Stable carbon isotope records of black carbon on Chinese Loess Plateau since last glacial maximum: An evaluation on their usefulness for paleorainfall and paleovegetation reconstruction

Abstract Stable carbon isotope composition of soil organic matter (δ 13 C SOM ) has been widely used to infer past environmental or vegetation (C 3 /C 4 ) changes. However, as a carbonaceous product of incomplete burning of biomass, the carbon isotope signature of black carbon (BC) has received little study and a limited usage on this purpose. Moreover, the environmental or vegetation indication of BC carbon isotope composition (δ 13 C BC ) in sedimentary records remains ambiguous although the δ 13 C BC is supposed to reflect carbon isotope composition of vegetation being burnt. This deserves site-specific studies. Here we analyzed δ 13 C BC and δ 13 C SOM on loess-paleosol samples from Lijiayuan and Yangling sections on Chinese Loess Plateau (CLP) spanning the last glacial maximum to decipher the environmental (or vegetation) meaning of δ 13 C BC at each study site. Opposite changing patterns were observed on the δ 13 C BC values for the two sections. The δ 13 C BC at Yangling (in southern part of the CLP) varied from − 19.18‰ to − 21.93‰ (mean: − 20.62‰) with more positive values occurred during the middle Holocene than those during the LGM, demonstrating more C 4 plants occupied in the region during the warm-humid middle Holocene. This is consistent with the changing pattern widely-documented in δ 13 C SOM records over the CLP during the same period. By contrast, the δ 13 C BC at Lijiayuan (in northwestern part of the CLP) changed within the range of − 21.83‰ ~ − 24.64‰ (mean: − 23.34‰) and displayed more negative values during the early-middle Holocene with respect to the LGM period. The δ 13 C BC at Lijiayuan were about 2.5‰ lower than those at Yangling, indicating a northward decrease of C 4 plants on the CLP. The anti-phased changes of δ 13 C BC at Lijiayuan were considered to reflect variations in paleorainfall because the vegetation is dominated by C 3 plants at the study site and carbon isotope compositions of C 3 plants decrease as rainfall increases. Compared with δ 13 C SOM record at the same profile, δ 13 C BC seem biased against capturing C 4 signal during mid-Holocene possibly due to dominance of C 3 plants at the study site during fire seasons and a potentially extra source of BC from forest fires occurred on surrounding mountains. By contrast, δ 13 C BC tend to exaggerate C 4 signal during the LGM attributed to a possible shift of fire season from spring to summer. These findings suggest that we probably underestimate the abundance of C 4 plants during mid-Holocene and overestimate C 4 biomass during the LGM using BC carbon isotope composition.