Effects of soil types and fertility management practices on soil silicon availability and banana silicon uptake

ABSTRACT Although silicon (Si) is one of soil abundant constituents and one of the key elements enhancing plant disease resistance, banana uptake of Si is limited by soil Si availability. It remains unclear what factors regulates soil Si availability and whether banana Si uptake increases with increasing soil Si availability in home garden soils. To test whether soil Si availability and banana leaf SiO2 contents are affected by soil types or fertility management practices, we measured (1) banana leaf SiO2 contents along the gradient of livestock inputs in Andisols of Tanzania and (2) water extractable Si, phosphate-buffer extractable Si, and banana leaf SiO2 in the home garden soils with different degree of volcanic ash admixing and with/without leguminous trees in Indonesia (East Java and East Kalimantan). Livestock dung application increased banana leaf SiO2 contents in the Tanzanian home gardens. Water extractable Si increased with increasing soil pH, but soil phosphate-buffer extractable Si and banana leaf SiO2 contents were strongly regulated by oxalate-extractable Si and andic properties in the Indonesian home gardens. Effects of companion trees (agroforestry) increased banana leaf SiO2 only in two of three Andisols (Java). These results suggest that banana leaf SiO2 contents could be primarily regulated by admixing degree of volcanic parent materials rich in weatherable minerals, but that banana leaf SiO2 contents could also be increased by inputs of livestock dung rich in Si (Tanzania) and agroforestry (Indoensia).


Introduction
Banana is important as fruit all over the world and a staple food in some tropical regions with subsistence agriculture. Banana is facing risks of diseases such as Black Sigatoka and Panama disease (Ploetz 2006;Soares et al. 2021). Especially, the latter is caused by pathogenic fungi (Fusarium oxysporum f.sp. cubense; FOC) (Ploetz 2006;Maryani et al. 2019). Compared to genetic and fungicide improvement, soil fertility management is one of cost-efficient strategies to increase plant resistance to diseases especially in noncommercial small home gardens.
As with some of Poaceae plants (rice or bamboo), banana is an accumulator of silicon (Si) or biogenic silica in plant tissues (Jones 2014;Vander Linden and Delvaux 2019). Si is accumulated in banana leaf as a result of evapotranspiration along with growth (Henriet et al. 2008). SiO 2 is most abundant constituent in soil, but Si solubility is low due to weathering resistance of crystalline structures in clay minerals (Ma 2004). The potential roles of Si in plants are enhancing photosynthetic activities and mitigating soil acidification, but plant Si also increases resistance to pathogen by forming cuticle-silica double layers (Wang et al. 2017). Cuticle-silica double layers protect penetration of fungal hyphae into the plant tissue to prevent diseases infected by FOC (Fortunato, da Silva, and Rodrigues 2014;Jones 2014). Our preliminary survey in Philippine showed the lower silica (SiO 2 ) contents in mature banana leaf with symptoms of black Sigatoka and Panama diseases than in healthy banana leaf in banana plantation ( Figure S1). Further study is needed to support the potential importance of Si in plant resistance to diseases.
The ultimate Si sources of banana are soil or organic amendments. Fertility management by organic amendments could affect plant SiO 2 contents (Jones 2014). Application of manure or livestock dung inputs and plant litters is hypothesized to increase soil Si availability and banana leaf Si contents, because fodder grass (e.g., Imperata spp.) is Si-accumulating plant (Komiyama, Kobayashi, and Yahagi 2013;Vander Linden and Delvaux 2019). Agroforestry (planting trees with banana) also has the potential to increase Si supply to the surface soil and banana through biological pumping (Lucas et al. 1993;Vander Linden and Delvaux 2019). In addition, soil types could also affect soil Si availability and banana leaf SiO 2 contents (Henriet et al. 2008). Weatherable minerals in volcanic soils could contribute to the higher Si solubility than crystalline minerals in non-volcanic or highly weathered soils (Henriet et al. 2008). Soil Si availability and banana leaf Si uptake are hypothesized to be enhanced by fertility management practices in volcanic soils, but not in the non-volcanic soils.
To test the first hypothesis that organic matter inputs (livestock dung or forest litters) increase banana leaf Si contents, we compared banana SiO 2 contents in the home-garden soils with different dose of livestock dung inputs in Tanzania and in agroforestry systems in Indonesia, respectively. To test the second hypothesis that banana leaf Si uptake is enhanced in volcanic soils, we examine the relationships between soil Si availability and soil chemical properties and between soil Si availability and banana leaf SiO 2 contents in the home-garden soils with different andic properties in Indonesia.

Experiment 1
To test the effects of livestock dung inputs on banana leaf Si contents, mature banana leaf [lamina of antepenultimate leaf (leaf III) from three individual trees (N = 3)] and soil samples (N = 5) were collected in January 2016 from five home gardens (TZ1, TZ2, TZ3, TZ4, TZ5) in Kilema Kaskazini ward located on the southern slopes on Mt. Kilimanjaro in northeastern Tanzania (S3°16,' E37°28; 1600 m a.s.l.). The mean annual air temperature is 19.0ºC, and the mean recorded annual precipitation is 1500-2000 mm yr −1 (Ichinose et al. 2020). Soils are classified as Andisols derived from volcanic parent materials (Soil Survey Staff 2014). Five home gardens were located within 1 km square to eliminate the effect of climate, topography, and parent material of soils. Five home gardens differ in livestock (cattle and/or goat) density and thus, dung inputs as follows: TZ1 (9.0 Mg ha −1 yr −1 ) > TZ2 (5.0 Mg ha −1 yr −1 ) > TZ3 (3.4 Mg ha −1 yr −1 ) > TZ4, TZ5 (0.6 Mg ha −1 yr −1 ) ( Table S1). The fodder of livestock was mostly derived from grassland outside home gardens. The detailed site information was given in Ichinose et al. (2020).

Experiment 2
To test the effects of volcanic ash parent materials on soil Si availability and mature banana leaf (lamina) SiO 2 contents, volcanic soil and banana leaf (lamina) samples were collected in April 2019 at home gardens in Senduro, Lumajang (LJ; S8°09,' E113°04), East Java Province, Indonesia ( Figure S2). The mean annual air temperature ranged from 16.1ºC (highland at 1500 m a.s.l.) to 25.9ºC (lowland, Lumajang), and the mean recorded annual precipitation is 1815 mm yr −1 at the metrological station of Lumajang. Soils are classified as Andisols derived from volcanic parent materials (Soil Survey Staff 2014). Non-volcanic home garden soils were collected in Bukit Soeharto (BS; S0°51,' E117° 06;' 99 m a. s. l.), East Kalimantan Province, Indonesia ( Figure S2). The mean annual air temperature is 26.8ºC, and the mean recorded annual precipitation is 2187 mm yr −1 . Soils are classified as Ultisols derived from sedimentary rocks (Soil Survey Staff 2014). In the home gardens, banana was cropped with and without leguminous trees (Paraserianthes falcataria L.) in LJ or Artocarpus spp. in BS, respectively. No chemical fertilizer or cattle manure was applied, except for woody litter application. The composite soil samples from three pits (N = 1) and mature banana leaf samples [lamina of antepenultimate leaf (leaf III) from three individual trees (N = 3)] were obtained at each site. For the paired sites, soil chemical properties and banana leaf Si contents were compared to examine the effects of companion trees (agroforestry) on banana leaf Si in agroforestry systems. So that soil samples can cover the wide range of physicochemical properties (e.g., pH and andic property) to analyze their relationships with soil Si solubility, we included the deeper soil samples as well as topsoil samples. We also took reference soil samples at the sites close to BS or LJ (BS2-BS6, LJ5) under different land uses that have the potential to be replaced by banana cultivation. The detailed site information of BS2 to BS6 was given in .

Physicochemical properties of soil samples
For experiment 2, soil samples were air-dried, and sieved using a 2-mm mesh. Soil pH was measured using a soil-to-solution (water or KCl) ratio of 1:5 (w/v) after 1 h of shaking. Total C and N concentrations were determined using a CN analyzer (vario MAX CN; Elementar Analysensystem GmbH, Langenselbold, Germany). Particle size distributions were determined using the pipette method, and exchangeable basic cation concentrations were extracted using the ammonium acetate (1 M, pH 7.0). Exchangeable Ca and Mg were measured using atomic absorption spectroscopy (AA-640-01; Shimadzu, Kyoto, Japan), and exchangeable Na and K were measured by flame photometry. The amounts of the short-range-order Fe and Al minerals and organo-Al/Fe complex (Fe o , Al o , and Si o ) in soils were estimated by extraction in the dark with acidic (pH 3) 0.2 M ammonium oxalate for 4 h (McKeague and Day 1966).

Silica contents in banana tissues
The collected plant materials were rinsed with distilled water, oven-dried at 70°C for 72 h, and milled. The SiO 2 contents in plant tissues were measured using the modified method of Umemura and Takenaka (2014). After nitric acid wet digestion, the digest solution with residues was filtered using membrane filter (ADVANTEC, cellulose mixed ester type, Pore Size 0.45 μm), washed with diluted HCl solution, and maffle at 600°C for 4 h, and remaining ash was weighed. Si concentration in the digest solution was measured with molybdenum-blue colorimetry. The sum of insoluble and soluble Si was presented as banana leaf SiO 2 contents.

Extractable silicon concentrations in soils
For experiment 2, soil Si availability was analyzed to assess relationship with banana Si uptake. Soil water-extractable Si was measured using the supernatant of soil extracts used for pH (water) measurement [soil-to-solution ratio of 1:5 (w/v)]. Soil available Si was measured by phosphate buffer method (Kato 1998), where 2 g soil in 50 mL polyethylene tube was shaken with 20 mL phosphate buffer [mixture of 0.04 mol L −1 NaH 2 PO 4 and 0.04 mol L −1 Na 2 HPO 4 (82:18), pH 6.2] for 5 min and stored at 40°C for 24 h. Then, after shaking for 5 min and centrifugation (3,500 rpm, 10 min), the supernatant was filtered with the filter paper (ADVANTEC, No. 6) and Si concentration in the supernatant was measured with molybdenum-blue colorimetry.

Statistics
All results are given as means � standard error (SE) of three to five replicates. The significance of differences in the mean values between groups (sites with/without companion trees) were evaluated using analysis of variance (ANOVA) at a P < 0.05 significance level. Pearson's correlation coefficients were calculated to examine the relationships between soil Si availability, banana leaf Si, and soil properties. These analyses were performed using Sigmaplot 14.5 software (SPSS Inc.).

Effects of livestock dung on banana leaf Si in the Tanzanian home gardens (experiment 1)
Banana leaf SiO 2 contents ranged widely from 5 to 19 g kg −1 in five home gardens of Tanzania (Figure 1(a,b)). When banana leaf SiO 2 contents were compared between the home-garden soils that differed in livestock dung inputs in Tanzania, banana leaf SiO 2 contents were positively correlated with livestock dung inputs (Figure 1(a)). The banana leaf SiO 2 contents appeared to reach saturation in the home garden soil with highest dose of livestock dung, although the range was not enough to assess manure dose dependency of banana leaf SiO 2 . In addition to manure inputs, banana leaf SiO 2 contents were also positively correlated with soil pH (P < 0.05) (Figure 1(b)). The banana stem or fruit SiO 2 contents were significantly (P < 0.05) lower than the respective leaf SiO 2 contents at all the sites except for TZ5 ( Figure S3). No correlations were found between livestock dung inputs and banana stem or fruit SiO 2 contents ( Figure S3).

Effects of soil types on soil Si availability and banana leaf Si in Indonesia (experiment 2)
Soil chemical properties indicate that the LJ soils exhibited higher pH, exchangeable basic cations, andic properties (Al o +1/2 Fe o ), and oxalate extractable Si (Si o ), compared to the BS soils (Table 1). When soil extractable Si were compared between different soil samples, phosphate buffer extractable Si in soils was positively correlated with oxalate extractable Si (P < 0.05) (Figure 2(a)) and andic properties or the amounts of oxalate extractable Al and Fe (P < 0.05) ( Figure S4), respectively. Water extractable Si in soils was positively correlated with soil pH (P < 0.05) (Figure 2(b)). Banana leaf SiO 2 contents ranged widely from 6 to 29 g kg −1 in the home gardens of Indonesia (Figure 3(a)). Banana leaf SiO 2 contents were positively correlated with oxalate extractable Si (P < 0.05) (Figure 3(a)) and phosphate buffer extractable Si (P < 0.05) (Figure 3(b)), respectively, but not with water extractable Si (P > 0.05) (Figure 3(c)). For the data including the data of Tanzania (Experiment 1) as well as Indonesia (Experiment 2), banana leaf SiO 2 contents increased with increasing amounts of oxalate extractable Al and Fe (P < 0.05) ( Figure S5). Note that banana leaf SiO 2 contents varied widely between Tanzanian home gardens (Experiment 1), despite the similar andic properties ( Figure S5).
When soil chemical properties and banana leaf SiO 2 contents were compared between the paired home gardens with/ without trees (LJ2, LJ3, LJ4, BS1), home garden soils with leguminous trees tend to have higher soil C concentrations and lower pH (H 2 O or KCl) in LJ2, LJ3, LJ4, and BS1 (Table 1). On the other hand, banana leaf SiO 2 contents in the agroforestry home gardens were significantly (P < 0.05) higher than the sites without leguminous trees in LJ3 and LJ4, but no difference was found in LJ2 and BS1 (Table 1)

Effects of livestock dung inputs on banana leaf Si contents in home gardens
Because fodder grass is generally rich in Si (Vander Linden and Delvaux 2019), livestock dung inputs [9.5% Si in cattle manure from Komiyama, Kobayashi, and Yahagi (2013)] could supply Si for banana. This leads to the first hypothesis that livestock dung manure inputs can increase soil Si availability and banana Si uptake (Experiment 1). Between Tanzanian home gardens, the wide variation of banana SiO 2 contents were found despite the similarity in andic properties ( Figure S5). A positive correlation between livestock dung inputs and banana leaf SiO 2 contents (Figure 1(a)) suggests the importance of livestock dung inputs for supply Si to banana. Although the potential importance of animal dung inputs for increasing soil Si availability and plant Si uptake has been addressed (De Tombeur, Roux, and Cornelis 2021), this is the first report confirming the roles of livestock dung inputs in banana leaf Si uptake in home-garden soils. Note that livestock dung manure inputs [cow dung pH of 6.9 to 7.1 from Abubakar and Ismail (2012) and Honest and Saria (2020)] on acidic soil could also increase soil pH (Fujii, Funakawa, and Kosaki 2012), as also seen in the positive correlation between soil pH and manure inputs in our study (R = 0.96, P < 0.05, N = 5; Table S1). Thus, livestock dung effects and soil pH effects could not be separately evaluated in this experiment. Both livestock dung inputs and original soil fertility (higher pH) could increase the differences in banana leaf SiO 2 contents among five home gardens of Tanzania (Table S1).

Effects of soil types and agroforestry on soil Si availability and banana leaf Si
Because of the abundance of weatherable minerals in volcanic soils, Andisols are hypothesized to have the higher Si solubility than non-volcanic Ultisols in Indonesia (Experiment 2). Shortrange-order Al and Fe minerals (allophane, imogolite, ferrihydrite, and organo-mineral complexes) have the higher Si equilibrium concentration than crystalline clay minerals such as kaolinite or quartz that is dominant minerals in the BS soils (Kittrick 1969). This leads to the higher Si solubility in the Andisols at higher elevation of LJ, compared to the BS soils and the lower elevation of LJ with less andic materials (Figure 2(a)). Soil pH (water extraction) regulates water Si solubility, as seen in a positive correlation between soil pH and water extractable Si in Indonesia soil samples (Figure 2(b)), but both factors are not directly related to banana leaf SiO 2 contents (Figure 3(c)). This suggests that banana Si uptake is not simply dependent on soil Si-pH chemistry, rather, banana Si uptake could be regulated primarily by the abundance of shortrange order minerals (Figure 3(a)). This is consistent with the lower Si uptake in Oxisols compared to Andisols in Guadeloupe (Henriet et al. 2008), highlighting the advantages of Andisols in supplying soil Si to banana. We also tested whether agroforestry lead to soil acidification and increase Si solubility or not, but significant increases in banana leaf SiO 2 contents were found only in two Andisol sites of LJ (Table 1). The differences in soil pH were still minor between agroforestry and monoculture cultivation, because soil acidification could be mitigated by high buffering capacities of short-range order minerals in volcanic soils (Fujii, Kanetani, and Tetsuka 2020). In addition, acidification of the highly weathered soil could promote weathering of Si-containing minerals (De Tombeur et al. 2020), but it also decreases soil Si solubility in the BS soils (Figure 2(b)). These findings highlight the importance of soil types in regulating soil Si availability and banana Si uptake.

Implications for soil fertility management for banana production
Phosphate buffer extraction is widely used for soil Si availability (Kato 1998). Our data suggest that soil phosphate-buffer extractable Si could roughly be estimated by oxalate extractable Si and andic property (Figure 2(a); Figure S4). This is consistent with the results in paddy soils affected by volcanic materials (Makabe et al. 2009;Yanai, Taniguchi, and Nakao 2016), but iron oxides and clays are also reported to be key properties regulating Si availability for non-volcanic soils (Huang and Hseu 2021). Key soil properties regulating Si availability could vary between soil types and land-uses (Yanai, Taniguchi, and Nakao 2016; Vander Linden and Delvaux 2019), but we found that the abundance of short-range order minerals, rather than pH, is important factor regulating soil Si availability at least for banana-growing volcanic soils (Figure 2(a)). Banana leaf Si contents could also vary between soil weathering stages and are also affected by biological Si cycles (Vander Linden and Delvaux 2019). Product removal of banana leaf and fruits and intensive mineral weathering could reduce Si availability in tropical soils (Vander Linden and Delvaux 2019), even if the original Si availability is high due to abundance of weatherable minerals in the volcanic parent materials. This supports the need of Si-containing organic amendment to maintain Si availability in soils. The manure inputs have direct influence on amounts of soluble Si and indirect influence on banana leaf Si through effects on soil pH (Figure 2(a,b)). Although we have confirmed livestock dung effects in Tanzanian Andisols (Figure 1(a)), manure effects or requirements might be higher in non-volcanic soils with the lower Si solubility ( Figure S5). Agroforestry could promote soil organic matter accumulation (Table 1), but the effects of agroforestry on mineral weathering and soil Si supply to banana could vary with soil properties related to the abundance of short-range-order minerals in the volcanic soils (Figure 3(a)).

Conclusions
We found that soil Si solubility could increase with increasing pH, but soil Si availability for banana and banana leaf SiO 2 contents are strongly regulated by parent materials or manure application. The soil Si availability for banana and banana leaf Si contents could increase with oxalate-extractable Si and andic properties in Indonesia home gardens, and they could also increase with increasing manure inputs in Tanzania home gardens. The agroforestry with leguminous trees increases soil C concentrations and banana leaf Si contents, although the additional paired samples are required to assess the impacts of agroforestry on banana leaf SiO 2 contents.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
The work was supported by the JSPS [20H03120].