Remote sensing of aboveground and belowground function in grassland ecosystems

Abstract

Grasslands are one of the most extensive biomes on Earth, are integral to global biogeochemical cycling, and are directly depended upon by human societies for food and fuel. Important functions of grassland ecosystems include carbon storage, biomass production and nutrient cycling, which are driven by closely interlinked grassland plant and soil communities. Measuring and monitoring these communities across large spatial extents is vital in order to understand and predict the impacts of climate and land-use changes which affect grasslands currently and are forecast into the future. However, current field-based methods are intensive and time-consuming to carry out, and therefore limited in their spatial and temporal coverage. Remote sensing technologies including imagery from airborne and satellite platforms have transformed capabilities to measure the biosphere at large scales. However, grassland ecosystems are under-represented in remote sensing research, and the application of remote sensing to belowground ecosystem functions remains relatively unexplored. The aim of this thesis is to investigate how remote sensing data can be used to retrieve important information about the above- and belowground portions of grassland ecosystems, across large spatial extents. This aim is achieved by pairing in-situ soil and plant community data from large-scale ecological monitoring networks with hyperspectral and multispectral imagery from airborne and satellite platforms, to answer three specific research objectives. 1) to predict variation in soil microbial communities from airborne hyperspectral imagery, 2) to retrieve community-level plant functional traits from multispectral satellite data and 3) to predict ecosystem multifunctionality using long-term (decadal) timeseries of satellite-derived vegetation indices. All objectives are addressed using observations from varied natural grassland communities distributed at continental and global scales. The results of the three research chapters demonstrate that soil microbial communities, in particular microbial community structure, can be retrieved with high accuracy (up to R2 0.68, NRMSE 9%) from hyperspectral sensing. Satellite multispectral data was able to retrieve variation in 13 out of 20 foliar traits, in some cases with comparable accuracy to recent hyperspectral mapping (up to R2 0.76, NRMSE 12%). The ability of satellite systems to facilitate a multitemporal perspective on the ecosystem was valuable for predicting ecosystem multifunctionality, where past vegetation dynamics and stability have important legacy effects on contemporary functioning. Overall, the thesis shows that remote sensing technologies have the potential to contribute to understanding of above- and belowground grassland functions by capturing ecological properties of the vegetation surface which are not able to be measured in the field, and by expanding the spatial and temporal scales across which relationships can be found. Remote sensing should be incorporated into the design of future large-scale ecological monitoring efforts, with a view to understanding and preserving grassland ecosystem functions for the future.

Date of Award	27 Jun 2023
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Richard Bardgett (Co Supervisor), Tim Allott (Co Supervisor) & Angela Harris (Main Supervisor)