Soil sampling and chemical analysis

Soil samples were collected from the 0.00-0.10 and 0.00-0.20 m layers with a cutting shovel, according to the method of sampling crosswise to the sowing line, in a width corresponding to the spacing of the last crop, with the line located in the middle of the sampled width (^{CQFSRS/SC, 2004}). In the liming experiments, sampling was carried out annually between the growing seasons (after harvesting the summer and prior to sowing of the winter crop). In turn, in the calibration and fertilization response experiments, samples were taken every six months immediately prior to fertilization and seeding of the crop under study.

In the liming experiments, base saturation values (V) were calculated from the ratio of the exchangeable bases (Ca, Mg and K) and cation exchange capacity at pH 7.0 (CEC_{pH 7.0}). In the experiments involving P and K application, the available levels of these nutrients in the soil were quantified by Mehlich-1 extractor. All analyses were performed by established methods described by ^{Tedesco et al. (1995)}.

Evaluation of crop yields

The grain yield was evaluated in all crops and experiments after mechanical harvesting (combine harvester) of the plots, corresponding to an area of 20-30 m² in the experiments of liming and P and K calibration, and 3-7 m² in the fertilization response experiments. Grain yield was expressed for a moisture content of 13 %.

For the combined yield analysis of the different growing seasons, crops and locations, the results were expressed as relative grain yield (RY), which was based on the yield of the control treatment (no lime in liming treatments; no P in P experiments and no K in K experiments) compared to the maximum yield of the crops (RY = 100) in the respective seasons and locations, calculated by equation 1, as described in detail by ^{Vieira (2014)} and ^{Vieira et al. (2015)}.

Results analysis and interpretation

The suitability of the two layers (0.00-0.10 or 0.00-0.20 m) studied for the evaluation of soil fertility in the experiments was analyzed based on two criteria: (i) analysis of variance, to verify whether the base saturation and levels of available P and K differed between the soil layers, by the LSD test (p<0.05); and (ii) the significance or not of the correlation between crop yields with these soil fertility properties between the layers studied. Due to the non-linearity of these relationships (^{Vieira et al., 2015}), the curves were linearized by logarithmic transformation of the V values and available P and K in the soil, to determine the linear correlation coefficients (Pearson) between the relative crop yields and the logarithm of the V values and available P and K in the 0.00-0.10 and 0.00-0.20 m layers.

Soil fertility properties after surface application of lime and fertilizers

Base saturation increased in the 0.00-0.10 m layer after only one year of surface liming in all experiments (Table 2). This effect lasted until the 10^th year in the two longest-running experiments. In the mean, base saturation increased from 49.6 to 69.6 %. Although there was an increase of this attribute in the 0.00-0.20 m layer, this increase occurred gradually and varied between the five experimental areas, although it occurred in the first year in one area. In this area, the V value increased from 10 to 38 %; in the others, the effect of surface liming in the 0.10-0.20 m layer occurred from the second year (one area) and third year (two areas). On average for all evaluations, base saturation in the 0.10-0.20 m layer increased from 45.4 to 56.1 %. This represents an increase of 24 % which, compared to the 0.00-0.10 m layer, was similar to the results after longer periods of liming (Table 2).

The increase in V values in both layers (0.00-0.10 and 0.00-0.20 m), in the mean of the years, was proportional to the limestone rates (Experiment 5). In this experiment, the increase in this attribute was five-fold, whereas in the other experiments, it ranged from 24 to 37 % in the 0.00-0.10 m and from 12 to 20 % in the 0.10-0.20 m layer (Table 2). Consequently, it was confirmed that the depth effect of surface lime application on soil acidity indicators is related to the time period after liming and that improvements in acidity indicators occurred after 12 months in the 0.00-0.10 m layer and after 24 to 30 months in the 0.10-0.20 m layer in Oxisols under long-term NT, as similarly stated by ^{Caires et al. (2005)}. These results showing an increase in V values in the 0.10-0.20 m layer reinforced the results of the two longer-duration experiments, with increases of up to 13 % in the first year and 48 % after 10 year of liming (Table 2). Importantly, the effects of liming on the deeper soil layers under long-term NT were not restricted to the upper 0.20 m of soil, since a decrease in soil acidity was observed to a depth of 0.60 m after only 3 year of liming in Oxisols under NT for more than 20 year under NT (^{Caires et al., 2008}; ^{Fontoura et al., 2012}). The physical movement of limestone particles through soil biopores may provide an explanation for the amelioration of soil acidity in the 0.00-0.20 m layer, as confirmed by ^{Amaral et al. (2004)}, in a Haplumbrept under NT.

With regard to the values of available P in the soil, as expected, the effect of surface fertilization was more pronounced in the 0.00-0.10 m layer, as shown by the higher and significant (p<0.05) coefficients of determination (R²) of linear regression (Table 3). However, a slight increase in this content in the 0.10-0.20 m layer in the experiments of Guarapuava and Candói in the first year is worth mentioning. Although some increases in available nutrient concentrations may occur over time in this layer, they are not expected to be relevant, since the soils within this study area (Humic Hapludox) have a high nutrient retention capacity, by forming inner sphere complexes, determined by high Al and Fe oxide contents (^{Inda Junior et al., 2007}).

Therefore, the absence of soil tillage favors P accumulation in the soil surface layer, even in the first years of NT, especially by broadcast fertilization (^{Eltz et al., 1989}; ^{Bayer and Mielniczuk, 1997}; ^{Schlindwein and Anghinoni, 2000}). However, in long-term NT experiments (>15 years), in Oxisols of Paraná (^{Eltz et al., 1989}; ^{Ciotta et al., 2002}; ^{Pauletti et al., 2005)}, as well as in an Ultisol of Rio Grande do Sul (^{Costa et al., 2010)} an increase in the available P content below a depth of 0.10 m was observed. Indeed, it seems that there is no vertical movement of the nutrient, since the explanation would be the fertilizer application at the bottom of the planting furrow with direct drilling. According to some authors (^{Dick, 1983}; ^{Rheinheimer and Anghinoni, 2003}; ^{Costa et al., 2010)}, there seems to be an increase in the form of P derived from the decomposition of crop residues left on the surface and from roots that have their growth increased in the area of phosphate fertilization (^{Anghinoni and Barber, 1980)}.

For soil available K, the levels in the layers 0.00-0.10 and 0.10-0.20 m increased in the first year in all three experiments, demonstrating a quick movement of K applied to the surface of the soil profile. After 5 year of K application, the increase of this nutrient in the 0.10-0.20 m layer was linear in both experiments, while at only one location (Pinhão) was the effect only observed in the 0.00-0.10 m layer, which may be related to the sequence of intense K fertilization in the past, resulting in very high (4.5 %) K saturation in the CEC.

There was a greater increase in available K (Table 4) than in available P (Table 3) when comparing the 0.00-0.20 m with the 0.00-0.10 m layer, as expected. This is due to the rapid release from crop residues after plant senescence (^{Giacomini et al., 2003}) and the greater mobility of K in the soil profile.

Relationship between soil properties and crop yields

The relationship between soil properties in the 0.00-0.10 and 0.00-0.20 m layers, and the relative crop yield (RY) were evaluated to identify which soil layer is most correlated with nutrient supply and plant development. In the liming experiments, correlation coefficients between RY of the crops (soybean, corn, wheat, barley, and oat) (n = 473) and base saturation were significant (p<0.01), but similar between the layers studied (0.50 in 0.00-0.10 m and 0.51 in 0.00-0.20 m) (Table 5). Similar relationships between grain yield, water pH and base saturation were also reported by ^{Nolla and Anghinoni (2006)} in the layers 0.00-0.10 and 0.00-0.15 m and by ^{Nicolodi et al. (2008)} for the layers 0.00-0.10 and 0.00-0.20 m. These results indicate that any of the studied layers can be used for acidity diagnosis of lime recommendation in soils under long-term NT.

In view of the different relationships between grain yield and the contents of available P and K in the soil of the crops (^{Lantmann et al., 1996}; ^{Vieira et al., 2013}), the species were grouped into winter cereals (oat, wheat and barley) and summer crops (soybean and corn).

Correlations between the concentration of available K and winter crop yield and the system as a whole (crop rotation) were low and similar for the two layers assessed, although significant (p<0.05) (Table 5). For summer crops, the correlations were very low and not significant. This is probably due to the low response of these crops to the addition of K, since its availability in the soil was high due to the high content in the soil (0.14-0.87 cmol_c dm^-3) and its rapid movement in the soil profile after surface application (Table 4).

In the evaluation of relationships between soybean and corn yield with available P, the correlation coefficients, although both significant (p<0.01), were higher for the 0.00-0.20 m (r = 0.69) than the 0.00-0.10 m layer (r = 0.58) (Table 5). The correlation coefficients between the yields of winter crops (wheat, barley and oat) and available P were significant (p<0.01) and similar: 0.68 in the 0.00-0.20 m and 0.66 in the 0.00-0.10 m layer. However, for the crop rotation system (winter + summer), although the significance was the same (p<0.01), the coefficients were higher for the 0.00-0.20 m layer (0.66 in 0.00-0.20 m and 0.58 in the 0.00-0.10 m layer).

The summer crops, representing the main income source of the producers in the South-Central region of Paraná in a rotation system of the crops under study, must be emphasized in the discussion of the results. Thus, the better correlation with P content in the 0.00-0.20 m than in the 0.00-0.10 m layer indicates the importance of including the sub-surface layer (0.10-0.20 m) in the evaluation of the availability of P in soils under long-term NT. This disagrees with the highest correlations between available P content in the topsoil and the crop yields observed by ^{Sá (1999)} and ^{Schlindwein and Gianelo (2008)}, probably for the shorter period under NT of the areas studied by these authors in comparison with the present study (17-27 year).

Due to the higher nutrient concentrations in the surface layer of NT soil, particularly of P, it has been suggested that samples for the evaluation of soil fertility in some regions of Brazil should be collected from the 0.00-0.05 and 0.05-0.10 m (^{Ribeiro et al., 1999}) or the 0.00-0.10 m layer (^{CQFSRS/SC, 2004}), rather than the 0.00-0.20 m, as this layer is usually indicated for systems with mechanical tillage and soil stirring.

The long-term cultivation of soils under no-tillage (Table 1) is an important aspect that reinforces the results of this study. Thus, and according to ^{Sá (1999}; ²⁰⁰⁴), during the first years of this system, the soil undergoes re-aggregation and structuring, during which the beneficial effects such as accumulation of organic matter, increased CEC and increase in inorganic and organic forms of labile P are not yet detectable. These characteristics, combined with surface applications of fertilizers and liming, result in nutrient accumulation in the topsoil in the initial years, justifying a higher correlation of the indicators in the 0.00-0.10 m layer with crop yields, as reported in various studies (^{Petrere et al., 1996}; ^{Sá, 1999}; ^{Schlindwein and Anghinoni, 2000}; ^{Schlindwein and Gianello, 2008}), and justifying the indication of this layer for soil sampling in the initial stages of system structuring (<10 year). However, with increasing the time of soil under NT, the effects of surface application are not limited to the surface soil layer (0.00-0.10 m), but also occur in the sub-surface layer (0.10-0.20 m), which strongly recommends sampling of the 0.00-0.20 m layer for evaluation the soil fertility in long-term no-tillage systems.