ by testate amoebae, comparable to that of
phytoliths. Consequently, testate amoebae can be considered a temporary
pool of Si on a monthly scale, considering their relatively short
generation times. However, the total protozoic Si pool and its
dissolution rates remain poorly understood \citep{Puppe2020}.

Soil microfauna, such as earthworms and termites, can also play a role
in the dynamics of Si. \citet{Bityutskiietal2016} analyzed
water-extracted DSi in Russian earthworm casts and reported
concentrations that were 2--12 times higher than those in surrounding
soils. The authors suggested that passage through the earthworm
digestive created finer particles and accelerated the dissolution of
silicate minerals. Empirical indicators of plant-available-silicon
(PAS) have been developed, which have been shown to be proportional to
the silicon concentration in plants. The concept of PAS will be further
discussed in Section~\ref{sec5}. \citet{Jouquetetal2024} analyzed the
role of earthworms in Vietnamese rice paddies with PAS measurements
using 0.5 M acetic acid (Si$_{\mathrm{AA}}$). They reported, in
agreement with the results of \citet{Bityutskiietal2016}, an increase
in the Si$_{\mathrm{AA}}$ concentration in the casts. The authors also
reported an increase in phytoliths in the casts, which could explain
their higher Si$_{\mathrm{AA}}$ values. Another explanation for the
high Si$_{\mathrm{AA}}$ values was proposed by \citet{Huetal2018}, who
isolated silicate-solubilizing bacteria in the digestive tract of
earthworms. These authors also inoculated these bacteria into pots
filled with Chinese soils and reported that they had a positive effect
on DSi and the amount of Si absorbed by maize.

The effects of termites on the PAS have also been documented 
\citep{Albertetal2025}. These authors reported higher 
Si$_{\mathrm{CaCl}_{2}}$ (PAS using CaCl\tsub{2} extractant)
concentrations in non-fungus-growing termite mounds than in surrounding
soils and a positive correlation between PAS and phytolith
concentrations. This type of termite accumulates residues of fecal
material rich in poorly degraded leaf and stem plant debris, which
explains the richness of the phytoliths that control
Si$_{\mathrm{CaCl}_{2}}$.

Large animals can also impact the silicon cycle.
\citet{Vandevenneetal2013} studied the impact of large herbivores on
the solubility of phytoliths. They reported that compared with
phytoliths extracted from hay, phytoliths from horse, cow, and sheep
feces dissolve 3--5 times faster. \citet{Schoelyncketal2019} quantified
the effect of hippopotamuses (in Kenya), which have the ability to
ingest large quantities of food from the grassland savanna at night and
rest during the day in the hippo pools of river meanders, where they
evacuate their fecal matter. The authors were able to quantify the Si
transferred from soils to rivers and reported that 76\% of the total Si
flux was affected by hippopotamuses. Large herbivores thus have a
significant effect on the turnover of Si in soils and on the balance of
Si delivery to aquatic systems.

\subsection{Physicochemical factors}\label{ssec23}

Since the 1960s, the adsorption of Si on the surfaces of soil minerals
has been known to increase with increasing pH
\citep{BeckwithReeve1963}. Laboratory analyses have shown that the
amount of desorbed Si decreases over time \citep{HaynesZhou2020},
suggesting that over time, a fraction of adsorbed DSi becomes immobile.
Although the mechanisms are not yet well understood, these results
could lead to a decrease in DSi in alkaline soils.

The effect of aggregation on the solubilization of phytoliths was
evaluated by \citet{Lietal2022} on mixtures of materials composed of
organic matter, aluminosilicates, oxides, and phytoliths extracted from
rice leaves. Their results revealed that aggregation can decrease the
release of DSi from phytoliths by up to 3.9 times depending on the type
of material used. Other factors can influence the dissolution of
phytoliths, such as morphology \citep{Cabanesetal2011}, age
\citep{Liuetal2023}, redox conditions \citep{Koebernicketal2022}, and
fire \citep{Cabanesetal2011}.

\subsection{Si isotopes as tracers of the sources of dissolved silicon
(DSi)}\label{ssec24}

In addition to the use of the Ge/Si ratio for tracing the source of
DSi, the use of Si isotopes has been developed. Si has three stable
isotopes: ${}^{28}$Si (92.23\%), ${}^{29}$Si (4.67\%), and ${}^{30}$Si
(3.1\%). Measurements of isotopic ratios are generally expressed in
${\updelta}$ notation:
{\begin{eqnarray*}
\updelta{}^{30}\mathrm{Si} &=&
[(\mbox{}^{30}\mathrm{Si}/\mbox{}^{28}\mathrm{Si~sample})/
(\mbox{}^{30}\mathrm{Si}/\mbox{}^{28}\mathrm{Si~standard}) 
- 1] \\
&& \times\, 1000
\end{eqnarray*}}\unskip
with NBS28 used as the international standard.

In nature, $\updelta{}^{30}$Si values range from
${-}$5.7${\permil}$ for silcretes \citep{BasileDoelschetal2005}
to ${+}$6.1${\permil}$ for rice grains \citep{Dingetal2005}.
Parent rocks, including crustal, magmatic, and sedimentary rocks, are
characterized by values of approximately~0 \citep{Fringsetal2016}.
Fractionation in soils can be explained by water--rock interaction
mechanisms or biological processes \citep{OpfergeltDelmelle2012}. Thus,
the Si isotopic composition of clay minerals formed during weathering
is lower than that of parent minerals. In soils, low $\updelta{}^{30}$Si
values are indicators of strong weathering \citep{Ziegleretal2005}.
Furthermore, this phenomenon is amplified by the adsorption of Si on
secondary Fe minerals, which is accompanied by the preferential
retention of light isotopes. Si uptake by plants is accompanied by an
enrichment in light isotopes, and during the transfer to the shoots,
heavy isotopes are preferentially incorporated into the phytoliths.
Depending on their location in the plant, the $\updelta{}^{30}$Si values
of phytoliths range from ${-}$2.3 to ${+}$6.1${\permil}$
\citep{OpfergeltDelmelle2012}. In an analysis of the different
compartments of a tropical forest ecosystem in South India,
\citet{Riotteetal2018} reported that the heavy $\updelta{}^{30}$Si values
of water infiltrating through the soil were similar to those of
phytoliths, thus revealing that the latter controlled DSi. However, in
acidic brown soils in France \citep{Cornelisetal2010},  the results
differed, with the light $\updelta{}^{30}$Si values of forest floor
leachates being controlled more by the dissolution of clays than by
phytoliths. In the temperate Black Forest (Germany),
\citet{Steinhoeferetal2017} reported that $\updelta{}^{30}$Si values in
streams were lower in spring than in late summer. During spring, the
increase in dissolved organic carbon triggers a decrease in the pH and
preferential dissolution of clays and a decrease in $\updelta{}^{30}$Si
values. During summer, the formation of secondary minerals and Si
uptake by plants lead to positive $\updelta{}^{30}$Si values. The use of
silicon isotopes remains a promising tool for constraining the dynamics
of silicon in soil--plant\break interactions.

\section{Improving the quantification of Si pools and fluxes}
\label{sec3}

Progress in quantifying plant--soil silicon transfers has been achieved
through the publication of data on Si concentrations in plants,
absorbed Si fluxes, and concentrations and fluxes of phytoliths in
soils. \citet{Hodsonetal2005} compiled data on 735 species from 125
publications and reported that, in general, ferns, gymnosperms, and
angiosperms contain less Si than nonvascular plants and horsetails do.
Furthermore, compared with other monocot clades, Poales and Arecales
accumulate more Si in their shoots. The fluxes of absorbed Si,
expressed in kg Si ha$^{-1}{\cdot}$yr$^{-1}$, have been estimated in
most terrestrial ecosystems, both natural and anthropogenic. In a
bibliographic synthesis, \citet{VanderLindenDelvaux2019} reported a
considerable range of values, between 0.7 and 1470~kg Si 
ha$^{-1}{\cdot}$yr$^{-1}$. Using such analyses and the net primary
production of terrestrial biomes, \citet{CareyFulweiler2012a} reported
that the annual amount of Si fixed depends on biome types and was
higher in tropical woodland savanna and temperate steppe than in
tropical forest, 22.19, 13.26, and 4.48 Tmol Si yr$^{-1}$,
respectively, thus \mbox{highlighting} the importance of grasses in the global
Si cycle. The Si production of cultivated land reaches \mbox{almost} 30 Tmol
Si yr$^{-1}$, which is 35\% of the production of all the biomes
combined. The high proportion of cultivated land is explained by the
fact that 8 of the 10 most cultivated crops in the world are Si-rich
plants: sugarcane, maize, rice, wheat, soybeans, sugar beet, barley,
and tomatoes \citep{Guntzeretal2012a}. Given the predictable increase
in cultivated areas to feed future populations,
\citet{CareyFulweiler2012a} hypothesized an increase in the transfer of
bioavailable Si to aquatic environments, with possible consequences for
the development of\break diatoms.

The fluxes of absorbed Si depend not only on the types of plants but
also on the types of soils and the soil weathering stage 
\citep{CornelisDelvaux2016,VanderLindenDelvaux2019}. Phytolith pools
have been measured in many ecosystems \citep{Haynes2017a}. Natural
grassland ecosystems contain a higher reservoir of biogenic silica
(BSi) than forest ecosystems, with values ranging from 20 to 
690~T~Si~ha$^{-1}$  (for soil depths between 0.2 and 1~m) and 0.6--17~T Si
ha$^{-1}$ (for soil depths between 0.2 and 0.85~m), respectively, which
is in good agreement with the absorbed Si fluxes presented earlier.

Finally, the contribution of atmospheric silicon (Si) inputs to soils
remains poorly documented. It is generally assumed that Si inputs are
neglected due to their low values \citep{Sommeretal2006}. For instance
in a temperate forest ecosystem, \citet{Turpaultetal2018} reported
values of dust and DSi respectively of 0.2 and 6~kg Si
Ha$^{-1}{\cdot}$yr$^{-1}$ while uptake by plant represented
95--157~kg${\cdot}$yr$^{-1}$. \citet{Xietal2022} reported values of
approximately 1~kg Si ha$^{-1}{\cdot}$yr$^{-1}$ in Chinese terrestrial
ecosystems from 2010--2020. These values are approximately 200 times
lower than the Si uptake by rice shoots (219~kg
ha$^{-1}{\cdot}$yr$^{-1}$). Therefore, the input of DSi from rainfall
does not appear to be negligible, as it would take only 4--10 years to
meet the needs of rice if the straw is exported. These results warrant
further validation in other regions worldwide.

\section{Impact of human activities}\label{sec4}

The question of the impact of anthropogenic activities, such as
intensive farming, deforestation, or urbanization, on soil silicon (Si)
dynamics has been the subject of innovative research since the 2000s.
With respect to deforestation, \citet{Conleyetal2008} documented the
effects on dissolved silicon (DSi) in stream discharges at the Hubbard
Brook experimental site (USA). They reported that DSi exports were
greater in deforested areas where biomass was left in place. According
to the authors, the excess DSi was due to the dissolution of excess
phytoliths from the biomass in place and not to an increase in the
chemical weathering of primary minerals. Deforestation followed by the
establishment of crops is known to disrupt soils and promote erosion,
but no work has focused on phytoliths. \citet{Struyfetal2010} analyzed
the fluxes of DSi and amorphous silica (ASi) in waters from 51
watersheds in Belgium. Their conclusion differed from that of
\citet{Conleyetal2008}, as they reported that exports of reactive
silica, TSi ${=}$\ DSi ${+}$ ASi, were higher in forested areas than in
agricultural areas. The authors suggested that the decrease in TSi in
cultivated areas was due to the long-term depletion of phytoliths in
the soils, whereas the results of \citetalias{Conleyetal2008} applied to the
short term (${<}$20 years). Several studies have sought to verify this
hypothesis in the 2010s. 

\citet{Clymansetal2011} measured ASi stocks along a gradient of soils
from forestland--pasture--arable land in Sweden. ASi was estimated by
destructive alkaline extraction, called ASia. The results revealed that
the total ASia values were greater in forests. These results were
confirmed in another study along a gradient of land use change in
temperate zones in Belgium using continuous extraction measurements of
Si and Al by NaOH \citep{Vandevenneetal2015}. In natural environments,
such as forests, the quantity of phytoliths decreases with depth, as is
the case for carbon \citep{Alexandreetal1997}. In a cultivated field
with biomass export, a decrease in the stock of phytoliths in the
surface horizon was expected. This hypothesis was documented at Hubbard
Brook \citep{Sacconeetal2008}. These authors reported that after
deforestation, ASi stocks in the O horizon decreased by 34\%. At
Rothamsted Research, \citet{Guntzeretal2012b} documented a decrease in
ASi in a cultivated field with partial or total export of straw after
1844. It has thus been clearly established that soil phytolith stocks
can decrease when litterfall is disturbed by human interventions in the
field.

Another aspect of the perturbations of the Si cycle by human activities
has been evaluated: the role of urban systems. Indeed, natural cycles
are strongly disrupted by urbanization, and this also concerns silicon,
as several studies have shown. The export and transformation of plant
materials can occur in urban areas and they can be released into rivers
through wastewater \citep{Vandevenneetal2011}.
\citet{Sferratoreetal2006} analyzed the impact of urban discharge on
the DSi balance in the Seine River (France). Their study revealed that
the concentration of DSi in urban runoff was at least twice as high as
that in rainwater, thus indicating that impervious urban surfaces
contain alterable siliceous materials. \citet{CareyFulweiler2012b}
analyzed DSi exports in a gradient of watersheds ranging from forested
to urbanized areas in the USA and reported lower values as forest cover
increased. These authors also analyzed the effect of seasonality and
reported a decline in DSi in forested areas between February and April,
which was lacking in urban areas. The authors attributed this decline
not to volumetric dilution or to the development of river diatoms but
to trees preferentially absorbing DSi during this period.
\citet{MaguireFulweiler2016} specifically estimated DSi release in
urban areas in Boston (USA) and reported that DSi release was greater
at the outlets of wastewater treatment plants than in infiltration
water and runoff. These results confirmed the importance of domestic
consumption on DSi fluxes in urban\break areas. 

\section{Si in agroecosystems}\label{sec5}

The beneficial role of silicon in the growth of cereals has been known
for more than a century \citep[see review in][]{Epstein1999}. Over
the past thirty years, a significant number of studies have documented
the positive impact of Si not only on yields but also when the plant is
subjected to biotic and abiotic stresses \citep{deMelloPrado2023,
Liangetal2015}. The mechanisms of these beneficial effects are still
poorly understood, but since the work of \citet{Maetal2006}, it has
been established that uptake can be active, meaning triggered by genes,
in Si-accumulating plants (for instance, rice, barley, maize, wheat,
and horsetail). Proteins from the aquaporin family are involved. The
silicon accumulated in plants can also be explained by passive
absorption controlled by water flux (transpiration). In cultivated
plants such as rice and maize, both mechanisms coexist
\citep{Liangetal2006}.

The accumulation of Si in plants depends on the pool of bioavailable Si
in the soils that supply DSi. However, measurements of DSi in solutions
are rare and difficult to implement; thus, another approach is used,
namely, chemical extractants as plant-available silicon indicators or
PAS \citep{Saueretal2006}. For example, \citet{HaysomChapman1975}
extracted Si from Australian soils using a 0.02~N CaCl\tsub{2} solution
and reported that the concentrations were proportional to the sugarcane
yield. This method has since been confirmed on other soils and plants,
such as Indian soils cultivated with rice
\citep{NarayanaswamyPrakash2010}.  The CaCl\tsub{2} method is used to
characterize the easily soluble Si pool and is thus directly
mobilizable by plants. Other more powerful extractants capable of
desorbing Si, such as acetic acid 
\citep[see the review in][]{Tubanaetal2016}, have also been proposed. Measurements of
DSi by different extractants are correlated with each other
\citep{NarayanaswamyPrakash2010}. The practical interest of these
techniques is that they can be used to determine the critical
thresholds of Si for a given cultivated plant. Below the critical
threshold, an input of Si fertilizer to the soil would ensure optimal
yield \citep{Payeetal2018}.\looseness=1

However, these PAS do not characterize the real pool of bioavailable
Si, which remains poorly understood. By combining Si analyses with
other soil properties (pH, particle size, amorphous phases, CEC, etc.),
several studies have attempted to characterize the mineral phases of
soils that control PAS. There is a positive correlation between the PAS
and pH. In highly weathered acidic soils of South Africa and South
India, the PAS are lower than those in soils close to neutrality, which
are richer in alterable silicate minerals \citep{Milesetal2014,
Meunieretal2018}. \citet{Klotzbucheretal2018} also reported such a
correlation in chernozems in Germany, suggesting the role of phytoliths
and sorption as factors controlling PAS. Indeed, the solubility of
phytoliths as well as adsorption increases with increasing pH. Two
studies on more than 2000 soil samples in France 
\citep{Landreetal2018, Caubetetal2020} helped to specify the role of
pedological factors on PAS using the CaCl\tsub{2} method
(Si$_{\mathrm{CaCl}_{2}}$). \citet{Landreetal2018} reported that
Si$_{\mathrm{CaCl}_{2}}$ values were negatively correlated with total
Si concentrations when all soil types were combined ($r=-0.32$,
$p<0.001$).  In regard to podzols, this negative correlation was
stronger ($r=-0.75$, $p<0.001$) and was explained by the presence of
quartz, a mineral that is recalcitrant to dissolution. In contrast, the
authors observed a slightly positive correlation between these same
parameters ($r=0.12$, $p<0.001$) for carbonate soils, which are known
to be poor in quartz.

\citet{Caubetetal2020} reported that Si$_{\mathrm{CaCl}_{2}}$
concentrations were positively correlated with clay content 
(${<}2~{\upmu}$m fraction) and pH. \citetalias{Caubetetal2020} also reported
higher Si$_{\mathrm{CaCl}_{2}}$ concentrations in cultivated soils than
in forest soils. This result can be explained by the liming commonly
practiced in cultivated soils, which not only increases the soil pH but
also soil PAS. To better explain these results, another study by the
same team focused on characterizing Si pools in pairs of cultivated
soils over two centuries and their neighboring forest soils 
\citep{Cornuetal2022}. The authors showed that allophanes (short-range
order aluminosilicates), included in the ${<}2~{\upmu}$m fraction,
control PAS  and not the phytoliths. The role of clays as a source of
Si for plants was the subject of a pot study with wheat and rice 
\citep{Kelleretal2021}. The authors reported that a soil consisting of
5\% amorphous silica (in the form of diatomite) provided wheat with the
same amount of Si as a soil consisting of 25\% vermiculite would. The
results showed that the identification of the source of DSi requires
good characterization of the soil properties.

In regions of the world where soils are highly weathered and
desilicified, phytoliths can constitute the main source of dissolved
silica (DSi) \citep{CornelisDelvaux2016,deTombeuretal2020}. As
previously mentioned, intensive cereal cultivation combined with straw
removal can deplete soils of phytoliths. The same is true for PAS, as
demonstrated by a study conducted in rice paddies in China 
\citep{Yangetal2021}. The authors compared soils sampled in the 1990s
with those collected between 2016 and 2018 and reported an
approximately 14\% decrease in NaAC-extractable Si. They also reported
that 66\% of China's paddy fields are deficient\break in PAS.

Since the 1980s, a decline in rice yields has been documented across
Asia. This phenomenon could result from multiple factors, one of which
is a deficiency in PAS \citep{Savantetal1997}. This hypothesis is
supported by numerous experimental studies---conducted under
hydroponic, pot, and field conditions---that have tested the effects of
silicon fertilization on soils \citep{Barao2022, Liangetal2015,
Haynes2017b, Tubanaetal2016}. Unlike nitrogen, the beneficial effects
of silicon on crop plants are not systematic. A relevant example was
provided by \citet{Sandhyaetal2018}, who \mbox{conducted} pot experiments
using acidic, neutral, and alkaline soils from southern India under
both field capacity and submergence conditions, with silicon fertilizer
applied in the form of diatomaceous earth (DE). Under submerged
conditions, DE application improved both Si uptake and grain yield in
acidic soils. In contrast, for neutral soils, no significant effect on
Si uptake was observed, and a yield increase was noted only at high
application rates (600~kg/ha DE). Acidic soils, which are generally low
in PAS, benefit from silicon inputs that improve yield. Conversely,
neutral soils, which are richer in PAS, are less responsive. The
observed yield increase without a corresponding increase in Si uptake
may be due to the presence of other nutrients in the amendment. In
alkaline soils, Si supplementation helps plants better cope with stress
caused by alkalinity and salinity, thereby increasing rice growth
\citep{deMelloPrado2023}. Furthermore, under field capacity conditions,
Si addition did not significantly increase Si uptake or grain yield in
acidic soils. This study underscores the importance of water
availability in controlling the release of DSi from fertilizers,
highlighting water as a key limiting factor for optimal Si uptake by
plants. Many studies have also used materials derived from mineral
extraction---such as the diatomaceous earth mentioned earlier---as well
as industrial by products \citep{Liangetal2015}. The  effectiveness of
Si fertilization on crop yields therefore requires a thorough
understanding of the soil's physicochemical and agronomic\break parameters.

Another strategy to increase the soil PAS concentration is to retain
crop residues (straw) onsite. This hypothesis was tested in a long-term
field experiment in Germany \citep{Puppeetal2021}. The authors compared
the PAS levels as Si$_{\mathrm{CaCl}_{2}}$ in soils with and without
straw application from 1976 to 2018. The results revealed a significant
increase in Si$_{\mathrm{CaCl}_{2}}$, as well as an accumulation of
soil phytoliths, over time. Another example of long-term analysis was
provided by a 36-year study in China \citep{Yangetal2020},
where the authors showed that the so-called ``labile'' fraction of
Si---measured as the sum of extractions using CaCl\tsub{2}, acetic
acid, H\tsub{2}O\tsub{2}, oxalate, and Na\tsub{2}CO\tsub{3}---was
higher in fields that received phytolith-rich straw. Additional
evidence of the positive effects of plant-derived Si inputs on PAS was
provided by  \citet{Songetal2014}, who used Si-rich manure at
experimental sites in China. After five years of \mbox{application,} a
significant increase in Si extracted using the acid-Na acetate method
was observed down to a depth of 30~cm.

Although the direct application of silicon fertilizers has proven
effective at improving crop yields, its use remains limited in
agricultural practices. This can be attributed to several factors,
including insufficient knowledge about the physiological role of Si and
a lack of local data on PAS levels in soils \citep{Zellneretal2018}.

\section{Conclusions}\label{sec6}

The continental biogeochemical cycle of silicon has become more complex
since the work of \citet{Bartoli1983} (Figure~\ref{fig3}). Phytoliths,
which reflect the recycling of soil silicon by plants, are now
considered major drivers of this cycle, as they dissolve more rapidly
than most other aluminosilicate minerals in\break soils do.

\begin{figure}
\vspace*{4pt}
\includegraphics{fig03}
\vspace*{4pt}
\caption{\label{fig3}A modern picture of the biogeochemical cycle of Si
in natural and cultivated terrestrial ecosystems 
\protect\citep[modified from][]{Meunieretal2022}.}
\vspace*{4pt}
\end{figure}

In addition to plants, other biological factors have been shown to
influence the cycle, ranging from microorganisms (e.g., testate
amoebae) to large herbivores and soil engineers. Human activities
disrupt the cycle, with the most well-documented impact being the
decline in phytoliths in the surface horizons of soils. 

The hypothesis that silicon may be a limiting nutrient for crop growth
has spurred research on the factors governing the mobilization of
silicon in solution. Indicators of bioavailable silicon (PAS) have
become more widespread, with practical applications in agronomy, as
threshold values are now available.

The use of silicon isotopes and Ge/Si ratios makes it possible to
constrain the silicon cycle at the ecosystem level, even though the
results of such studies are not always straightforward.

To predict the appropriate doses of silicon fertilizers to apply to
soils to improve crop yields or to forecast changes in the global
silicon cycle, it would be desirable to develop mechanistic models that
consider the multiple factors that have been documented in recent
years.

\section*{Acknowledgements}

\vspace*{2pt}

I would like to thank the anonymous reviewers whose
comments and suggestions helped improve both the content and the form
of the manuscript.

\vspace*{2pt}

\section*{Declaration of interests}

\vspace*{2pt}

The author does not work for, advise, own shares in, or receive funds
from any organization that could benefit from this article, and has
declared no affiliations other than their research organization.

\back{}

\printbibliography
\refinput{crgeos20250938-reference.tex}

\end{document}