Abstract
Aquatic and terrestrial ecosystems are deeply coupled and inherently connected through spatial flows of organisms and resources, modulating the flow or exchange of organisms thereby the land-water linkage of biodiversity at local to global scales. Although there is a plethora of studies about aquatic-terrestrial cross-ecosystem linkages, these researches normally only focus on a few taxa and in a few local catchments. This lack of a systematic and comprehensive understanding of the spatial scale and magnitude of linkages is partly due to the laborious and costly field sampling especially in aquatic ecosystems, impeding effective aquatic biodiversity protection strategy. Recent advances in environmental DNA (eDNA) metabarcoding and remote sensing (RS) have made large-scale sampling possible in a time- and cost-efficient manner, with abundant and booming amount of biodiversity data available. RS is advantageous in monitoring terrestrial ecosystems, while eDNA is especially effective in assessment of aquatic biodiversity. When combined together, they can provide a holistic view of biodiversity at the landscape level. Based on these two technologies, this thesis studies the land-water linkage of biodiversity at regional to global scales. Briefly, two dimensions—spatial range and magnitude—were adopted to depict such linkages within drainage catchments, which were the basic spatial unit for analysis. Specifically, I combined satellite remote sensing imagery and eDNA extraction from river water across a temperate mountainous catchment and identified a characteristic spatial land-water fingerprint, attesting that biodiversity patterns in the river can be linked to the functional diversity of surrounding terrestrial ecosystems. The spatial scale of this linkage was measured to be 2.0 km upstream from sampling sites with a peaking signal at a 400 m distance. Further analysis using high-resolution airborne imaging spectroscopy and light detection and ranging (LiDAR) point clouds showed a variation in spatial scales of the land-water linkage across major aquatic functional feeding groups (FFGs). I detected the scale of the linkage ranging from a few hundred meters to more than 10 km, with collectors and filterers, shredders, and small invertebrate predators having local-scale association, while invertebrate eating fish, grazers and scrapers having more regional-scale associations. The magnitude as well as spatial range of the land-water linkage of biodiversity were assessed by developing a spatially explicit biodiversity modeling framework. In a large-scale subtropical river catchment characterized by remarkable biodiversity yet intense anthropogenic alterations, I linked eDNA sampling of fish species with a contemporary terrestrial land use and land cover (LULC) map and estimated a spatial range of LULC effects to be 19 km upstream from sampling sites. This model explained nearly 60% of the variance in the observed fish species richness, associated with a relative positive effect from rainfed cropland yet relative negative effects from forest and urban area. The global analysis using the same model illustrated diverse spatial ranges and magnitudes in the fish species richness-land cover relationships. I found large-scale associations in North America, South America, and Europe, yet more local-scale associations in Asia, Oceania and Africa partly due to the more fragmented landscape of the sampling catchments and limited number of sampling rivers. Plus, high magnitudes of LULC effects on fish diversity were found in croplands and forests, and a high uncertainty was found in LULC effects from urban areas. River dams produce as much as 14.3% of global electricity supply yet causing a reduction in biodiversity in rivers especially for fish species. Based on the global fish eDNA database and satellite-derived global dam locations, I estimated dam effects on fish distributions between upstream and downstream reaches across continents. Results showed that, compared with free-flow river channels, dams caused a reduction in fish community similarity between upstream and downstream reaches, thereby an isolation effect on riverine fish distributions because of habitat fragmentation. This thesis provides a systematic and in-depth perspective about how aquatic biodiversity is associated and interlinked with the surrounding terrestrial landscape. I provide specific estimations of spatial scales and magnitudes which are key information to a better understanding of land-water linkages and a refence information to aquatic biodiversity protection. As the combination of eDNA and remote sensing and the approach developed in this thesis do not require specific taxonomic knowledge, it allows an adequate scalability for biodiversity scientists. With a possibility to project riverine biodiversity under global change scenarios, this thesis also offers a quantitative tool for conservation design and management that can ultimately also be applied worldwide.
Zhang, Heng