『Abstract
　It is practically impossible to measure soil properties continuously at each location across the globe. Therefore, it is necessary to have robust systems that can predict soil properties at a given location. That is needed in many tropical countries where the dearth of soil property measurements is large. This paper reviews the use of pedotransfer functions (PTF) for predicting properties of soils in the tropics. First, the guiding principles of prediction and the type of predictors are discussed, including laboratory data, field description and soil morphology, electromagnetic spectrum, proximal and remote sensed data. In the subsequent section, PTFs are discussed for soil physical and chemical properties followed by infrared spectroscopy, proximal sensing and remote sensing. An analysis of ISRIC (mainly tropical) and USDA (mainly temperate) soil databases showed that soils in the tropics have higher clay content, lower cation exchange capacity, higher bulk density, lower water content at - 10 kPa and - 1500 kPa than soils in the temperate regions. Various methods developed in temperate regions can be applied for the soils in the tropical regions although calibration and careful selection of predictors remains necessary. It is concluded that PTFs are an important tool to overcome the dearth of soil data in many tropical countries.

keywords: soils of the tropics; soil data; pedotransfer functions; soil prediction; infrared spectroscopy; soil inference systems』

Contents
1. Introduction
2. A brief history of pedotransfer functions in the tropics
3. The principles of prediction
　3.1. For the user and the developer
　　3.1.1. Do not predict something that is easier to measure than the predictor
　　3.1.2. When predicting a variable, there should be a physical basis for the predictors
　　3.1.3. Developer should explicitly list the statistics of their training variables
　　3.1.4. Do not use PTFs unless you can evaluate the uncertainty, and for a given problem, if a set of alternative PTFs is available, use the function with the lowest variance
4. Predictors
　4.1. Laboratory data
　4.2. Field description and soil morphology
　4.3. Infrared spectroscopy
　4.4. Proximal sensing
　4.5. Remote sensing
5. Pedotransfer functions in the tropics
　5.1. Predicting soil physical properties
　5.2. Predicting soil chemical properties
6. Statistical approaches
　6.1. For the PTF developer
7. Discussion and conclusions
　7.1. For the user and the developer
　7.2. Hydropedology
　7.3. Digital soil mapping
References