Association of urban forest landscape characteristics with biomass and soil carbon stocks in Harbin City, Northeastern China

Background: Urban forests help in mitigating carbon emissions; however, their associations with landscape patterns are unclear. Understanding the associations would help us to evaluate urban forest ecological services and favor urban forest management via landscape regulations. We used Harbin, capital city of the northernmost province in China, as an example and hypothesized that the urban forests had different landscape metrics among different forest types, administrative districts, and urban-rural gradients, and these differences were closely associated with forest carbon sequestration in the biomass and soils. Methods: We extracted the urban forest tree coverage area on the basis of 2 GF-1 remote sensing images and object-oriented based classification method. The analysis of forest landscape patterns and estimation of carbon storage were based on tree coverage data and 199 plots. We also examined the relationships between forest landscape metrics and carbon storage on the basis of forest types, administrative districts, ring roads, and history of urban settlements by using statistical methods. Results: The small patches covering an area of less than 0.5 ha accounted for 72.6% of all patches (average patch size, 0.31 ha). The mean patch size (AREA_MN) and largest patch index (LPI) were the highest in the landscape and relaxation forest and Songbei District. The landscape shape index (LSI) and number of patches linearly decreased along rural-urban gradients (p < 0.05). The tree biomass carbon storage varied from less than 10 thousand tons in the urban center (first ring road region and 100-year regions) to more than 100 thousand tons in the rural regions (fourth ring road and newly urbanized regions). In the same urban-rural gradients, soil carbon storage varied from less than five thousand tons in the urban centers to 73-103 thousand tons in the rural regions. The association analysis indicated that the total forest area was the key factor that regulates total carbon storage in trees and soils. However, in the case of carbon density (ton ha-1), AREA_MN was strongly associated with tree biomass carbon, and soil carbon density was negatively related to LSI (p < 0.01) and AREA_MN (p < 0.05), but positively related to LPI (p < 0.05). Discussion: The urban forests were more fragmented in Harbin than in other provincial cities in Northeastern China, as shown by the smaller patch size, more complex patch shape, and larger patch density. The decrease in LSI along the rural-urban gradients may contribute to the forest carbon sequestrations in downtown regions, particularly underground soil carbon accumulation, and the increasing patch size may benefit tree carbon sequestration. Our findings help us to understand how forest landscape metrics are associated with carbon storage function. These findings related to urban forest design may maximize forest carbon sequestration services and facilitate in precisely estimating the forest carbon sink.