Site-specific management demands the identification of subregions with homogeneous characteristics (management zones). However, zoning is difficult due to the complex relationships and spatial variability of soil properties, which affect spatial patterns of crop yields. In the present work we evaluate if the apparent electrical conductivity (ECa) works as a potential estimator of the properties of soil and nutrients, and a tool for the delimitation of management zones. The CEa was mapped in 58 ha in two production lots, located in the south of the province of Córdoba, close to the town of Canals, composed of Haplustoles. Soil properties and ECa were analyzed using descriptive statistics, simple correlations, and ANOVA. The contents of fine particles (clay and silt), sand and organic matter showed high correlations with ECa (r>0.6; p<0.001), while the content of P and pH were lower (r= -0.47 and 0.42, respectively ). The correlations between the ECa and the electrical conductivity of the extract (ECext), N-NO3 – and S-SO 4 -2 were weak and inconsistent. The ECa measurement successfully delimited three management zones based on the contents of fine particles (clay and silt), sand and OM. While this cannot be considered to differentiate management zones based on the contents of N-NO 3 – and S-SO 4 -2 and ECext because the values were very low and there was little variation between ECa classes. These results suggest that plot-scale ECa maps have the potential to apply site-specific management of crops (MSEC).
Keywords: Precision Agriculture; Management zones; Spatial variability; Nutrients.
ABSTRACT
Relationship between apparent electrical conductivity with soil properties and nutrients
Site-specific management demands the identification of small areas within fields with homogeneous characteristics (management zones). However, determination of these homogeneous sub-fields is difficult because of complex relationships and spatial variability of soil properties, which affect spatial patterns of yield crop. In this paper, we evaluated apparent electrical conductivity (ECa) functions as potential estimators of soil properties and nutrients, and their possible use as tools for the delimitation of homogeneous areas. ECa mapping of a total of 58 ha was performed on two Haplustoll production fields, located south of the province of Cordoba, near the town of Canals. Soil, nutrients and ECa were analyzed using descriptive statistics, simple correlations and ANOVA. Fine particles (clay, silt), Coarse sand and soil organic matter (OM) content exhibited higher correlation with ECa (r>0.6; p<0.001), whereas P content and pH were lower (r= -0.47 and 0.42, respectively). The correlation among ECa and extract electrical conductivity (ECext), N-NO3 – and S-SO 4 -2 were weak or inconsistent. ECa measurements successfully delimited three homogeneous soil zones associated to spatial distribution of fine particles (clay and silt), sand and OM content. However, it would not be advisable to differentiate management zones based on N-NO 3 -, SSO 4 -2 content and ECext because the values were very low and there was little variation between ECa classes. These results suggest that field-scale ECa maps have may be a useful tool for site-specific crop management (SSCM).
Key words. Precision agriculture; Management zones; spatial variability; Nutrients.
INTRODUCTION
The physical-chemical properties of the soil (texture, organic matter content, salt concentration, soil pH, among others), present spatial variability within the lot, which can influence the growth and development of crops and; therefore, the spatial distribution of grain yield (Johnson et al., 2001; Sudduth et al., 2003; Corwin et al., 2005). The spatial variability of soils is caused by interactions between physical, chemical and biological processes that act simultaneously with different intensity (Mallarino & Vittry, 2004). It must be considered that the uniform management of the lots does not take into account the existing variability, consequently, it is not an efficient management strategy (Moral et al., 2010). Thus, an understanding of the spatial distribution of the physical-chemical properties of the soil is important to apply site-specific management of crops (MSEC). It is defined as a subdivision of the lots in homogeneous areas to apply differential management depending on the potential of the site (Bullock et al., 2007), with the aim of improving productive efficiency and optimizing the use of inputs in agronomic terms, economic and environmental (Corwin & Lesch, 2010). A method used to sample spatial variability is grid sampling, which requires a dense quantity (0.2 ha) of samples, raising operating costs (Mallarino & Vittry, 2004). Another option is to find a cheaper and simpler method. The sampling by zones (MZ) would be a method to identify the variability of the soil; it assumes that the lot can be classified into homogeneous areas (referring to zones) that reflect differences between soil properties (Johnson et al., 2001). The MZ allows to reduce operating costs and improve efficiency, without losing information about the variability of the batch (Shaner et al., 2008). A method that is receiving a lot of attention to carry out the MZ is the georeferenced measurement of the Apparent Electrical Conductivity of the soil (EC The MZ allows to reduce operating costs and improve efficiency, without losing information about the variability of the lot (Shaner et al., 2008). A method that is receiving a lot of attention to carry out the MZ is the georeferenced measurement of the Apparent Electrical Conductivity of the soil (EC The MZ allows to reduce operating costs and improve efficiency, without losing information about the variability of the lot (Shaner et al., 2008). A method that is receiving a lot of attention to carry out the MZ is the georeferenced measurement of the Apparent Electrical Conductivity of the soil (ECa ) (Corwin et al., 2003; Moral et al., 2010).
The ECa is influenced by a combination of physicochemical properties of the soil, such as soil texture, organic matter content, soil moisture, cation exchange capacity, salinity, pH, Ca +2 and Mg +2 ., soil types, among others (Corwin & Lesch, 2005; Sudduth et al., 2005; Serrano et al., 2010; Terrón et al., 2011; Peralta et al., 2013). An additional advantage of this method is that the spatial distribution patterns of ECa do not change over time, so the delimited areas are repeatable over time, even under different soil conditions (Veris Technologies, 2001; Sudduth et al. al. 2001; Sudduth et al. 2003; Farahani et al., 2007). In Argiudoles and Paleudoles of Southeast Buenos Aires, Peralta et al., (2012, 2013) observed that the CEa is a potential estimator of the spatial variability of the clay content and soil moisture. Considering that the clay content is stable over time,a have the potential to delimit potential management zones to apply MSEC.
However, the application of CE a in MSEC may show weak and inconsistent relationships with soil characteristics (Corwin et al., 2003; Sudduth et al., 2005). These inconsistent relationships may be due to complex interactions between EC a and some soil properties. Therefore, it is necessary to understand which are the factors or properties of the soil that significantly influence the variation of EC a to characterize the spatial variability of the soil to apply the MSEC technology. The objective of this work was to evaluate if (i) the georeferenced measurement of the CEa is a potential estimator of soil properties (clay, silt, sand, OM, pH, and CEext) and nutrients (N-NO 3 – , S-SO 4 -2 and P); (ii) if the CE allows the delimitation of potential management areas in production lots in the Southeast of Córdoba. This information may be essential to successfully use EC georeferenced measurement to delimit production areas or management zones to allow producers to apply MSEC technology. Achieving thus, increase the efficiency in the use of inputs, improve the sustainability of the company, the protection of the environment and the economic benefit to the producer (Dinnes et al., 2002).
MATERIALS AND METHODS
Experimental sites
The mapping of the EC a and the soil sampling were carried out in July 2009, prior to sowing the wheat crop (Triticum aestivum). For this work, 2 plots located in the south of the province of Córdoba, close to the town of Canals, named: F1 (38 ha) and F2 (20 ha) were selected ( Fig. 1 ) . The 2 plots are cultivated under the direct sowing system since 2002, with a soybean (Soybean)-corn (Zea mays) rotation system for summer crops and in the winter season with wheat (Triticum aestivum), as cover. The predominant soils at the plot scale are Udortentic Haplustolls in the lower positions of the toposequence and Enthic Haplustols on the hills and foothills (Table 1 and Table 2 ). The climate of this region is characterized by a thermal regime with an average annual temperature of 17 °C and an amplitude of 14 °C. Average annual rainfall is 871 mm and the seasonal distribution is monsoonal (Ghida Daza et al., 2009). The two lots are located within the same Geographical area, and essentially represent two replicates in space for this region. The lots have little undulation with an average slope for lot F1 and F2 of 0.89 and 1.08%, respectively.
Figure 1. Experimental sites (lots) located in the Southeast of Córdoba, La Unión department . White dots within each batch represent apparent electrical conductivity (ECa) measurements.
Figure 1. Experimental sites (fields) located in Southeast Cordoba, La Union department. White points in each field represent apparent electrical conductivity (CEa) sample points.
Table 1. Soil classification.
Table 1. Soil classification.
INTA (1986). Soil map of the province of Córdoba (1:50000). Ministry of Agriculture, Livestock and Fisheries. (USDA-Soil Taxonomy V.2006).
Table 2. Depth and granulometry of each soil horizon (INTA, 1986).
Table 2. Depth and size distribution of each soil horizon.
Apparent Electrical Conductivity (ECa) Mapping
In each batch, ECa measurement was performed using the Veris 3100® (Veris 3100, Division of Geoprobe Systems, Salina, KS). The Veris is a contact device (the electrodes are in contact with the soil surface), it consists of 6 disc-shaped metal electrodes that penetrate approximately 5 cm into the soil (Fig. 2 ) . The two central discs emit a continuous electric current and simultaneously the other two pairs of electrodes detect the current gradient, given by transmission through the ground (resistance). The two central discs measure the EC at 0 to 30 cm depth; while the two extreme discs measure the ECfrom 0 to 90 cm. The Veris Data Logger performs the conversion from resistance to conductivity (1/resistance=conductivity). For the elaboration of the CE a maps, the CE at 0-30 cm (called CE a ) was used , coinciding with the depth of the georeferenced soil samples. The Veris unit was pulled by a pick-up and simultaneously it was measuring and georeferencing the EC to with a differential GPS (Trimble 132, Trimble Navigation Limited, USA) with submeter measurement precision and configured to take satellite position every 1-s. Each lot was covered in the direction of the sowing furrows in parallel transects spaced between 15-20 m because greater than 20 m generate measurement errors and loss of information (Farahani et al., 2007). The average forward speed ranged between 15 and 20 km h -1 .
Figure 2. Rear view of the veris 3100 for ECa determinations in the working position.
Figure 2. Rear end view of the Veris 3100 coulter-based apparent soil electrical conductivity sensor.
Classes of apparent Electrical Conductivity and determination of sampling points
The soil sampling carried out was by zones, based on three classes of ECa. Previous investigations in different soils recommend the division into three classes for the delimitation of homogeneous management zones because very little information is obtained using a larger number of classes (Fleming et al., 2000; Peralta et al., 2012). The values and width of the CE a classes were classified into quantiles (quartiles) of equal areas using the Geostatistical Analyst in ArcGIS 9.3.1. The sampling points were centered within the different zones from CE toto avoid transition zones. The average value of the ECa was determined within a diameter of 20 m from the sampling point. The ArcGIS 9.3.1 “Buffer” tool (Environmental System Research Institute, Redlands, CA) was used to create the 20-m diameter and obtain the average EC value a . The objective of this way of determining the sampling is to obtain data from the soil analysis that cover the entire range of the spatial variability of the ECa, in order to obtain reliable correlations between variables.
Soil sampling and analysis
Sampling was carried out with the sampler at a depth of 30 cm guided to said points manually with a Juno GPS (Trimble Navigation Limited, USA). The soil samples were dried in an oven at 60 ºC with forced air circulation for 10 to 16 hours, depending on the humidity of the sample. They were ground and sieved through a 2mm mesh. Subsequently, the particle size distribution was determined by the Bouyoucos method (Dewis & Freitas, 1970), the OM content by the Walkley & Black (1934) wet digestion method, the pH and the Electrical Conductivity of the extract at Saturation (ECext, ds/m) in a 1:2.5 ratio (soil-water). The content of N-NO 3 – was determined by the 2,4-phenoldisulfonic acid colorimetric method (Bremner, 1965). P and S-SO4 -2 content were quantified by extracting the soil solution with Mehlich-3 extractant (Mehchenlich, 1984) and analyzed with the Perkin Elmer Plasma System (PerkinElmer, Wellesley, MA).
Spatial variability of the Apparent Electrical Conductivity (CEa)
The spatial correlation of the CEa was quantified with semivariograms (Eq. 1). The semivariogram is a basic function that describes the spatial variability of a phenomenon of interest (CEa) and was estimated using the equation (Isaaks et al., 1989):
where; ã ∗ ( h )= value of the semivariogram in intervals of distance h ; z ( xi ) = value of the variable of interest at the point xi , in which there are data xi and xi + h ; N ( h)) is the total number of pairs of points within the distance interval. The semivariogram shows the degradation of the spatial correlation between two points in space, as the separation distance increases. In Samper and Carrera (1990) there is a discussion regarding the characteristics and conditions that they must meet. The best semivariogram model that adjusted to the spatial structure of the CEa was the spherical one (Peralta et al., 2013). For the classification of the spatial structure of the CE tothe partial plateau:nugget (partial sill: nugget) relationship was used, ie proportion of variance explained by the model with respect to the total variance (C1/(C0+C1)), adopting three classes proposed by Cambardella et al. (1994): strong (> 0.75), moderate (0.25-0.75), and weak (< 0.25). Subsequently, the CEa data was interpolated with the ordinary kriging procedure, which quantifies the spatial structure of the data using the semivariograms and statistically predicts them assuming that the data closest to a known point have greater weight or influence on the data. interpolation, an influence that decreases as you move away from the point of interest (Bullock et al., 2002).
Statistical analysis
To evaluate if the georeferenced measurement of the ECa allows delimiting homogeneous areas within the lots, we compared the differences in the averages of the soil properties (clay, Ac; silt, Li; sand, Ar; OM; pH; and ECext ) and nutrients (P, N-NO 3 -, S-SO 4 -2 ) in the different classes of ECa using the ANOVA of PROC MIXED (SAS Institute, 2002), with the classes of EC aas fixed effects, batches as random effects, and sampling points within each CEa class as a replicate (Littell et al., 1996). The comparison of means of soil properties was made with a significance level of p<0.05, using the LS-MEANS procedure (SAS Institute, 2002). We analyzed each batch as a location, each AEC class as a treatment in a randomized complete block design. Descriptive statistics for soil properties were calculated using the MEANS procedure (SAS Institute, 2002) and simple regressions between soil properties-CEa using the CORR procedure (SAS Institute, 2002).
RESULTS AND DISCUSSION
Structural analysis of CEa
The spatial dependence of CEa decreased progressively (equivalent to an increase in semivariance) with distance. CE a showed higher ranges of spatial dependence in lot F2 than in F1, and CE0-90 presented a higher spatial correlation than CE0-30 ( Table 3 ). The difference in the distance of the ranges can be observed in the size of the homogeneous zones of the CEa maps, where the largest zones are found in the CE0-90 map ( Fig. 3 ) . The similarity in the spatial patterns of the CE a measurements in both strata (0-30 and 0-90) ( Fig. 3), is attributable to the fact that the spatial variability is due to characteristics inherent to the soil of the study area (Sudduth et al. 2001; Sudduth et al. 2003). According to the classification proposed by Cambardella et al. (1994), the special dependence of ECa within each batch can be considered moderate to strong ((C1/ (Co+C1)> 0.5)), meaning that the models were mainly explained by spatial variability and not by sampling errors (Chang et al., 1999).
Table 3. Parameters of the models of the semivariograms for the apparent electrical conductivity (ECa) in each lot.
Table 3. Parameters of the semivariogram models for apparent electrical conductivity (CEa) at each field.
Figure 3. Map of apparent electrical conductivity (ECa) for each batch. Figure 3. Apparent electrical conductivity (CEa) map for each field.
Descriptive statistics of the ECa and soil
properties The descriptive statistical analysis was carried out to know the variation of the ECa, soil properties and nutrients ( Table 4 ). The average EC a for batch F1 and F2 was 47.22 (mS m -1 ) and 45.17 (mS m -1 ), respectively. Similar EC values between lots is attributable to the fact that they are composed of the same types of soil (Shaner et al., 2008). Peralta et al. (2013), mentioned that differences in the average values of the CE at, are associated with different types of soil, mainly due to the difference in the size of soil particles. The CV for batch F1 and F2 was 27.56 and 30.53%, respectively. The variation in ECa confirms the possibility of using it to detect soil variability, as has already been widely verified in different locations in other countries (Corwin et al., 2006; Moral et al., 2010).
Table 4. Average value (Average), coefficient of variation (CV), minimum (Min) and maximum (Max) of the soil and nutrient properties for each lot.
Table 4. Average values (Average), coefficient of variation (CV), minimum (Min) and maximum (Max) values of soil properties and nutrients at each field.
ECa: Apparent Electrical Conductivity; Ac: clay; Li: slime; Ar: sand; P: Phosphorus Bray; ECext: Electrical Conductivity of the extract at saturation.
The criteria suggested by Wilding et al. (1994) was used to characterize the magnitude of the variability of soil and nutrient properties, with a coefficient of variation (CV) of 0 to 15, 15 to 30 and 35 to 100% characterizing low, medium and high variability, respectively. . All soil and nutrient properties presented medium to high variability, except soil pH ( Table 4). These results are within the CV reported in the bibliography. Alesso et al. (2012) in a study of macroplots obtained a CV lower than < 5% for soil pH. The concentration of P presented a CV > 35% for both batches. In the last 10 years, the tillage system has been direct sowing and the application of P to cover the requirements of the crops was carried out in bands. The residual P in the application bands generates a wide variation (Kitchen et al., 1990; Mallarino et al., 1996), since it is not homogeneously applied throughout the lot. The variation in the properties of the soil and nutrients is an indicator that the uniform management within the lots may be inefficient (Corwin et al., 2003).
Correlation between ECa and soil properties Simple correlation analysis is used to analyze ECa
data in Precision Agriculture to determine the predominant soil factor influencing ECa within the field (Corwin et al. ., 2003; Kitchen et al., 2003). The contents of fine (Ac+Li), coarse (Ar) and OM particles were the soil properties that presented highly significant correlations (p<0.001, Table 5 ) in each lot. The content of fine particles was positively correlated with the ECa ( Table 5). The conduction of electricity in soils takes place mainly through continuous, water-filled macro- and micropores between soil particles. Soils with a higher percentage of fine particles have a significant particle-particle contact and a greater number of small pores that retain water more strongly and for a longer time, thus allowing them to better conduct electricity (Rhoades et al., 1989), compared to with the soils that have a higher content of sand particles (Farahani et al., 2007; Shaner et al., 2008;). The positive relationship between CE to and fine particle content is consistent with previous studies (Sudduth et al., 2003; Corwin et al., 2006). In addition, the fine particle contents were positively associated with the OM content of the soil (0.71; p<0.001). Some OM components are mainly responsible for the formation and stabilization of soil aggregates, generating continuous pores and macropores (Lal, 2004), increasing the soil’s ability to conduct electric current. Martínez et al., (2009), explained that the strong correlation between CE to– OM is due to the fact that OM plays a significant role in the maintenance of the physical properties of the soil and, that OM is associated with an accumulation of nutrients and water retention, factors that are directly related to EC a .
Table 5. Correlation (r) between Apparent Electrical Conductivity (ECa), soil properties and nutrients for each plot.
Table 5. Correlations (r) between apparent electrical conductivity (CEa), soil properties and nutrients at each field.
Significance level: *, **, ***, p< 0.05; 0.01; 0.001, respectively. ns, differences not significant. ECa: Apparent Electrical Conductivity; Ac: clay; Li: slime; Ar: sand; P: Phosphorus Bray; ECext: Electrical Conductivity of the extract at saturation.
The relationship between EC a with EC ext and soil pH were weakly significant and positive ( Table 5 ). The relationships found by other authors in non-saline soils are generally low and inconsistent (Peralta, pers. comm.; Johnson et al., 2001). While in saline soils the spatial variability of the salt concentration is the main soil factor that affects the measurement of EC a (Corwin et al., 2005).
The concentrations of the anions (N-NO3 – and S-SO 4 -2 ) did not show an association with CE a , except for S-SO 4 -2 in lot F1 ( Table 5). The poor correlation is probably because other anions may be dominating the ECa measurement (Johnson et al., 2001; Fox, 2004). Zhang et al., (2002) and Corwin et al., (2005) found strong correlations between CE a – N-NO 3 – and CEa – S-SO 4 -2 , working in soils with higher concentrations and variations. The P content presented a negative and significant correlation with the ECa ( Table 5 ). The identification of correlations greater than 0.6 indicate that the ECa can be used successfully to estimate the variability of soil and nutrient properties (Heiniger et al., 2003). As has been seen, the CE a showed strong associations (r> 0.6) with fine (Ac+Li), coarse (Ar) particles, OM content, and, to a lesser extent, with ECext, pH, and P content. CE a was not correlated with the contents of N-NO 3 – and S-SO 4 -2 .
Delimitation of potential zones
To evaluate if the georeferenced measurement of ECa allows delimiting homogeneous zones within the lots, we compared the averages of soil and nutrient properties in the classes of EC a using the PROC MIXED ANOVA. All soil properties (Ac, Li, Ar, MO, pH, and ECext) and nutrients (P, N-NO3 -, S-SO4 -2) showed interaction between location (lots) and EC classes a ( p < 0.05, Table 6), therefore, we analyze the behavior of these variables by batch. The content of Ar and fine particles presented significant differences between the classes of CEa in each batch. The content of fine particles and Ar presented differences in three classes of CE a , except Ar in F2, where there were no differences between the medium-high classes. Previous investigations reported that the EC a is influenced by the content of fine particles and Ar, which reflects the water retention capacity of the soil, and therefore, the spatial variation of grain yield of crops (Herber, 2011). ). The OM percentage of the soil presented the same trend as the fine particles, but with smaller differences between ECa classes (Table 6 ). The low class of CE a was associated with areas of the lot with higher contents of fine particles and OM; and the high class of CE a associated with high sand and low OM contents. These results are due to the fact that fine particles were positively related to OM content ( Table 5 ) (Peralta et al., 2013). By increasing the water retention capacity of soils, as a result of finer soil particles, their OM content increases as a consequence of the greater contribution of residues they receive and due to the protective effect of fine particles on OM (Parton et al. al., 1993). The ECext and pH only exhibited significant differences in the high class of EC to, except for the pH in F1, which did not present significant differences ( Table 4 ). The highest values of these soil properties were associated with the high class of CE a ; and vice versa. Peralta et al., 2012, reported that for soils in the Southeast of Buenos Aires, without salinity problems, the salt content and pH are not soil properties that significantly affect the EC a variability .
Table 6. Average of the soil and nutrient properties for the different classes of Apparent Electrical Conductivity (ECa).
Table 6. Soil properties and nutrient means for different classes of apparent electrical conductivity (CEa) at each field.
Ac: clay; Li: slime; Ar: sand; P: Phosphorus Bray; ECext: Electrical Conductivity of the extract at saturation.
On the other hand, the concentrations of the anions (NNO 3 – and S-SO 4 -2 ) did not show significant differences between classes of CE a ( Table 6) , making it difficult to separate zones because the transformation of these nutrients in the soil It is controlled by moisture content, biological activity, composition, and amount of organic matter. These soil characteristics have an impact on the immobilization and washing or mineralization processes that define the levels of N-NO 3 – and S-SO 4 -2in the soil (Stevenson, 1982; Ericksen, 1997a). In contrast, the content of P showed significant differences between classes of CEa ( Table 6 ), finding the highest content of P in the low class of CE a ; and vice versa. The lowlands from CE toare associated with areas with sandy hills (Entic Haplustoles) with low water retention capacity (Urricariet et al., 2011) and OM, generating areas with low yield potential, and therefore, low P extraction. While the high class of CEa, was associated with low areas of the land, with greater potential fertility and yield, generating areas with an extractive history of P. Bermúdez (2011), observed in lots in the West of Buenos Aires, sandy hills with 20-22 ppm, while the lower areas of the lot (with higher contents of fine particles) presented 12-14 ppm.
This possibility of detecting the variability of the soil, using the EC toit provides an effective basis for delimiting soil properties or attributes that are interrelated and offers a very useful framework for soil sampling, reflecting spatial heterogeneity (Moral et al., 2010; Peralta et al., 2012).
CONCLUSION
The content of fine particles (Ac+Li), Ar, MO, P and the pH values were the soil properties that presented significant correlations with the georeferenced measurement of EC a . Consequently, the EC a is a potential estimator of the spatial variability of these soil properties. While the ECext, the N-NO3 – and S-SO4 -2 contents presented weak and inconsistent correlations with the CEa, making the CEa a weak estimator of said properties.
Whereas the ECallows to characterize the spatial variability of the contents of fine particles (Ac+Li), which are stable over time and are related to the physical-chemical fertility of the soils; These results suggest that the CE a maps have the potential to delimit three MSEC zones and design a sampling by zones, managing to reduce sampling costs without losing information on the spatial variability of the soils.
The application of variable doses of inputs, either fertilizer and/or seed, can be carried out, reducing the inputs in the least productive zone (lower class from CE to−associated with high sand contents and low water retention capacity−), as a strategy to increase the efficiency in the use of inputs, and the economic benefit to the producer, compared to the uniform handling of lots (Dinnes et al. , 2002). In addition, other studies can be carried out to compare this purpose of delimiting management zones, with yield maps to better understand the agronomic importance of this classification.
THANKS
The authors express their gratitude to Ing. Agr. Julián Muguerza and Sebastián Storti, and the Aceitera General Deheza for their help in compiling the data in the lots where the study was carried out. In addition, we want to thank the postgraduate scholarship program of the National Council for Scientific and Technical Research (CONICET); INTA Project, Development and Application of Precision Agriculture Technology for Crop Management (AEAI3722), and to the two reviewers who with their suggestions allowed us to substantially improve the work.
BIBLIOGRAPHY
- Alesso, CA; MA Pilatti; SC Imhoff & M Grilli. 2012. Spatial variability of chemical and physical attributes in a soil from the Santa Fe plains. Soil Science 30(1): 85-93. [ Links]
- Bermúdez, M. 2011. How to implement site-specific N and P management on a large scale. Fertility Symposium Pp. 47-54. [ Links]
- Bullock, D.S.; N Kitchen & D. G. Bullock. 2007. Multidisciplinary Teams: A Necessity for Research in Precision Agriculture Systems Crop Sci. 47: 1765-1769. [ Links]
- Cambardella, CA; T. B. Moorman; JM Novak, TB Parkin; RF Karlen & AE Konopka. 1994. Field scale variability in central Iowa soils Soil Sci. Soc. Am. J. 58: 1501-1511. [ Links]
- Chang, J.; DE Clay; CG Carlson & S Clay. 1999. Precision farming protocols: Part 1. Grid distance and soil nutrient impact on the reproducibility of spatial variability measurements. Precise. Agric. 1:227-289. [ Links]
- Corwin, D.L.; SM Lesch; PJ Shouse; R Soppe & JE Ayars. 2003. Identifying soil properties that influence cotton yield using soil sampling directed by apparent soil electrical conductivity. Agron. J. 95(2): 352-364. [ Links]
