Nonlinear imprints of forest coverage on the relationships between gross primary production (GPP) and landscape patterns

• Published in: Ecological indicators 2023-02, Vol.146, p.109783, Article 109783
• Main Authors: Gao, Haiqiang, Liu, Shuguang, Feng, Shuailong, Peng, Xi, Ning, Ying, Shi, Yi, Wang, Zhao, Wei, Baojing, Yan, Wende
• Format: Article
• Language: English
• Subjects: Afforestation; Forest coverage; Forest management; Gross primary production; Landscape metrics; Nonlinear effects
• ISSN: 1470-160X; 1872-7034
• DOI: 10.1016/j.ecolind.2022.109783

•Annual GPP peaked in region with 70–80% forest cover.•Annual increase of GPP in the highest forest-cover region (>80 %) was the lowest.•The sensitivity of GPP to landscape metrics was the highest in regions with 70% forest cover.•Regulating the composition and configuration of forest helps forest management. Land use and land cover change (LUCC) has yielded significant impacts on ecosystem structure and functions. Gross primary production (GPP), the largest flux in the terrestrial carbon cycle, has been drastically changed by various LUCC activities. However, the indirect impacts of LUCC on GPP (i.e., local environment changes induced by LUCC), particularly those related to landscape dynamics, remain unclear. Here, we explored the relationships between GPP and landscape metrics of forest ecosystems in China along a forest coverage gradient from 2001 to 2019, using an up-to-date GPP dataset derived from Solar-Induced Chlorophyll Fluorescence and a widely used landcover product MCD12Q1. Results showed that forest landscape features (e.g., core area index, CAI) synchronized with GPP spatially as shown by the distribution of their hot/cold spots on correlations. Furthermore, two nonlinear change patterns of GPP and landscape metrics along forest coverage gradient were found. One is a hump-shaped pattern, which indicated that GPP and some landscape features (e.g., annual median GPP, annual change rate of forest composition) peaked in medium (about 70–80%) forest-cover regions. Another is a super-linear curve, revealing the rapid and accelerated growth of some landscape features (e.g., annual median CAI) in high forest-cover regions. Finally, forest coverage affected the sensitivity of GPP to landscape features in a nonlinear fashion, characterized as an inverted ‘hook’ curve peaking in intermediate (around 70%) forest-coverage regions. Our study proposed the optimal landscape regulation hypothesis to explain the nonlinear pattern: forest GPP is most responsive to the landscape pattern regulation at intermediate to low levels of disturbance, implying that forest coverage has various impacts on landscape features; together they affect GPP nonlinearly. Regulating the forest landscape composition and configuration can enhance the performance of forest ecosystems in addition to managing forest coverage. Thus, we recommend the adoption of suitable region-specific forest management practices to improve GPP and other ecosystem services of forest ecosystems via manipul