Silicon recycling in tropical agrosystems: study of the ability of biochar forSilicon transfer from Si-rich to Si-poor soils

http://lib.uliege.be https://matheo.uliege.be

Silicon recycling in tropical agrosystems: study of the ability of biochar for Silicon transfer from Si-rich to Si-poor soils

Auteur : Collard, Laurie

Promoteur(s) : Cornelis, Jean-Thomas; de Tombeur, Félix Faculté : Gembloux Agro-Bio Tech (GxABT)

Diplôme : Master en bioingénieur : sciences et technologies de l'environnement, à finalité spécialisée Année académique : 2018-2019

URI/URL : http://hdl.handle.net/2268.2/7734

Avertissement à l'attention des usagers :

Tous les documents placés en accès ouvert sur le site le site MatheO sont protégés par le droit d'auteur. Conformément aux principes énoncés par la "Budapest Open Access Initiative"(BOAI, 2002), l'utilisateur du site peut lire, télécharger, copier, transmettre, imprimer, chercher ou faire un lien vers le texte intégral de ces documents, les disséquer pour les indexer, s'en servir de données pour un logiciel, ou s'en servir à toute autre fin légale (ou prévue par la réglementation relative au droit d'auteur). Toute utilisation du document à des fins commerciales est strictement interdite.

Par ailleurs, l'utilisateur s'engage à respecter les droits moraux de l'auteur, principalement le droit à l'intégrité de l'oeuvre et le droit de paternité et ce dans toute utilisation que l'utilisateur entreprend. Ainsi, à titre d'exemple, lorsqu'il reproduira un document par extrait ou dans son intégralité, l'utilisateur citera de manière complète les sources telles que

mentionnées ci-dessus. Toute utilisation non explicitement autorisée ci-avant (telle que par exemple, la modification du document ou son résumé) nécessite l'autorisation préalable et expresse des auteurs ou de leurs ayants droit.

Silicon recycling in tropical

agrosystems: study of the ability of biochar for silicon transfer from Si-rich

to Si-poor soils

Laurie Collard

TRAVAIL DE FIN D’ETUDES PRESENTE EN VUE DE L’OBTENTION DU DIPLOME DE MASTER BIOINGENIEUR EN SCIENCES ET

TECHNOLOGIES DE L’ENVIRONNEMENT

Année académique 2018-2019

Co-promoteurs:

Professeur Jean-Thomas Cornélis

Assistant-doctorant Félix de Tombeur

© Toute reproduction du présent document, par quelque procédé que ce soit, ne peut

Silicon recycling in tropical

agrosystems: study of the ability of biochar for silicon transfer from Si-rich

to Si-poor soils

Laurie Collard

TRAVAIL DE FIN D’ETUDES PRESENTE EN VUE DE L’OBTENTION DU DIPLOME DE MASTER BIOINGENIEUR EN SCIENCES ET

TECHNOLOGIES DE L’ENVIRONNEMENT

Année académique 2018-2019

Co-promoteurs:

Professeur Jean-Thomas Cornélis

Assistant-doctorant Félix de Tombeur

“Pour ce qui est de l’avenir, il ne s’agit pas de le prévoir, mais de le rendre possible.”

– Antoine de Saint Exupéry, Citadelle, 1948

Remerciements

Au moment d’écrire les dernières lignes de ce TFE, je souhaiterais prendre le temps de remercier les personnes qui m’ont, de près ou de loin, apporté du soutien, des encourage- ments et des précieux conseils.

Tout d’abord, je souhaiterais remercier le Prof. Jean-Thomas Cornélis pour m’avoir offert l’opportunité de travailler sur un sujet si passionnant. Je souhaiterais également remercier Félix de Tombeur pour avoir été un encadrant à l’écoute, toujours disponible pour répondre à mes questions. Merci de m’avoir soutenue au quotidien. Grâce à ces deux co-promoteurs, j’ai pu m’épanouir dans ce travail laborieux mais au combien enrichissant.

Merci à toute l’équipe du GP pour la bonne ambiance de travail durant ces 6 mois à vos côtés. Un merci spécial à Emilie Marit, Pauline Biron, Raphaël Tarantino et Françoise Toussaint pour leurs conseils et leur encadrement durant les labos. Merci à Victor Burgeon de m’avoir partagé ses connaissances concernant le biochar.

J’aimerais également remercier le Prof. Hassan B. Nacro ainsi que toute son équipe pour l’accueil chaleureux et les conseils avisés reçus lors de notre séjour à Bobo-Dioulasso.

Votre expérience du terrain m’a grandement aidé lors de ce travail.

Je souhaiterais également profiter de l’occasion qui m’est donnée pour remercier ceux qui m’ont permis de vivre six années extraordinaires au sein de cette si belle faculté. Tout d’abord, merci à Maxence, mon petit Maxou, pour tous ces moments mémorables passés à tes côtés. Merci à Chewbie, Oriane, Marc, Mélanie, Fanny et les autres de m’avoir soutenu durant ces six mois de dur labeur mais également pour m’avoir accompagné dans mes plus grands périples bibistiques.

Enfin, merci à ma famille pour l’affection et le soutien que vous affichez pour tous les projets que j’entreprends, aussi fous soient-ils. Merci à toi papa, d’avoir toujours cru en moi et de m’avoir permis de faire ces belles études. Merci à toi Simone, pour toutes ces petites choses que tu fais et qui embellissent notre quotidien. Merci à vous, Doriane, Axelle et Lorah, pour être des soeurs toujours prêtes à écouter mes péripéties gembloutoises et à me soutenir dans leur réalisation.

Merci à vous tous.

Table of contents

Remerciements i

Table of contents iii

Table of tables iv

Table of figures v

Abstract vi

Résumé vii

Part I : Article 2

1. Introduction 2

2. Material and methods 3

2.1. Studied site . . . 3

2.2. Soil sampling . . . 3

2.3. Silicon fertilizers : biochars and wollastonite . . . 3

2.4. Pot experiment . . . 4

2.5. Laboratory analyses . . . 4

2.6. Data analyses . . . 7

3. Results 8 3.1. Soils physico-chemical characterization . . . 8

3.2. Crops residue and biochar chemical characterization . . . 12

3.3. Treatments Si release . . . 13

3.3.1. CaCl₂-Si kinetic release . . . 13

3.3.2. CaCl₂-Si release efficiency . . . 14

3.4. Pot experiment . . . 15

3.4.1. Rice plants chemical characterization . . . 15

3.4.2. Rice plants functional traits . . . 15

4. Discussion 17 4.1. Soils fertility characterization . . . 17

4.2. Fertilizer potential of biochars . . . 17

4.3. Effect of amendments on Si bioavailability in soil . . . 18

4.4. Effect of amendments on rice plants Si concentration . . . 19

4.5. Effect of amendments on rice plants functional traits . . . 21

5. Conclusion 24

Part II : Appendix 26

Appendix 1 : state of the art 26

1. Biochar . . . 26

2. Silicon . . . 28

Appendix 2 : additional information 32 1. Experimental plant . . . 32

2. Biochars analysis report . . . 33

3. Nutrients rice plants concentration . . . 37

4. Plants specific leaf area . . . 38

Bibliography 39

Table of tables

1 Treatments applied during the experiment . . . 4 2 Lixisol and Gleysol physico-chemical characterization. Parameter value and STD are

presented. Concerning p-value, *** = <0.001, ** = <0.01 and * = <0.05 . . . 9 3 Nutrients concentration in crops residue. Mean and STD are presented. Concerning

p-value, *** = <0.001, ** = <0.01 and * = <0.05 . . . 12 4 Biochars physico-chemical characteristics. . . . 13 5 CaCl2-Si release efficiency (RSi) with respect to treatments. Mean and STD are pre-

sented. Letters show treatments of equal means, according to Fisher’s test (p-value <

0.05). . . . 14 6 Plant Si uptake efficiency (USi) with respect to treatments. Mean and STD are pre-

sented. Letters show treatments of equal means, according to Fisher’s test (p-value <

0.05). . . . 20 7 Nutrients plants concentration with respect to treatments. Mean and STD are presented.

Letters show treatments of equal means according to Fisher’s test (p < 0.05). . . . 37

Table of figures

1 Leaf arc measurement. Modified from [Antônio Zanão Júnior et al., 2010] . . . 7 2 Lixisol X-ray diffraction (XRD) pattern. . . . 10 3 Gleysol X-ray diffraction (XRD) pattern. . . . 11 4 Kinetic release of CaCl2-Si (6h, 24h, 8 and 32 days) with respect to treatments. All

treatments are presented (A) as well as a zoom on the lowest (B). Means and STD are presented. Letters show treatments of equal means after 32 days, according to Fisher’s test (p-value < 0.05). . . . 14 5 Plant Si concentration (A) and Si mineralomass (B), with respect to treatments. Mean

and STD are presented. Letters show treatments of equal means according to Fisher’s test (p < 0.05). . . . 15 6 Rice plants functional traits with respect to treatments. Aboveground total biomass

(A), plant height (B), resistance to traction (C), penetration (D) and leaf arc (E) are displayed. Mean and STD are presented. Letters show treatments of equal means according to Fisher’s test (p < 0.05). . . . 16 7 Relation betweenCaCl₂-Si content and pH with with respect to treatments. All treat-

ments are presented (A) as well as a zoom without W5 and W30 treatments (B). . . . 19 8 Relation betweenCaCl2-Si content (after 32 days) and total Si concentration in plants,

with respect to treatments. All treatments are presented (A) as well as a zoom without W5 and W30 treatments (B). . . . 21 9 Relation between functional trait and Si plant concentration. Aboveground total biomass

(A), plant height (B), resistance to traction (C), penetration (D) and leaf arc (E) are displayed. Regression lines are presented when correlation is significant. . . . 23 10 Silicon in different soil phases [Tubaña and Heckman, 2015]. . . 28 11 (A) SLA with respect to treatments. Mean and STD are presented. (B) Relation

between SLA and Si plant concentration. . . . 38

Abstract

In order to reduce the loss of soil fertility, which particularly affects highly weathered tropical soils, one of the solutions could be the use of biochar in agriculture. It could also contribute to increase bioavailable Si content in soils. Silicon, once uptaked by plants, plays an important role against many biotic and abiotic stresses.

This work aims to study the relevance of using biochar as a slow-release silicon fertilizer in tropical agrosystems. More specifically, this research focuses on the establishment of a regional loop in Burkina Faso between highly weathered soils developed on the plateaus and relatively less weathered soils of the valley. Biochar would therefore be used to transfer silicon from Si-rich soils to Si-poor soils.

During a pot experiment, Si-rich rice-biochar is compared with Si-poor cotton-biochar and wollastonite. All treatments are applied in two different ratio, 5 and 30 tons per hectare, on a rainfed rice culture. Among others, the release of bioavailable Si in soil and the concentration of Si in plants were measured in order to compare the effect of the different amendments.

Through this study, it appears relevant to use rice-biochar in tropical agrosystems as a slow-release Si fertilizer since it is a more efficient fertilizer than cotton-biochar in terms of quantity released and wollastonite in terms of effect duration. Therefore, the creation of a regional loop between the less fertile soils of the plateaus and the more fertile soils of the valleys therefore makes sense if rice-biochar is used to transfer silicon between both types of soil.

Keywords : Burkina Faso, tropical soils fertility, biochar, wollastonite, silicon, rainfed rice culture.

Résumé

Afin de réduire la perte de fertilité des sols qui touchent particulièrement les sols trop- icaux hautement altérés, une des solutions pourrait être l’utilisation du biochar dans l’agriculture. Celui-ci pourrait également contribuer à l’augmentation de la teneur en Si biodisponible dans le sol. Cet élément, une fois assimilé par les plantes, joue un rôle important contre de nombreux stress biotiques et abiotiques.

Ce travail vise à étudier la pertinence de l’utilisation du biochar dans les agrosys- tèmes tropicaux comme fertilisant à relargage lent de silicium. Plus spécifiquement, cette recherche se concentre sur l’établissement d’une boucle régionale au Burkina Faso entre les sols fortement altérés, développés sur les plateaux, et les sols relativement moins altérés de la vallée. Le biochar serait donc utilisé pour transférer le silicium des sols plus riches en silicium vers les sols plus pauvres en silicium.

Au cours d’une expérience en pot, le biochar-riz, riche en silicium, est comparé au biochar-coton, pauvre en silicium, et à la wollastonite. Tous les traitements ont été appliqués en deux quantités différents, 5 et 30 tonnes par hectare, sur une culture de riz pluvial. Le relargage du silicium biodisponible dans le sol et la concentration en silicium dans les plantes ont notamment été mesurés afin de comparer l’effet des différents amendements.

A travers cette étude, il apparaît pertinent d’utiliser le biochar-riz dans les agrosystèmes tropicaux comme fertilisant à relargage lent de silicium. En effet, il s’avère plus efficace que le biochar-cotton en terme de quantité relarguée et plus efficace que la wollastonite en terme de durée d’action. Par conséquent, la création d’une boucle régionale entre les sols moins fertiles des plateaux et les sols plus fertiles des vallées fait sens si l’on utilise du biochar-riz comme outil de transfert du silicium entre les deux types de sols.

Mots-clés : Burkina Faso, fertilité des sols tropicaux, biochar, wollastonite, silicium, culture de riz pluvial.

Part I : Article

1. Introduction

The impact of human activity on the environment is no longer in doubt in light of recent IPCC reports on climate change and its consequences on terrestrial ecosystems [IPCC et al., 2018]. Among these, the loss of soil fertility and the limitation of carbon storage within the soil constitute issues that particularly affect highly weathered tropical soils such as those in Burkina Faso [Bationo et al., 2007]. Finding solutions to these two major problems, therefore, becomes a great challenge for the scientific community.

One of the solutions to the issues mentioned above could be the increase of biochar use in agriculture. Biochar is a stable organic material created by pyrolysis of fresh biomass [Verheijen et al., 2010]. As an aromatic form of black carbon, biochar has high stability towards decay [Atkinson et al., 2010]. Thus, its application to soils could be a means of mitigating climate change by sequestering organic carbon and improving soil physico- chemical properties [Lehmann et al., 2011]. Furthermore, during the pyrolysis process, renewable energy is produced, forming by this way a win-win-win scenario [Laird, 2008].

Recently, several papers show that biochar application could also improve bioavailable Si content in soils [Houben et al., 2013, Li and Delvaux, 2019, Liu et al., 2014]. This role played by biochar is important for highly weathered or desilificated soils. These latter, even if they have a high total silicon content, have a low content of bioavailable Si. Besides that, silicon plays beneficial roles for plants. For example, silicon reduces biotic stress such as water and salt stress [Ma and Yamaji, 2006]. For monocotyledonous plants, such as rice, this results in an increase in yield. Furthermore, by deposition in leaves, silicon precipitates in phytoliths and creates physical protection against herbivorous insects, fungi and diseases [Bakhat et al., 2018]. Finally, silicon improves functional plant traits such as the quantity, rigidity and erectness of the biomass produced [de Oliveira et al., 2016].

This work aims to study the relevance of using biochar as a slow-release silicon fertilizer in tropical agrosystems. More specifically, this research focuses on the establishment of a regional loop in Burkina Faso between highly weathered soils developed on the plateaus and relatively less weathered soils of the valleys. Managing and improving the fertility soils from plateaus could involve the use of Si-rich biochar, produced from rice husks in the valleys. This biochar is compared with Si-poor biochar (cotton straws) and wollastonite during rainfed rice cultivation, this latter being generally carried out on the soils of the plateaus.

2. Material and methods

2.1. Studied site

Si-depleted soil used for the experiment comes from the village of Koumbia, province of Tuy, in the southwest part of Burkina Faso. The coordinates of the field are 11°13’5,5”N 3°43’17,6”W. The weather is qualified as a Sudanian climate [Tirogo et al., 2016] including a rainy season between May and September. The mean annual rainfall varies between 1000 and 1200 mm [SP/CONEDD, 2010]. The mean monthly temperature in this region varies between 22°C and 35°C [SP/CONEDD, 2010].

In this location, the soil is defined as a FLIPP according to theBureau national des sols (BUNASOL) of Burkina Faso [BUNASOL, 1990]. According to the WRB classification, this soil is a Lixisol [FAO, 2015]. The studied soil is developed on a granitic parent material [Burgeon, 2017] and shows a ferruginous crust between 20 and 40 cm depth [SP/CONEDD, 2010].

2.2. Soil sampling

Sampling occurred in January 2019. On the same plot, four subplots were randomly defined to collect a total of 124 kilograms of soil, representing the required quantity for the pot experiment. Sampling was realized up to 30 cm depth, corresponding to those of the ferruginous crust. Concretions made of iron and manganese oxides were abundant.

Hence, after being air-dried, samples were sieved through a mesh screen of 10mm to reject biggest concretions and so to avoid bias during the pot experiment.

2.3. Silicon fertilizers : biochars and wollastonite

The two types of biochar studied were produced locally from rice husks and cotton straws in January 2019, at Nazi Boni University, Bobo-Dioulasso. The production was realized by pyrolysis at approximately 450°C, during 80 minutes. A conventional Top-Lit UpDraft (TLUD) oven was used [Burgeon, 2017].

Rice husks come from the Kou Valley (11°22’45,6”N 4°23’31,5”W), 100 km away from Koumbia, where the soil is considered as an HPGS [BUNASOL, 1990] or a Geysol ac- cording to the WRB classification [FAO, 2015]. The biochar produced is called hereafter rice-biochar.

Cotton straws come from the region of Koumbia and were collected less fertile soils from plateaus (FLIS, FLIPP and HPGS) [BUNASOL, 1990]. The biochar produced is called hereafter cotton-biochar.

Rice-biochar and cotton-biochar were compared to wollastonite, a conventional Si fer- tilizer (CaSiO₃). The W10 wollastonite was provided by Vanderbilt Chemicals, LLC (Norwalk, CT, USA).

2.4. Pot experiment

The experiment was conducted in a greenhouse at Gembloux Agro-Bio Tech, ULiège, Gembloux. Rainfed rice culture was achieved by using the FRK45N variety. This latter is commonly cultivated in Burkina Faso. Its growth cycle lasts 95 days. The seeding rate is 70 to 90 kilograms per hectare [FAO, 1997], which in the experiment represents one plant per pot. Dibbling seeding was realized, followed by a thinning after 10 days to select the toughest plant.

The experiment was carried out in pots whose dimensions are 10.5*10.5*22 cm, ar- ranged according to a randomized complete block design, described in Appendix 2. Seven treatments were evaluated and each treatment was applied with five replicates (Table 1). In order to reflect as much as possible field culture conditions, 150 kg/ha of NPK (14-23-14) and 100 kg/ha of urea (46% N) were applied after respectively 20 and 35 days of experiment. Pots were watered every day with 150 mL of tap water.

Table 1 – Treatments applied during the experiment

Treatment Amendment Replicates Application ratio NPK Urea (t/ha) (kg/ha) (kg/ha)

R5 Rice-biochar 5 5 150 50

R30 Rice-biochar 5 30 150 50

C5 Cotton-biochar 5 5 150 50

C30 Cotton-biochar 5 30 150 50

W5 Wollastonite 5 5 150 50

W30 Wollastonite 5 30 150 50

T Control 5 0 150 50

2.5. Laboratory analyses

2.5.1. Soil characterization

Each analysis described below was carried out both plateaus soil sampled for the ex- periment and on valley soil, where the rice-biochar comes from.

The soil texture was determined by gravimetric sedimentation following the NF-X 31- 107 standard. Prior to this analysis, organic matter was removed from soil samples by using hydrogen peroxide.

pH KCl was analyzed in a 1:5 ratio (v/v) with a solution of KCl 1M, by using a pHmeter according to the NF ISO 10390 standard. pH H20 was analyzed similarly but with deionized water in a 1:5 ratio (v/v).

Total Organic Carbon (TOC) content was determined by dry combustion after decar- bonation with HCl solution. The method used derived from ISO 10694 standard. Total Nitrogen content was also determined by dry combustion, according to the ISO 13878 standard. Finally, C/N ratio was computed.

The cation exchange capacity (CEC) was measured by percolating soil columns with 1M ammonium acetate. The excess ammonium was then rinsed with ethanol. The solution obtained was alkalized with sodium hydroxide (NaOH - 50%) and diluted in distilled water.

Finally, the solution was titrated with hydrochloric acid (HCl 0,1N) [Metson, 1956].

The pool of so-called "bioavailable Si" was determined by CaCl₂ extraction. 50 mL of CaCl₂ 0.01M were added to 5g of soil before 16h of shaking [Sauer et al., 2006]. This latter was realized at very slow speed to avoid quartz abrasion [McKeague and Cline, 1963]. Si content was then quantified using a Varian Vista-MPX Inductively Coupled Plasma- Optical Emission Spectrometer (ICP-OES)

Bioavailable nutrients (Mg_av, K_av, Ca_av and P_av) were quantified after extraction using ammonium acetate 0.5 N and EDTA 0.02M (pH 4.65). Mav, Kav and Caav were then measured by AAS and P_av by spectrophotometry [Lakanen and Ervio, 1971].

Soil total elements concentration (Mg_tot, K_tot, Na_tot, Ca_tot, P_tot, Si_tot, Fe_totand Al_tot) was determined after alkaline fusion. First, 0.5 g of soil was heated at 450°C for 24 hours. Ash content was then weighted. Finally, alkaline fusion was realized at 1000°C with 1.6 g of Li-metaborate and 0.4 g of Li-tetraborate [Chao and Sanzolone, 1992]. The quantification was performed by using ICP-OES. Total reserve in bases (TRB) corresponds to the molar sum of Mg_tot, Na_tot, K_tot and Ca_tot.

Crystalline soil minerals were identified by X-ray diffraction (XRD) on bulk soil grinded at250µmand on clay-size fraction (<2µm). The analysis was carried out by a Bruker D8- Advance Eco diffractometer generating CuKα radiation. Minerals were identified thanks to EVA 3.2 software and then quantified by Topas software.

2.5.2. Biochar characterization

Nutrients concentrations were determined in the plants residue used to produce biochar.

Before analyses, the biomass was dried in the proofer during 2 days at 65°C and pounded.

To process the mineralization, 30mL of a 50:50 vol. mixture of nitric acid (HN O₃ 65%) and perchlorid acid (HClO₄ 70%) were added to 2 g of plants residue. After heating and evaporation of the liquid phase, 5 mL of hydrochloric acid (HCl 10%) were added and the solution was filtered. Ca, Mg, and K were then quantified using AAS and P was determined by spectrophotometry. Furthermore, Sitot, Ptot and Fetot concentration was determined by using ICP-OES after alkaline fusion [Chao and Sanzolone, 1992].

Biochar characteristics were determined according to the European Biochar Certificate (EBC) by Eurofins Umwelt, an accredited test laboratory.

2.5.3. Si kinetic release

The Si kinetic release of each treatment studied was measured through aCaCl₂ extrac- tion. The soil-amendment ratio of each treatment was reproduced in triplicates. Then, 50mL of CaCl₂ 0.01M were added to 5 g of this soil-amendment ratio. The solution obtained was homogenised by handshaking twice per day [Sommer et al., 2013].

At each defined time step (6h, 24h, 8 days and 32 days), the solution was centrifuged at 4750g for 5 minutes and supernatant was filtered. pH was measured and then, the extract was acidified by addition of 50 µL of HN O₃ 65% prior to analysis. Finally, CaCl₂-Si, CaCl2-Al andCaCl2-Fe concentrations were analyzed by using ICP-OES.

In order to standardize the results obtained by the total silicon concentration of the different amendments used, efficiency of Si release (R_Si) was computed according to the following equation :

R_Si = (CaCl₂-Si treatment)−(CaCl₂-Si control)

Total Si input from treatment ×100 (1) whereCaCl₂-Si is the amount of Si released from each treatment in soil after 32 days.

Total Si input is the amount of Si initially contained in each treatment.

2.5.4. Rice plants analyses

Total major nutrients (Si, Mg, Ca, K, P, Fe and Al) contained in rice plants were determined by using ICP-OES after alkaline fusion [Chao and Sanzolone, 1992].

Some mechanical and physical properties were measured on rice plants. First, plant height was measured. Then, the highest leaf of each plant was sampled to measure the resistance to traction and to penetration [Onoda, 2011]. These two properties were determined thanks to a SMS texturometer. On the third-highest leaf plant, leaf arc was measured [Antônio Zanão Júnior et al., 2010] (Figure 1).

Figure 1 – Leaf arc measurement. Modified from [Antônio Zanão Júnior et al., 2010]

2.6. Data analyses

2.6.1. Statistical analyses

Prior to statistical analyses, normal distribution within each treatment was tested thanks to the Ryan-Joiner’s Normality Test. Similarly, homogeneity of variances within each treatment was tested via Bartlett’s Test. One-way variance analyses (AV1) were performed with Minitab®.

For each treatment response, mean and standard deviation (STD) were determined. In order to compare difference between treatments, Fisher’s Test was used to define groups of equal means. Finally, Pearson comparison was used as a correlation test. For each test mentioned above, p-value < 0.05 was defined as the significance threshold.

3. Results

3.1. Soils physico-chemical characterization

Results of the Lixisol and Gleysol physico-chemical characterization are shown in Table 2. Means difference signifiance is presented for the parameters of interest, further used in the discussion. Lixisol is characterized by a dominant sandy texture, with73.3±2.7% of sand, 14.5±1.6 % of silt and12.2±1.1% of clay. Its TOC content equals3.4±0.2g/kg, its CEC 2.9±0.1 meq/100g and its TRB 117.9±8.9 cmolc/kg. Lixisol can be qualified as a sandy loamy soil according to the USDA soil texture triangle [U.S.D.A., 2017].

Gleysol has 34.0±0.6 % of sand, 37.0±0.7 % of silt and 29.0±0.2 % of clay. Its TOC content equals 9.1±0.3g/kg, its CEC4.2±0.3meq/100g and its TRB144.8±10.7 cmolc/kg. All these values are significantly higher than those of Lixisol. Gleysol can be qualified as a clay loamy soil according to the USDA soil texture triangle [U.S.D.A., 2017].

Lixisol and Gleysol X-ray diffraction (XRD) patterns for bulk soil are respectively presented in Figure 2 and Figure 3. XRD results on clay-size fraction are not presented since they did not provide additional information about mineralogical characterization.

Results show that, for both studied soils, quartz is the major constituent (72-73%). The second constituent is kaolinite (23%). Minor constituents are K-feldspar (2%) and anatase (2%). In addition, Lixisol contains mica/illite (1%). Thus, no significant mineralogical difference can be highlighted between these two soils.

Table 2 – Lixisol and Gleysol physico-chemical characterization. Parameter value and STD are presented. Concerning p-value, *** = <0.001, ** = <0.01 and * = <0.05

Parameter Unit Lixisol Gleysol

P-value Value ±STD Value ±STD

Sand [%] 73.3±2.7 34.0±0.6 ***

Silt [%] 14.5±1.6 37.0±0.7 ***

Clay [%] 12.2±1.1 29.0±0.2 ***

pH KCl [-] 4.84±0.05 4.52±0.02

pH H20 [-] 6.14±0.49 6.04±0.03

Total Organic Carbon [g/kg] 3.5±0.2 9.1±0.3 ***

Total Nitrogen [%] 0.03±0.002 0.08±0.002

C/N [-] 11.51±0.15 11.86±0.05

CaCl2-Siav [mg/kg] 18.33±0.45 6.27±0.55 ***

Mg_av [mg/100g] 3.69±0.40 3.91±0.61

Kav [mg/100g] 3.91±0.61 2.51±0.01

Ca_av [mg/100g] 19.76±2.82 37.22±0.49

Pav [mg/100g] 0.13±0.05 0.07±0.01

CEC [meq/100g] 2.9±0.1 4.2±0.3 **

Mgtot [g/kg] < dl < dl

K_tot [g/kg] 22.68±2.33 31.23±1.51

Natot [g/kg] < dl < dl

Ca_tot [g/kg] 12.00±0.73 13.01±1.37

TRB [cmolc/kg] 117.9±8.9 144.8±10.7 *

P_tot [g/kg] < dl < dl

Sitot [g/kg] 342.51±6.57 144.80±10.67

Fe_tot [g/kg] 7.96±0.30 21.93±0.89

Al_tot [g/kg] 22.75±1.21 50.19±1.73

Figure 2 – Lixisol X-ray diffraction (XRD) pattern.

10

Figure 3 – Gleysol X-ray diffraction (XRD) pattern.

11

3.2. Crops residue and biochars chemical characteriza- tion

3.2.1. Crops residue

Nutrients concentration in crops residue is presented in Table 3. Si concentration is higher in rice husks than in cotton straws. This difference is highly significant (p < 0.001).

Rice husks also contain more potassium (K), aluminium (Al) and iron (Fe) than cotton straws.

Nevertheless, magnesium (Mg), calcium (Ca) and phosphorus (P) concentration are higher in cotton straws. These latter also contain more carbon (C) and nitrogen (N) but since no STD value is available, the significance of means difference is impossible to characterize.

Table 3 – Nutrients concentration in crops residue. Mean and STD are presented. Concerning p-value, *** = <0.001, ** = <0.01 and * = <0.05

Rice Husks Cotton straws P-Value

Si mean 69.39 1.60

[g/kg] STD 4.36 0.07 ***

Mg mean 0.33 0.81 [g/kg] STD 0.02 0.004 ***

Ca mean 2.62 7.48

[g/kg] STD 0.16 0.77 **

K mean 21.94 18.88

[g/kg] STD 0.93 1.07 *

P mean 0.32 0.83

[g/kg] STD 0.03 0.01 ***

Fe mean 0.63 0.13 [g/kg] STD 0.06 0.001 ***

Al mean 1.48

< dl [g/kg] STD 0.13

C mean 37.13 44.42

[%]

N mean 3.37 7.87 [g/kg]

3.2.2. Biochars

Main rice-biochar and cotton-biochar physico-chemical characteristics are presented in Table 4. Values come from analyses on fresh matter. Complete analysis can be found in Appendix 2.

Ash content, total organic carbon content and specific surface are obtained by micro wave pressure digestion. Ash content and specific surface are significantly higher in rice- biochar than in cotton-biochar. Conversely, total organic carbon content is higher in cotton-biochar. Oxides content (SiO₂, MgO, CaO, K₂O, Fe₂O₃) are obtained by borate digestion of ash 550°C. SiO₂ content is significantly higher in rice-biochar since this latter contains 7 times more SiO2 than cotton-biochar.

Table 4 – Biochars physico-chemical characteristics.

Parameter Unit Rice-biochar Cotton-biochar

Ash content (550°C) [% (w/w)] 52.8 7.3

Total Organic Carbon [% (w/w)] 40.9 70.9

Specific surface [m²/g] 8 1.7

SiO₂ content [% (w/w)] 41.1 0.9

P₂O₅ content [% (w/w)] 0.5 0.2

MgO content [% (w/w)] 0.3 0.4

CaO content [% (w/w)] 0.3 1.9

K₂O content [% (w/w)] 0.8 1.0

Fe₂O₃ content [% (w/w)] 1.9 0.2

3.3. Treatments Si release

3.3.1. CaCl₂-Si kinetic release

Results ofCaCl₂-Si extraction for each treatment and for the 32 first days are presented in Figure 4. Over time, Si content released by each treatment continually increases.

Furthermore, CaCl₂-Si release increases between treatments as follows: T < C5 < C30

< R5 < R30 < W5 < W30.

After 32 days,CaCl₂-Si content released range from11.2±0.6mg/kg in T to189.2±8.4

Figure 4 – Kinetic release of CaCl2-Si (6h, 24h, 8 and 32 days) with respect to treatments. All treatments are presented (A) as well as a zoom on the lowest (B). Means and STD are presented.

Letters show treatments of equal means after 32 days, according to Fisher’s test (p-value < 0.05).

3.3.2. CaCl₂-Si release efficiency

CaCl₂-Si release efficiency (RSi) results, presented in Table 5, are computed from equa- tion 1. Significant differences can be highlighted between the amendments studied. First, rice-biochar (R) is the lowest efficient amendment after 32 days of experiment. This efficiency difference with other amendments is significant.

Secondly, wollastonite is significantly more efficient than cotton-biochar when it is ap- plied by 5 tons/hectare. In contrast, this difference does not exist when the application ratio is 30 tons/hectare. More generally, Si release efficiency is significantly higher when treatments are applied by 5 t/ha than when they are applied by 30 t/ha.

Table 5 –CaCl2-Si release efficiency (RSi) with respect to treatments. Mean and STD are presented.

Letters show treatments of equal means, according to Fisher’s test (p-value < 0.05).

Treatment Value [%] STD Group

R5 1.5 0.09 d

R30 0.8 0.01 d

C5 10.4 2.0 b

C30 7.2 1.2 c

W5 30.5 2.2 a

W30 6.8 0.3 c

3.4. Pot experiment

3.4.1. Rice plants chemical characterization

The concentration of total silicon in plants range from19.0±6.7g/kg in T to 62.6±5.2 mg/kg in R30 (Figure 5A). Furthermore, this concentration increases as follows : T

< C5 < W5 < C30 < R5 < W30 < R30. Compared to T, plant Si concentration is significantly higher in response to rice-biochar (R) application, by 156% for 5t/ha and by 190% for 30t/ha application ratio (Figure 5). C30 and W30 treatments also show a significant increase in total Si concentration, by respectively 59% and 176%. Other nutrients concentration (Mg, Ca, K, P, Al and Fe) can be found in Appendix 2. Conversely to values obtained for Si, differences between the two soils are clearly less marked for the other nutrients.

Si mineralomass, quantified by multiplying plant Si concentration by rice plants biomass, follows the same trend as developed above, except for W30 (Figure 5B). Indeed, this treat- ment is not significantly different from the control treatment (T).

Figure 5 – Plant Si concentration (A) and Si mineralomass (B), with respect to treatments. Mean and STD are presented. Letters show treatments of equal means according to Fisher’s test (p <

0.05).

3.4.2. Rice plants functional traits

Concerning the aboveground total biomass produced, no difference can be highlighted between the control and the majority of treatments, except R30 and W30 (Figure 6A).

Indeed, these latter produced significantly less aboveground biomass.

Plant height (Figure 6B) is significantly higher for R5, C30 and W5. Conversely, as well as for biomass, the W30 treatment value is significantly lower than the others. Remaining treatments, R30 and C5, are not significantly different from T.

Measures regarding the leaf resistance to traction was realized (Figure 6C). They show that R5, R30 and C5 have a significantly higher resistance to traction than T. Since they

The resistance to penetration was also measured on the rice plants (Figure 6D). C5 is the only treatment where this resistance is significantly higher than T resistance. R30 and W30 treatments even have a significantly lower resistance to penetration than T.

Finally, leaf arc was measured on plants (Figure 6E). T, C5 and C30 treatments have the most important leaf arc, significantly higher than those of other treatments. Then, R5 and W5 form another group of equal means and the last group contains R30 and W30, which have the lowest leaf arc values. Thus, leaf arc decreases as follows : T > C30 > C5

> R5 > W5 > W30 > R30.

Figure 6 – Rice plants functional traits with respect to treatments. Aboveground total biomass (A),

4. Discussion

4.1. Soils fertility characterization

Gleysol was sampled in a valley, where a river is diverted to allow the accumulation of water for rice cultivation. As the soil is wetter and richer in alluvial deposits, it has a higher TOC content. Its clay content is also more important. Logically, its CEC value is therefore higher (Table 2). This proves that Gleysol is a more fertile soil than Lixisol.

To go further in the interpretation, weathering stage of both soils can be compared by using their respective TRB values. Gleysol has a higher weatherable minerals content than Lixisol (Table 2). Therefore, Lixisol can be qualified as less fertile but also more altered than Gleysol. This confirms the interest of using Lixisol during the experiment in order to have contrasting conditions of fertility between control and amendments treatments.

Although Gleysol is a more fertile soil, the content of bioavailable CaCl₂-Si is higher in Lixisol (Table 2). This result seems absurd but could be explained by the agitation duringCaCl₂-extraction. Even if it was slight, it may have caused abrasion of the quartz contained in the Lixisol and thus, artificially increased the bioavailableCaCl₂-Si content.

Nevertheless, these different values remain in the same order of magnitude, which is consistent with the almost identical DRX results obtained for both soils (Figures 2 and 3) [Cornelis and Delvaux, 2016].

Therefore, using rice cultivated on Gleysol to produce biochar remains consistent in lights of its greater fertility and above all, in order to study the relevance of the regional loop.

4.2. Fertilizer potential of biochars

CEC is used, as above, to characterize soil fertility. Furthermore, this paramater can also be used to characterize the potential of biochar to adsorb nutrients. Among other parameters, a high specific surface leads to a larger CEC [Lehmann, 2007]. In this case, the specific surface of rice-biochar (8 m²/g) is markedly higher than the one of cotton biochar (1.7 m²/g) (Table 4). Rice-biochar should therefore be able to adsorb more nutrients than cotton-biochar.

With regards to Si fertilizer potential, rice-biochar has a SiO₂ content (41.1%) much larger than the one of cotton-biochar (0.9%) (Table 4). This means that the silicon pool that can be made available to plants is higher in the rice-biochar. These results are in line with those obtained for crops residue used to produce both biochars (Table 3). Although climate and soil type can play a role [Hodson et al., 2005], this difference is mainly due to the intrinsic characteristics of the two studied species. Indeed, rice is known

Guntzer et al., 2012]. It is qualified as Si high-accumulator plant. For its part, cotton is a dicotyledonous plant that accumulates less silicon than rice [Hodson et al., 2005].

Thus, both characteristics developed above allows to say that rice-biochar has a better fertilization potential than cotton-biochar and more specifically concerning silicon fertil- ization.

4.3. Effect of amendments on Si bioavailability in soil

Kinetic release curves show that wollastonite releases a significantly higher amount of CaCl₂-Si in solution compared to rice-biochar and cotton-biochar (Figure 4). For comparison, after 32 days of experience, W30 release 7 times more CaCl₂-Si than R30, the only biochar treatment significantly different from the others.

This major difference between wollastonite and biochars can be explained by two main hypothesis. Firstly, Si is contained in the form of phytoliths in biochar. As a re- sult, the Si release is slowed down by an intermediate step which is the dissolution of these phytoliths. On the other hand, wollastonite is applied in powder form (100µm) [Vanderbilt Minerals, 2012] which makes is more reactive and more able to release easily its silicon content.

Secondly, phytoliths solubility is controlled by the solution pH [Fraysse et al., 2006, Fraysse et al., 2009]. Alkaline pH enhances phytoliths dissolution [Li and Delvaux, 2019].

However, after 32 days of experiment, biochar treatments pH remains acidic (from 5.90 to 6.71) (Figure 7). As a result, wollastonite, with higher pH values (7.19 for W5 and 7.42 for W30), is able to release more easily its Si content than biochars.

The study of RSi support results developed above, by stating that wollastonite is more efficient than biochars during the 32 days of experiment (Table 5). For comparison, RSi of W5 is 20 times higher than the one of R5 and almost 3 times higher than the one of C5.

However, in the longer term, this difference between wollastonite and biochar should be alleviated. Indeed, biochar is known to be a slow-release fertilizer [Schmidt et al., 2017], which allows it to release silicon in the solution for a long time. Conversely, wollastonite will arrive faster at its maximum release capacity.

Figure 7 – Relation between CaCl₂-Si content and pH with with respect to treatments. All treat- ments are presented (A) as well as a zoom without W5 and W30 treatments (B).

By focusing the discussion on the two types of biochar, it appears that rice-biochar releases more CaCl₂-Si in the solution than cotton-biochar. This difference can be eas- ily explained by the higher Si concentration in rice-biochar than in cotton-biochar (Ta- ble 4). Furthermore, specific surface of both biochars can be also used to develop a second explanation. Indeed, a larger specific surface improves phytoliths dissolution [Li and Delvaux, 2019]. Since rice-biochar has a bigger specific surface (Table 4), its phytoliths are more easily dissolved and thus, CaCl₂-Si release in solution is higher.

However, by comparing biochars RSi values, rice-biochar seems less efficient than cotton- biochar. Indeed, rice-biochar has a larger Si reserve than the one of cotton-biochar (Table 4). Hence, during the same time period (32 days), rice-biochar releases only a small por- tion of its overall Si content (1.5% for R5 and 0.8% for R30), compared to cotton-biochar (10.4% for C5 and 7.2% for C30). This confirms that rice-biochar is a better fertilizer than cotton-biochar, due to its ability to release its silicon content over a longer period of time.

4.4. Effect of amendments on rice plants Si concentration

When plant Si concentration and Si mineralomass results are interpreted together, it appears that both rice-biochar treatments have highest values, with R30 significantly higher than R5 treatment (Figure 5). The statement made above concerning the fertilizer potential between the two biochars studied are thus confirmed.

Besides that, it is important to emphasize the difference between both studied parame- ters for W30. Indeed, this treatment allows one of the highest plant Si concentration but has a Si mineralomass value almost as low as the one of T. This difference is explained by limited growth of the plants that received the W30 treatment (Figure 6B), which decreases Si mineralomass value.

To support the discussion, further interpretations of the results were realized. First, efficiency of plant Si uptake (U_Si) was computed according to the following equation :

U_Si = Plant Si concentration × plant biomass

Soil CaCl₂-Si content ×soil weight ×100 (2) where soil weight corresponds to the quantity of soil contained in each pot experiment.

Results obtained by using equation 2 are presented in Table 6. Since the equation numerator and denominator were computed on different time scales (91 days for plant Si concentration in pot and 32 days for soil CaCl2-Si content in centrifuge tube), results are only interpreted qualitatively. These latter show that wollastonite treatments are highly unefficient in terms of plant Si uptake. It means that from the important pool of CaCl2-Si available in soil (Figure 4), only a small part was indeed assimilated by the plant. One hypothesis put forward to explain the result is the loss of wollastonite in solid form by leaching due to excessive watering. Conversely, rice-biochar and cotton-biochar were significantly more efficient.

This USi comparison highlights the pertinence of using biochar as an Si amendment, compared to the use of a conventional fertilizer such as wollastonite. By combining these results with those of RSi, it can also be stated that biochar will fertilize the soil for a much larger number of years than wollastonite.

Table 6 – Plant Si uptake efficiency (USi) with respect to treatments. Mean and STD are presented.

Letters show treatments of equal means, according to Fisher’s test (p-value < 0.05).

Treatment Value [%] STD Group

T 85.5 17.3 c

R5 146.3 17.1 a

R30 105.0 6.1 b

C5 107.8 10.9 b

C30 106.3 16.9 b

W5 9.4 2.2 d

W30 6.1 0.6 d

Afterwards, relation between plant Si concentration and CaCl₂-Si content (after 32 days) was plotted (Figure 8). As above, results are only interpreted qualitatively due to the different time scales used. Figure 8A clearly shows the difference between wollas- tonite and biochars, already put forward in USi values comparison. When the W5 and W30 treatments are removed from the chart (Figure 8B), plant Si concentration becomes significantly correlated (Pearson p-value <0.05 and R = 0.9372) with CaCl₂-Si content.

This correlation means that Si released in soil solution by biochars was actually uptaked by the plant and less lost by leaching, compared to wollastonite.

Figure 8 – Relation between CaCl2-Si content (after 32 days) and total Si concentration in plants, with respect to treatments. All treatments are presented (A) as well as a zoom without W5 and

W30 treatments (B).

4.5. Effect of amendments on rice plants functional traits

In order to interpret results of functional traits measurement, they were put in relation with plant Si concentration (Figure 9). Aboveground total biomass is highly negatively correlated (Pearson p-value < 0.001 and R = -0.6991) with plant Si concentration (Figure 9A). For its part, plant height is not correlated with plant Si concentration (Figure 9B).

These results concerning aboveground biomass and plant height are opposed to those commonly found in literature [Jeffery et al., 2011, Li et al., 2018, Wang et al., 2015].

However, the biomass results can be explained by the average specific leaf area (SLA) value obtained for each treatment (Figure 11A, in Appendix 2), associated with the height results. In general, all plants have the same height and number of leaves. Their total surface area can be considered equivalent. The significant increases in SLA for the R5, R30 and W30 treatments therefore reflect a smaller biomass.

Another general hypothesis that can be put forward to explain the lack of height differ- ence between treatments is the size of the experiment pot (10.5*10.5*22cm). This latter could be a limiting factor for root growth and therefore for plant development. A more specific hypothesis to explain W30 very low values (Figure 6) could be the increase in calcium content, intrinsic to wollastonite amendment. Ca promotes the precipitation of phosphorus in the form of Ca₃(P O₄)₂, making it unavailable to the plant, which limits its development.

Leaf resistance to traction is correlated (Pearson p-value < 0.05 and R = 0.4262) with plant Si concentration (9C). An increase in silicon concentration generates an increase in phytoliths deposition in the leaves. Therefore, it creates a physical structure which plays a significant role in the leaf resistance to traction and more generally improves leaf overall mechanical strength [Ma and Yamaji, 2006, Tubaña and Heckman, 2015].

Leaf resistance to penetration is highly negatively correlated (Pearson p-value < 0.001 and R = -0.6024) with plant Si concentration (9D). This means that an increase in silicon concentration goes with a decrease in resistance to penetration. These results are con- sistent with those obtained by calculating the specific leaf area (SLA) of each treatment.

Indeed, the increase in Si concentration goes with a highly correlated SLA increase (Pear- son p-value < 0.001 and R = 0.4320) (Figure 11B). This means that for these treatments, shoots have a larger surface area and therefore a smaller thickness. Hence their lower re- sistance to penetration. These results do not corroborate those concerning the resistance to traction. Indeed, an increase in resistance to traction should be related to an increase in resistance to penetration [Edwards et al., 2000], which is no the case in this study.

Finally, leaf arc results are highly negatively correlated (Pearson p-value < 0.001 and R = -0.8076) with plant Si concentration (9E). This relation clearly show that a Si supply reduces the curvature of leaves, and therefore increases their erectness. This increase allows the plant to intercept the photosynthetic radiation more efficiently, in favor of a potentially better biomass production [Antônio Zanão Júnior et al., 2010].

Figure 9 – Relation between functional trait and Si plant concentration. Aboveground total biomass (A), plant height (B), resistance to traction (C), penetration (D) and leaf arc (E) are displayed.

Regression lines are presented when correlation is significant.

5. Conclusion

Through this study, it appears relevant to use rice-biochar as a slow-release Si fertilizer in tropical agrosystems, compared to the other studied amendments. Firstly, although in the short term wollastonite releases a higher amount of silicon in the soil, rice-biochar can release its Si content over a much longer period of time than wollastonite. Secondly, compared to cotton-biochar, rice-biochar contains significantly more silicon that can be made available for plants. Rice-biochar is therefore a more efficient fertilizer than cotton- biochar in terms of quantity and wollastonite in terms of duration.

The increase in Si concentration in rice plants due to the application of rice-biochar proves that the latter can be used to increase the fertility of highly weathered soils.

Since rice-biochar is produced from rice husks in the valley, the creation of a regional loop between the less fertile soils of the plateaus and the more fertile soils of the valleys therefore makes sense. Through this loop, the fertility of weathered soils would be improved while adding value to residues from local crop.

6. Perspectives

To improve the interpretations realized during this study, aCaCl₂-extraction of bioavail- able Si should be performed on the soil of each treatment after the pot experiment. In this way, all components of the USi equation would be measured on the same time scale. USi values obtained could therefore be interpreted quantitatively and not only qualitatively.

To more clearly identify the role played by the soil and by the crop residues nature in the biochar quality, two experiments could be carried out in parallel. First, a comparison between rice-biochar produced by using crop residues from more fertile soil and by using crop residues from less fertile soil. Secondly, a comparison between rice-biochar and cotton-biochar from the same soil.

In order to obtain results even more in line with field reality, it would be interesting to carry out the same type of experiment on a rainfed rice cultivation run on a highly weathered soil. The regional loop could thus be tested under real conditions and not on an experimental basis. Still concerning the establishment of the regional loop, its technical, economical and social feasibility should be studied to go further in the project. Indeed, this loop could be the source of new social and economical links between farmers in the region concerned.

Part II : Appendix

Appendix 1 : state of the art

1.Biochar

Biochar is a stable organic material created by thermal degradation (pyrolysis) of biomass [Verheijen et al., 2010], between 300°C and 1000°C and in absence or limited supply of oxygen. Biochar application to soils, largely studied in recent years, is currently being considered as a means of mitigating climate change by sequestering C, while con- currently improving soil physico-chemical properties [Lehmann et al., 2011]. These two major roles can be associated with a third one, the production of renewable energy during biomass pyrolysis, forming by this way a win-win-win scenario [Laird, 2008].

The use of biochar in agriculture has its origin in the discovery of Anthropogenic Dark Earths (ADE) or terra preta de Indios, in South America, dating back to pre-Colombian period [Erickson, 2003]. Due to high weathering process and the anthropogenic over- exploitation, Amazonian ecosystems are normally characterized by a low fertility level [Glaser and Birk, 2012]. Nevertheless, terra preta is characterized by large stocks of or- ganic matter and high nutrient levels [Glaser et al., 2001]. Its formation has clear anthro- pogenic origin [Neves et al., 2003] via accumulation of household waste, including biochar [Glaser and Birk, 2012]. Among other residues, this latter was retained as one of the easiest option to obtain similar soils [Burgeon, 2017].

In the following sections, biochar physico-chemical properties are presented. These latter mostly depend on the feedstock nature but also on the pyrolysis temperature and its duration [Lehmann and Joseph, 2009]. Furthermore, biochar effects will be in- fluenced by the soil type on which it is applied and by the environmental conditions [Hardie et al., 2014].

1.1. Physical properties

Biochar structure is characterized by a large pore size distribution, from nano- (< 0.9 nm) to macro-pores (> 50 nm) [Downie et al., 2009]. These latter are a key in the biochar very porous structure, with a high surface area, which is considered to improve a range of soil physical properties, including soil porosity, density and water-holding capacity [Hardie et al., 2014].

Soil amendment with biochar logically increases the soil total porosity [Xiao et al., 2016].

This increase is generally proportional to biochar application rates [Omondi et al., 2016].

Furthermore, a decrease in soil density had been observed in line with an increase in soil porosity [Basso et al., 2013]. On the one hand, biochar amendment appears to improve loamy soils drainage and aeration by decreasing its total bulk density [Downie et al., 2009].

On the other hand, by increasing its surface area, biochar increases sandy soils water-

1.2. Chemical properties

While biochar macro-pores play a key role in its physical properties, smaller pores are involved in biochar chemical properties [Downie et al., 2009]. Indeed, these micro- pores (< 2 nm) confer to biochar high nutrients adsorptive capacities [Joseph et al., 2010].

Thanks to this latter, biochar thus reduces nutrient leaching in soils [Steiner et al., 2007].

The addition of biochar in soils rises the soil organic matter (SOM) content. This results in an increase of soil Cation Exchange Capacity (CEC) [Liang et al., 2006]. In addition, biochar chemical properties, such as CEC, are expected to evolve over time due to biotic or abiotic processes on the biochar surface [Cheng et al., 2006].

Biochar application also induces an increase in soil pH, more or less important de- pending on the initial pH soil value [Lehmann and Joseph, 2009]. As a consequence of this pH change, nutrients cycle can be impacted as well as their bioavailability in soils [Steiner et al., 2007].

1.3. Biological properties

The porous structure of biochar also has a beneficial role for soil biota. Indeed, biochar macro-pores prove to be a suitable habitat for microorganisms, including bacteria and fungi [Lehmann et al., 2011].

Furthermore, by raising the soil pH, biochar application creates alkaline conditions, op- portune for the microbial community development [Lehmann et al., 2011]. Nevertheless, this pH rise may, in specific cases, encourage the dominance of some specialized species, at the expense of other species. Hence, soil biodiversity could be reduced [Burgeon, 2017].

1.4. Carbon sequestration

In literature, biochar is considered as a means of mitigating climate change by se- questering organic carbon in soils [Lehmann et al., 2011]. Indeed, by being produced via pyrolysis, biochar is an aromatic form of black carbon, which has a high stability towards decay [Atkinson et al., 2010]. This high stability allows biochar to play a significant role of long-term organic carbon storage. The residence time of this latter in soils depends on the half-life of biochar, controlled by several factors, including feedstock nature and environmental conditions [Lehmann et al., 2006].

Despite its recalcitrance to decay, over time, biochar will be mineralized and will release CO₂ in atmosphere. However, the biochar residence time in soils is much longer than the one of others organic carbon forms such as crops residue [Baldock and Smernik, 2002].

1.5. Bio-energy production

Besides biochar production during the pyrolysis process, several gaseous components are also released in addition to heat [Czernik and Bridgwater, 2004]. With a properly

electricity. Furthermore, bio-oil can also be used as a diesel substitute after an appropriate treatment. [Woolf et al., 2010]

Depending on the biomass residence time, the temperature and the level of oxygen in the oven, respective proportions of bio-oil and biochar produced vary [Burgeon, 2017].

Indeed, during a fast pyrolysis where biomass stays a few seconds in the oven, more bio-oil is generated than biochar. The reverse occurs when the pyrolysis process lasts for hours or days [Woolf et al., 2010].

2. Silicon

2.1. Silicon in soils

Silicon is considered as the second most abundant element in the Earth’s crust, after oxygen [Epstein, 1994]. In soils, silicon is generally grouped into three different fractions:

the solid phase, the liquid phase and the adsorbed phase [Sauer et al., 2006]. A description of these different fractions is given in Figure 10 and is detailed below.

Figure 10 – Silicon in different soil phases [Tubaña and Heckman, 2015]

2.1.1. Solid phase

In solid phase, silicon forms are divided into three classes : crystalline forms, poorly or microcrystalline forms and amorphous forms. Crystalline forms include primary silicates, secondary silicates and silica materials. Primary silicates, derived from parent material, are found in sand and silt fractions. Secondary silicates, such as phylosilicates and Al-Fe oxides/ hydroxides [Cornelis et al., 2009], are mainly found in clay particles and originated from pedogenic processes. Silica materials consist mostly of quartz and disordered silica.

Poorly crystalline forms of silicon correspond to short-range ordered silicates, e.g. al- lophane and imogolite. Microcrystalline forms are represented by chalcedony and sec- ondary quartz [Allen and Hajek, 1989]. These minerals ensue mainly from soil formation

Amorphous forms are divided in two categories : litho/pedogenic forms and biogenic forms. This latter mainly include phytoliths issued from plant residues and remains of microorganisms [Sauer et al., 2006]. Litho/pedogenic forms consist of silicon complexes with Al, Fe, heavy metals and soil organic matter (SOC) [Farmer et al., 2005]. Further- more, pedogenic forms also consist of non-crystalline inorganic fraction represented by, inter alia, opal A, silica glass and opal coatings on secondary minerals [Drees et al., 1989].

2.1.2. Liquid phase

In liquid phase, silicon occurs as monosilicic, oligosilicic or polysilicic acid. Monosilicic acid (H₄SiO₄) is the only available silicon form for plants [Epstein, 1994]. The primary source of H₄SiO₄ in soil solution is soil silicate minerals. Thence, Si concentration in soil solution is significantly affected by the solubility of the different Si forms in solid phase [Tubaña and Heckman, 2015]. Easily weatherable components such as amorphous silica have larger contributions than those from quartz for instance. Indeed, the latter is more stable and thermodynamically more resistant to weathering [Drees et al., 1989].

Transformation of amorphous silica SiO₂ into silicic acid H₄SiO₄ happens according to the following equation :

(SiO2)n+ 2nH2O =nSi(OH)4 (3) This release of H₄SiO₄ is dependent on several physico-chemical properties, such as pH, temperature, particle size, redox potential, organic matter and water content [Savant et al., 1997]. Once available in the soil solution, Si can be combined with other elements to form clay minerals, be released into natural waters (rivers and seas) or be used for uptake by plants and microorganisms.

2.1.2. Adsorbed phase

A part of dissolved silicon can be adsorbed by secondary clay minerals and more im- portantly by Fe and Al oxyhydroxides [Beckwith and Reeve, 1964], which have strong adsorption capacities. This adsorption of H4SiO4 by oxyhydroxides, which makes Si non- available for plants, is influenced by pH, soil redox potential and the type of metal involved [Liang et al., 2015].

2.2. Silicon in plants

The essentiality of Si in the plant growth cycle has not been demonstrated to date [Epstein, 1994]. Nevertheless, Si is beneficial for numerous higher plants, by enhanc- ing strength and rigidity and by improving defense against abiotic and biotic stresses [Ma, 2004]. It also promotes photosynthesis by keeping leaves erect. Its beneficial effects on plant growth are thus based on several physical and chemical mechanisms, including the formation of a protective outer layer composed of amorphous silica (i.e. phytoliths), or its reactivity with heavy metals ions and other compounds [Tubaña and Heckman, 2015].

2.2.1. Silicon uptake and transport

Silicic acid transport from soil solution into plant roots is provided by three possi- ble mechanisms : active Si uptake - higher than water uptake -, passive Si uptake -

in soil solution could decrease, remain the same or increase [Liang et al., 2015]. These three mechanisms also allowed to categorize plant species into high-, intermediate- or non- accumulators [Takahashi et al., 1990] if, respectively, active, passive or rejective mecha- nism is used.

From the soil to the shoots, Si transport is carried out by 2 types of transporters. Lsi1, an influx transporter, allows the entry of silicic acid into the roots symplasm. Then, Lsi2, an efflux transport, transfers H₄SiO₄ in the apoplasm [Ma et al., 2007].

Once the xylem loaded, silicic acid is translocated upwards to the shoots via transpira- tion mechanisms. In some species, including rice, xylem unloading in shoots area is also an active process achieved by another influx transporter, named Lsi6 [Ma et al., 2011].

In shoots,H4SiO4 concentration increases due to water loss via transpiration. When this concentration overshoots the silicon solubility limit (2 mM), silicic acid polymerizes into amorphous silica (SiO₂)_nxH₂O , also known as opal or phytolith [Sommer et al., 2006].

2.2.2. Phytoliths

Phytoliths contain about 90% of silica and can, during their formation, occlude from 1 to 6% of organic carbon, water and trace amounts of other components such as aluminium and iron [Parr and Sullivan, 2011]. This occluded C can persist in the soil for millennia due to its strong resistance to decomposition [Li et al., 2013] and phytoliths thus play a major role in the carbon cycle.

Phytoliths concentration varies with species, cultivars and plant age. Indeed, species like rice, wheat, maize, and sugarcane, bundled in monocotyledons category, accumulate much more phytoliths than non-monocotyledon species [Hodson et al., 2005]. In addition, accumulation of phytoliths is higher in older tissues because Si is no longer mobile once polymerized within plants [Ma and Yamaji, 2006].

If the plants are not harvested, phytoliths return to soil either through plant litter fall or root decomposition. Once weathering processes achieved, they contribute to the biogeochemical cycle of Si by resupplying the Si soil content [Cornelis and Delvaux, 2016, Guntzer et al., 2012]. They can even accumulate in soil if their return flux is higher than erosion and dissolution fluxes [Cornelis et al., 2011].In crops, a major proportion of phytoliths produced in plants is exported. Only some phytoliths present in plants straw or roots remain on the site and return to the soil.

2.2.3. Abiotic stress alleviation

Silicon contained in plants plays a valuable role against several abiotic stresses, includ- ing chemical stresses (salt, metal toxicity, nutrient imbalance) and physical stresses (lodg- ing, drought, radiation, high temperature, freezing, UV) [Ma and Yamaji, 2006]. Most of these beneficial effects are attributed to silica deposition in cell walls [Coskun et al., 2019].

Against water stress, this deposit reduces water loss by restricting the cuticular transpi- ration [Matoh et al., 1986]. Furthermore, transpiration through stomata is also limited thanks to SiO2. Indeed, the latter regulates potassium concentration, nutrient involved

In order to reduce salt stress, silicon is able to reduce apoplastic bypass flow. This flow reduction decreases uptake and translocation of salt from the roots to the shoots [Ma and Yamaji, 2006]. To alleviate metal toxicity, reduction of apoplastic bypass flow is also one of the two mechanisms used, qualified as the internal mechanism. The external mechanism consists of soil pH change induced by Si to reduce the bioavailability of metals [Liang et al., 2015].

2.2.4. Biotic stress alleviation

Biotic stresses induced by pests and diseases can also be alleviated by silicon. Indeed, silica deposits in leaves, in the form of phytoliths, constitute a physical barrier to pests.

By increasing the leaves roughness and the number of spines and hairs, Si mitigates dam- ages created by herbivores [Bakhat et al., 2018]. This mitigation could be explained by two main mechanisms. First, the enhanced abrasiveness could increase the wear and tear of insects mandibles. Secondly, silica deposits in cell walls could decrease the digestibil- ity of plant leaves by preventing access to nitrogen and carbohydrates during digestion [Massey et al., 2006].

Beside the mechanical barrier, also useful against fungal pathogen induced diseases, sili- con can also protect the plant by triggering chemical processes able to boost plant defenses, such as accumulation of lignin, flavonoids and phenolic compounds [Debona et al., 2017].

Moreover, silicon, applied on plants as a preventive measure, reduces the severity of fungal diseases by improving the activity of defense related enzymes [Bakhat et al., 2018].

2.3. Potential Si fertilizers

Considering all the beneficial roles played by silicon for the plant, it is clear that the silicon fertilization can not be neglected. Several fertilizers are used to play this role including wollastonite.

Wollastonite is an anisotropic calcium silicate (CaSiO₃) from the pyroxene mineral group. Its key characteristics are a low water and oil absorptivity, a high brightness, a thermal and chemical stability and a biological inertia [Kodal et al., 2015]. Thanks to these qualities, wollastonite has multiple applications in various industrial areas, from painting and plastic production to cluchs and brakes design.

Its formation stems from the reaction of calcite and silica at high temperature in the Earth’s crust. Most deposits form at depths between 2 and 15 km below the surface [Hawley, 2010]. Biggest producer countries in the world are China, Mexico, India and United States. According to estimates, world reserves of wollastonite exceed 1 megatonne [Haque et al., 2019] and the world annual production in 2008 was about 600 kilotonnes [Hawley, 2010].

In agriculture, it is commonly used as a silicate amendment or a lime alternative [Liang et al., 2015].

Appendix 2 : additional information

1. Experimental plant