{"title":"The effects of climate and soil properties on the magnitude of the visual soil quality indicators: a logistic regression approach","authors":"F. Teixeira","doi":"10.3934/geosci.2023027","DOIUrl":null,"url":null,"abstract":"Understanding how different climates and soil properties affect the soil processes requires quantifying these effects. Visual soil quality indicators have been proposed to assess the robustness of the soil processes and infer their ability to function. The scores of the visual soil quality indicators covary with climate features and soil properties, and their magnitude is different in acid-to-neutral and alkaline soils. These variables show collinearities and interactions, and the assessment of the individual effect of each variable on the scores of the visual indicators and the selection of the best set of explanatory variables can only be made with a definite set of variables. Logistic regression was used to calculate the effects of six climate variables and four soil properties, and their interactions, on the scores of eight visual soil quality indicators. Simple models featuring climate and soil variables explained a substantial part of the variation of the visual indicators. Models were fitted for each visual indicator for acid-to-neutral and alkaline soils. The sample size needed was calculated, and the method and its validity were discussed. For two possible outcomes, the sample size using the events per variable (EPV) criterium ranges between 62 and 183 observations, while using one variable and a variance inflation factor, it ranges between 22 and 234. Except for the model of soil structure and consistency for acid-to-neutral soils, with a C statistic of 0.67, all others had acceptable to excellent discrimination. The models built are adequate, for example, for the large-scale spatial outline of the soil health indices, to couple with soil morphological-dependent pedotransfer functions, and so on. Future models should consider (test) other explanatory variables: other climate variables and indices, other soil properties and soil management practices.","PeriodicalId":43999,"journal":{"name":"AIMS Geosciences","volume":null,"pages":null},"PeriodicalIF":0.9000,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"AIMS Geosciences","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.3934/geosci.2023027","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"GEOSCIENCES, MULTIDISCIPLINARY","Score":null,"Total":0}
引用次数: 0
Abstract
Understanding how different climates and soil properties affect the soil processes requires quantifying these effects. Visual soil quality indicators have been proposed to assess the robustness of the soil processes and infer their ability to function. The scores of the visual soil quality indicators covary with climate features and soil properties, and their magnitude is different in acid-to-neutral and alkaline soils. These variables show collinearities and interactions, and the assessment of the individual effect of each variable on the scores of the visual indicators and the selection of the best set of explanatory variables can only be made with a definite set of variables. Logistic regression was used to calculate the effects of six climate variables and four soil properties, and their interactions, on the scores of eight visual soil quality indicators. Simple models featuring climate and soil variables explained a substantial part of the variation of the visual indicators. Models were fitted for each visual indicator for acid-to-neutral and alkaline soils. The sample size needed was calculated, and the method and its validity were discussed. For two possible outcomes, the sample size using the events per variable (EPV) criterium ranges between 62 and 183 observations, while using one variable and a variance inflation factor, it ranges between 22 and 234. Except for the model of soil structure and consistency for acid-to-neutral soils, with a C statistic of 0.67, all others had acceptable to excellent discrimination. The models built are adequate, for example, for the large-scale spatial outline of the soil health indices, to couple with soil morphological-dependent pedotransfer functions, and so on. Future models should consider (test) other explanatory variables: other climate variables and indices, other soil properties and soil management practices.