Comissão 3 . 2-Corretivos e fertilizantes EFFECT OF ALTERNATIVE muLTINuTRIENT sOuRCEs ON sOIL CHEmICAL PROPERTIEs

The current high price of potassium chloride and the dependence of Brazil on imported materials to supply the domestic demand call for studies evaluating the efficiency of alternative sources of nutrients. The aim of this work was to evaluate the effect of silicate rock powder and a manganese mining by-product, and secondary materials originated from these two materials, on soil chemical properties and on brachiaria production. This greenhouse experiment was conducted in pots with 5 kg of soil (Latossolo Vermelho-Amarelo distrófico Oxisol). The alternative nutrient sources were: verdete, verdete treated with NH4OH, phonolite, ultramafic rock, mining waste and the proportion of 75 % of these K fertilizers and 25 % lime. Mixtures containing 25 % of lime were heated at 800 oC for 1 h. These sources were applied at rates of 0, 150, 300, 450 and 600 kg ha-1 K2O, and incubated for 45 days. The mixtures of heated silicate rocks with lime promoted higher increases in soil pH in decreasing order: ultramafic rock>verdete>phonolite>mining waste. Applying the mining waste-lime mixture increased soil exchangeable K, and available P when ultramafic rock was incorporated. When ultramafic rock was applied, the release of Ca2+ increased significantly. Mining subproduct released the highest amount of Zn2+ and mn2+ to the soil. The application of alternative sources of K, with variable chemical composition, altered the nutrient availability and soil chemical properties, improving mainly plant development and K plant uptake, and are important nutrient sources.


INTRODuCTION
Since Brazil has a large area of crop production, and the nutrient availability of the soils is low, the country has become a major consumer of highly soluble mineral fertilizers such as potassium chloride (KCl), most of which is imported (DNPM, 2011).
In Brazil, there are K-rich rocks (Resende et al., 2006), but with lower solubility than KCl.Alkaline feldspars, feldspathoids and micas are alternative sources of K for fertilizer production in form of salts, thermophosphates, or for direct application to the soil (Nascimento and Loureiro, 2004).Among the rocks studied earlier by Resende et al. (2006), biotite schist, alkaline breccia, carbonatite, phlogopite, and alkaline ultramafic rock had the greatest capacity to release K.These rocks are composed of several chemical elements that play a role as plant nutrients.
Silicate rocks and mining or steel industry mining wastes represent alternatives, because they contain plant nutrients in appreciable amounts and availability, depending on the minerals in their composition.These rock components contribute to increase soil fertility in the medium to long-term, according to their solubility and reaction with soil.They contribute to increase soil fertility and, aside from P, K, Ca, and Mg, also include essential micronutrients, such as Zn, and Mn (van Straaten, 2007).Additionally, the nutrient release from rocks is slow and gradual, which reduces losses by leaching and favors a long-term release.Some materials have an alkalizing effect, acting as soil conditioners (Resende et al., 2006).Thus, it is essential to know the soil mineralogy and forms of K, among other proprieties, which may contribute to the prediction of supply, fixation and availability of this nutrient for crops.
This study aimed to evaluate the application of alternative potassium sources on some soil chemical properties, and Marandu grass (Brachiaria brizantha cv.Marandu) production, analyzing their efficiency after 45 days of incubation of the rocks and mining waste in the soil.
The rock powders and a mining by-product were originated by different processes in different regions: phonolite (Plateau of Poços de Caldas, MG) originally from alkaline volcanic rock (with high Na 2 O + K 2 O contents) are frequently vitreous to sub-vitreous; ultramafic alkaline rock (Lages, SC), formed by an igneous intrusion composed of ferromagnesian
Palavras-chave: acidez do solo, disponibilidade de nutrientes, fracionamento do potássio, pó de rocha.minerals, plagioclase and carbonate, was collected in a disused quarry, formerly used for construction material; waste from metallurgical processing of Mn mining from (Sete Lagoas, MG), in which K is removed from Fe ore and concentrated in the mining by-product; verdete (Cedro de Abaeté, MG) is a light greenish slate with a clayey matrix with Fe oxide in glauconite.
Potassium fractionation was performed by a specific method for each form.Total K was extracted by microwave digestion, EPA3052 method (USEPA, 1998), exchangeable K by Mehlich-1 (Embrapa, 1997), and non-exchangeable K by 1 mol L -1 boiling nitric acid solution (Pratt, 1973).Finally, soluble K + was extracted by boiling water (Brasil, 2006) and determined by flame photometry.
After the incubation period, soil samples were collected to determine the complementary nutrient requirements of brachiaria plants.Macro (N and P 300, Mg 30 and S 50 mg dm -3 ) and micronutrients (B 0.5, Mn 5.0, Zn 5.0, Mo 0.1 and Cu 1.5 mg dm -3 ) were mixed to the soil in form of pure chemical solutions (Malavolta, 1980).Nitrogen and KCl were split in three applications.Liming was applied only to the control and KCl was used to raise base saturation to 50 %, as recommended by CFSEMG (1999) at a Ca:Mg ratio of 3:1.
Thirty seeds per pot were sown, which sprouted after seven days.Ten days after emergence, plants were thinned to five per pot, when ⅓ of the KCl of each rate was applied.Sixty days after thinning, the plants were harvested for analysis.
Plants were divided in leaves, and stem + sheath.All collected material was washed in distilled water and dried to constant weight in an oven with forced air circulation at 65 °C.Subsequently, the material was weighed and ground in a Willey mill.After grinding, samples of each shoot part were sent to a laboratory for determination of K contents (Malavolta et al., 1997).Potassium was determined by flame photometry after nitric-perchloric digestion.
The relative agronomic efficiency (RAE) of each treatment was calculated based on the K accumulation in brachiaria shoots, according to the following expression: For statistical analysis, the results were subjected to the mean test and regression analysis by statistical software SISVAR® version 5.3 (Ferreira, 2008), using mathematical models to optimize the equation and correlation analysis by software SigmaPlot 11.0.

REsuLTs AND DIsCussION
Rates of potassium, rock types, and rock-lime interactions significantly influenced nutrient availability and altered chemical properties of the incubated soil (p<0.05).
In general, pH values increased proportionally to the increase of applied K and according to the source (Figure 1).
Calcined silicate rocks and lime mixtures promoted an increase in soil pH, in descending order: 25:75U>25:75V>25:75P>25:75T, with RNVRNV of 29, 17, 30, and 48 %, respectively, and relative efficiency (RE %) higher than 70 %.Moreover, rocks without lime caused a linear increase in soil pH, in compliance with K rates, except for pure and treated verdete, which did not cause changes in soil pH as a function of K rates (Figure 1).The CEC of the soil used in this study was low (T) (3.78 cmol c dm -3 ); thus, the buffering capacity is small and tends to re-equilibrate when mixed with neutralizing or acidic materials.
Calcium availability and sum of bases (SB) increased with increasing K 2 O rates for most treatments (Figure 2a,b) with the exception of phonolite, verdete, verdete treated with NH 4 OH, and mining waste.After incubation, fertilizers promoted increases in Ca 2+ from 0.2, natural soil, to 2.90 cmol c dm -3 ).Ultramafic rock + lime (75 U: 25 L) passed (based on criteria of CFSEMG, 1999) from a "very low" to an "ideal" class (from 2.41 to 4.00 cmol c dm -3 ).Among the non-calcined treatments, only ultramafic rock contributed to increase Ca 2+ in the soil.
Magnesium availability increased in soil incubated with ultramafic rock, verdete and the mixture 75 % ultramafic rock with 25 % lime.(Figure 2c).
Diversity in rock mineralogy explains the trends in nutrient release among treatments.Release rates of K + increased after 45-day incubation with multi-nutrient sources, except for ground verdete, mixed with lime, and treated with NH 4 OH (Figure 2d).Although this silicate is rich in K, mineral solubility is low, since their structures are not easily disrupted by natural means, being required a more energetic treatment for K extraction.
Results showed higher K release from the mining by-product (229 mg dm -3 ) and its mixture 25:75 T (233 mg dm -3 ) than from the other treatments, due to the physic-chemical treatments of this material originated from manganese mining in a metallurgical process.The K release in decreasing order in treated soil was mining waste>25:75T>25:75P>ultramafic rock>phonolite>25:75U.
Exchangeable K content in soil was 22.17 mg dm -3 before treatments were applied (Figure 2d).This value is near the minimum required for plant growth and is considered "low" (16-40 mg dm -3 ) K content by CFSEMG (1999).After the treatments, even at the lowest K rates applied, materials such as ultramafic rock, mining waste, 25:75 P and 25:75 T, reached the "medium" level (41-70 mg dm -3 ), and with increasing K rates reached "high" levels of K (71-120 mg dm -3 ) (Figure 2d).
The increased availability of Ca 2+ , Mg 2+ and K + with increasing K rates, altered soil base saturation (V) (Figure 3).Highest values were obtained by lime mixture application, confirming the acidity-correction effect of these treatments, as indicated by the RNV for each treatment.
The mixture 25:75U (RNV = 29 %) induced highest base saturation (75 %).Ribeiro et al. (2010) reported that alkaline ultramafic rock has proved promising for acidic soils, releasing K + , Ca 2+ , and Mg 2+ .This rock powder can also influence the uptake of other nutrients, such as Fe, Mn and Zn, promoting plant growth and development, and play a role as soil conditioner.
Contrarily, Al saturation decreased substantially when treatments were applied (Figure 3b).Ground verdete and verdete treated with NH 4 OH did not change soil pH and Al saturation significantly (Figure 1).
Aside from providing nutrients and improving soil physical-chemical properties, silicate rocks also provide silicon (Si), which is essential for the crop development (Figure 4a).Additionally, other positive effects of silicates are also related to Si-P interactions, enhancing P use (Carvalho et al., 2001).Ultramafic rock addition resulted in highest Si content in soil.Phosphate competes with silicate for the same adsorption sites, so that the latter can move (desorb) the first, and vice versa, from the solid to liquid phase.According to Prado and Fernandes (2001), Si occupies P adsorption sites and, thus, increases P availability in soil solution.
Available P (Mehlich-1) (Figure 4b) in soil treated with ultramafic rock and its mixture with lime increased from 0.47 mg dm -3 (zero-rate) to 7.9 mg dm -3 (450 kg ha -1 K 2 O) and 4.5 mg dm -3 (600 kg ha -1 ).This is due to several factors; the pH, for instance, influences P availability, and in this soil type, characterized by variable charges, increases in pH raise the P availability.Ribeiro et al. (2010) observed that ultramafic rocks also contributed to an increase in exchangeable K, as well as P, Na, pH and base saturation in a Latossolo Amarelo (Oxisol).
Phosphorus extraction by resin in the soil treated with ultramafic rock, and its mixture, increased with increasing soil pH (Figure 1c).Release of P may be a result of Al and Fe precipitation, reduced adsorption of phosphate ions, as well as displacement of P from soil due to greater silicic acid activity.
When 25:75P and 25:75U were applied, there was a decrease in remaining-P levels with increasing K 2 O rates (Figure 4d).This reduction may have resulted from the increased concentration of Ca in soil (Figure 2a), which can complex with P.
Manganese mining waste, and its mixture (25:75T), were the treatments that released the highest amount of micronutrients, such as Zn 2+ and Mn 2+ (Figure 5).Extracted Zn 2+ and Mn 2+ from soil raised, respectively, from 2.6 and 1 mg kg -1 (untreated soil), with the treatments mining waste and its mixture, respectively, to 61 and 34, and 396 and 273 (highest applied rates); treatments: mining waste and 25:75T did not exceed permissible limits of Zn for agricultural soils, according to Cetesb (2005).Zinc is present in several basic and acidic rocks, due to isomorphic substitution of Mg 2+ by Zn 2+ , common in silicate rocks.However, mining waste is very rich in Zn, as a result of Mn extraction, which concentrates Zn in the mining waste.

Potassium fractionation in soil
Total soil K content differed significantly among treatments.Higher levels of total K were found in treatments with the highest amounts of applied K, except for treatments 25:75V, ultramafic rock, and 25:75U that differed only from untreated soil (Table 1).Total K content among all treatments ranged from 233 to 825 mg dm -3 (Table 1).Total K contents in soil incubated with rocks and limerock mixtures were similar to those found in the clay fraction of Oxisols by Melo et al. (2005) (549, 810 and 960 mg dm -3 ), which were considered low.Generally, younger soils have higher K levels (4,220;  4,191; 9,412 mg dm -3 ), which are significant, mainly, due to the higher content of primary minerals in the clay fraction.
Several studies have shown high correlations between soil K extracted by nitric acid with K uptake by corn, wheat and eucalyptus (Simard et al., 1992;Melo et al., 2005).In this study, nonexchangeable K forms extracted by nitric acid and exchangeable K values extracted by Mehlich-1 were positively correlated.Significant differences were observed for exchangeable K among treatments, ranging from 24, 26, 29, and 30 mg dm -3 K for verdete, to 83, 146, 160, and 233 mg dm -3 K for mining waste (Table 1).

Plant response to treatments
Leaf dry matter weight (LDM) increased linearly with increasing rates in treatments, after 60-day cultivation with 25:75P, 25:75T, ultramafic rock, and mining waste.Nonetheless, for 25:75U, rates of 200 kg ha -1 K 2 O and higher promoted a decrease in LDM.The lowest LDM values were obtained in all treatments containing verdete, being non-significant in relation to rates in the treatments 25:75V and phonolite (Figure 6a).Stem and sheath production were not significantly altered by increasing rates in the treatments 25:75 V, calcined verdete, verdete treated with NH 4 OH, and 25:75P.The LDM was highest in soil treated with mining waste, phonolite, 25:75U, ultramafic rock, and 25: 75T, respectively with positive linear behavior (Figure 6b).
Leaf, stem and sheath production was lowest in the verdete treatments (pure, mixed with lime, and treated with NH 4 OH), unlike found by Eichler and Lopes (1983).These authors tested a fertilizer obtained through calcination at 1,100 °C of verdete and lime (in equal parts), resulting in a shoot dry matter weight of corn equivalent to that obtained with KCl in the first crop cycle and greater in the following.This difference may be related to the heating temperature and time, and the mixture components and their proportions.Calcination with lime reduces mixture melting temperature, which, according to Kirsch (1972), favors structure alterations of the original mineral, and subsequent formation of other compounds, releasing part of the mineral-bound K.
At each stage of forage development, the leaf, stem and dead material ratios of plant dry matter differ.This means that the plant structure changes over time.Animals have preferences for some plant parts, for example leaves over stems (L'Huillier et al., 1986).
According to Pinto et al. (1994), a critical limit for a weight ratio (leaves:stem) of 1.0 has been considered.Lower weight ratio leaves:stem would mean a drop in forage quantity and quality.In this study, we found a mean ratio of 1.71.Carvalho et al. (1991) evaluated Brachiaria decumbens responses to N and K fertilization in a Latossolo Vermelho-Amarelo (Oxisol) treated with 43.5 mg dm -3 K for two years (three growing cycles).They found visual deficiency symptoms in leaves in treatments without K, which did not occur in our study, even when the forage was grown in untreated soil (22 mg dm -3 K).
Differences among treatments induced variations in soil-available K and, consequently, variations in forage yield.The treatments that provided more K to the soil (Figure 2d) were also those in which the plants absorbed more, that is, the treatments with mining by-product and 25:75T (Figure 7).

Figure 1 .UFigure 2 .
Figure 1.Effect of the different treatments and K rates on soil pH, after a 45-day incubation.*significant at 5 % by the Scott Knott test.Non-significant for verdete and verdete mixed with NH 4 OH.

RRFigure 4 .
Figure 3. Values of: (a) base saturation (V) and (b) aluminum saturation (m) of soil as affected by the treatments, after 45-day incubation.

Figure 7 .
Figure 8. Relative Agronomic Efficiency Index (RAE) of the multi-nutrient sources based on K accumulation in Brachiaria shoots, as affected by the applied rates.

Table 1 . Potassium fractionation in soil after a 45-day incubation, as affected by the different rates and treatments
16 cA 21 cA 36 bB 36 bA 68 aA 22 dA 98 cA 137 bA 146 bA 233 aA 31 eA 131 dA 193 cA 235 bA 290 aA 233 cA 429 bB 520 aB 561 aB 560 aCPhonolite 16 aA 16 aA 26 aC 26 aB 16 aC 22 bA 36 bC 43 bC 63 aC 74 aC 31 bA 65 aA 77 aC 87 aD 108 aD 233 dA 483 cA 626 bA 599 bB 718 aB Means followed by the same uppercase letter in columns and lowercase letter in rows do not differ from each other at 5 % of significance by the Scott-Knott test.