High-resolution images and drone-based LiDAR reveal striking patterns of vegetation gaps in a wooded spinifex grassland of Western Australia

Abstract

Context Vegetation patterns in hummock grasslands of Australia’s arid interior can be very complex. Additionally, the grasslands are interspersed with variable amounts of trees and shrubs. Objectives To better understand the spatial arrangement of this vegetation structure, and in particular the unvegetated bare-soil gaps, we analyzed the scale-dependent patterns of gaps, trees, and shrubs. Methods We focused on two size categories of grassland gaps, large gaps ≥ 4 m2 known as fairy circles (FCs) and small gaps 1 to < 4 m2, and on trees and shrubs. We mapped four 200 m × 200 m study plots located east of the town of Newman in Western Australia, using drone-based aerial images and LiDAR. The RGB images were converted into binary images and the gaps and woody plants were automatically segmented. The spatial patterns of the four vegetation components were analyzed, as well as the shape properties of the vegetation gaps. Results The most striking result was that small gaps appeared consistently at about 5 m distance away from the FCs, which are known as the most water-depleted locations in the grassland. The FCs were also rounder than the small gaps and this symmetry underlines their function as an extra source of water for the surrounding matrix vegetation. Trees and shrubs had spatial patterns that were unrelated to FCs, which likely results from their water uptake in deeper sub-soil layers. Conclusions The consistent distance of small gaps to FCs is further support that the Australian fairy circles are a self-organized vegetation pattern that results from ecohydrological feedbacks.

Inferring plant–plant interactions using remote sensing

“…The picture was taken from a balloon. (b) Small gaps and labyrinth‐like pattern of Triodia basedowii , also known as Spinifex grass, near Newman in Pilbara region in Western Australia (360 mm/year; see Getzin et al, 2022; 23.4413S, 119.8379E). (c) Large rings of Paja brava grass ( Festuca orthophylla ) showing clonal growth in the high and remote Altiplano in Bolivia (100 mm/year; 4900 m above sea level; 21.8952S, 66.8640 W).…”

“…However, many outcomes of key plant-plant interactions in shaping plant traits, community structure, ecosystem states, and landscape patterns can be reflected in remote sensing signals (e.g., Getzin et al, 2022;Xu, Holmgren, Van Nes, Maestre, et al, 2015).…”

“…Clearly, interactions per se cannot be directly captured by sensors. However, many outcomes of key plant–plant interactions in shaping plant traits, community structure, ecosystem states, and landscape patterns can be reflected in remote sensing signals (e.g., Getzin et al, 2022; Xu, Holmgren, Van Nes, Maestre, et al, 2015). Therefore, many interaction attributes (e.g., direction and strength) can be indirectly inferred through remote sensing of interaction outcomes.…”

Correction of UAV LiDAR-derived grassland canopy height based on scan angle

“…High-density canopy was considered to be a main cause in forests, which hamper the penetration of laser pulses to reach the ground ( Hu et al., 2020 ). Although grasslands may have a higher canopy density than forests, the gaps between grass individuals combined with high point cloud density allow UAV LiDAR to obtain reliable ground elevation information ( Getzin et al., 2021 ; Zhao et al., 2022 ). The height loss at canopy was proved to be the main cause which greatly reduced the accuracy of grassland height-related structural traits ( Miura et al., 2019 ).…”

“…The underestimation would affect the estimation of grassland ecosystem functions related to canopy height and its heterogeneity. For example, Da Costa et al (2021) found that grassland aboveground biomass estimated based on UAV LiDARderived canopy height was lower than that measured on the ground. High-density canopy was considered to be a main cause in forests, which hamper the penetration of laser pulses to reach the ground (Hu et al, 2020).…”