Data
Vegetation surveys were carried out on Marion Island from 2018 to 2019.
A total of 464 vegetation plots were placed throughout the island in a
random stratified design to represent each of the geologies (i.e., red
scoria, black lava and grey lava) (Rudolph et al., 2021). Sampling was
carried out in 3 x 3 m plots and in the centrally nested 1 x 1 m plots
(Appendix Figure A1). The identity of all vascular plant species in the
main and nested plot was recorded. Azorella selago was excluded
from the vascular plant species richness count in each plot, but its
presence was recorded as its occurrence would be used as a predictor of
species richness in subsequent analyses.
A digital elevation model (DEM, produced at 1 m resolution but resampled
to 20 m here) was obtained from the surveyor general (South Africa).
DEMs describe terrain and elevation for a given area, enabling studies
on habitat suitability, landscape connectivity, hydrological processes,
and climate-related ecological patterns. They prove especially valuable
in challenging, inaccessible locations on Marion Island, where physical
measurements are difficult to obtain. Using this DEM, the following
parameters were extracted for each plot: elevation, topographical
wetness index (TWI) which was used as a proxy for soil moisture,
hillshade, slope, northness, distance to the nearest drainage line,
distance to the coast, and elevation. TWI was calculated using the SAGA
GIS tool (Böhner and McCloy, 2006): TWI = ln (AS/tan β) where AS is the
drainage area (in m²) and β is the local slope gradient (in %) (Beven
and Kirkby, 1979). Slope and hillshade were calculated with the tools
Slope and Hillshade respectively in ArcMap 10.8.1. The hillshade
function is a proxy for potential incident solar radiation (Najafifar et
al., 2019). Northness was calculated using the formula northness =
cos(aspect) in radians, where aspect was calculated with function Aspect
in ArcMap from the DEM. The distance from the nearest drainage line was
calculated using the topographic database from the DEM and the distance
to the coast was calculated using the Euclidian Distance tool in ArcMap.
A layer of mean land surface temperature was created by averaging
monthly temperatures from Leihy et al. (2018). Because temperature was
not available for some coastal cells, these were interpolated using the
Kriging downscaling method. This temperature layer was then resampled
using the bilinear technique to 20 x 20 m resolution, and the mean
temperature for each plot extracted. Geology was extracted from Rudolph
et al. (2021). Distance of plots to the nearest research base or field
hut was calculated as a proximity for the intensity of human activities.
Although elevation can be used as a proxy for temperature, we included
both temperature and elevation in our models since elevation also
accounts for a broad range of factors beyond temperature alone. For
example, elevation on Marion Island may also influence factors such as
habitat structure, incident solar radiation, and soil characteristics,
which can impact species richness independently of temperature.
Moreover, as elevation increases, the influx of biotic inputs from
mammal and bird species on the island diminishes, resulting in soils at
higher elevations exhibiting reduced nutrient content (Haussmann et al.,
2013).
Predictor variables were selected as proxies for groups of species
richness drivers: mean temperature, elevation, distance to coast,
northness and hillshade as proxies for temperature and energy inputs;
TWI and distance to nearest drainage line as proxies for water
availability; slope, geology, and distance to nearest base/hut as
proxies for topographical and environmental heterogeneity; and
presence/absence of A . selago as a biotic interaction, asAzorella selago , a keystone species on the island, serves as a
favourable establishment site for other plant species, particularly in
higher elevations and on dry, nutrient-poor soils – areas other plant
species would not colonise independently (le Roux and McGeoch, 2004, le
Roux and McGeoch, 2010).