Landscape metrics of post-restoration vegetation dynamics in wetland ecosystems

摘要

Context To monitor wetlands at regional scales, it is pivotal to identify metrics that show rapid and predictable responses to restoration interventions. Remote sensing can monitor such metrics at high frequency and low cost but remains underutilized in practice. Objectives This study sought to find a set of landscape metrics most responsive to restoration and vegetation dynamics across 11 years and 20 restored wetlands in the Sacramento-San Joaquin Delta of California, USA. Methods Breakpoint analysis was used to detect phases in the development of vegetated pixels as estimated from the enhanced vegetation index (EVI) derived from Landsat (2004-2017). Landscape metrics were then generated from land cover classifications based on high resolution aerial images from the National Agricultural Inventory Program (NAIP). Using hierarchical clustering, we grouped phases showing similar temporal characteristics. We selected a subset of landscape metrics that best described the spatial structure of vegetation and its dynamics in each phase type. Results We identified four phases in vegetation development: (1A) rapid increase; (1B) decrease; (2A) low change; (2B) low change with fluctuations. Landscape metrics showed a significant response to vegetation dynamics in our sample, suggesting their potential to expand current monitoring practices at low cost. Young sites and sites experiencing a rapid increase in greenness were characterized by a lower density of small patches, while older sites, reference sites, and low variability sites were characterized by large, clustered patches. Conclusions Our study demonstrates that open source remote sensing can detect patterns in wetland response to restoration and help identify factors promoting their recovery.