Introduction
Various prokaryotes and eukaryotes are able to synthesize hydrated amorphous
silica (SiO2 ⚫ nH2O) structures from monomeric silicic acid
(H4SiO4) in a process called biosilicification (Ehrlich et al.,
2010). In terrestrial biogeosystems, biogenic silicon (BSi) synthesized by
bacteria and fungi, plants, diatoms, testate amoebae and sponges can be found
forming corresponding microbial, phytogenic, protophytic, protozoic and
zoogenic BSi pools, respectively (Puppe et al., 2015; Sommer et al., 2006).
BSi has been recognized as a key factor in the control of Si fluxes from
terrestrial to aquatic ecosystems as it is in general more soluble compared
to silicate minerals (e.g., Fraysse et al., 2006, 2009). These fluxes
influence marine diatom production on a global scale (Dürr et al., 2011;
Sommer et al., 2006; Struyf and Conley, 2012). Marine diatoms in turn can fix
large quantities of carbon dioxide via photosynthesis, because up to 54 %
of the biomass in the oceans is represented by diatoms; thus diatoms have an
important influence on climate change (Tréguer and De La Rocha, 2013;
Tréguer and Pondaven, 2000).
While the importance of phytogenic Si pools for global Si fluxes has been
recognized for three decades (e.g., Bartoli, 1983; Meunier et al., 1999;
Street-Perrott and Barker, 2008), information on the other BSi pools is
comparatively rare (Clarke, 2003). However, in recent publications the potential
importance of diatoms, testate amoebae and sponge spicules in soils for Si
cycling has been highlighted (Aoki et al., 2007; Creevy et al., 2016; Puppe
et al., 2014, 2015, 2016). Furthermore, evidence arises that BSi pools are in
disequilibrium at decadal timescales due to disturbances and perturbations
by humans, e.g., by changes in forest management or farming practices
(Barão et al., 2014; Keller et al., 2012; Vandevenne et al., 2015). As a consequence, BSi accumulation and BSi dissolution are not balanced, which
influences Si cycling in terrestrial biogeosystems, not only on decadal but
also on millennial scales (Clymans et al., 2011; Frings et al., 2014; Sommer
et al., 2013; Struyf et al., 2010). Sommer et al. (2013), for example, found
the successive dissolving of a relict phytogenic Si pool to be the main
source of dissolved Si in soils of a forested biogeosystem. Due to the fact
that the continuous decomposition of this relict phytogenic Si pool is not
compensated by an equivalent buildup by recent vegetation the authors
concluded that a BSi disequilibrium occurred on a decadal scale. On a millennial scale
Clymans et al. (2011) estimated the total amorphous Si storage in temperate
soils to be decreased by approximately 10 % since the onset of
agricultural development about 5000 years ago. This decrease does not only
have consequences for land–ocean Si fluxes but also influences agricultural used
landscapes, because Si is a beneficial element for many crops (e.g., Epstein,
2009; Ma and Yamaji, 2008).
For a better understanding of BSi dynamics, chronosequence studies are well
suited, because they allow us to analyze time-related changes in BSi pools
during biogeosystem development. In the present study we analyzed various BSi
pools in differently aged soils of an initial artificial catchment (Chicken
Creek; Hühnerwasser) in a post-mining landscape in NE Germany. Chicken Creek represents a
study site with defined initial conditions and offers the rare opportunity to
monitor BSi dynamics from the very beginning. Former studies at this site
revealed (i) a formation of protophytic (diatom frustules), protozoic
(testate amoeba shells) and zoogenic (sponge spicules) Si pools within a
short time (< 10 years) and (ii) a strong relation of spatiotemporal
changes in protistic (diatoms and testate amoebae) BSi pools to the
vegetation, because plants provide, e.g., rhizospheric micro-habitats
including enhanced food supply (Puppe et al., 2014, 2016). From these results
it can be concluded that vegetated spots in particular represent hotspots of BSi accumulation of various origin at initial biogeosystem
sites (compare Wanner and Elmer, 2009). Furthermore, construction work with large machines
resulted in differently structured sections of Chicken Creek with slight
differences in abiotic conditions (for details see Sect. 2.1.) (Gerwin et
al., 2010). These differences in turn lead to section-specific vegetation
dynamics at Chicken Creek (Zaplata et al., 2010).
Knowledge about BSi accumulation dynamics is crucial for the understanding of
Si cycling in terrestrial biogeosystems. We regard water-extractable Si as a
useful proxy for desilication and biological uptake (plants, testate amoebae
etc.). In addition, we used an alkaline extractant (Tiron) to detect eventual
short-term changes in the amorphous Si fraction. We hypothesized that (i) BSi
pools influence short-term changes in water-soluble Si in initial soils
but not short-term changes in amorphous Si fractions, (ii) the phytogenic Si
pool is the most prominent one in size and thus the biggest driver of
short-term changes in water-soluble Si, and (iii) BSi pool changes are
section-specific, i.e., related to vegetation dynamics. The aims of the
present study were (i) to quantify various BSi pools, i.e., protophytic,
protozoic, zoogenic and phytogenic Si pools, during initial soil and
ecosystem development; (ii) to analyze potential section-specific short-term
changes in these BSi pools after a decade of ecosystem development; and
(iii) to evaluate the influence of different BSi pools on water-soluble Si in
these soils.
Material and methods
Study site
The study site Chicken Creek (51∘36′18′′ N,
14∘15′58′′ E) represents an artificial catchment in a
post-mining landscape located in the active mining area of Welzow South
(lignite open-cast mining, 150 km southeast of Berlin) in the state of
Brandenburg, Germany (Kendzia et al., 2008; Russell et al., 2010). Climate at
Chicken Creek is characterized by an average air temperature of
9.6 ∘C and an annual precipitation of 568 mm comprising data from
1981 to 2010 (Meteorological Station Cottbus, German Weather Service).
Contents of skeleton (> 2 mm), fine earth (< 2 mm),
sand, silt and clay fractions (upper 30 cm) at the sampling points in
western, eastern and southern sections at Chicken Creek (t0,
calculations based on data of Gerwin et al., 2010). Minimal (min.) as well as
maximal (max.) values, means (x‾) and standard
deviations (SD) are given.
Section
> 2 mm
< 2 mm
Sand
Silt
Clay
%
%
West
Min.
9
80
77
7
5
Max.
20
91
88
13
10
x‾
13
87
83
10
7
SD
5
5
4
2
2
East
Min.
2
77
69
6
4
Max.
23
98
91
20
11
x‾
13
87
82
11
7
SD
7
7
9
6
3
South
Min.
0.2
84
78
7
4
Max.
17
99.8
89
17
8
x‾
8
92
83
11
6
SD
8
8
4
4
2
To construct the ∼ 6 ha sized catchment a 1–3 m thick base
layer (aquiclude) of Tertiary clay was covered by a 2–3 m thick sandy,
lignite- and pyrite-free Quaternary sediment serving as a water storage layer
(aquifer) (Gerwin et al., 2010; Kendzia et al., 2008). Quaternary material
was taken from a depth of 20 to 30 m during lignite mining process and its
texture is classified as sand to loamy sand (Table 1) with low contents of
carbonate (Gerwin et al., 2009, 2010; Russell et al., 2010). Dumping of
material and construction work with large machines (e.g., stackers and
bulldozers) resulted in differently structured sections of Chicken Creek.
Generally, the catchment area can be divided into four sections: (i) an
eastern part (ca. 1.8 ha), (ii) a western part (ca. 1.6 ha), (iii) a
central trench (ca. 0.9 ha) separating the eastern from the western part and
(iv) a southern part (ca. 1.5 ha) with a pond at the lowest point (Fig. 1).
Construction work was completed in September 2005 (time zero, t0).
Analyses subsequent to catchment completion indicated slight differences in
abiotic conditions between the eastern and the
western parts (in soil pH, conductivity, skeleton content with soil particle
diameter > 2 mm, proportions of sand, silt and clay,
concentration of organic and inorganic carbon; Gerwin et al., 2010). The primary mineral component in all
particle size fractions at t0 was quartz (only small amounts of
K-feldspar, plagioclase). Calcite comprised 0.5–4.5 % of the initial
sediment, dolomite was only detectable in a few samples with contents of
0.5 %, and magnesite (MgCO3) was not detectable by mineralogical
analysis (W. Schaaf, personal communication, 2011). For detailed information
on the site construction and initial ecosystem development see Gerwin et
al. (2010) and Schaaf et al. (2010), respectively.
Soil sampling
We used samples taken shortly after the construction of Chicken Creek (2005, t0)
and after an ecosystem development period of about 10 years (2015,
t10). For t0 (no vegetation detectable) we assumed that
biogenic siliceous structures were homogenously distributed across the whole area of
Chicken Creek, i.e., no section-specific distribution of BSi (BSi t0
east ≈ BSi t0 west ≈ BSi t0 south) at the
beginning of ecosystem development (Puppe et al., 2016). This is why we did
not sample all different sections of the catchment but took soil samples in
six field replicates to quantify BSi pools at t0. However, for t10
we hypothesized section-specific differences in BSi pool quantities related
to section-specific vegetation dynamics. To evaluate these differences after
a decade of ecosystem development and to cover the biggest possible BSi
accumulation in soil we focused on spots where Si-accumulating plant species,
i.e., Calamagrostis epigejos and Phragmites australis,
became dominant (Zaplata et al., 2010). Thus we took samples in the eastern
(C. epigejos dominant) and western (mainly C. epigejos
dominant, one spot with P. australis) and southern section
(P. australis dominant) of Chicken Creek.
For an accurate description of changes in abiotic soil conditions and related
phytogenic Si in every section, we took soil and plant samples in eastern,
western and southern sections at t0 as well as t10. Erosion and
deposition processes were clearly evident in the Chicken Creek catchment
during the first years without plant cover. Substantial surface changes
resulted from rill erosion, as aerial photographs (rill network) and a
comparison of photogrammetry-based digital elevation models showed (Schneider
et al., 2013). Interrill erosion did not lead to surface changes larger than
about 20 cm during the first 5 years. Afterwards the establishment of an
area-wide plant cover substantially reduces interrill erosion. Because all
soil data at t0 referred to a depth increment of 30 cm we reasonably
assumed the same soil conditions for the sampled t0 spots during the
first years. Furthermore, we carefully selected sampling points at t10
to be not influenced by erosion, i.e., at spots with low surface roughness
and outside rills. Soil samples for the determination of soil properties and
plant samples were taken in five (western and southern section) and six
(eastern section) field replicates at t0 and t10 (Fig. 1). At every
sampling point three undisturbed soil cores were taken with a core cutter
(diameter = 3.4 cm, depth = 5 cm) and transferred into plastic
bags. Bulk densities were calculated from dividing the weight of dried
(105 ∘C) soil samples by their corresponding volumes.
Map of Chicken Creek (W is the western section, CT is the central trench,
E is the eastern section, S is the southern section with pond). Triangles indicate
the sampling points used for BSi analyses at t0 (n = 6). Circles
indicate the sampling points used for measurements of soil parameters (at
t0 and t10) and plant analyses (only at t10) (W, n = 5; E,
n = 6; S, n = 5). Empty and filled circles represent sampling points where
Calamagrostis epigejos and Phragmites australis became
dominant. Note that the size of sampling points is not to scale.
Determination of basic soil properties
Soil samples were air dried and sieved and the fine earth fraction
(< 2 mm) was used for laboratory analyses. Soil pH was measured
based on the DIN ISO method 10390 (1997) in 0.01 M CaCl2 suspensions
at a soil-to-solution ratio of 1:5 (w/v) after a 60 min equilibration
period using a glass electrode. The total carbon content was analyzed by dry
combustion using an elemental analyzer (Vario EL, Elementar Analysensysteme,
Hanau, Germany). Carbonate (CaCO3) was determined conductometrically
using the Scheibler apparatus (Schlichting et al., 1995). Organic carbon
(Corg) was computed as the difference between total carbon and
carbonate carbon. Analyses of basic soil properties were performed in two lab
replicates per sample.
Water-extractable Si (SiH2O)
Water-extractable Si was determined based on a method developed by
Schachtschabel and Heinemann (1967). Ten grams of dry soil (< 2 mm)
was weighed and put into 80 mL centrifuge tubes, and 50 mL distilled water was
added with three drops of a 0.1 % NaN3 solution to prevent
microbial activity. Total extraction time was 7 days in which tubes were
shaken by hand twice a day for 20 s. Mechanical (constant) shaking
by using, e.g., a roll mixer, was avoided to prevent abrasion of mineral
particles from colliding during shaking (McKeague and Cline, 1963). The solutions
were centrifuged (4000 rpm, 20 min), filtrated (0.45 µm polyamide
membrane filters) and Si was measured with ICP–OES (ICP-iCAP 6300 DUO, Thermo
Fisher Scientific Inc). Analyses of water-extractable Si were performed in
two lab replicates per sample.
Tiron-extractable Si (SiTiron), aluminum
(AlTiron) and iron (FeTiron)
The Tiron (C6H4Na2O8S2 ⚫ H2O)
extraction followed the method developed by Biermans and Baert (1977) and
modified by Kodama and Ross (1991). It has been used to quantify amorphous
biogenic and pedogenic Si (Kendrick and Graham, 2004), although a partial
dissolution of primary minerals is well known (Kodama and Ross, 1991; Sauer
et al., 2006). The extraction solution was produced by dilution of 31.42 g
Tiron with 800 mL of distilled water, followed by addition of 100 mL sodium
carbonate solution (5.3 g Na2CO3 + 100 mL distilled water)
under constant stirring. The final pH of 10.5 was reached by adding small
volumes of a 4M NaOH solution. For the extraction, 30 mg of dry soil were
weighed into 80 mL centrifuge tubes and a 30 mL aliquot of the Tiron
solution was added. The tubes were then heated at 80 ∘C in a water
bath for 1 h. The extracted solutions were centrifuged at 4000 rpm for
30 min and filtrated (0.45 µm polyamide membrane filters, Whatman NL
17), and Si, Al and Fe were measured with ICP–OES. Analyses of Tiron-extractable Si,
Al and Fe were performed in three lab replicates per sample.
Microscopical analyses of diatoms, sponge spicules and testate
amoebae
Fresh soil samples were homogenized by gentle turning of the plastic bags
before air drying. Afterwards 2 g of fresh soil was taken per sample and
stored in 8 mL of formalin (4 %). Subsequently, biogenic siliceous
structures, i.e., diatom frustules, testate amoeba shells and sponge spicules
(Fig. 2a–d), were enumerated in soil suspensions (125 mg fresh mass – FM)
received from serial dilution (1000–125 mg soil in 8 mL of water each)
using an inverted microscope (OPTIKA XDS-2, objectives 20:1 and 40:1,
equipped with a digital camera OPTIKAM B9).
Determination of phytoliths in soil samples
Ten grams of dry soil material (< 2 mm) was processed in four steps
(adapted from Alexandre et al., 1997). First organic matter was oxidized using
H2O2 (30 Vol.%), HNO3 (65 Vol.%) and HClO4
(70 Vol.%) at 80 ∘C until the reaction subsided. Secondly,
carbonates and Fe oxides were dissolved by boiling the sample in HCl
(10 Vol.%) for 30 min. Thirdly, the < 2 µm
granulometric fraction was removed by dispersing the remaining solid phase
of step 2 with 2 Vol.% sodium hexametaphosphate solution (6–12 h),
centrifugation at 1000 rpm for 2–3 min and subsequent decantation.
Finally, the phytoliths were separated by shaking the remaining solid phase
of step 3 with 30 mL of sodium polytungstate
(Na6(H2W12O40) ⚫ H2O) with a density of
2.3 g cm-3 and subsequent centrifugation at 3000 rpm for 10 min.
Afterwards, the supernatant was carefully pipetted and filtered using
5 µm Teflon filters. This step was repeated three times. The filter
residue was washed with water, bulked, dried at 105 ∘C and
weighted.
Micrographs (light microscope) of biogenic silica structures found at
Chicken Creek. (a) Pennate diatom (valve view), (b) testate amoeba shell
(Euglypha cristata), (c) and (d) sponge spicules (fragments),
(e) elongate phytolith and (f) bilobate phytolith. All scale bars:
50 µm.
Quantification of biogenic Si pools
In general, biogenic siliceous structures consist of hydrated amorphous
silica (SiO2 ⚫ nH2O). We assumed an average water
content of about 10 % for these structures to avoid an overestimation of
BSi pools (Mortlock and Froelich, 1989).
Protophytic Si pools (represented by diatom frustules) were quantified by
multiplication of Si content per frustule with corresponding individual
numbers (see Puppe et al., 2016). Protozoic Si pools (represented by testate
amoebae) were quantified by multiplication of silica contents of diverse
testate amoeba taxa (Aoki et al., 2007) with corresponding individual numbers
(living plus dead individuals, for details see Puppe et al., 2014, 2015).
Zoogenic Si pools (represented by sponge spicule fragments) were calculated
by multiplying volumes (µm3) of the found spicule fragments
with the density of biogenic Si (2.35 g cm-3) and summing up the
results. Volume measurements were conducted using a laser scanning microscope
(Keyence VK-X110, magnification 200–2000x) (details in Puppe et al., 2016).
For laser scanning microscopy spicule fragments were taken from soil
suspensions by micromanipulation, washed in distilled H2O and placed on
clean object slides. Afterwards, air-drying images of spicule fragments were
acquired (software Keyence VK-H1XVD) and analyzed (software Keyence
VK-H1XAD).
Measured soil parameters (upper 5 cm, means
(x‾) with standard deviation – SD) at the different
sections of Chicken Creek.
Age
Section
SiH2O
SiTiron
AlTiron
FeTiron
Corg
CaCO3
pH
g m-2
t0
West
x‾
0.70
524
312
249
237*
88
7.9
SD
0.10
95
24
33
156
72
0.1
t10
West
x‾
1.73
552
254
239
556*
101
7.4
SD
0.22
300
154
104
167
93
0.1
t0
East
x‾
0.87*
503
268
261
123
91
8.1
SD
0.48
281
151
130
38
79
0.2
t10
East
x‾
1.50*
196
122
151
396
30
7.1
SD
0.57
49
27
29
54
18
0.2
t0
South
x‾
0.84
399
232
238*
160*
174
8.3
SD
0.06
154
112
65
131
109
0.1
t10
South
x‾
2.24
317
147
157*
474*
126
7.4
SD
0.33
149
62
57
258
40
0.1
Significant differences between t0 and
t10 are each stated in bold (Mann–Whitney U test, p < 0.05) or
marked with asterisks (p < 0.1) for the western, eastern and
southern sections.
Phytogenic Si pools were estimated by multiplying the numbers of found
phytoliths with corresponding mean volumes ( µm3) of
phytoliths, multiplying these results with the density of biogenic Si
(2.35 g cm-3) and summing up the results. Volume measurements with the
laser scanning microscope of 30 typical elongate (Fig. 2e) and 30 typical
bilobate phytoliths (Fig. 2f) resulted in mean volumes of
3765 µm3 and 707 µm3, respectively. For laser
scanning microscopy extracted phytoliths were placed on clean object slides
and images were acquired and analyzed analogous to sponge spicules. For
bilobate phytoliths we measured the upper half per phytolith and doubled the
result to obtain the corresponding total volume; thus we assumed bilobate
phytoliths to be symmetric. We assumed phytoliths to consist of 95 %
SiO2 and 5 % other elements, e.g., carbon (Song et al., 2012) and
elements like iron, aluminum or calcium (Buján, 2013).
BSi pools (mg m-2) were calculated considering bulk density
(g cm-3), thickness (5 cm) and – for protistic and zoogenic Si pools
– water content ( % of fresh mass) per soil sample. Silica
(M = 60.08 g mol-1) pools were converted to Si
(M = 28.085 g mol-1) pools by multiplication with 28/60 (details
in Puppe et al., 2014, 2015, 2016).
Plant analyses
Plant and litter samples of C. epigejos and P. australis
were collected in the summer of 2015. In general, monomeric silicic acid
(H4SiO4) enters the plant via its roots and is carried in the
transpiration stream towards transpiration termini. When water evaporates,
silicic acid becomes supersaturated and is precipitated as hydrated silica in
the form of phytoliths. The vast majority of Si in plants is located at the
transpiration termini (e.g., leaves) in the aerial plant parts, while
considerably less Si can be found in other plant portions like stems, roots
and rhizomes. Sangster (1983), for example, found no significant Si
depositions in rhizomes of P. australis. Consequently, we only
analyzed the aboveground vegetation (including transpiration termini and
stems). The collected plant material was washed with distilled water to
remove adhering soil minerals and oven-dried at 45 ∘C for 48 h.
Spearman's rank correlations between measured soil parameters and
total BSi (upper 5 cm, n = 6) at Chicken Creek.
SiH2O
SiTiron
AlTiron
FeTiron
Corg
CaCO3
pH
BSi
SiH2O
1.000
SiTiron
-0.257
1.000
AlTiron
-0.600
0.829
1.000
FeTiron
-0.486
0.771
0.943
1.000
Corg
0.714
0.086
-0.371
-0.486
1.000
CaCO3
0.200
0.086
-0.086
-0.029
0.029
1.000
pH
-0.600
0.200
0.486
0.543
-0.771
0.543
1.000
BSi
0.941
-0.213
-0.577
-0.577
0.880
0.152
-0.698
1.000
Significant correlation
coefficients are given in bold (p < 0.05).
Surface areas, volumes and surface-to-volume ratios (A / V) of
different biogenic siliceous structures found at Chicken Creek.
Surface area (µm2)
Volume (µm3)
A / V ratio
Min.
Max.
Min.
Max.
Range
Mean (SD)
Bilobate phytoliths
216
3730
36
2046
0.7–9.8
2.8 (1.8)
Elongate phytoliths
2302
22 203
390
14 649
0.6–5.9
2.6 (1.1)
Diatom frustules*
351
9901
347
28 024
0.3–3.3
0.9 (0.5)
TA shells*
1229
5085
900
15 812
0.2–2.7
0.8 (0.7)
Sponge spicules*
305
16 963
291
59 744
0.3–1.6
0.8 (0.4)
Spicule fragments*
2828
17 268
5255
34 812
0.5–0.6
0.5 (0.03)
* Data taken from Puppe et al. (2016).
Total Si content in plant materials
Plant samples were milled using a knife mill (Grindomix GM 200, Retsch) in
two steps: 4000 rpm for 1 min and then 10 000 rpm for 3 min. Sample
aliquots of approximately 100 mg were digested under pressure in PFA
digestion vessels using a mixture of 4 mL distilled water, 5 mL nitric acid
(65 %) and 1 mL hydrofluoric acid (40 %) at 190 ∘C using a
microwave digestion system (Mars 6, CEM). A second digestion step was used to
neutralize the hydrofluoric acid with 10 mL of a 4 % boric acid solution
at 150 ∘C. Silicon was measured with ICP–OES (ICP-iCAP 6300 Duo,
Thermo Fisher Scientific Inc) with an internal standard. To avoid
contamination, plastic equipment was used during the entire procedure.
Analyses of total Si content were performed in three lab replicates per
sample.
Determination of phytoliths in plants and litter
Plant material was washed with distilled water and oven-dried at
45 ∘C for 48 h. Removal of organic matter was conducted by burning
the samples in a muffle furnace at 450 ∘C for 12 h. Next, the
material was subject to additional oxidation using 30 % H2O2
for 12 h. The obtained material was filtered through a Teflon filter with a
mesh size of 5 µm. The isolated phytoliths and siliceous cast
(> 5 µm) were subject to analysis via polarized light
microscopy (Nikon ECLIPSE LV100 microscope) for full characteristics. We used
laser scanning microscopy for measurements of the surface area
(µm2) of the 30 typical bilobate and 30 typical elongated
phytoliths used for volume measurements (see Sect. 2.6) and calculated
corresponding surface-area-to-volume ratios (A / V ratios) as an indicator
of the resistibility of these siliceous structures against dissolution. Higher
A / V ratios indicate a bigger surface area available for dissolution
processes.
Statistical analyses
Correlations were analyzed using Spearman's rank correlation (rs).
Significances in two-sample (n = 2) cases were verified with the
Mann–Whitney U test. For k-sample (n > 2) cases the
Kruskal–Wallis analysis of variance (ANOVA) was used followed by pairwise
multiple comparisons (Dunn's post hoc test). Statistical analyses were
performed using software package SPSS Statistics (version 19.0.0.1, IBM
Corp.).
Results
Basic soil parameters
Soils at the initial state (t0) showed organic carbon
contents (Corg) in the upper 5 cm between 1.1 and 4.4 g kg-1 in the western
section, 0.8 and 1.8 g kg-1 in the eastern section and 0.2 and
3.3 g kg-1 in the southern section. This corresponded to mean carbon
stocks of 237 g m-2 (west), 123 g m-2 (east) and
160 g m-2 (south, Table 2). After 10 years (t10) of ecosystem
development the Corg stocks increased up to a factor of 3
(396–556 g m-2 in the upper 5 cm) from corresponding values
at t0. This resulted in a surprisingly high mean annual CO2-C
sequestration rate of 27–32 g m-2 (upper 5 cm). Hereby the largest
Corg stock changes were found in the western section of the area
followed by the eastern section and the southern section (Table 2).
The carbonate contents (CaCO3) at t0 varied between means of
1.0 g kg-1 (west), 0.9 g kg-1 (east) and 1.8 g kg-1
(south). The corresponding stocks were 88 g m-2 (west),
91 g m-2 (east) and 174 g m-2 (south, Table 2). The carbonate
pools in the western and eastern section were very similar, while the high
carbonate values in the southern section were due to the original soil
properties. At t10 the distribution of carbonate was as follows: in the
western section there was an increase of about 17 % (from 88
to 101 g m-2), in the eastern part a distinct decrease of about
67 % (from 91 to 30 g m-2) was detected and in the
southern section again a decrease of about 28 % (from 174 to
126 g m-2) was identified.
At t0 the pH values of the soils showed a range between 7.9 and 8.3
(Table 2) with relatively low variation between the different sections. After
10 years the pH values decreased to 7.1–7.4 in all sections.
Water and Tiron extractions
The mean water-soluble Si (SiH2O) contents in the upper 5 cm showed low
variation between the different sections at t0:
7.3 mg kg-1 (west), 7.2 mg kg-1 (east) and
8.6 mg kg-1 (south). The corresponding stock values were
0.7 g m-2 (west), 0.87 g m-2 (east) and 0.84 g m-2
(south) for all sections at t0 (Table 2). After 10 years (t10) an
overall significant increase of SiH2O from t0 was found
in each of the different sections. The corresponding stock values were
1.7 g m-2 (west), 1.5 g m-2 (east) and 2.2 g m-2 (south,
Table 2).
At t0 the mean Tiron-extractable Si contents in the upper 5 cm varied
between 5.5 g kg-1 (west), 5.2 g kg-1 (east) and
4.1 g kg-1 (south). The related stock values were 524 g m-2
(west), 503 g m-2 (east) and 399 g m-2 (south, Table 2). After
10 years (t10) the Tiron-extractable Si content showed a slight
increase in the western section to 6.5 g kg-1 (552 g m-2),
while the concentration in the eastern section decreased significantly to
2.6 g kg-1 (196 g m-2, Table 2). In the southern section only a
slight decrease to 3.8 g kg-1 (317 g m-2) was found. The Al and
Fe-extractable Tiron contents followed the distribution of the Si
concentrations with one exception in the western section, where contrary to
Si the Al and the Fe contents slightly increased at t10 (Table 2).
Si / Al ratios ranged between 1.6 and 2.2 at Chicken Creek. Tiron-extractable Si and Al fractions as well as Tiron-extractable Al and Fe
fractions were strongly correlated (Table 3).
Total biogenic Si pools in soils (means ± standard deviation,
upper 5 cm) at Chicken Creek at the end of construction work (t0) and
after 10 years of ecosystem development (western, eastern and southern
sections, t10). Significant differences are indicated by different
letters (p < 0.05, Kruskal–Wallis ANOVA with Dunn's post hoc test).
Box plots (top, middle and bottom lines of the boxes show the
25th, 50th and 75th percentiles and whiskers
represent 1.5× the interquartile ranges) of biogenic Si pools in
soils (upper 5 cm) at Chicken Creek at the end of construction work
(t0) and after 10 years of ecosystem development (western, eastern and
southern sections, t10). (a) Phytogenic Si pools (phytoliths),
(b) protophytic Si pools (diatom frustules), (c) zoogenic Si pools (sponge
spicules) and (d) protozoic Si pools (testate amoeba shells). Significant
differences are indicated by different letters (p < 0.05,
Kruskal–Wallis ANOVA with Dunn's post hoc test). Circles and asterisks
indicate outliers and extreme values, respectively. Note different scales for
diagrams (a) and (b) and (c) and (d).
Proportions of phytoliths (PHY), sponge spicules (SPO), diatom
frustules (DIA) and testate amoeba shells (TA) to total BSi in soils (upper
5 cm) at Chicken Creek at t0 and t10. Note that total BSi pools
differ in size (see Fig. 3).
Biogenic Si pools in soils
In general, total biogenic Si pools increased in every section after 10
years of ecosystem development with statistically significant differences
between t0 (11.6 ± 6.5 mg Si m-2) and the southern section
at t10 (96.0 ± 87.2 mg Si m-2) (Fig. 3). Total BSi showed
strong positive and statistically significant correlations to water-soluble
Si (Table 3). Phytogenic (phytoliths > 5 µm) Si pools
ranged from 0 to 18 mg m-2 (mean: 6.6 mg m-2) at
t0 and
significantly increased to means of 20.7 mg m-2 (range:
7–52 mg m-2) and 12.9 mg m-2 (range: 14–15 mg m-2) at
the eastern and southern sections over 10 years, respectively (Fig. 4a).
Protophytic Si pools (diatom frustules) ranged from 0 to 7 mg m-2 (mean:
2.6 mg m-2) at t0 and increased up to a mean of
47.4 mg m-2 (range: 0.1–162 mg m-2) at t10 (southern
section) (Fig. 4b). At t0 no sponge spicules were found with one
exception representing an extreme value (12.7 mg m-2). After one
decade of ecosystem development zoogenic Si pools increased to a maximum of
46 mg m-2 in the southern section (t10) (Fig. 4c). Protozoic Si
pools were zero at t0, with one exception representing an extreme value
(1.8 mg m-2), and significantly increased to 4.6 mg m-2 (range:
1–11 mg m-2) and 11.5 mg m-2 (range: 2–36 mg m-2) in
the eastern and the southern sections at t10, respectively (Fig. 4d).
Comparison of water-soluble Si (SiH2O), amorphous Si
(SiTiron) fractions and total BSi in soils (means ± standard deviation, upper 5 cm), where Calamagrostis epigejos (a)
and Phragmites australis (b) became dominant. Data are given for
t0 (no vegetation) and t10 (C. epigejos, P. australis). For t10 total plant Si contents, extracted phytogenic Si
(phytoliths) contents and Si pools for C. epigejos and P. australis (plants and litter) are stated in addition. Paintings are from
Cornelia Höhn, Müncheberg.
Micrographs of fragile phytogenic Si structures (arrows) of
C. epigejos (a) and P. australis (b).
At t0 most BSi (> 50 %) is represented by phytoliths
> 5 µm followed by diatom frustules, sponge spicules
and testate amoeba shells (Fig. 5). After 10 years of ecosystem development
the proportion of the different BSi pools to total BSi changed. While the
proportion of protozoic Si pools increased in all sections at t10, the
other BSi pools showed more variable changes over time. The proportion of
phytogenic Si pools either increased (western section) or decreased (eastern
and southern sections). In contrast, the proportion of protophytic Si pools
decreased in the western section and increased in the eastern and southern
sections. The proportion of zoogenic Si pools decreased in the western and
eastern sections but increased slightly in the southern section at t10.
Phytoliths and total Si content in plant materials
The total content of Si was determined for two Si-accumulating plant species,
Calamagrostis epigejos and Phragmites australis, which
dominate distinct catchment sections. For C. epigejos the mean total content
of Si was 2.25 % (range: 1.8–3.1 %), whereas for P. australis a mean total Si content of 2.70 % (range: 2.0–3.2 %) was
determined (Fig. 6a, b). For litter we found mean total Si contents of
3.1 % (range: 2.8–3.3 %) and 2.9 % (range: 1.7–3.2 %) for
C. epigejos and P. australis, respectively.
Phytoliths > 5 µm were also isolated from both plants, showing mean phytolith contents of 0.37 %
(range: 0.31–0.46 %) and 0.43 % (range: 0.37–0.50 %) for C. epigejos and P. australis, respectively (Fig. 6a,
b). Regarding the total Si content of plants only about 16 % of phytogenic Si were represented by the extracted phytoliths.
Thus, small-scale (< 5 µm) and/or fragile
(siliceous structures mostly thinner than 5 µm, but up to several
hundred micrometers long, Fig. 7) phytogenic Si represented about
84 % of total phytogenic Si in C. epigejos and P. australis, respectively. Mean extracted phytolith contents in plant litter
were 0.47 % (range: 0.35–0.70 %) and 0.51 % (range:
0.41–0.59 %) for C. epigejos and P. australis.
Surface areas of 30 typical bilobate and 30 typical elongate phytoliths were
in the ranges of 216 to 3730 µm2 and 2302 to
22 203 µm2 (Table 4). The corresponding volumes
of bilobate and elongate phytoliths were in the ranges of 36 to
2046 µm3 and 390 to 14 649 µm3.
Surface-to-volume ratios of bilobate and elongate phytoliths were in the
ranges of 0.7 to 9.8 and 0.6 to 5.9 with means of 2.8 and 2.6.
BSi and Si fractions under Calamagrostis epigejos and Phragmites australis
Water-soluble Si fractions increased by 99 and 163 % and total BSi by
281 and 660 % after 10 years of ecosystem development in soils under
C. epigejos and P. australis (Fig. 6a, b). In
contrast, SiTiron decreased by 42 and 1.4 % from
t0 to t10 in soils under C. epigejos and P. australis. If we assume mean dry biomasses of 115 and
186 g m-2 for C. epigejos and P. australis (M.
Wehrhan, personal communication, 2017) about 2.6 and 5.0 g Si m-2 are
stored in the aboveground biomass at Chicken Creek at t10.
For C. epigejos and P. australis litter (mean dry
biomasses of 59 and 94 g m-2 at t10; M. Wehrhan, personal
communication, 2017), we calculated corresponding pools of about 1.8 and
2.7 g Si m-2 at t10.
Discussion
Drivers of short-term changes in water-soluble Si at Chicken
Creek
In general, weathering of silicates represents the ultimate source of
Si(OH)4 in terrestrial biogeosystems in the long term (Berner, 2003). In
this context, the long-term accumulation of BSi can influence the total
amorphous (Tiron-extractable) Si as it is known from forested catchments or
old chronosequence soils (Conley et al., 2008; Kendrick and Graham, 2004;
Saccone et al., 2008). Contrary, short-term changes in BSi pools likely do
not influence Tiron-extractable Si in initial soils (total BSi represents
only 0.002–0.03 % of Tiron-extractable Si at Chicken Creek). Thus, the
major proportion of Tiron-extractable Si at Chicken Creek seems to be of
pedogenic origin (e.g., Si included in Al / Fe oxides / hydroxides). This is
supported by relatively low Si / Al ratios (< 5) indicating a
minerogenic origin of Tiron-extractable Si instead of BSi as a source of
SiTiron (Bartoli and Wilding, 1980). We further exclude changes
in Tiron-extractable Si as the main driver of water-soluble Si at Chicken
Creek in the short term, because (i) SiTiron and
SiH2O showed no statistical relationship at all and (ii) a
significant change of the Tiron-extractable Si fraction occurred only in the
eastern section, whereas in the western and southern section
SiTiron did not change significantly over time. We assume that
these changes in SiTiron in the eastern section are related to
abiotic conditions (soil pH, conductivity, skeleton content, proportions of
sand, silt and clay, concentration of organic and inorganic carbon), which
were slightly different to the conditions of the western section at
t0 (Gerwin et al., 2010). Furthermore, we excluded atmospheric inputs as
potential drivers of short-term changes in water-soluble Si at Chicken Creek.
On the one hand, dust depositions (dry deposition) at Chicken Creek are very
low (73–230 mg m-2 d-1) and only slightly above the annual
average (70–90 mg m-2 d-1) measured in the state of Brandenburg
(Wanner et al., 2015). On the other hand, the total input of Si (as a
lithogenic element) from precipitation (wet deposition) is negligible as well
(< 1 kg Si ha-1 yr-1, Sommer et al., 2013).
Our results indicate a strong relationship between water-soluble Si and total
BSi. In this context, two different causal chains can be discussed: either
SiO2-synthesizing organisms are drivers of the amount of Si(OH)4 in
the soil or – vice versa – the amount of water-soluble Si in the soils is
the main driver of SiO2-synthesizing organisms as biosilicification is
limited by Si(OH)4. Laboratory studies revealed that
SiO2-synthesizing organisms, i.e., testate amoebae, can deplete the
amount of Si(OH)4 in culture media due to biosilicification (Aoki et
al., 2007; Wanner et al., 2016). However, Wanner et al. (2016) also showed
that culture growth of SiO2-synthesizing testate amoebae was dependent
on Si concentration in the culture media. Furthermore, in situ analyses
showed that marine diatom blooms can deplete Si(OH)4 concentrations in
the oceans (Hildebrand, 2008). In forested biogeosystems Puppe et al. (2015)
found high individual numbers of SiO2-synthesizing testate amoebae at
study sites with low amounts of Si(OH)4 and vice versa. However, it is
unlikely that testate amoebae depleted amounts of Si(OH)4 at these
sites, because corresponding protozoic Si pools are relatively small compared
to phytogenic ones (Puppe et al., 2015; Sommer et al., 2013). Regarding
vegetation and corresponding phytogenic Si pools, their influence on the
amount of Si(OH)4 in soils has been shown in several studies (e.g.,
Bartoli, 1983; Farmer et al., 2005; Sommer et al., 2013). On the other hand,
phytolith production is probably more influenced by the phylogenetic position
of a plant than by environmental factors like temperature or Si availability
(Hodson et al., 2005; Cooke and Leishman, 2012).
From our results and the discussion above we conclude short-term changes in
water-soluble Si to be mainly driven by BSi. However, total BSi represents
only small proportions of water-soluble Si at t0 (< 2 %)
and t10 (< 4.5 %). From this result a question arises:
where does the major part of the increase in water-soluble Si at Chicken
Creek come from? We will discuss this question in Sect. 4.2 below.
Sources of water-soluble Si at Chicken Creek
From former results of BSi analyses in forested biogeosystems, we assumed the
phytogenic Si pool to be the most prominent in size. In this context, results
of Sommer et al. (2013) and Puppe et al. (2015) showed that phytogenic Si
pools in soils of forested biogeosystems were up to several hundred times
larger than protozoic Si pools. However, phytogenic Si pools in soils are
surprisingly small compared to other BSi pools at Chicken Creek. Our findings
can be attributed to at least two factors. Firstly, phytogenic Si is stored
in a developing organic litter layer where it is temporarily protected
against dissolution, and secondly, the used methods were not able to
accurately quantify the total phytogenic Si pool, but only the larger
(> 5 µm) and more stable part.
Total Si and phytolith contents of litter samples at Chicken Creek did not
differentiate from total Si and phytolith contents of plants. This fact
indicates that litter decomposition and related Si release into the subjacent
soil are relatively slow processes and we interpret our findings as an indication
of a developing compartment of dead plant tissue above the mineral soil
surface. Esperschütz et al. (2013) showed in a field experiment in
initial soils near Chicken Creek that after 30 weeks only 50 % of the
C. epigejos litter was degraded, whereby degradation rates were
highest in the first 4 weeks. Estimations of biomasses of C. epigejos and P. australis at Chicken Creek via remote sensing with
an unmanned aerial system showed that the relation between phytogenic Si
pools of plant biomass and litter biomass are almost the same for both plant
species (factor about 1.5, based on the total area of Chicken Creek); i.e.,
Si in the plants was about one-third higher than in litter (M. Wehrhan,
personal communication, 2017). At the sampling points about 1.8 and
2.7 g Si m-2 were stored in the litter of C. epigejos and
P. australis at t10, respectively, which is in the range of
published data for annual Si input through litterfall in a short grass steppe
(2.2–2.6 g Si m-2 yr-1, Blecker et al., 2006).
Altogether, these results clearly underline our interpretation of a
developing organic layer where litter accumulates and phytogenic Si is
temporarily stored and protected against dissolution. Thus Si release is
delayed and biologically controlled, as it can be observed at forested
biogeosystems (Sommer et al., 2013). The Si pools in the aboveground biomass
of C. epigejos (2.6 g Si m-2) and P. australis
(5.0 g Si m-2) at Chicken Creek at t10 are comparable to
reported values of Great Plains grasslands (2.2–6.7 g Si m-2 in the
aboveground biomass) (Blecker et al., 2006) and reach about 30 %
(C. epigejos) or 59 % (P. australis) of published data
for a beech forest (8.5 g Si m-2 in the aboveground biomass of
Fagus sylvatica trees) in northern Brandenburg, Germany (Sommer et
al., 2013), after (only) 10 years of ecosystem development.
Regarding methodological shortcomings of the used phytolith extraction
procedure there are several aspects to be discussed. Wilding and
Drees (1971), for example, showed that about 72 % of leaf phytoliths of
American beech (Fagus grandifolia) are smaller than 5 µm.
This is in accordance with our findings.
Phytoliths > 5 µm only amounted to about 16 % of
total Si contents of plant materials of C. epigejos and P. australis; thus about 84 % of phytogenic Si (< 5 µm
and/or fragile phytogenic Si structures) are not quantified by the used
phytolith extraction method. Watteau and Villemin (2001) found even smaller
(5–80 nm) spherical grains of pure silica in leaf residues in topsoil
samples of a forested biogeosystem. In addition, silica depositions can be
found in intercellular spaces or in an extracellular (cuticular) layer
(Sangster et al., 2001), whereat no recognizable phytoliths are formed. These
structures might be too fragile for preservation in soils and are likely lost
to a great extent in the used phytolith extraction procedure due to
dissolution. Meunier et al. (2017) analyzed different phytolith morphotypes,
e.g, silica bodies originating from cells of the upper epidermis, silica
casts of trichomes or parenchyma/collenchyma cells and of durum wheat plant
shoots. They found fragile subcuticular silica plates (2–4 µm
thick, up to several hundred micrometers long and wide) to be the second most
common phytolith morphotype. This is corroborated by our own findings as the
biggest part (about 84 %) of total plant Si is represented by small-scale
(< 5 µm) and/or fragile phytogenic Si in C. epigejos and P. australis. If we assume that total Si contents of
plants at Chicken Creek are one-to-one reflected by phytogenic Si pools in
soils, we can easily calculate these small-scale and fragile pools resulting
in about 130 and 100 mg m-2 (84 % of total, i.e.,
156 and 119 mg m-2, phytogenic Si each) under C. epigejos and P. australis, respectively. These calculated
phytogenic Si pools are about 13 (diatom frustules), 38 (testate amoeba
shells) and 45 (sponge spicules) or 3 (diatom frustules) and 10 (testate
amoeba shells, sponge spicules) times bigger than the other BSi pools at
C. epigejos and P. australis sampling points.
If we further assume an input of this phytogenic Si for at least 7 years
(Zaplata et al., 2010) phytogenic Si might be the main driver of short-term
changes in water-soluble Si at Chicken Creek. This is supported by relatively
high surface-to-volume ratios of bilobate and elongate phytoliths. These
ratios are about 3 times higher compared to ratios of other biogenic
siliceous structures, i.e., testate amoeba shells, diatom frustules and
sponge spicules.
In addition, Si pools represented by single siliceous platelets of testate
amoeba shells have to be considered as well, as these platelets can be
frequently found in freshwater sediments, for example (Douglas and Smol,
1987; Pienitz et al., 1995). Unfortunately, there is no available information on the
quantity of such platelet pools in soils, but it can be assumed
that these platelets can be frequently found in soils, as they are used by
some testate amoeba genera (e.g., Schoenbornia, Heleopera)
for shell construction (Meisterfeld, 2002; Schönborn et al., 1987). In
general, it can be assumed that phytogenic Si structures
< 5 µm and single testate amoeba platelets (about
3–12 µm in diameter, Douglas and Smol, 1987) are highly reactive
due to their relatively high surface-to-volume ratios. However, to the best of
our knowledge there is no publication available dealing with corresponding
physicochemical analyses or dissolution kinetics of these siliceous
structures. In general, experiments with phytoliths
(> 5 µm) showed that surface areas and related
dissolution susceptibilities are, for example, age-related due to changes in
specific surface areas and the presence of organic matter bound to the
surface of phytoliths (Fraysse et al., 2006, 2009).