Reforestation improves vegetation coverage and biomass, but not spatial structure, on semi-arid mine dumps

• Published in: Ecological engineering 2022-02, Vol.175, p.106508, Article 106508
• Main Authors: Yang, Yongjun, Tang, Jiajia, Zhang, Yiyan, Zhang, Shaoliang, Zhou, Yongli, Hou, Huping, Liu, Run
• Format: Article
• Language: English
• Subjects: Aridity; Back propagation networks; Biomass; Community composition; Composition; Diversity; Ecosystem restoration; Ecosystem services; Ecosystems; Evolutionary algorithms; Mathematical models; Mining; Neural networks; Optimization; Parameters; Propagation; Reforestation; Regeneration; Regeneration (biological); Rehabilitation; Remote sensing; Restoration; Restoration success; Spatial structure; Species composition; Species diversity; Structure-function relationships; Substrates; Vegetation; WorldView-2
• ISSN: 0925-8574; 1872-6992
• DOI: 10.1016/j.ecoleng.2021.106508

Modern mine rehabilitation aims at thorough restoration of an ecosystem with emphasis on not only land area covered by vegetation but also the structure and function of that vegetation. However, assessment of current restoration success reveals a lack of attention to the spatial structure and its relationships with vegetation coverage and biomass. A forest's spatial structure is an important attribute of structural diversity. Complex spatial structures mean diverse species composition and spatial dissimilarity, which can provide a base for self-sustaining and regeneration. This study uses WorldView-2 images and field data to train the mind evolutionary algorithm-back propagation neural network (MEA-BP) model for the purpose of mapping three parameters (coverage, biomass, and spatial structure) across mine dumps. The results show that the spectral and textural features could effectively assess the coverage, biomass, and spatial structure, with an R2 of 0.91, 0.86, and 0.62, respectively. The coverage is positively correlated with biomass, while the spatial structure is negatively correlated with coverage and biomass. Pure Hippophae rhamnoides L. had high coverage but very low spatial structure, while the mixed community dominated by Populus L. and Pinus tabulaeformis Carr had high coverage, high biomass, and medium structure at around 10 years. The results suggested that the artificial reforestation improves vegetation coverage and biomass but did not synchronously increase spatial structure. The initially planted species composition, substrates, and succession process have a significant influence on the forest parameter relationships. The future reforestation and optimization of community assemblages should take the relationships and their influence factors and effects on ecosystem services into account. The remote sensing-based data and model has potential and advantages in dynamically guiding the rehabilitation and monitoring of the restored ecosystem. •WorldView-2 image and artificial neural network model were used for mapping forest parameters.•Trade-offs exist among vegetation coverage, biomass, and spatial structure during mined land restoration.•Artificial reforestation improves vegetation coverage and biomass but not synchronously increase spatial structure.•Optimization of community assemblages should consider the forest parameter relationships and effects.