Spatial variation of tundra soil organic carbon along the coastline of northern Alaska Fugen Dou a, ⁎, Xian Yu b , Chien-Lu Ping c , Gary Michaelson c , Laodong Guo d , Torre Jorgenson e a International Arctic Research Center, University of Alaska Fairbanks, Fairbanks, AK99775, United States b Department of Mathematical Sciences, University of Texas at Dallas, Richardson, TX 75083, United States c Department of Animal, Plant, and Soil Science, University of Alaska Fairbanks, Palmer, AK 99645, United States d Department of Marine Science, University of Southern Mississippi, Stennis Space Center, MS 39529, United States e ABR Inc., Fairbanks, AK 99708, United States abstract article info Article history: Received 17 February 2009 Received in revised form 24 July 2009 Accepted 31 October 2009 Available online 4 December 2009 Keywords: Coastline erosion Tundra soil Soil organic carbon Geostatistical model Gaussian variogram Geological information system 1-D (dimension) or 2-D (dimension) model Coastal erosion plays an important role in the terrestrial–marine–atmosphere carbon cycle. This study was conducted to explore the spatial variation of soil organic carbon (SOC) and other soil properties along the coastline of northern Alaska. A total of 769 soil samples, from 48 sites along over 1800-km of coastline in northern Alaska, were collected during the summers of 2005 and 2006. A geological information system (GIS) and a geostatistical method (ordinary kriging) were coupled to investigate the spatial variation of SOC along the coastline. SOC have a big variation ranging from 0.8 to 187.4 kg C m -2 with the greatest value observed in the middle and lowest in the northeastern coastline. Compared to the 1-D model or the 1-D model with shortcut distance, the 2-D model was more reasonable to describe SOC along the coastline. The Gaussian correlation structure model had less prediction error than other examined geostatistical models. All mapping results also indicate that soils of the northwestern coastline stored greater SOC than those of the northeastern coastline. The estimation of total SOC along the coastline of northern Alaska was 6.86 10 7 kg m -1 . The prediction errors indicated that greater errors were observed in both ends of the coastline than were observed in other fractions, although the range was from 0.739 to 0.779. Our study suggests that the isotropic 2-D model without a trend, with the nugget effect and the Gaussian correlation structure is a useful tool to investigate SOC in large scale. Results of stable isotope of organic matter indicate that SOC are mainly derived from C3 plant, which ranged from -30‰ to -22‰. © 2009 Elsevier B.V. All rights reserved. 1. Introduction Coastal erosion plays an important role in the terrestrial–marine– atmosphere carbon cycle (Chen et al., 2003). For the Arctic Ocean, coastal erosion is more important than for other oceans. First, proportionally, the Arctic Ocean has a greater ratio of coastline to land area than other oceans (Dittmar and Kattner, 2003). Second, greater erosion rates were observed for the Arctic Ocean (Dolan et al., 1983). Third, the eroding soils store more organic carbon (OC) than that in other coasts (Michaelson and Ping, 1996; Chen et al., 2003; Jorgenson and Brown, 2005). Many studies have reported that the climate in the Arctic regions has warmed appreciably (Osterkamp, 2003; Chapin et al., 2005; Jorgenson et al., 2006). In a warming climate, the magnitude of carbon during coastal erosion would be more significant. First, the period of open sea (not covered by sea ice) would be longer. Second, the temperature of both soil and sea water would be higher. Third, the sea level would be higher. These three consequences of warming climate would accelerate the coastal erosion rate, extend the coastal erosion time, and accelerate the decomposition of SOC during exposure of buried soil and in sea water (Bird, 1985; McCarthy et al., 2001). To estimate the SOC input into the Arctic Ocean accurately due to coastal erosion, information on OC distribution and coastal erosion rates or shoreline change should be obtained. Shoreline change is usually estimated through four methods: repeated shoreline surveys, aerial photography, historical maps and charts, and nonmetric photography (hand-held cameras) (Dolan et al., 1983). Compared to the coastal erosion rate, information about OC distribution along a coastline is limited. Arctic tundra soil has sequestrated lots of OC. Michaelson and Ping (1996) reported that OC ranged from 36 to 82 kg m -3 and 82 to 94 kg m -3 for some sites near the Barrow and Prudhoe Bay coastal plains, respectively. In addition, the variation of OC in the active layer and upper permafrost layer is significant. For example, OC ranged from 21 to 42 kg m -3 for an active layer and 5 to 55 kg m -3 for an upper permafrost layer at several sites at Prudhoe Bay. Similar results Geoderma 154 (2010) 328–335 Abbreviations: SOC, soil organic carbon; STL, seasonal thawed layer; OK, ordinary kriging; GIS, geological information system; GPS, global position system. ⁎ Corresponding author. Texas AgriLife Research and Extension Center, Texas A&M University, 1509 Aggie Drive, Beaumont, TX 77713, United States. Tel.: + 1 409 752 2741x2223. E-mail address: f-dou@aesrg.tamu.edu (F. Dou). 0016-7061/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.geoderma.2009.10.020 Contents lists available at ScienceDirect Geoderma journal homepage: www.elsevier.com/locate/geoderma