Abstract

Background: Our objective in this work was to determine the variation of key physico-chemical properties of TRF soils across a wide altitudinal gradient in Sri Lanka. We aimed to identify the climate controls of TRF soil properties by relating their altitudinal variation to the climatic variation with altitude. Methods: We extracted nine soil variables, viz. pH, organic carbon (SOC), cation exchange capacity (CEC), percentages of sand, silt and clay, bulk density (BD), soil water contents at suctions of 0.033 MPa (Field Capacity, FC) and 1.5 MPa (Permanent Wilting Point, PWP), at five soil depths (0-5, 5-15, 15-30, 30-60 and 60-100 cm) from the national-level digital soil database for a series of permanent sampling plots in TRFs established from 117 to 2235 m above sea level. We quantified the contributions to the observed variance of each soil property from variations of altitude, soil depth and altitude × depth interaction, via a random effects general linear model analysis. We applied a factor analysis to extract key underlying composite soil variables and a subsequent multiple linear regression analysis to determine the influence of long-term climatic variables on the extracted composite soil variables. Results: The altitude effect contributed most (51% - 81%) to the variance of pH, percentages of sand, silt and clay, BD, FC, PWP and available soil water, ASW, calculated as the difference between FC and PWP. Soil depth effect was strongest on the variance of SOC and CEC. Field capacity, ASW, CEC and silt% showed significant linear increases with altitude whereas BD showed a significant linear decrease. Soil organic carbon and CEC showed significant linear decreases with soil depth. Bulk density showed significant negative correlations with SOC, CEC, FC and ASW. Cation exchange capacity showed significant positive correlations with SOC and silt%, but a significant negative correlation with sand%. Factor analysis extracted an underlying soil variable, Factor 1, incorporating CEC, Silt%, FC, ASW, PWP and BD, which explained 74% of the observed soil variation, plus another two, Factors 2 and 3, incorporating pH and clay% explaining 13% and 12% respectively. Multiple regression analysis showed that Factor 1 decreased with increasing long-term (1970-2018) annual means of temperature and rainfall but increased with diurnal temperature range (DTR). Factor 3 increased with decreasing DTR and rainfall, but Factor 2 was climate-insensitive. Cluster analysis separated the TRFs at different altitudes to three clusters (i.e. up to 1100 m, 1100 – 1800 m and > 1800 m) at a maximum linkage distance of 1.0. Conclusion: In this study, we demonstrate the complex interactions between soil properties and climate in the TRF ecosystems in Sri Lanka. The relationships between the composite factors and long-term climate are important to predict how future climate change would influence key soil physico-chemical properties and thereby the TRFs in Sri Lanka.