CharaCterization and nutrient release from siliCate roCks and influenCe on ChemiCal Changes in soil (1)

summarY the expansion of Brazilian agriculture has led to a heavy dependence on imported fertilizers to ensure the supply of the growing food demand. this fact has contributed to a growing interest in alternative nutrient sources, such as ground silicate rocks. it is necessary, however, to know the potential of nutrient release and changes these materials can cause in soils. the purpose of this study was to characterize six silicate rocks and evaluate their effects on the chemical properties of treated soil, assessed by chemical extractants after greenhouse incubation. the experimental design consisted of completely randomized plots, in a 3 x 6 factorial scheme, with four replications. the factors were potassium levels (0-control: without silicate rock application; 200; 400; 600 kg ha -1 of k 2 o), supplied as six silicate rock types (breccia, biotite schist, ultramafic rock, phlogopite schist and two types of mining waste). the chemical, physical and mineralogical properties of the alternative rock fertilizers were characterized. treatments were applied to a dystrophic red-Yellow oxisol (ferralsol), which was incubated for 100 days, at 70 % (w/w) moisture in 3.7 kg/pots. the soil was evaluated for ph; calcium and magnesium were extracted with kCl 1 mol l -1 ; potassium, phosphorus and sodium by mehlich 1; nickel, copper and zinc with dtPa; and the saturation of the cation exchange capacity was calculated for aluminum, calcium, magnesium, potassium, and sodium, and overall base saturation. the alternative fertilizers affected soil chemical properties. ultramafic rock and soil ph, while the mining byproduct (mB) led to high k levels. zinc availability was highest in the treatments with mining byproduct and Cu in soil fertilized with Chapada and mining byproduct.

soil ph, while the mining byproduct (mB) led to high k levels. zinc availability was highest in the treatments with mining byproduct and Cu in soil fertilized with Chapada and mining byproduct. introduCtion the Brazilian production of potash salts covers only 8 % of the country's requirements (oliveira, 2008). the remainder has been imported from Canada, russian federation, Belarus and Germany. the increasing use of nutrients is a result of the rapid growth of high-technology and effective agriculture, especially in savanna regions (anda, 2011). the world-wide demand has caused considerable price rises of conventional fertilizer sources, affecting the agricultural development, especially in tropical countries (manning, 2010).
to face this constraint, nutrient-rich rocks can be ground and used to reduce the dependence on imported sources. the technique consists of applying rock powder directly onto the soils; the competitiveness of this practice is directly related to the nutrient contents of the material and to the distance from the mining areas. these rocks might vary in quantity and sources of mineral elements they contain, but contribute to raise soil fertility, in the mid-to-long term, depending on the solubility and reaction with soil (hilsinger et al., 1996; van straaten, 2007). rocks and their byproducts are, in general, sources of phosphorus (P), potassium, calcium (Ca) and magnesium (mg), often including trace-elements, such as zinc (zn) and copper (Cu), which are essential in plant nutrition (leonardos et al., 1991Wilpert & lukes, 2003;leonardos & theodoro, 1999;theodoro & leonardos, 2006;fyfe et al., 2006 the breccia, biotite schist, ultramafic rock, mining byproducts, and phlogopite schist rock powder and mining wastes were selected for their levels of k, micronutrients and na. When applied to soils, they become a major k source, even though feldspar-rich rocks have shown to be more efficient in raising k availability (resende et al., 2006; manning, 2010). these alternative fertilizers may however bring other undesirable elements into the soil, e.g., sodium (na) and, if applied at high quantities, alter the soil ph. after their chemical reaction with soils, changes in ph occur, contributing to mineral imbalance and availability of toxic elements (moreira et al., 2006). other studies have been conducted with basalt powder (microcrystalline basalt and olivinebasalt), carbonatite, pyroclastic rock, biotite schist, and alkaline ultramafic rock, carbonatite, adularia, showing increases in soil k levels (escosteguy & klant 1998;Bakken et al., 2000;andrade, 2002;oliveira et al., 2006;nogueira et al., 2012).
the origin and composition of rocks are factors that influence k solubility and, as a consequence, k availability to plants. these factors are, in turn, related to previous thermal-or chemical rock treatments, soil characteristics, incubation time, plant species being grown and synergetic action of microorganisms, such as bacteria (harley & Gilkes, 2000;Wang et al., 2000;stanford et al., 2008).
Given the scarcity of information and procedures to compare rocks and byproducts with respect to solubility, soil property interferences, nutrient availability and plant uptake, further research is needed to assess their potential as fertilizers. the purpose of this study was to characterize six silicate rocks and evaluate their effects on the chemical properties of treated soil, assessed by chemical extractants after greenhouse incubation.

material and methods
soil of a sandy-clay loam, dystrophic red-Yellow oxisol (ferralsol) was sampled in itutinga, mG, Brazil, under native savannah vegetation, in the 0-20 cm layer. the following physic-chemical characteristics were determined by soil analysis: ph (water) = 5.4; om = 0.8 dag kg; k + = 22 mg dm -3 ; s = 5.4 mg dm -3 ; P = 0.9 mg dm -3 ; Ca 2+ = 0.1 cmol c dm -3 ; mg 2+ = 0.1 cmol c dm -3 ; al 3+ = 0.1 cmol c dm -3 ; (h + al) = 1.7 cmol c dm -3 ; sB = 0.3 cmol c dm -3 ; CeC = 0.4 cmol c dm -3 ; CeC at ph 7.0 = 2.0 cmol c dm -3 ; fe = 27.4 mg dm -3 ; zn = 0.5 mg dm -3 ; Cu = 0.7 mg dm -3 ; B = 0.0 mg dm -3 ; mn = 0.4 mg dm -3 ; sand = 600 g kg -1 ; silt = 170 g kg -1 ; clay = 230 g kg -1 . the extractors were: kCl 1 mol l -1 for Ca 2+ , mg and al; mehlich -1 for k, P and na; dtPa for ni, Cu and zn. the treatments were applied to the dried and homogenized soil, which was filled into 3.7 l pots and placed in a glasshouse of the soil science department, federal university of lavras (ufla). the experimental design consisted of completely randomized plots, in a 3 x 6 factorial scheme, with four replications. the factors were k levels (0-control: without silicate rock application; 200; 400; 600 kg ha -1 of k 2 o), supplied as six alternative fertilizer rocks (breccia, biotite schist, ultramafic rock, Chapada mining byproduct and phlogopite schist). the amount of ground rock was defined on the basis of respective potassium oxide (k 2 o) concentration (table 1). each experimental unit, a pot containing rocktreated soil, was randomly placed on a bench in a glasshouse and incubated for 100 days. moisture in the pots was maintained at 70 % (w/w) at a mean temperature of 25 o C. the ground mining wastes were applied in their original sieving as provided by the suppliers (table 2). this is possibly the form to be used by farmers, with no additional preparation cost, as has been defined for lime.
Photomicrographs of rocks were taken with a camera attached to a petrographic microscope (figure 1).  the ground rock treatments, for which the respective k contents were calculated, carry traceelements in variable quantities (table 3). the soil and rocks were analyzed by X-ray diffractometry with a vertical goniometer and q geometry for mineralogical fraction determination. the equipment was adjusted at an angular variation of 2-100° (2θ), radiation of Cukα (l= 1.5450) and speed of 1°2θ min -1 (Bigham et al., 2001;azzone & ruberti, 2010).
after the incubation period, soil samples were collected for analysis of ph (h 2 o), Ca, mg, k, P, na, ni, Cu, and zn, following the previous procedure for soil analysis, to assess soil reactions to the treatments. the al saturation rate (m), cation saturation (Ca, mg, k and na) of potential cation exchange capacity (CeC at ph 7.0) and soil base saturation rate were calculated. the data were subjected to analysis of variance and the means compared by the tukey test (p = 0.05), using the statistical package sisVar 5.3 ® (ferreira, 2008). the rate effect and interactions were compared by regression analysis, using mathematical models for best fit equation.

results and disCussion
the k rates, ground rock types, and the respective interactions influenced (p = 0.05) the nutrient availability and some soil chemical properties (figure 2). in general, the ph values increased proportionally to the rate of ground rock and were related to the type of nutrient sources. ultramafic rock and mining byproducts resulted in considerable soil ph rises, as a result of the amount and neutralizing power of rock. moreover, the CeC of the red-Yellow oxisol (ferralsol) was low, and was little buffered by acidity-neutralizing minerals (souza et al., 2011).
other glasshouse studies with ultramafic rock and phlogopite schist, from the states of Goiás, santa Catarina and Bahia, Brazil, all with particles > 0.3 mm, also reported changes in ph (ribeiro et al., 2010), similar to the oxisol studied. moreover, in a typical savannah, clayey dystrophic red oxisol, the ph and potential acidity were altered by breccia, ultramafic rock and biotite schist rock powder.
the variations in ph can also be explained by the different values of effective calcium carbonate (eCC) (table 4). in the case of biotite schist, breccia and phlogopite schist, the eCC values are very low, with less effect on ph changes. it should be emphasized that none of these mining wastes is specifically indicated for use as soil amendment, to regulate soil acidity and increase Ca and mg supply (Brazilian regulations require a minimum CCe = 67 %; minimum Cao + mgo = 38 %; total eCC = 45 %).
Calcium availability and saturation increased with the rate of rock powder and were related to the type of nutrient sources (figure 2b,c). Prior to the addition of rocks and mining wastes, Ca 2+ availability was very low (0.1 cmol c dm -3 ) in native soil, whereas after incubation for 100 days, Ca 2+ increased in all treatments. Calcium soil content was in the range of "very low availability" (0.1 cmol c dm -3 ; CfsemG, 1999) before the application of ground rocks, and after incubation of 100 days, only the soil treated with ultramafic rock reached the range of "low availability". a CeC of 40-60 % saturated by Ca 2+ is most suitable for oxisols (lopes, 1998). even though none of the rock treatments reached this level, the data showed that the ground rocks and mining wastes under study, at high rates, released Ca 2+ in addition to k + , creating conditions to ameliorate soil acidity.
the Ca 2+ increase is related to clay-mineral 2:1 (Chapada mining byproduct), pyroxene and feldspar, present in ultramafic rock and breccia powder (figure 3). the differences between pyroxene and feldspar, in terms of nutrient release, have been compared and reviewed, confirming the dynamics observed in this experiment (hilsinger et al., 1996; manning, 2010). Pyroxenes are altered in their constitution when placed in soils in powder form, releasing Ca 2+ , mg 2+ and fe 2+ , while feldspars can be potassic (kalsio 3 ) or calco-sodic (naalsi 3 o 8 and Caal 2 si 2 o 8 ); in reaction with soil, they release k + , Ca 2+ and mg 2+ (Brady & Weil 2002;meurer, 2006;resende et al., 2007). this increases the relative availability of these elements in soil.
in the Chapada mining byproduct, the presence of 2:1 clay-mineral, quartz (sio 2 ) and rutile (tio 2 ) was detected, while in biotite schist, of the phyillosilicate group, biotite and mica were present. this is a possible cause for the release of k + , Ca 2+ , and mg 2+ from rock to soil. for the mining byproducts, these data were not obtained due to the interference of low cristallinity material in the samples.
the main compounds in ultramafic rock were 2:1 clay-minerals and olivine (mg,fe) 2 sio 4 , pyroxene and feldspar. the latter is a neosilicate, common in igneous and metamorphic rocks, pyroxenes of single-chain inosilicate and in feldspar.
the types of minerals, the k + arrangement in the rock mineral structure, the particle-size distribution (100 % 0.150 mm and 45 % < 0.075 mm mesh to assure total reactivity) and possible action of synergetic microorganisms can hamper k + release from rocks (Wang et al., 2000;moreira et al., 2006;stanford et al., 2008).
the most important minerals detected in the rock powder, which are potential k sources to soils, were phlogopite schist (phlogopite schist); illite and biotite; feldspars (ultramafic rock), feldspars and phlogopite schist (breccia) (figure 3).
With the exception of the mining waste, the feldspar-containing rocks led to high k availability (ultramafic rock and breccia). the k they released was not readily exchangeable, due to strong binding to sio 4 and alo 4 tetrahedron, compensating charge deficiency in mineral structure. the final k release into soil depends on cation exchange in acid medium as follows: k alsi 3 o 8 + h + → halsio 8 + k + (Curi et al., 2005).
in breech mineral phase, phlogopite schist and feldspars were found; the latter are tectosilicates and rather important nutrient sources when altered they supply k + and Ca 2+ , na + and fe 2+ (meurer, 2006; resende et al., 2007).
in the case of biotite schist and phlogopite schist, the main potassium sources were illite, biotite (dark mica) and phlogopite schist. mica layers, consisting of tetra-and octahedral components, retain k + in the space in-between, strongly bound by oxygen molecules. this binding hinders layer expansion, causing low k + release into the soil. from biotite and phlogopite schist, with tri-octahedral blades components, k tends to be released faster than from muscovite (Curi et al., 2005).
in the tri-octahedral mica-type structure, the k _ o binding is longer (0.3 nm) than in the dioctahedral (0.285 mm), making the binding to the first weaker. in addition, the mica-type structure has a repulsion force caused by octahedral al 3+ over h + from oxydril that diverts h in the octahedron direction, driving k farther away (Brown et al., 1978). Consequently, repulsion and binding forces are weaker in the proximity of h and k ions in trioctahedral than in di-octahedral micas (Bigham et al., 2001;Brady & Weil, 2002;Wilson, 2004;Curi et al., 2005).
Potassium saturation of potential CeC varied between 3.9 and 23.7 % (figure 4a). the adequate value for this characteristic is between 2 and 5 % (lopes, 1998), according to field crop requirements. although CeC is highly occupied by k + , the studied soil had a very low CeC (2.0 cmol c dm -3 ), meaning that even a small amount of k + can occupy a great extent of the exchangeable sites. ultramafic rock also increased the na + soil levels considerably, with a maximum of 121.13 mg dm -3 , followed by breccia and mining byproduct, with 20.23 and 13.20 mg dm -3 , respectively (figure 4b,c). the high sodium availability induced by ultramafic rock was reflected in the CeC saturation, reaching a maximum of 18.4 % at 600 kg ha -1 k 2 o. the high na + levels resulting from the ultramafic rock treatments can be explained by release from na-rich minerals present in these rocks, such as plagioclase feldspars (naalsio 3 ), from which the element becomes available in soil (meurer, 2006). sodium saturation rates of 15 % and higher may affect some of the physical properties of soils, among them hydraulic conductivity (richards, 1954). this happens because na + increases the diffuse double layer, promoting colloid dispersion and movement, obstructing pores, thus altering conductivity, water movement, aeration and fertility in soils. therefore, the na level in rocks must be taken into consideration before using them as fertilizers.
the increased availability of Ca 2+ , mg 2+ , k + and na + changed the base saturation in soils (figure 4d), proportional to the application rates, except for phlogopite schist. highest values were obtained by applications of ultramafic rock, mining byproduct and mining byproduct, confirming the liming effect of these fertilizers. the mg 2+ levels and respective base saturation were influenced by the alternative nutrient sources ( figure 5a,b). the mg 2+ availability (figure 5a) and mg saturation (figure 5b) were higher after application of breccia and other ultramafic rock. Before applying the alternative fertilizers, the mg 2+ content (0.10 cmol c dm -3 ) in the soil was in the class "very low" (CfsemG, 1999), and after addition of ultramafic rock the availability class changed to "low"(0.18 cmol c dm -3 ).
CeC saturation by the element, which was 5 %, did not increase to the expected values of 10-20 % and the highest value was obtained with breccia and ultramafic rock at 6.59 and 6.46 %, respectively. the rocks of this study contained the following minerals: phlogopite schist and talc (phlogopite schist); clay-minerals 2:1 (Chapada mining byproduct); biotite schist (biotite); clay-minerals 2:1, pyroxenes, feldspars and olivine (ultramafic rock); and feldspars and phlogopite schist (breccia). these are potential mg sources that may become available after reaction in soils (resende et al., 2007; meurer 2006).
al saturation dropped considerably as a result of fertilizer sources and rates (figure 5c,d). Biotite schist promoted the least reduction in al saturation of all ground rocks (figure 5c). Biotite schist led to a small increase in ph (figure 2a) and had low eCC factors that contributed to a change in al saturation in comparison with the other silicate rocks.
in addition to k + , ultramafic rock also contributed to the increase in available P in a dystrophic Yellow latosol (oxisol) (ribeiro et al., 2010). the high availability was explained by the naturally P-rich compounds in mafic rocks (turekian & Wedepohl, 1962); as they react in soils, P is released, contributing to offset deficiencies (table 1). mining byproduct was the fertilizer that most released zn 2+ , while the other nutrient sources released low amounts of the element into the soil (figure 6b). the zn contribution to the soil was as follows (in descending order): mining byproduct > Chapada mining byproduct = ultramafic rock = breccia = phlogopite schist = biotite. the total zn 2+ (124 mg kg -1 ) level in the Chapada byproduct was lower than in biotite schist (151 mg kg -1 ) and phlogopite schist (149 mg kg -1 ); (table 1), although it released higher levels into the soil. since the ground rocks were applied on the basis of their respective k 2 o content, the amount of phlogopite schist was lower than in the byproduct and biotite schist, partly explaining these differences. Biotite contains greater zn 2+ quantities and was applied at higher levels than the Chapada byproduct, although releasing less zn 2+ to the soil. this could be related to zn 2+ sorption to amorphous and 2:1 type soil particles, present in the Chapada byproduct (figure 3). as already pointed out (figure 1a), the ph was influenced by the rock powder in the soil; its increase results in higher ph-dependent negative charges, increasing zn 2+ retention.
zinc is present in various basic and acidic rocks, due to isomorphic mg substitution by zn, common in silicate types. the element also appears in primary minerals, e.g., in olivine, in ultramafic rock, hornblende, augite, biotite, and magnetite. other minerals containing zn are sulphides (spharelite), carbonates, and phosphates (souza & ferreira, 1991; raij, 1991). the zn 2+ soil levels were found to be above the critical level (> 1.2 mg dm -3 ) in the treatments that received mining byproduct, at all k 2 o rates. Copper release from to the soil was intensive when Chapada byproduct and mining byproduct were applied (figure 6c), explained by their high Cu contents and Cu amounts applied to the soil (tables 1, 3). in general, Cu increase was proportional to the applied rock rates. the lowest Cu availability (0.0307 mg dm -3 ) was recorded after 200 kg ha -1 k 2 o in biotite schist, and the highest level after Chapada byproduct application, at 600 kg ha -1 k 2 o (0.7067 mg dm -3 ).
the ph increase related to rock powder application (figure 1a) increases zn 2+ and Cu 2+ adsorption in soil, possibly explaining the lower Cu 2+ extraction in most treatments compared to the control. in general, complexing with soil organic matter occurs in the descending order: Cu > zn > mn (abreu et al., 2007).
the mining byproduct has a higher Cu 2+ content than the Chapada byproduct, but was applied at lower rates (tables 1, 3). this could be the cause for the higher Cu 2+ release.
finally, no significant changes in ni were observed in soil interaction with the k 2 o rates, ascribed to the naturally low ni levels in the silicate rocks used here.

ConClusions
1. the varied chemical composition of the six silicate rocks tested in this study recommends them as alternative lime sources, while additionally supplying k and other nutrients.
2. the application of rock powder alters the soil ph, nutrient and na availability and other chemical properties.
3. high rates of ultramafic rock and Chapada mining byproduct cause a ph increase and ionic imbalance with high na saturation CeC of with high na saturation. 4. the mining byproduct raises the levels of available k, and increases available zn and Cu considerably. Cooper also increases when mining byproducts are applied to the soil. aCknoWledgements this study was financially supported by the Brazilian Council for scientific and technological development (CnPq) and minas Gerais research support foundation (faPemiG).