Evaluation of the agricultural potential of the serpentinite rock as a soil remineralizer

Magnesium is one of the essential elements for the plant growth. However, when the supply of magnesium is required exclusively, few economically feasible options are available. Serpentinite represents an alternative source of magnesium, although little is known about its potential and effi ciency under tropical soil conditions. This work aimed to evaluate the use of serpentinite as a soil remineralizer, as well as magnesium fertilizer. The study was conducted in a greenhouse, using a completely randomized design, with seven treatments and four replications, as follows: three levels of serpentinite, mix of serpentinite and phonolite, and the controls with dolomitic limestone and without fertilization. Two plant species (Zea mays L. corn hybrid BRS – 1055 and Phaseolus vulgaris L. common bean variety BRS – Estilo) and two contrasting soils (clayey and sandy texture), were used in pots. Results showed that serpentinite’s free silica and toxic element contents fi tted the legal requirements. No statistically signifi cant difference was observed for the plant dry matter weight production in the serpentinite and dolomitic limestone control, as well as in the pure serpentinite and the mix with phonolite treatments. The serpentinite was able to supply and to fullfi l magnesium requirements for growth and development of corn and bean plants.


INTRODUCTION
Magnesium is one of the essential elements for the plant growth, with many functions in metabolic processes (Hawkesford et al. 2012). Its supply is necessary for the expression of the genetic potential of plants, especially in highly weathered soils, with low available nutrient content. The most important source of magnesium is dolomitic limestone, used for soil acidity reduction and calcium and magnesium supply. However, in specifi c soil conditions, the supply of magnesium is required exclusively, especially when pH correction is not necessary or calcium content is adequate or in excess. In this condition, few economically feasible options are available, and the use of limestone may cause calcium/magnesium disorders. Unbalanced rates may be induced by the exportation of nutrients by high yield crops, once magnesium exportation is higher than calcium in plants like corn, for instance. Recent studies showed that high yield hybrids export 0,03 kg of Ca and 0,8 kg of Mg for tonne of grain, an exportation ratio of Mg/Ca of 27, approximately (Silva 2009). The most prominent alternative sources for the magnesium supply are the ultramafic rocks, formed by iron and magnesium minerals, and their metamorphic equivalents. Among the latter, the serpentinite group is a potential source due to the high magnesium content of the protolith (Bucher & Grapes 2011). The use of serpentinites in agriculture has been studied for the past decades in temperate climate countries. Chittenden et al. (1967) studied the performance of serpentinites and of dunites as magnesium sources in New Zealand. McNaught et al. (1968) evaluated the effects of many magnesium fertilizers, including serpentinite, in the forage yield in New Zealand. Hanly et al. (2005) studied the effect of the use of serpentinite and of its products of acidification in pastures of New Zealand. Błońska et al. (2016) studied the effect of the use of serpentinite and of other nutrients in a pine forest in Poland. Serpentinite based commercial products are available abroad, e.g. the fertilizer supplier in New Zealand: http:// www.fertco.co.nz/products/magnesiumfertiliser-products/serpentinite. Some works have already been done in Brazil, aiming to evaluate the performance of the serpentinite as a silicate agromineral and magnesium fertilizer. Luz et al. (2010) analyzed the agricultural potential of the serpentinites in Brazil, mostly as a byproduct of mining activities. Carmignano (2014) evaluated the performance of serpentinite from Minas Gerais in soybean crops. Blaskowski et al. (2016) evaluated the agricultural potential of the materials on mining waste piles from chromite mining in Bahia, which are mainly composed of serpentinites and dunites. Alovisi (2018) studied the response of forages to the use of serpentinite as a source of silicon. The measured reserves of the industrial serpentinite rocks in Brazil are above 140 million t, and are concentrated in the States of Minas Gerais, Goiás and Paraná (DNPM 2010). In Minas Gerais, one of the main occurrences of the serpentinites is in the Córrego dos Boiadeiros Body (CBB), an association of metaultramafic and subordinated metamafic rocks, at the central area of the Quadrilátero Ferrífero (QF). The CBB is a maficultramafic intrusive body in the Rio das Velhas greenstone belt (Fernandes 2016). Laboratory analyses of samples from the Mostardas mine showed that serpentinite is composed of 39.9% of MgO, mostly in the minerals lizardite (Mg3Si2O5(OH)4) and talc (Mg3Si4O10(OH)2), with the secondary minerals: chromite (FeCr 2 O 4 ) and magnetite (Fe 3 O 4 ) (Ferreira et al. 2019). This work aimed to evaluate the use of serpentinite as a soil remineralizer, as well as a magnesium fertilizer.

MATERIALS AND METHODS
The experiment was carried out in a greenhouse at Embrapa's experimental farm in Sete Lagoas - 44° 11.7' W). The experimental design was based on a completely randomized scheme, with seven treatments, two soils, two plant species and four replications (112 experimental units). Two soils with contrasting granulometry were selected: clayed Ustox (Latossolo Vermelho Amarelo distrófico -LVAd) and a Quartzipsamments (Neossolo Quatzarênico órtico -RQo). Both presented low exchangeable magnesium contents (Table  I). The sandy soil is an almost pure quartzose sand (developed from the Espinhaço group quartzites), and the clayed soil is a highly weathered Oxisol, presumably dominated by gibbsite and kaolinite in the clay fraction (based on similar soils of the region, e.g. De Brito Galvão & Schulze 1996) (Teixeira et al. 2012). The mix of the phonolite to the serpentinite aimed to increase the total K content in the final product, to fulfill the remineralizer legislation demands (MAPA 2016). The calcitic and the dolomitic limestone in powder and the fertilizers were commercial grade products. The limestone doses were equivalent to 5.0 Mg/ha for the LVAd soil, and to 2.0 Mg/ha for the RQo soil, accordingly to Minas Gerais State recommendations (Alvarez & Ribeiro 1999). Before planting, fertilizers were applied and mixed to the soils as follow: LVAd soil -urea (40 kg N/ha), superphosphate (240 kg P 2 O 5 /ha), KCl (200 kg K 2 O/ha) plus the micronutrients source (100 kg FTE-BR12/ha), and RQo soil -urea (40 kg N/ha), superphosphate (120 kg P 2 O 5 /ha), KCl (100 kg K 2 O/ha) plus the micronutrients source (100 kg FTE-BR12/ha). After that, 5-liter pots were filled with the soils. An additional urea dose was supplied 20 days after the germination, equivalent to 100 kg of N/ha. The treatments were: 1 -Control without serpentinite, limestone or fertilizers (TEST); 2 -Control without serpentinite plus dolomitic limestone and fertilizers (the doses were accordingly to soil fertility analysis) (CALC_D); 3 -Control without serpentinite and with calcitic limestone and fertilizers (CALC_C); 4 -Treatment with serpentinite (dose 2) plus phonolite (3:1 in dry weight), plus calcitic limestone and fertilizers (SERP_F); 5 -Treatment with serpentinite (dose 1) plus calcitic limestone and fertilizers (SERP_D1); 6 -Treatment with serpentinite (dose 2) plus calcitic limestone and fertilizers (SERP_D2); 7 -Treatment with serpentinite (dose 3) plus calcitic limestone and fertilizers (SERP_D3). The doses applied are in the Table  II. The serpentinite doses were equivalent to the nominal MgO content (14%) in the reference (CALC_D), so that the dose 1 is equivalent to half the required dose of limestone, the dose 2 is equivalent to the full limestone dose, and the dose 3 is equivalent to one and a half the limestone dose. The pots were saturated with deionized water before being sown and kept close to the field capacity throughout the experiment. Based on previous tests, the amount of applied water was selected to be just enough to keep the soil moisture constant and to avoid leaching of excess water. Three seeds were used for each pot, and the plantlets were  kept until the harvest of the plants. After 40 days of growth, the whole plants (above ground) were harvested and packed in paper bags, sent to oven dry at 60 °C to measure the dry weight and ground in a rotary mill for the chemical analyses. The soil from each pot was sampled air dried and sieved in a 2.0 mm sieve and sent for the fertility analyses. The serpentinite and the mixture serpentinite/phonolite were analyzed according to the EPA 5032 reference method for the semi-total element content of the oxides of K, Ca, Mg, B, Zn, Cu, Fe, Mn, Mo, Co, Ni, Se, Cl, As, Cd and Pb, according to the EPA 7074-A reference method for the semi-total element content of Hg. The reference methods referred at the IN-SDA Nº003/2015 MAPA (Cap I -B and Cap V) were followed for the granulometric analysis of the ground materials. A petrographic thin section of the serpentinite rock was evaluated by optical microscopy to measure the free silica content. The test was performed to fulfill the legal requirements. The legislation is vague concerning this issue (both the nature of the 'free' silica and the method to be used) and leaves to the researcher the choice of the method. The analysis was performed by point counting in the thin section, which is presumably to count basically the quartz grains. Nevertheless, this type of rock is not supposed to bear quartz, as is a metamorphosed peridotite. The soil samples and the plant materials were analyzed by the standard fertility methods and granulometric analysis (Silva 2009). The pH was measured in a soil/water proportion of 1:2.5. The exchangeable cations (Ca, Mg and Al) were extracted with the 1 M KCl solution. The Mehlich 1 solution was used for the extraction of P, K and the micronutrients. The exchangeable acidity (H + + Al 3+ ) was extracted with the calcium acetate solution (0.5 M). The plant analysis was done by the wet acid method (hot concentrated solution of HNO 3 + HClO 4 3:1, Silva 2009). The results were subjected to the statistical analyses using the R system's (R Core Team 2018) packages "MASS" (Venables & Ripley 2002) and "agricolae" (Mendiburu 2017). Table II. Values of applied doses of the minerals for the pot essay.

RESULTS AND DISCUSSION
The petrographic analysis of the rock indicated that the free silica content (as quartz) is less than 5%, which is below the maximum value of 25%, by the legislation (MAPA 2016). Results of the semi-total element content are presented in the Table III. The serpentinite powder had high magnesium content (>38% MgO), and low K and Ca content, but the sum of the bases (CaO, MgO, K 2 O) was above 39%, higher than the minimum requirement in the legislation (9%). Higher potential as source of boron (0.7%) and cobalt (68 mg/kg) was also evidenced. The potentially toxic elements (As, Cd, Hg and Pb) were all below the maximum levels established by the law (MAPA 2016), and below the detection limits of the method (Table III). Thus, the material is safe for use in agriculture as a silicate agromineral. Also, other elements, such as Cu, Zn, Mn and Ni, for which no limit is established, did not present levels that indicate potential risks for agricultural use. The mix of serpentinite and phonolite produced a material with element contents very similar to the pure serpentinite powder (Table  IV), but with higher K content (>2%) and lower content of Mg and of the rest of the elements, by dilution. The mix product's K content was above the minimum requirement of the law, 1% (MAPA 2016), and it was the only nutrient that had a higher total content after the mixing. The mix fitted the granulometric requirements of the law to be used as a powder (Table III). The pot soil analyses indicated that the fertilization was effective to raise the nutritional status and the soil's pH. The exchangeable contents of the Ca, K and sum of bases were higher than the control (TEST) in both soils. The differences between the soils are related to the initial contents, granulometry and mineralogy. There was no statistically significant effect on phonolite application (treatment SERP_F) in the K exchangeable content, compared to the treatments without phonolite (Tables V and VI). This result is in accordance with the expected low solubility of the K sources in the minerals of this rock, regarding the short time of experiment and despite the use of a strong acid extraction solution for the analysis (Mehlich 1). The mix of the phonolite reduced the total content of the nutrients without any evident benefit to their availability (Table IV). The treatments with the same dose of serpentinite (treatment SERP_F and treatment SERP_D2), with the sole difference of the phonolite input in the former, statistically did not differ overall, including the soil fertility parameters and the total plant dry matter weight production (Tables V, VI, VII and VIII). These results indicated that its use did not have any practical effect in the short time range and the mix only increases the costs of production and freight per unit of the nutrients per tonne of product. Potential long term benefits could not be accessed by this experiment. The pH in water and the exchangeable acidity results (Tables V  and VI) showed the effect of the liming on the soils. The lower elevation of the pH in the LVAd soil, compared to the RQo soil, is related to its higher buffering capacity due to the greater clay content (82%). There was statistically significant effect of the treatments on the pH results, without a clear pattern. The exchangeable Mg content was not statistically different among treatments with serpentinite, with calcitic limestone and the control TEST, but they were statistically different of the control CALC_D. This result is expected as the rock powders are less soluble than the limestone and the time span of the experiment was not long enough to allow the biochemical reactions that release the Mg of the minerals (Crawford et al. 2000) to reach the same level produced by the ready soluble carbonate. This result is also coherent with the Mg extraction by the plants. The doses of the serpentinite were chosen to be equivalent to the ones expected to be use in real agronomic situations, and the reference methods for soil analyses may be underestimating the availability of the Mg from the lower solubility fertilizers, which is indicated by the plant dry matter production. The differences in exchangeable Mg content did not relate to the plant dry matter, which indicate that there was no limitation for the plant production. The results of the nutrient content in the plant dry matter of corn and common bean clearly indicated the effect of fertilizer use, compared to the control TEST (Tables IX and X). The effect depends on the nutrient and soil characteristics, and it should be pointed that both soils have no previous report of fertilization or liming, and both have originally very low Mg content (Table I). The values of the corn dry matter weight were statistically different for the controls CALC_D and CALC_C and for the control TEST, for the LVAd soil (Table VII). The control CALC_C did not differ from the treatment SERP_D2, but differed from the other treatments with serpentinite. For RQo soil, there was no statistically significant difference among treatments, except for the control TEST, which was statistically different of all but the control CALC_C. The values of the common bean dry matter weight were not statistically different, except for the control without fertilizer, for the LVAd soil (Table VIII). There were statistically significant differences for the RQo soil. The treatments control CALC_D, SERP_F and SERP_D1 did not differ, the four treatments with serpentinite (SERP_F, SERP_D1, SERP_D2 and SERP_D3) did not differ, the control CALC_C and the SERP_D3 did not differ, and the control TEST differed from the others treatments. The regression equations fitted for the effect of serpentinite dose in the dry matter weight indicated a trend for the increment of the plant weight with the applied dose, not significant at the probability of 5% (Tables XI to XIV (Tables IX and X). For the RQo soil, the same trend was observed, but only the control CALC_D was statistically different from the rest of treatments. The Mg content values in the common bean dry matter weight were in the medium level class for the serpentinite treatments, they were in the low class for the control CALC_C and the control TEST, and the control CALC_D was above the adequate recommended levels in the LVAd soil. The latter was the only statistically different from all the others. For the RQo soil, the control CALC_D was above the adequate recommended levels, the treatments SERP_D3 and SERP_F were in the medium class and the rest were in the low class. The control CALC_D was the only statistically different from all the others. These results are coherent with the soil exchangeable Mg results, discussed above.
In their work, comparing the effect of Mg fertilization in five common bean varieties, Canizella et al. (2015) reported that the BRS Estilo was included in the group of varieties with the greatest content of Mg, regardless of the rate applied. There was no effect of the type of Mg source in the plant dry matter weight, and this element was not the limiting factor. The corn plant dry matter weight of the treatments with serpentinite (SERP_F, SERP_D1 and SERP_D2 and SERP_D3) were not statistically different from the control CALC_D. The higher Mg plant content in the control CALC_D may be attributed to the luxury consumption of this nutrient, as its content is 3 times the content of the treatment with serpentinite (Table IX), and well above the recommended content. Madhok & Walker (1969) reported the luxury consumption of Mg by common sunflower in nutrient solution, with Table V. Content of potassium and magnesium in soils cultivated with corn. (The means of four replicates followed by the same letters in columns are not statistically different at a probability of 5%).

CONCLUSIONS
Chemical and mineralogical analyses showed that the serpentinite presented high total MgO contents. The free silica and the potentially toxic elements (As, Cd, Hg and Pb) contents are all below the legal requirements. Also, other measured elements, for which no limit is stipulated by law, presented no potential risks regarding its use. No statistically significant difference was reported from the plant dry matter weight of the evaluated plant species among the treatments with the use of the serpentinite and the control CALC_D. Differences among the treatments with the use of the serpentinite and the control CALC_D, compared to the control without fertilization, did occur.
On the other hand, the treatments with the use of the pure serpentinite and the mix with the phonolite presented similar behavior, as well as considering the plant dry matter weight and Mg content in the soil and plant dry matter weight.
No loss or benefit could be inferred from the mix of the serpentinite with the phonolite, other than the reduction of the total nutrient content by dilution, with exception of the K. The exchangeable Mg in the soil and the Mg in the plant dry matter weight were statistically different only for the control treatment with the use of the dolomitic limestone, compared to all the others. This difference was not related to the plant dry weight, for common bean and for corn. It is inferred that the tested remineralizers  (pure serpentinite and the mix with phonolite), were able to supply the Mg and to fulfil the plant requirements, and the higher plant Mg content in the control with use of the dolomitic limestone was attributed to the luxury consumption of the element.