Multivariate modeling of some metals concentrations in agrarian soils: distribution and soil fertility implications in the tropics

Predicting metals concentration in agricultural soils is a sine qua non in establishing environmental policies and evaluating the soils’ agricultural potentials in an area. The relevance of metals to ecological health, agriculture and pollution has sprung a lot of related studies. This study was setup to determine the concentration and profile distribution of aqua regia (AR) extractable Fe, Al, Mn, Mg and K in agricultural soils, and to predict AR extracted elements via Al2O3 (Alx), K2O (Kx), physical and chemical properties for soil fertility interpretations. One soil pit was randomly sited in each slope transition obtained via digital elevation models (DEM), resulting in 27 composite soil samples. Soil samples meant for AR and X-ray florescence were analyzed in triplicate. The soils were dominated by AR extractable Fe with mean concentrations showing the trend; Fea > Ala > Mga > Mna ≈ Ka and ranges of 639.09–125,719.46, 1252.63–14,895.13, 67.61–2408.36, 4.51–2162.91 and 161.84–1356.23 mg/kg, respectively. The distribution of AR metals in the entire soils was quite similar, however, higher values of soluble Fe occurred in the 0–37 cm depth of IH1P1. Multiple linear regression functions were within acceptable and best prediction criteria (R2 = 0.55–0.77). The best performing models were Ka and Mna, with lower errors. The models selected Kx, Mg and CEC which contributed 89.9, 79.9 and 73.4%, respectively to the 44.2% contribution of PC1 to data variation. The dominance of Kx and Alx with ranges of 2381.0–50,401.0 and 57,766.67–119,433.35 mg/kg, respectively, over Ka and Ala is due to limitations associated with AR extraction of elements in silicate minerals, hence the necessity for extracting soil mineral elements by more than one method