29(2)_12+Maiz.indd Recceived for publication: 17 May, 2010. Accepted for publication: 2 June, 2011. 1 Physiology online, Centro de Investigaciones del Banano (Cenibanano); Asociación de Bananeros de Colombia (Augura). Apartadó (Colombia). 2 Department of Agronomy, Faculty of Agronomy, Universidad Nacional de Colombia. Bogota (Colombia). 3 Corresponding author. jsanchez@augura.com.co Agronomía Colombiana 29(2), 265-274, 2011 Spatial variability of soil chemical properties and its effect on crop yields: a case study in maize (Zea mays L.) on the Bogota Plateau Variabilidad espacial de las propiedades químicas del suelo y su efecto sobre el rendimiento del cultivo: estudio de caso en maíz (Zea mays L.) para choclo en la Sabana de Bogotá Jaiver D. Sánchez T.1, Gustavo A. Ligarreto M.2, and Fabio R. Leiva2 ABSTRACT RESUMEN To evaluate the effect of soil chemical properties on the crop yield of corn, in the context of site-specific fertilization, it was characterized the spatio-temporal variability of these properties and crop yield in a lot at the Centro Agropecuario Marengo of the Universidad Nacional de Colombia (Mosquera, Colombia). Using a systematic sampling grid of 32 points (25 x 25 m), soil samples were taken before crop sowing and 60 days after sowing (das) to determine soil pH, N (%); Ca, K, Mg, Na, Al, H (cmol+ kg-1), P, Cu, Fe, Mn, Zn and B (mg kg-1). At 162 das, harvest and yield components were evaluated by site. The data was processed using multivariate procedures, descriptive analysis and geostatistical analysis. Emergent properties were obtained from the original chemical variables using principal component analysis (PCA); these new variables were evaluated using geostatistical analysis to show spatial distribution and its correlation with crop yield. The PCA allowed the finding of three patterns of spatial variability in the soil corresponding to the variables related to soil fertility OC, Ca, Mg, K, CIC and B, the availability of nutrients by soil redox potential, and the variability associated with salinity explained by the Na content and soil electrical conductivity. The first group of variables largely explains the spatial variability of crop yield of corn. Para evaluar el efecto de las propiedades químicas del suelo sobre el rendimiento del cultivo del maíz para choclo, en el contexto de la fertilización específica por sitio, se caracterizó la variabilidad espacio temporal de esas propiedades y del rendimiento del cultivo, en un lote del Centro Agropecuario Marengo de la Universidad Nacional de Colombia (Mosquera, Colombia). Usando una red sistemática de muestreo de 32 pun- tos (25 x 25 m), referenciados dentro del lote, antes de siembra del cultivo y 60 días después de siembra (dds), se tomaron muestras de suelos para determinar pH; N (%); Ca, K, Mg, Na, Al, H (cmol+ kg-1); P, Cu, Fe, Mn, Zn y B (mg kg-1). A los 162 dds se evaluó la cosecha y los componentes de rendimiento por punto. El procesamiento de los datos se realizó mediante procedimientos multivariados, análisis descriptivos y análi- sis geoestadísticos. Se obtuvieron propiedades emergentes a partir de las variables químicas originales mediante análisis de componentes principales (ACP); a esas nuevas variables se les realizó análisis geoestadístico para conocer su distri- bución espacial y su correlación con el rendimiento. El ACP permitió encontrar tres patrones de variabilidad espacial en el suelo y que corresponden a: las variables relacionadas con la fertilidad del suelo CO, Ca, Mg, K, CIC y B; la disponibilidad de nutrientes por potencial redox del suelo; y, la variabilidad relacionada con la salinidad explicada por el contenido de Na y conductividad eléctrica del suelo. El primer grupo de variables explicó en gran medida la variabilidad espacial del rendimiento del cultivo del maíz. Key words: fertilization, geostatistical analysis, productivity, soil. Palabr as clave: fer t i l i z ac ión, a ná l i sis ge oe st ad í st ico, productividad, suelo. Introduction The variability that occurs in crop development and yield is caused by specific ecological conditions of each site, by genetic factors specific to the planting material and crop management conditions that depend on the management and/or cultural aspects of the producers (Srinivasan, 2006; Leiva, 1998; Cerri et al., 2004; Jin and Jiang, 2002; Leiva, 2006). The spatial variability of soil occurs due to pedogenetic fac- tors and their use and management (Rodenburg et al., 2003; Viera and Gonzalez, 2003), and is expressed in physical and chemical properties (Cerri et al., 2004; Jin and Jiang, 2002; Corwin et al., 2003; Bailey et al., 2001), mineralogy (Sovik and Aagaard, 2003), moisture content and field capacity (Reichardt et al., 2001), in the organic matter content and mineralization of carbon (Amador et al., 2000). 266 Agron. Colomb. 29(2) 2011 Spatial and temporal variability occurs on different scales. The level of a climate variable can be relatively homo- geneous within a cultivation site and depending on the topography of the region, vary widely with distance. In contrast, the levels of physical and chemical variables of soil may switch to sub-meter scales (Srinivasan, 2006; Leiva, 2006; Cerri et al., 2004; Jin and Jiang, 2002). The variability of the biological, physical and chemical properties of soil is a major reason for the variability of crop production (Jin and Jiang, 2002; Rodriguez et al., 2008). Recognizing the importance of quantifying, understanding and managing the variability that occurs in agricultural systems in Colombia, different approaches have been used to implement site-specific management, from work at the regional level to determine homogeneous agro-ecological zones where two spatially distant sites can belong to the same homogeneous management zone, and share similar characteristics to smaller scale projects where sites charac- terized by spatial continuity are necessary for the analysis of spatial correlation between sampling site (Isaacs et al., 2004; Rubiano, 2005; Leiva, 2006; Peña et al., 2009). Due to the high number of interactions in ecological pro- cesses, a considerable part of the mechanisms that contribu- te to the variability of a variable, such as performance or quality of the harvest site, remain uncertain, this is because in complex systems, such as agricultural, there are attri- butes that cannot be explained by individual variables, due to diverse relationships and the fact that they do not act in isolation. These attributes are called emergent properties, so a challenge to the analysis of information management is site-specific data to be analyzed in a comprehensive manner, with appropriate methodologies and techniques to reduce system complexity and improve their understanding by using multivariate analysis. The present study aimed to determine the effect of spatial variability of soil chemical properties on the yield of maize on the Sabana de Bogotá, for possible implementation of site-specific fertilization. For this purpose we evaluated the spatial variability of crop production of corn and chemical properties of the soil, and then analyzed the correlation between these two types of variability. Materials and methods The research was conducted on a commercial lot of 2 ha at the Centro Agropecuario Marengo (CAM) of the Uni- versidad Nacional de Colombia, Bogotá, located in the municipality of Mosquera, Cundinamarca (Colombia), coordinates 4º 42’ N; 74º 12’ W, 20 km from Bogotá DC. The altitude is 2,547 m a.s.l.; the average yearly temperature is 12.9ºC, average precipitation of 820 mm year-1, average relative humidity of 78% and sunshine of 4.8 h d-1. The soil was classified, according to USDA taxonomy as Fluventic Humic Dystrudepts. In the experimental plot, a systematic network of samples using a regular grid (grid) 25 x 25 m was laid out, with sampling points referenced by cartesian coordinates (X and Y). A total of 32 points were taken, avoiding the edge effects of the lot. At each referenced point, soil samples were taken in two periods for chemical analysis of organic carbon (OC) by Walkley-Black method, exchangeable bases: calcium (Ca), potassium (K), magnesium (Mg) and sodium (Na) by extraction with 1N ammonium acetate pH 7; CIC analyzed by the displacement of NH4 exchanged with 1M NaCl, pH, phosphorus (P) by the Bray II method, copper (Cu), iron (Fe), manganese ( Mn) and zinc (Zn) by extraction with DTPA and boron (B) by extraction with phosphate monobasic (azomethine-H). The first take was done before planting the maize, without any tillage, the second at 60 das. The only crop fertilization was performed at 20 das, using 15-15-15 for the equivalent of 25 kg ha-1 N, 25 kg ha-1 of P2O5 and 25 kg ha-1 of K2O. Corn variety ICA V-508 was planted: starchy, yellow and adapted to cold weather (Moreno, 1984), the planting dis- tance used was 0.80 cm between rows and 0.25 cm between plants, for a density of 50,000 plants/ha. We performed the commercial management used by corn producers in the area, applied uniformly throughout the crop. At 162 das, the corn was harvested. To evaluate the performance, an area of 9 m2 was denoted around each point and all plants within this area were harvested and quantified for performance. Each ear was weighed, including the mats and grains cob, according to market standards. The measurements were taken as kg m-2. Assuming that the variables of soil and yield of maize have spatial structure, geostatistical analysis was performed for each of these variables. Experimental semivariograms were calculated for all variables and were adjusted to theoreti- cal semivariance models using GS+™ software version 9. Interpolation was performed using the ordinary Kriging method and the spatial trend was visualized with contour maps (Giraldo, 2002). Multivariate analysis was performed by groups of variables, to reduce the dimensionality of the data and to facilitate the interpretation of the information contained in the data 267Sánchez T., Ligarreto M., and Leiva: Spatial variability of soil chemical properties and its effect on crop yields: a case study in maize (Zea mays L.) on the Bogota Plateau using the covariance matrix of principal component analysis (PCA) in SPSS™ version 15 ( Diaz-Monroy, 2002; Hidalgo, 2003). Bivariate analysis was also performed with Pearson’s linear correlation between soil variables and per- formance between the factorial coordinates of the major components and performance. Results and discussion Spatial variability of crop yield and soil chemical properties The corn yield had a coefficient of variation (CV) of 50% and ranged between 1.82 and 0.11 kg m-2, with an average of 0.84 kg m-2 (Tab. 1). The high CV of performance, consistent with the same variety with a uniform crop management under similar weather conditions in the field of culture, suggests that this variability can be explained by other factors, including the various soil variables analyzed. Tab. 1 shows that some variables showed considerable vari- ability with coefficients of variation above 30%. This was the case during the two sampling periods for EC, K, Na, Al, H, before planting, for P and B, and 60 das for Cu, Mn and Zn. In particular, it highlights the high variability of exchangeable Al and H, in contrast to the low variability presented by the pH. In the case of Al, however, it is im- portant to note that the values found are not limiting for the crop (Tab. 1). TABLE 1. Descriptive analysis of maize yield and nutrient content in the soil at crop establishment and 60 das on the Bogota Plateau. Variable Unit Period Min Max Mean SD Bias Kurtosis CV Rdto kg m-2 0.11 1.82 0.84 0.42 0.25 -0.59 50 EC dS m-1 1 0.47 1.87 0.75 0.28 2.37 7.27 38 EC Ds m-1 2 0.44 3.03 1.25 0.5 1.62 4.23 41 pH 1 4.81 5.36 5.00 0.12 0.89 1.08 2 pH 2 4.77 5.62 5.17 0.19 -0.15 0.14 4 OC % 1 1.76 4.3 3.06 0.52 0.17 0.61 17 OC % 2 2.25 4.58 3.27 0.6 0.26 -0.64 18 N % 1 0.15 0.37 0.26 0.05 0.17 0.61 17 N % 2 0.19 0.39 0.28 0.05 0.22 -0.68 19 Ca cmol+ kg-1 1 8.28 13.75 10.93 1.31 0.34 0.18 12 Ca cmol+ kg-1 2 8.00 13.43 10.69 1.34 0.17 -0.52 13 Mg cmol+ kg-1 1 2.46 4.01 2.99 0.36 1.31 1.93 12 Mg cmol+ kg-1 2 2.13 4.38 3.00 0.39 0.93 4.46 13 Na cmol+ kg-1 1 0.77 2.60 1.17 0.37 2.02 6.01 32 Na cmol+ kg-1 2 0.46 3.65 1.06 0.53 3.93 19.59 49 K cmol+ kg-1 1 0.39 1.95 0.88 0.43 1.21 0.51 49 K cmol+ kg-1 2 0.40 2.58 0.98 0.52 1.83 3.19 52 Al cmol+ kg-1 1 0.03 0.50 0.15 0.10 1.86 4.62 67 Al cmol+ kg-1 2 0 0.64 0.14 0.11 2.82 12.63 80 H cmol+ kg-1 1 0.03 0.25 0.13 0.05 0.25 -0.41 40 H cmol+ kg-1 2 0 0.28 0.12 0.06 0.18 0.48 52 CIC cmol+ kg-1 1 21.26 29.75 24.85 2.27 0.56 -0.38 9 CIC cmol+ kg-1 2 21 30.60 24.50 2.31 0.86 0.27 9 P mg kg-1 1 36.80 140.10 65.29 20.66 1.77 4.53 32 P mg kg-1 2 77.60 144.30 129.16 18.59 -1.51 1.11 14 Mn mg kg-1 1 3.25 15.75 11.88 3.26 -1.31 1.23 27 Mn mg kg-1 2 1.24 11.66 5.31 2.84 0.82 -0.08 53 Zn mg kg-1 1 10.48 39.5 27.4 5.66 -0.45 1.70 21 Zn mg kg-1 2 3.74 36.18 14.12 6.78 1.10 2.17 48 Fe mg kg-1 1 440.00 605.00 523.19 47.95 -0.04 -0.90 9 Fe mg kg-1 2 402.00 490.00 436.69 21.44 0.78 0.27 5 Cu mg kg-1 1 0.90 3.28 2.66 0.61 -1.51 1.46 23 Cu mg kg-1 2 0.34 2.26 0.93 0.45 1.51 2.38 49 B mg kg-1 1 0.17 0.72 0.34 0.13 1.36 1.90 37 B mg kg-1 2 0.37 0.92 0.56 0.12 1.14 1.17 22 Period 1: before sowing; period 2: 60 das. 268 Agron. Colomb. 29(2) 2011 TABLE 2. Models fitted to experimental semivariograms for yield variables of maize, the nutrient content at the time of crop establishment and 60 das in a lot on the Bogota Plateau. Variable Period Model Nugget Sill Range* R2** Yield Gaussiano 5.820 31.630 164 0.99 H 1 EPP 0.003 0.003 105 0.38 H 2 Spherical 0 0.011 37 0.01 Al 1 Spherical 0 0.009 41 0.75 Al 2 Exponential 0.001 0.004 89 0.60 B 1 Gaussiano 0.01131 0.03932 329 0.96 B 2 Exponential 0.002 0.021 166 0.98 Ca 1 Gaussiano 0.826 2.692 206 0.98 Ca 2 Gaussiano 0.672 3.354 217 0.96 N 1 Spherical 0.041 0.447 216 0.95 N 2 Gaussiano 0.062 0.400 90 0.99 EC 1 Spherical 0.001 0.078 42 0.50 EC 2 Exponential 0.019 0.263 63 0.76 CIC 1 Gaussiano 2.610 9.229 244 0.82 CIC 2 Spherical 0.330 5.653 117 0.99 Cu 1 Gaussiano 0.100 0.507 101 0.99 Cu 2 Exponential 0.016 0.206 52 0.38 P 1 Exponential 1 352.3 69.9 0.99 P 2 Spherical 76.000 453.500 160 0.99 Fe 1 EPP 2117.74 2117.741 - 0.80 Fe 2 Exponential 67.000 534.600 120 0.98 Mg 1 Spherical 0.031 0.130 96 0.99 Mg 2 Exponential 0.014 0.173 140 0.99 Mn 1 Spherical 4.790 11.990 108 0.99 Mn 2 Exponential 0.660 8.255 29 0.09 pH 1 Spherical 0.005 0.016 87 0.96 pH 2 Exponential 0.007 0.043 109 0.90 K 1 Exponential 0.014 0.198 100 0.62 K 2 Gaussiano 0.126 0.313 125 0.98 Na 1 Exponential 0.008 0.142 69 0.65 Na 2 Spherical 0.000 0.306 64 0.66 Zn 1 Gaussiano 12.200 55.400 211 0.97 Zn 2 Exponential 0,1 43,96 48,9 0,79 Period 1: before planting; Period 2: 60 days after sowing. EPP: pure nugget effect. * Effective range of the model. ** R2 of the semivariogram. FIGURE 1. Spatial distribution of the variable fresh weight of the yield of the corn (kg m-2) in a lot on the Bogota Plateau. 0 25 50 75 100 125 150 175 Length (m) 0 25 50 75 Le ng th (m ) Yield (kg/m2) 1,81 1,39 0,97 0,54 0,12 effect-EPP (ratio between the sill and the nugget of 100%), indicating lack of spatial correlation. The variables Al, Mn, and Cu H (60 das) R2 values were lower in the models, so that maps of each variable estimated with these models may not adequately represent its spatial distribution in the field. The adjustment of other variables to dimensional models indicates that the average values of the variables do not show clear trends in any direction, so that they meet the assumption of isotropy and use of these models to make estimates at unsampled sites provides adequate spatial representation of what happens in the field with each of these variables (Viera and González, 2003). To determine the degree to which spatial variability in yield can be explained by the chemical properties of the soil, we Theoretical models of semivariance set are shown in Tab. 2 and the spatial distribution of yield (kg m-2) are presented in Fig. 1. For the contents of Fe and H before planting it was not possible to set a theoretical model, showing a pure nugget Yield (kg m-2) 1.81 1.39 0.97 0.54 0.12 269Sánchez T., Ligarreto M., and Leiva: Spatial variability of soil chemical properties and its effect on crop yields: a case study in maize (Zea mays L.) on the Bogota Plateau performed bivariate correlation analysis using Pearson’s method (Tab. 3). TABLE 3. Pearson correlation coefficients between the variables of soil and crop yield of corn in Lot 7 at (CAM). Variables Correlation coefficients with yield Before sowing 60 das CO 0.69** 0.77** Ca 0.52** 0.55** Zn 0.40* 0.34 Mg 0.35 0.58** K 0.28 0.56** P 0.14 0.63** Cu -0.11 0.21 Mn -0.08 -0.15 B -0.03 0.52** Fe -0.03 0.30 Na 0.16 0.26 Al -0.22 -0.24 H -0.02 0.41* CIC 0.45* 0.67** pH -0.39 * -0.54** CE 0.24 0.51** das, days after sowing. * Significant (P≤0.05), ** Highly significant (P≤0.001). Due to many significant correlations found between soil variables and yield, as well as the diversity of patterns found when variables are spatially displayed with contour maps, it is difficult to make a management decision by allowing site-specific implementing in the field, taking into account all variables, so we used principal component analysis (PCA) to reduce the dimensionality, facilitating the inter- pretation of the information and grouping of the variables that explain the crop behavior (Diaz-Monroy, 2002). This independent analysis was performed for the two data sets. In analyzing the content of elements in the soil, four com- ponents explained 77% of the variability of the soil before planting and 79% of the variability at 60 das (Tab. 4). Using the information from the main components, geo- statistical analysis was performed. Tab. 5 presents the semivariance models adjusted for these components. EPP was found for component IV before planting, showing absence of spatial correlation of the emergent property that accounts for this component. The other components were adjusted to semivariance dimensional models, but model adjustment of component IV at 60 das was low (R2 = 0.18), so it may not properly represent spatial behavior in the field. For the first three components, in the analysis of the vari- ables of soil before planting and 60 das, the models show spatial dependence structure, in components I and II spatial autocorrelation occurs at a greater distance compared to component III that presented lower ranges of semivariance function. This indicates that soil properties explained by each of the components have different spatial behavior TABLE 4. Variance explained by each principal component characteristic of the soil before planting and 60 das. Component Variances Before sowing 60 das Total Variance (%) Cumulative variance (%) Total Variance (%) Cumulative variance (%) 1 5.513 34.455 34.455 6.317 39.480 39.480 2 3.309 20.683 55.137 3.378 21.113 60.593 3 1.963 12.271 67.408 1.797 11.229 71.821 4 1.590 9.937 77.345 1.187 7.418 79.239 das, days after sowing. TABLE 5. Theoretical models fitted to experimental semivariograms of the principal components obtained for each set. Period Comp Model Nugget Sill Range* R2** 1 I Gaussiano 0.37900 2.43900 194.7 0.96 1 II Spherical 0.19800 1.11400 97.9 0.99 1 III Gaussiano 0.04400 1.00100 41.4 0.71 1 IV EPP 0.86456 0.86456 104.9 0.01 2 I Spherical 0.00600 1.22300 135.9 0.99 2 II Exponential 0.45900 1.26500 157.2 0.98 2 III Spherical 0.04600 0.99900 54.8 0.99 2 IV Exponential 0.00300 0.94200 46.2 0.18 Period 1: before planting; Period 2: 60 days after sowing. EPP: pure nugget effect. * Effective range of the model (m). ** R2 of the semivariogram. 270 Agron. Colomb. 29(2) 2011 within the lot, which confirms the importance of grouping variables through the ACP. To establish the relationship between the principal compo- nent and crop yield, a correlation analysis was performed (Tab. 6). The results show significant correlations between yield and the main component I of the properties of the soil before planting and 60 das. Other components showed no correlation with yield. In addition, the main components I obtained before and at 60 das correlated with each other in a highly significant manner (0.746 P≤0.01). Emergent variables in the soil The emergent property of component I may be called ‘soil fertility’ because it is caused by the contents of Ca, Mg, K, N, P and CIC before planting and 60 das for the same variables and B (Tab. 7). By comparing the variables that explain these components, the values themselves and their signals and spatial distribution, we find that these positively affect yield and are responsible to a greater de- gree for spatial variability in the range of that found. This important result is consistent with that found in research in conditions similar to the present study (Ospina, 1999; Fenalce, 2002; Rodriguez et al., 2008). Figs. 2 and 3 shows the spatial distribution of the components I in the two pe- riods, within the lot areas with higher values in the color scale being the most fertile soil. FIGURE 2. Spatial distribution of component I associated with the fertility of the soil before corn crop establishment in a lot on the Bogota Plateau. TABLE 6. Bivariate correlation between crop yield and main components obtained for each set of variables. Before sowing 60 das I II III IV I II III IV Correlation coefficient 0.463 0.066 0.06 0.261 0.723 -0.315 -0.067 -0.008 Significance 0.008 0.719 0.746 0.15 0 0.079 0.715 0.964 das, days after sowing. TABLE 7. Vectors of the main components per element before planting and 60 das. Before sowing 60 das I II III IV I II III IV Al -.703 .219 .480 -.534 -.496 -.012 .070 B .368 -.308 -.115 .463 .842 .092 .153 .106 Ca .955 .876 .063 -.125 -.290 CE .495 .244 .806 .451 -.525 .344 .188 CIC .777 -.297 -.231 .275 .823 -.426 -.198 -.173 Cu -.367 .834 -.204 .486 .687 .078 .221 Fe -.781 -.202 -.149 .389 -.008 -.823 .208 H -.104 .542 .246 -.489 -.153 .658 K .830 -.121 -.112 .879 -.039 .003 .310 Mg .868 .237 .163 .823 -.064 .407 -.223 Mn -.257 .820 -.187 .112 .854 .161 .299 N .821 -.282 .209 .811 -.328 -.338 -.070 Na .250 .944 .388 -.240 .756 .100 P .708 .514 -.275 .682 .199 -.106 -.465 pH .105 -.320 -.150 -.825 -.198 .850 -.070 -.021 Zn .477 .759 -.174 .674 .538 -.021 .174 das, days after sowing. 0 25 50 75 100 125 150 175 Length (m) 0 25 50 75 Le ng th (m ) PC I s 1,18 0,70 0,22 -0,15 -0,74 0 25 50 75 100 125 150 175 Length (m) 0 25 50 75 Le ng th (m ) PC I - 60 das 1,66 1,00 0,34 -0,32 -0,98 FIGURE 3. Spatial distribution of component I associated with the fertility of the soil at 60 days corn crop on the Bogota Plateau. Other studies in maize also agree with the findings in this investigation where nitrogen is the element that most often limits the growth and crop yield. This is because plants require relatively high amounts of N (between 1.5 and 3.5% 1.81 0.70 0.22 -0.15 -0.74 1.66 0.00 0.34 -0.32 -0.98 271Sánchez T., Ligarreto M., and Leiva: Spatial variability of soil chemical properties and its effect on crop yields: a case study in maize (Zea mays L.) on the Bogota Plateau of plant dry weight) and because they usually do not have enough soil N available to maintain adequate levels of pro- duction. This deficiency can reduce grain yield and quality (Rincón and Ligarreto, 2008; Below, 2002), in addition, the metabolism of N has an impact on the growth and develop- ment of corn in two general functions: (1) establishing and maintaining photosynthetic capacity, (2) development and growth of reproductive sinks (Below, 2002). On the other hand, the deficiency of N in maize has been described by its effect on the determination of number of grains per plant, because its availability encourages differences in the rate of growth of the ear (critical time) and the number of grains per plant is related to the rate of growth in the criti- cal period (D’Andrea et al., 2008). Calcium absorption in corn grown in tropical conditions occurs gradually in the vegetative stage at the beginning of floral differentiation. Like other nutrients, Ca does not accumulate during the full bloom stage, but its absorption reaches a marked increase from grain filling. This accu- mulation may be due to high demand during the process of cell division in the steps outlined, so that its availability influences the crop yield potential (Rengel, 2003). In maize, potassium deficiency causes accumulation of sugars in the stem, it appears that the low content of K in the cell limit the translocation of sugars to the ear, which interrupts the growth cycle and hinders the normal devel- opment of the grain; during the grain development period the presence of K is important, to promote the transport of the products of photosynthesis from the leaves to the ears (Rengel, 2003). Boron deficiency affects many plant physiological processes such as the transport of sugars, synthesis and structure of cell wall lignification, carbohydrate metabolism, RNA metabolism, AIA, phenols and ascorbate, respiration and plasma membrane integrity. Bazinga et al. (2002) explain that the abortion of the ears occurs when assimilate supply is below the threshold required for normal development and this happens when the levels of soil fertility is low, making CIC an indicator of soil fertility because it is associated with an adequate base saturation. On the other hand, the variability does not explain chemical properties and performance is contained in other components may explain other emergent proper- ties that characterize the dynamics that occurs in the floor of the test group. Component II in the soil before sowing (bs) is explained by the values of Mn, Cu and Fe, while component II in the soil after sowing is explained by Mn and pH (Tab. 7). By comparing the coordinates of the principal components factor before planting and 60 days after, it was found that they are significantly correlated (0.456 P≤0.001). When analyzing the variables that explained component II of the chemical properties of soil, we can deduce that the emergent property of the second component is the redox potential, this assertion is supported by the fact that this potential is correlated with the solubility of mineral nutrients such as Mn, Cu and Fe which explains their values within these components (Marschner, 2002), this agrees with the fact that during the present study ponding and drying presented the lot in relatively short periods of time and according to a study of soils on the Bogota Plateau, conducted by IGAC, approximately 44% of soils suitable for agriculture have constant water logging caused by restricted natural drain- age (Universidad Nacional de Colombia, 1970). Figs. 4 and 5 show the spatial distribution of the compo- nent II, the highest values of the scales presented can be associated with the highest or lowest redox potential under reducing conditions. The redox potential is a quantitative measure of a particular system to reduce susceptible sub- stances, is positive and high in strongly oxidizing systems and negative and low in strongly reducing systems. The variability in the observed patterns between the two seasons for this emergent property can be explained by the FIGURE 4. Spatial distribution of component II associated with the re- dox potential of the soil before corn crop establishment on the Bogota Plateau. FIGURE 5. Spatial distribution of component II associated with the redox potential of the soil after corn crop establishment on the Bogota Plateau. 0.66 0.22 -0.23 -0.68 -1.12 1.03 0.61 0.18 -0.25 -0.67 272 Agron. Colomb. 29(2) 2011 higher accumulated rainfall that occurred before the first soil sampling (32.62 mm for the 7 d preceding the sam- pling) and with respect to the second (11.55 mm for this cumulative), this coupled with poor natural drainage of the soil, led to limiting conditions before crop establishment compared with those presented at 60 das. From the spatial point of view, in places where reducing conditions are more severe, the extractability of Fe is greater, this is because the oxides of manganese and iron are subject to dissolution and precipitation processes that are associated with changes in redox conditions (Tack et al., 2006) and after the organic matter, in soil with Mn oxides are the following electron acceptors, as well in acid soils, such as in the study with high contents of Mn and low nitrate , there may be significant levels of Mn2+ (Marschner, 2002; Santiago et al., 2008). With lower redox potential and iron-rich soils Fe3+ to Fe2+ has been reduced, preventing the precipitation of ferric phosphate and releasing P to the environment (Santiago et al., 2008). The relationship found between Mn and pH is also ex- plained by the reduction conditions of soils and under con- ditions of water logging in acid soils the soil pH increases (Snyder and Slaton, 2002; Domínguez-Vivancos, 1988). The redox potential is an intensity factor and is closely related to the soil aeration system, as well as pH, since both are conditioned on microbial activity and the type of reactions are carried out in the soil. Water influences these processes by modifying the distribution of air in the soil, and thus the diffusion of O2 and CO2 concentration (Santiago et al., 2008; Marschner, 2002). The fact that the redox potential is a cause of variability may help explain the general decline in performance in all cul- ture sites. This compares with results reported by Rodriguez et al. (2008) under similar soil conditions, indicating that excess moisture during the growing cycle of corn reduced corn yield in some parts of the production lot. The slowdown in growth is a typical symptom of species not adapted to water logging. In corn, limited aeration in the root in certain stages of development can cause a low concentration of N, K and P in the area of stem elongation and reduced elongation (Marschner, 2002). Component III of the nutrient content before planting and 60 das were explained by the variability of CE and Na content and the soil salinity caused by the exchangeable sodium is in accordance with the emergent property of component III. This indicator showed somewhat different patterns in the two assessment periods but were correlated statistically (0.364 * P≤0.05), these differences are probably due to higher rainfall before the first sampling , and this led to a higher water table compared to the one at 60 das. The highest values in the contour maps correspond to areas with higher salinity levels (Figs. 6 and 7). The importance of identifying this pattern is that the salinity within a lot may limit water uptake by the plant by increasing the osmotic potential of the soil and cause nutritional imbalances causing reduced growth and devel- opment of plants (Carranza et al., 2009; Corwin and Lesch, 2005). On the other hand, it is important to determine the pattern of variability in the soils of the Bogota Plateau because the rise of salts is a limiting factor in agricultural production, which is caused by poor natural drainage and high groundwater levels. Conclusions The crop yield presented spatial variability within the field, which was explained jointly by the spatial distribu- tion of the soil content of nitrogen, calcium, potassium, magnesium and boron nutrients that are indicators of soil fertility and are important to development, growth and productivity of maize. The use of principal component analysis (PCA) allowed grouping of the chemical variables and observance of three FIGURE 7. Spatial distribution of component III associated with soil sali- nity at 60 das corn crop on the Bogota Plateau. FIGURE 6. Spatial distribution of component III associated with the sa- linity of the soil before corn crop establishment on the Bogota Plateau. 1.57 0.96 0.34 -0.28 -0.89 2.83 1.65 0.47 -0.72 -1.90 273Sánchez T., Ligarreto M., and Leiva: Spatial variability of soil chemical properties and its effect on crop yields: a case study in maize (Zea mays L.) on the Bogota Plateau emergent properties of the soil: fertility, redox potential and salinity, making it possible to understand the complexity of the interactions between the elements in the ground. Emergent properties related to the redox potential and soil salinity was not correlated with the spatial variability of the maize yield, but are credited with contribution to the reduction in general performance throughout the batch culture. Emergent properties of the soil: fertility, salinity, and redox potential presented spatial correlation structure shown in geostatistical analysis, but each property has a different spatial distribution pattern, therefore the manage- ment strategies for specific sites must be different, and that the cause of the variability of each pattern may be due to different circumstances. The combined use of multivariate techniques and geosta- tistical analysis makes it possible to take information from many agricultural systems, reduce complexity, relatively easily visualization and make practical decisions to im- prove the technical criteria for maize fertilization through management by location. Acknowledgements The authors are grateful to the Department of Science, Te- chnology and Innovation, Colciencias and the Universidad Nacional de Colombia, for funding this research developed within the framework of “sustainable land management in cropping systems under the concept of precision agricultu- re” from the Research Group for sustainable development and environmental management. Cited literature Amador, J., A. Wang, M.C. Savin, and J.H. Gorres. 2000. Fine-scale spatial variability of physical and biological soil properties in Kingston, Rhode Island. Geoderma 98, 83-94. Bailey, J.S., J.K.W. Crawford, and A. Higgins. 2001. Use of preci- sion agriculture technology to investigate spatial variability in nitrogen yields in cut grassland. Chemosphere 42, 131-140. Bazinger, M., G.O. Edmeades, and H.R. Lafitte. 2002. Physiological mechanisms contributing to the increased N stress tolerance of tropical maize selected for drought tolerance. Field Crops Res. 75, 223-233. Below, F. 2002. Fisiologia, nutricao e adubacão nitrogenada do milho. Informacoes Agronômicas 99, 7-12. Carranza, C., O. Lanchero, D. Miranda, and B. Chaves. 2009. Aná- lisis del crecimiento de lechuga (Lactuca sativa L.) ‘Batavia’ cultivada en un suelo salino de la Sabana de Bogotá. Agron. Colomb. 27(1), 41-48. Cerri, C.E.P., M. Bernoux, V. Chaplotb, B. Volkoff, R.L. Victoria, J.M. Melillo, K. Paustian, and C.C. Cerri. 2004. Assessment of soil property spatial variation in an Amazon pasture: basis for se- lecting an agronomic experimental area. Geoderma 123, 51-68. Corwin, D.L., S.R. Kaffka, J.W. Hopmans, Y. Mori, J.W. Groenigen, C. Kessel, S.M. Lesch, and J.D. Oster. 2003. Assessment and field-scale mapping of soil quality properties of a saline-sodic soil. Geoderma 114, 231-259. Corwin, D.L. and S.M. Lesch. 2005. Apparent soil electrical con- ductivity measurements in agriculture. Comput. Electron. Agr. 46, 11-43. D’Andrea, K.E., M.E. Otegui, and A.G. Cirilo. 2008. Kernel number determination differs among maize hybrids in response to nitrogen. Field Crops Res. 105, 228-239. Díaz-Monroy, L.G. 2002. Estadística multivariada: inferencia y métodos. Department of Statistics, Faculty of Sciences, Uni- versidad Nacional de Colombia. Bogota. Domínguez-Vivancos, A. 1988. Los microelementos en la agricul- tura. Ediciones Mundi, Madrid. Fenalce, Federación Nacional de Cultivadores de Cereales. 2002. Fertilización del maíz en Córdoba. Boletín Informativo de la Subgerencia Técnica 10(4), 4. Giraldo, H.R. 2002. Introducción a la geoestadística su teoría y aplicación. Universidad Nacional de Colombia, Bogota. Hidalgo, R. 2003.Variabilidad genética y caracterización de especies vegetales. In: Franco, T.L. and R. Hidalgo (eds.). 2003. Análisis estadístico de datos de caracterización morfológica de recursos filogenéticos. Bol. No. 8. Instituto Internacional de Recursos Filogenéticos (IPGRI), Cali, Colombia. Isaacs, E.C.H., C.V. Carrillo, A.E. Anderson, G.J. Carbonell, and U.B.V. Ortiz. 2004. Desarrollo de un sistema interactivo de información en web con el enfoque de agricultura específica por sitio. Serie Técnica No. 34. Centro de Investigación de la Caña de Azúcar de Colombia (Cenicaña), Cali, Bogota. Jin, J. and C. Jiang. 2002. Spatial variability of soil nutrients and site specific nutrient management in the P.R. China. Comput. Electron. Agr. 36, 165-172. Leiva, F.R. 2006. Aproximación metodológica al manejo por sitio especifico del suelo para la sostenibilidad y competitividad de cultivos transitorios en Colombia. Suelos Ecuat. 36(2), 49-56. Leiva, F.R. 1998. Sostenibilidad de sistemas agrícolas. Agron. Co- lomb. 15(2-3), 181-183. Marschner, H. 2002. Mineral nutrition of higher plants. 2nd ed. Academic Press, New York. Moreno, M.J.D. 1984. ICA V.508: Variedad mejorada de maíz tipo sogamoceño para clima frío. Bogota. Ospina, J.G. 1999. Tecnología del cultivo de maíz. Fondo Nacional Cerealista, Bogota. Peña, R., Y. Rubiano, A. Peña, and B. Chaves. 2009. Variabilidad espacial de los atributos de la capa arable de un Inceptisol del piedemonte de la cordillera Oriental (Casanare, Colombia). Agron. Colomb. 27(1), 111-120. Reichardt, K., J.C. Araújo-Silva, L.H. Bassoi, L.C. Timm, O.J. Mar- tins, O.O. Santos-Bacchi, and J.E. Pilotto. 2001. Soil spatial variability and the estimation of the irrigation water depth. Sci. Agric. 58(3), 549-553. Rengel, M.L. 2003. Crecimiento y dinámica de acumulación de nutrientes de maíz (Zea mays L.) en Venezuela. Informaciones Agronómicas 53, 5-8. 274 Agron. Colomb. 29(2) 2011 Rincón, A. and G.A. Ligarreto. 2008. Fertilidad y extracción de nutrientes en la asociación maíz-pastos en suelos ácidos del pi- edemonte Llanero de Colombia. Agron. Colomb. 26(2), 322-331. Ritchie, J.T. and G.l. Alagarswamy. 2003. Model concepts to express genetic differences in maize yield components. Agron. J. 95, 4-9. Rodenburg, J., A. Stein, M. Van-Noordwijk, and Q.M. Ketterings. 2003. Spatial variability of soil pH and phosphorus in rela- tion to soil run-off following slash-and-burn land clearing in Sumatra, Indonesia. Soil Till. Res. 71, 1-14. Rodríguez, J., A.M. González, F.R. Leiva, and L. Guerrero. 2008. Fer- tilización por sitio específico en un cultivo de maíz (Zea mays L.) en la Sabana de Bogotá. Agron. Colomb. 26(2), 308-321. Rubiano, Y. 2005. Geosoil sistema georreferenciado de indica- dores de calidad del suelo herramienta SIG para apoyo a la planificación, uso y manejo del suelo. Ph.D. thesis. Faculty of Agricultural Sciences, Universidad Nacional de Colombia, Palmira, Colombia. Santiago, A., J.M. Quintero, and A. Delgado. 2008. Long-term effects of tillage on the availability of iron, copper, manganese, and zinc in a Spanish Vertisol. Soil Till. Res. 98, 200-207. Snyder, C. and N. Slaton. 2002. Effects of soil flooding and drying on phosphorus reactions. News Views No. April. Potash and Phosphate Institute. Norcross, GA. Søvik, A.K. and P. Aagaard. 2003. Spatial variability of a solid porous framework with regard to chemical and physical properties. Geoderma 113, 47-76. Srinivasan, A. 2006. Precision agriculture: an overview. pp. 3-15. In: Srinivasan, A. (ed.). Handbookof precision agriculture principles and applications. Food Products Press; The Haworth Press, New York, YN. Tack, F.M.G., E. Van Ranst, C. Lievens, and R.E. Vandenberghe. 2006. Soil solution Cd, Cu and Zn concentrations as affected by short-time drying or wetting: The role of hydrous oxides of Fe and Mn. Geoderma 137, 1-2. Universidad Nacional de Colombia. 1970. Centro agropecuario marengo (CAM). Boletín Informativo de la Facultad de Agronomía. Bogota. Vieira, S.R. and A.P. González. 2003. Analysis of the spatial vari- ability of crop yield and soil properties in small agricultural plots. Bragantia 62(1), 127-138.