Assessment of soil group, site and climatic effects on soil organic carbon stocks of topsoil in Japanese forests. (9th June 2017)

• Record Type: Journal Article
• Authors: Nanko, K.; Hashimoto, S.; Miura, S.; Ishizuka, S.; Sakai, Y.; Levia, D. F.; Ugawa, S.; Nishizono, T.; Kitahara, F.; Osone, Y.; Kaneko, S.
• Is Part Of: European journal of soil science. Volume 68:Number 4(2017:Aug.)
• Page Start: 547
• Page End: 558
• Publication Date: 2017-06-09
• Subjects: Soil science -- Periodicals; 631.4
• DOI: 10.1111/ejss.12444
• Languages: English
• ISSNs: 1351-0754

Abstract

Abstract : The aim of this study was to assess the factors that account for the geographical variation in soil organic carbon stocks at the 0–30‐cm depth (SOC30 ) of forests in Japan. Boosted regression tree analysis was applied to 2157 points throughout Japan and to four regional geographical subdivisions with 16 environmental variables. The rank of predictor variables was different for Japan as a whole and among the regions. For Japan as a whole, soil group, air temperature, slope inclination, altitude and organic carbon stocks of litter were the most important factors that affected SOC30 stocks. Overall, SOC30 stocks decreased with air temperature, which was attributed to the decomposability of organic carbon. In addition, SOC30 stocks decreased with slope inclination because of instability of the topsoil on slopes, which, in turn, is related to the increase in rock fragment content and decrease in soil bulk density. The distribution of volcanic soil resulted in larger SOC30 stocks than was expected from climatic conditions. Precipitation was not important because of conflicting effects between the increase in soil organic carbon content with increasing net primary production and the decrease in mineral soil mass by the loss in topsoil. The regional analyses provide insight into the factors that cause variation in SOC30 stocks, which were obscured by the macroscale analysis of Japan as a whole, thereby illustrating the power of regional geographical analyses. Our results provide an improved basis for soil, forestry and biogeochemical models that require accurate estimates of SOC30 stocks. Highlights: We assessed factors that account for geographic variation in soil organic carbon (SOC) stocks. Effect and dependence of factors were estimated by a machine learning approach. SOC stocks were affected by soil type, climate, site‐specific location and organic matter input. Volcanic soil distribution, climate, slope steepness and historical overuse of forest affected SOC stocks.