Conversion of soil pH values and exchangeable base cation quantities determined by different methods

Date
2021-12
Journal Title
Journal ISSN
Volume Title
Publisher
Stellenbosch : Stellenbosch University
Abstract
ENGLISH ABSTRACT: Soil management recommendations based on chemical test results rely on comparisons with norm values developed using specific analytical methods and soil taxonomic classes. Conversions of soil pH and exchangeable base cation concentration between different methods are required for the standardization of data from scientific literature, soil test reports and soil databases. The combined effects of soil physicochemical properties on the accuracy and parsimony of soil pH and cation concentration conversions are unknown. Rapid soil analyses can be obtained by infrared (IR) spectroscopy, but few studies have evaluated the accuracy of IR spectral-based models to determine soil pH and exchangeable base cations in Western Cape soils. The aim of this study was to develop equations for conversion of soil pH values and exchangeable base cation concentrations between different methods of measurement. An additional aim was to develop IR spectral-based models for determination of soil pH and exchangeable base cation content of Western Cape soils. Soil pH was measured in different matrices (water, 0.1M KCl and 0.01M CaCl2) of varying soil: solution ratios (1:1, 1:2.5, 1:4 and 1:5). The exchangeable base cation content was measured according the ammonium acetate (1M, pH = 7), compulsive exchange, ammonium acetate (0.2M, pH = 7), Ambic- I, Mehlich-III, Bray-II and Citric acid (1%) methods. Models converting soil pH and cation content between values obtained by different methods were developed using simple linear regression (SLR). In addition, multivariate linear regression (MLR) equations incorporating soil physicochemical properties as parameters were developed. Spectral measurements of the soil samples were taken with three spectroscopic instruments in the near- to mid-IR range. IR spectral-based models predicting exchangeable base cation content and soil pH were calibrated using partial least squares regression. SLR equations for predicting soil pH (RMSE = 0.12-0.32) were more accurate than the conversion factors. MLR equations constructed by stepwise regression to convert exchangeable base cation content (RMSE = 1.4-1.5 cmolc kg-1, 0.4 cmolc kg-1, 16-20 mg kg-1 and 27-39 mg kg-1 for Ca, Mg, Na and K, respectively) were more accurate than SLR models. Exchangeable K content was accurately converted between the most extraction methods. Models converting base cation content may be chosen based on the purpose of application and relative size of the prediction error. IR spectroscopy is recommended to determine exchangeable Ca and Mg content in Western Cape soils (RMSE = 1.2-1.4 cmolc kg-1 and 0.37 cmolc kg-1 for Ca and Mg, respectively). Subsetting soil sample sets by similarities in texture and organic carbon content led to more accurate predictions of soil pH and exchangeable Ca content (RMSE decreased by up to 0.19 pH units and 0.79 cmolc kg- 1, respectively). In summary, use of the developed SLR as well as MLR models are recommended to convert soil pH and exchangeable base cation concentrations of Western Cape soils. Models should only be as complex as practical context requires. Moreover, IR spectroscopy may be used for determination of exchangeable Ca and Mg content in Western Cape soils.
AFRIKAANSE OPSOMMING: Grondbestuurshandleiding wat op chemiese toetsresultate gebaseer is, is afhanklik van vergelyking met kritiese drempelwaardes wat d.m.v. spesifieke ontledingsmetodes en grondtipes ontwikkel is. Omskakeling van grond pH-waardes en uitruilbare basiese katiooninhoud tussen verskillende metodes is vir standaardisering van data wat in die wetenskaplike literatuur, grondtoetsverslae en gronddatabasisse voorkom, nodig. Die saamgestelde effekte van grondfisiochemiese eienskappe op die akkuraatheid en parsimonie van pH- en katiooninhoud-omskakelings, is onbekend. Grondontledings kan vinnig m.b.v. infrarooi-(IR)-spektroskopie bepaal word, maar daar is min studies wat die akkuraatheid van IR modelle vir die bepaling van grond pH-waardes en uitruilbare katiooninhoud in Wes-Kaapse gronde geëvalueer het. Die doel van hierdie studie is om vergelykings vir die omskakeling van grond pH-waardes en uitruilbare basiese katiooninhoud tussen verskillende metingsmetodes, te ontwikkel. ’n Addisionele doel was om IR spektraal-gebaseerde modelle vir die bepaling van grond pH- waardes en uitruilbare basiese katiooninhoud in Wes-Kaapse gronde, te ontwikkel. Grond pH-vlak is in verskillende matrikse (water, 0.1M KCl en 0.01M CaCl2) met verskeie grond: oplossing verhoudings (1:1, 1:2,5, 1:4 en 1:5) gemeet. Die uitruilbare basiese katiooninhoud is volgens die ammoniumasetaat (1M, pH = 7), kompulsiewe uitruil, ammoniumasetaat (0.2M, pH = 7), Ambic-I, Mehlich-III, Bray-II en Sitroensuur (1%) metodes gemeet. Modelle is d.m.v. eenvoudige lineêre regressie ontwikkel vir die omskakeling van die grond pH-vlak of katiooninhoud tussen waardes wat deur verskillende metodes verkry word. Meerveranderlike lineêre regressie vergelykings wat verskillende grondfisiochemiese eienskappe insluit as veranderlikes, is ook ontwikkel. Spektrale metings van die monsters is met drie verskillende spektroskopiese instrumente in die nabye tot middel-IR omvang geneem. IR spektraal- gebaseerde modelle vir die voorspelling van uitruilbare basiese katiooninhoud en grond pH-vlak is d.m.v. gedeeltelike regressie gekalibreer. Eenvoudige lineêre vergelykings vir die bepaling van die grond pH-vlak (RMSE = 0.12-0.32) was akkurater as omskakelingsfaktore. Meerveranderlike modelle vir die omskakeling van uitruilbare basiese katione wat d.m.v. stapsgewyse regressie ontwikkel is (RMSE = 1.4-1.5 cmolc kg-1, 0.4 cmolc kg-1, 16-20 mg kg-1 en 27-39 mg kg-1 vir Ca, Mg, Na and K, onderskeidelik), was akkurater as die lineêre regressie modelle. Uitruilbare K inhoud was tussen die meeste ekstrasie metodes akkuraat omgeskakel. Modelle vir die omskakeling van basiese katiooninhoud kan gekies word op grond van die doel van toepassing en relatiewe grootte van die voorspellingsfout. IR spektroskopie word aanbeveel vir die bepaling van uitruilbare Ca- en Mg-inhoud in Wes-Kaapse gronde (RMSE = 1.2-1.4 cmolc kg-1 en 0.37 cmolc kg-1 vir Ca and Mg, onderskeidelik). Onderverstelling van grondmonsterstelle volgens soortgelykhede van grond tekstuur en organiese koolstof-inhoud, het gelei tot akkurater bepaling van grond pH-waardes en uitruilbare Ca-inhoud (RMSE het afgeneem tot en met 0.19 pH eenhede en 0.79 cmolc kg-1 onderskeidelik). Ter afsluiting word gebruik van die ontwikkelde lineêre regressie sowel as multiveranderlike modelle voorgestel vir die omskakeling van grond pH-vlak en uitruilbare basiese katione in Wes-Kaapse gronde. Modelle moet slegs so kompleks wees as wat praktiese konteks vereis. Boonop kan IR spektroskopie gebruik word vir die bepaling van die uitrtuilbare Ca- en Mg-inhoud in Wes-Kaapse gronde.
Description
Thesis (MScAgric)--Stellenbosch University, 2021.
Keywords
Soil chemistry, Soil fertility, Soil pH, Exchangeable cations, Soils -- Analysis, UCTD
Citation