Explaining the Soil Quality Using Different Assessment Techniques

,


Introduction
Soils in agriculture are an important part of the ecological system that produces food and fber for human consumption, but they are a limited and largely non-renewable resource [1,2]. Soils are a key enabling resource and essential to the production of a wide range of goods and services integral to ecosystems and human well-being [3,4]. Nonetheless, soil fertility depletion caused by a variety of factors (soil erosion, acidity, nutrient depletion, lack of soil fertility replenishment, nutrient mining, and lack of balanced fertilization) is a signifcant contributor to food insecurity [5,6].
Soil quality (SQ), which is defned as the capacity of soil to function within the ecosystem and land use boundaries to sustain biological productivity, maintain environmental quality, and promote plant, animal, and human health, is now highly related to sustainable and productive agriculture [2,7,8]. Good-quality soils will preserve natural ecosystems by improving air and water quality for improved food and fber production while also protecting the environment and human health [9].
Te SQ simultaneously addresses the issues of productivity and sustainability and makes it indispensable for developing countries such as Ethiopia [2,4]. A better understanding of the SQ and the factors that degrade the SQ is necessary to fully exploit the potential benefts of soil resources. For example, poor soil physical and chemical health is very likely to result in poor aggregate stability, a decline in soil OM, nutrient-related plant stresses, crop yield stagnation, and exacerbate soil degradation [10,11]. Tis suggests that SQ is linked to chemical properties, biophysical environments, and anthropogenic factors. Meanwhile, SQ cannot be measured directly in the feld or laboratory; rather, it is inferred from measured soil physical, chemical, and biological properties and is thus expressed in terms of soil quality index (SQI) [2,8,12].
Te SQI could be defned as a minimum set of parameters that provides numerical data about a soil's ability to perform one or more functions [13]. It aids in assessing overall soil condition and management response or resilience to natural and anthropogenic forces [1,7,14,15]. Expert opinion (subjective) or mathematical and statistical (objective) methods are used to select a minimum soil data set (MDS) [13,16]. Te use of multivariate techniques of principal component analysis (PCA) (multiple correlations and factor analyses) to reduce statistical data has become more common [12,17]. Tus, the SQI, which takes into account the physical, chemical, and biological properties of soils as well as their variability, is critical for long-term utilization and site-specifc management of soil resources [2,8,12,14,15].
Despite the importance of SQ assessment, very few studies have been conducted on smallholder arable lands in Ethiopia where traditional practices dominate soil management [2]. Tis emphasizes the importance of having adequate soil property information in order to intervene and prevent soil fertility degradation problems. Against this backdrop, the present study aimed to explore the soil quality status of farmlands belonging to diferent soil groups using diferent varied approaches.

Description of the Study Area.
Te study sites were Farawocha farm in Wolaita Zone and Kechi farm in Dawro Zone, Southern Ethiopia ( Figure 1). Farawocha farm lies between 7°6′34″N to 7°9′0″N latitude and 37°34′54″E to 37°37′33″E longitude. Te farm has 3.85 ha (cultivated land) within an average altitude of 1500 m.a.s.l and a slope of less than 3%. Ten years (2010-2019) mean annual precipitation is 1300 mm, and the monthly temperature fuctuates between 13.8 and 25.3°C with an average of 19.6°C ( Figure 2) [18]. Kechi farm lies between 7°1′7″N and 7°5′48″ N latitude and 36°57′5″E and 37 0 0′25″E longitude with an average altitude of 2090 m.a.s.l. It has a total area of 131.26 ha of which the cultivated land shares 32.66 ha, grass land (5.8 ha), and forest land (92.8 ha). Kechi farm lies from a gentle to the steep slope. Ten years (2010-2019) mean annual precipitation was 1502 mm and the monthly temperature fuctuates between 14.5 and 24.2°C with an average of 19.3°C (Figure 3) [18]. According to WRB [19] and FAO [20], the soil types of the Farawocha farm and Kechi farm are grouped under the Nitisols and Luvisols, respectively.

Soil Sampling Procedure and Analysis
2.2.1. Soil Sampling Procedure. Various tasks, including prefeld work, feldwork, and postfeld work stages, were completed prior to sample collection. Prior to sample collection, sample points to the study area shape fle were assigned in grid patterns using geographical information system (GIS). While conducting the survey, a geographical positioning system (GPS) receiver was used to fnd the sample locations. A total of 57 geo-referenced points were used to collect surface soil samples at a depth of 0-20 cm (18 from Nitisols and 39 from Luvisols). Ten subsamples from each sample were combined to create one kilogram of composited soil.

Soil Sample Preparation and Analysis.
Following the standard procedure outlined in Sahlemedhin and Taye [21], soil samples were processed (air-dried, ground, and passed through a 2 mm sieve), and some soil physicochemical properties were examined (2000). Tis includes soil pH, organic carbon (OC), total nitrogen (TN), available phosphorus (P) and sulfur (S), exchangeable bases (calcium (Ca), magnesium (Mg), and potassium (K)), soil micronutrients (boron (B), copper (Cu), iron (Fe), manganese (Mn), and zinc (Zn)), cation exchange capacity (CEC), and texture (particle size distribution). Soil pH (1 : 2.5 soil: water suspension) was measured with a glass electrode (ES ISO 10390 : 2014). Total N was determined by the wet-oxidation (wet digestion) procedure of the Kjeldahl method (ES ISO 11261 : 2015). Organic carbon (OC) was determined following the wet combustion method of Walkley and Black. Available P and S, exchangeable basic cations (Ca, Mg, and K), and extractable micronutrients (B, Cu, Fe, Mn, and Zn) were determined using the Mehlich-III multinutrient extraction method [22]. Te CEC was determined by using the 1 N ammonium acetate (pH 7) method. Particle size analysis was carried out by the hydrometer method as described by [21]. Textural classes were determined by Marshall's triangular coordinate system.

Soil Quality Assessment.
Since soil quality cannot be directly measured, it is inferred from other soil properties and expressed as the soil quality index (SQI) [8,12,23]. Te approaches discussed were used in the study to assess the soil quality: [7,9,24]. Te process involved three main steps: (i) selecting appropriate indicators; (ii) converting indicators into scores; and (iii) combining the scores into an index [13,25].

SQI Estimate Using an Additive System Based on Common Soil Parameters
where RSTC � assigned ranking values for soil textural class; RpH � assigned ranking values for soil pH; ROC � assigned ranking values for soil organic carbon; RNPK � assigned ranking values for nitrogen (N); phosphorus (P), and potassium (K) ( Table 1)  Applied and Environmental Soil Science are multiplied by 1, 2, and 3, respectively. If the index value is less than 1.67, the fertility status is low; if the index value is between 1.67-2.33, the fertility status is medium; and if the index value is greater than 2.33, then the fertility status is high [25].
where N L � number of samples in low category; N M � number of samples in the medium category; N H � number of samples in high category, and N T � total number of samples.

Principal Component Analysis (PCA) Based SQI (Statistical Model-Based SQI).
A statistics-based model is used to estimate SQI using PCA [17,26]. Te PCA method is more objective because it makes use of a variety of statistical tools (multiple correlation, factor, and analyses), which could prevent bias and data redundancy by selecting a minimal dataset (MDS) using formulas [12]. Te PCA model included all the original observations of each soil parameter.
Very poor Poor Fair Good Best Source: Bajracharya et al. [24], where C-clay;S-sand;CL-clay loam; SC-sandy clay; SiC-silty clay; Si-silt;LS-loamy sand; L-loam;SiL-silty loam; SL-sandy loam; LS-loamy sand; SiL-silty loam; SL-sandy loam; SiCL-silty clay loam; SCL-sandy clay loam; SQI-soil quality index. Note. Te ranges for which each of the parameter values are assigned are based upon corresponding ratings from low to high levels following the appropriate standard rating.
Te PCs with high eigenvalues represented the maximum variation in the dataset, while most studies have assumed to examine PCs only the variables having high factor loadings with eigenvalues >1.0 that explained at least 5% of the data variations were retained for indexing [12,17].
Under a given PC, each variable had a corresponding eigenvector weight value or factor loading. Only the "highly weighted" variables were retained in the MDS. Te "highly weighted" variables were defned as the highest weighted variable under a certain PC and absolute factor loading value within 10% of the highest values under the same PC [12,23]. However, when more than one variable was retained under a particular PC, a multivariate correlation matrix is used to determine the correlation coefcients between the parameters. If the parameters were signifcantly correlated (r > 0.70), then the one with the highest loading factor was retained in the MDS and all others were eliminated from the MDS to avoid redundancy.
Still, the normalized PCA of SQI would be calculated if more than one highest eigenvectors were retained in the MDS [12,23]. Te noncorrelated and highly weighted parameters under a particular PC were considered important and retained in the data. Each PC explained a certain amount of variation in the dataset, which was divided by the maximum total variation of all the PCs selected for the MDS to get a certain weightage value under a particular PC [12,26]. Tereafter, the SQI-3 (PCA) was computed using the following equation: where PC Weight is the weightage factor determined from the ratio of the total percentage of variance from each factor to the maximum cumulative variance coefcients of the PC considered; individual soil parameter score is the score of each parameter in the MDS.

Data Analysis.
Description of data analysis was performed using Microsoft Excel. All these values presented as mean, minimum, maximum, SD, CV, PCA, and MDS selections were performed using statistics-8 and Microsoft Excel software. In addition, Pearson correlation analysis on selected parameters was performed.

Soil Quality Index.
Based on the common soil parameter approach, the SQI values for the soils taken from Nitisols and Luvisols were 0.17 (Table 2) and 0.30 (Table 3). Te shares of each indicator in the Nitisols lands were 0.04 (soil textural class), 0.04 (soil pH), 0.08 (soil OC), and 0.005 (N-P-K) ( Table 2); and 0.08 (soil textural class), 0.06 (soil pH), 0.12 (soil OC), and 0.03 (N-P-K) in Luvisols lands (Table 3). According to Bajracharya et al. [24], the SQI was classifed as very poor quality if the ranking value was less than 0.2, poor if it was between 0.2 and 0.4, fair if it was between 0.4 and 0.6, good if it was between 0.6 and 0.8, and best if it was between 0.8 and 1 (Table 4). Te soil quality of lands in the Nitisols was rated very poor (<0.2), whereas the soil quality of Luvisols lands was poor (0.2-0.4) [7,9]. Te evaluation using the soil fertility/nutrient/index method also revealed values of 1.42 and 1.78 for Nitisols and Luvisols situated lands, respectively (Table 5). Te index value is rated low fertility status, if less than 1.67, medium [1.67 and 2.33], and high [>2.33]. Terefore, the soil fertility status of the lands found in the Nitisols was low in soil fertility while it was medium in the Luvisols lands [25].
Five principal components (PCs) with eigenvalues >1 were identifed by the PCA SQI in both Nitisols and Luvisols land, accounting for 89.3% and 81% of the total variation, respectively (Table 6). PC 1 and 2 accounted for 52.6% of the total variation (36.2% and 16.4%) in Luvisol lands and 59.6% of the total variation (39.1% and 20.5%) in Nitisols lands. Eight soil parameters, including S, Mg, Na, B, Cu, Fe, Mn, and Zn for the land in the Nitisols and fve soil parameters, including pH, Ca, PBS, B, and Fe in Luvisols land from PC 1, were correlated (bolded parameters) to observe their close interrelationship and to choose for the minimum data set (MDS) ( Table 6). Tus, the highest factor loadings from each PC analysis were found for six parameters, including silt, pH, OC, Ca, B, and Zn for samples taken from lands located in Nitisols and fve parameters, including TN, S, Ca, Mg, and Mn in the Luvisols lands (bold underlined) ( Table 7). Tese parameters were then retained in the MDS (Table 7). In addition, for normalized PCA-based SQI estimation, the MDS was retained following the approach indicated by Tesfahunegn [23]; Podwika et al., [12] (Table 8). Subsequently, the estimated SQI values following the PCA and normalized PCA techniques (Tables 7 and 8) for the soils belonging to the Nitisols revealed 0.42 and 0.36, whereas the values were 0.40 and 0.38 for the Luvisols, respectively (Table 8).
According to Li et al. [17], grading values for PCA and normalized PCA-SQI values were very low (<0.38), low (0.38-0.44), moderate (0.45-0.54), high (0.55-0.60), and very high (>0.60) soil quality. In view of this, the soil quality belonging to Nitisols using the PCA and normalized PCA approaches were rated very low and low, respectively, whereas the soil quality of lands belonging to Luvisols was qualifed as low level (0.38-0.44).
Overall, the estimated SQI values of both soil types using various techniques demonstrated poor-quality soils (Table 9). About 50% of the essential nutrients that come from the soil, N, P, S, Ca, Mg, and B, were low from soil samples taken in the Nitisols, and 36% of the nutrients, N, P, K, S, and B, from lands in the Luvisols were found to be inadequate. In both soil types, the soil pH was strongly acidic, which is problematic for nutrient availability and microbial activity.   A further restriction in both soil types was their low organic matter levels. A lack of organic matter input and the removal of basic nutrients caused the soil to become more acidic, which in turn accelerated nutrient defciencies [6,29]. According to the fndings, problem-focused soil management interventions are urgently required [12,13].

Conclusion
Diferent techniques, including principal component analysis (PCA), common soil parameters, and the soil fertility/ nutrient/index approach, were employed to estimate the soil quality. All evaluation techniques for the lands belonging to Nitisols consistently demonstrated comparable soil quality status, whereas PCA and common soil parameter techniques generated similar results for the Luvisols lands. Based upon the consistency of the outcomes generated in both soil types, the use of PCA and the common soil parameters approach could be taken as useful tools to assess soil quality. Furthermore, it was noted that low soil quality necessitates the use of management interventions.

Data Availability
Te data used to support the fndings of this study are available from the corresponding author upon request.