Current climate and land cover change threaten global mountaintops with increased spread of invasive species. Long-established plantations of exotic and invasive trees on these mountaintops can alter their surroundings, further increasing invader-facilitated invasion. Identifying the ecological conditions that promote such specific associations can help develop better management interventions. 

The Western Ghats's Shola Sky Islands (>1400m MSL) host vast stretches of exotic and invasive tree plantations that sustain colonisation of other invasive woody, herbaceous and fern species in their understories. Here we analysed vegetation and landscape variables from 232 systematically-placed plots in randomly selected grids using NMDS and Phi Coefficient approaches, to examine patterns of association (positive interactions) between understory invasive species with specific exotic and invasive overstory species. We also conducted GLMM with zero inflation to determine the influence of environmental variables where such associations occur. 

We find that understory invasion of multiple species under the canopy of other exotic invasives is widespread across the Shola Sky Islands. Stands of Eucalyptus host the colonisation of 70% of non-native invasive species surveyed across the Shola Sky Islands. In particular, Lantana camara invasion is strongly associated with Eucalyptus stands. 

We also found that climatic variables affect the colonisation of understorey woody invasive species, while invasion by exotic herbaceous species is associated with the density of road networks. Canopy cover impacts all invasives negatively, while incidence of fire was negatively associated with invasion by Lantana spp and the Pteridium spp.  While the restoration of natural habitats largely targets the highly invasive Acacia, less invasive Eucalyptus and Pine are often not included. Our study suggests that retaining such exotic species in natural habitats, particularly protected areas, can hinder ongoing restoration efforts by facilitating further invasions by multiple woody and herbaceous species.  

The present study was conducted across the high-elevation montane forests of the Nilgiris and Anamalai-Palani hills landscapes (10.12⁰N 77.60⁰E to  11.50⁰N 76.70⁰E). We classified wooded habitats above 1400m MSL into shola forests and plantations following Arasumani et al. 2019. A point lattice of spacing 100 m between the points was overlaid using ArcGIS, and we randomly selected 0.5% of points. Squares of 200mx200m were placed over the lattice, with their centres coinciding with the selected points. We chose 143 random squares and surveyed up to eight circular plots of radius 7m within the grids placed systematically. Since our objective was to assess native regeneration under exotic overstorey, we did not include plots with native trees in their overstory in our analyses. The final numbers of plot samples were 232.

Each grid cell consists of up to eight 7m-radius circular plots arranged systematically, and each plot has five circular sub-plots of radius 1m. The circumference at breast height (GBH) for all live trees and snags over 30 cm was measured within the circular plots. The trees were classified as native species, Acacia spp., Eucalyptus spp., pine spp., and others. The intensity of regeneration may depend on the amount of light reaching the floor, determined by the canopy cover, measured at five points inside the plot (four cardinal directions and the centre) using HabitApp, an android application (Bianchi et al., 2017). Within the subplots inside the plots, we measured the vegetation profile using a 5 m pole, calibrated at every 0.5 m. Calibrations at which vegetation touched the pole were noted. At each point, we also noted the presence of fire and counted the instances of invasive undergrowth. We identified the colonising invasive species at least up to the family level.

From the locations collected for each plot, we determined the following variables through satellite images: elevation, slope, aspect, topographic position index and topographic convergence index. TPI was collected at a local scale, and TCI was collected at a scale of 30m after resampling the DEM.

We used the software R to conduct our analyses. The file is in CSV format and can be opened in Microsoft Excel.