Halting biological invasions and rewilding extirpated fauna are conservation interventions to bolster biodiversity, species interactions, and ecosystems. These actions are often considered separately and the potential for reintroduced wildlife to facilitate invasive plants has been largely overlooked. Here, we investigate the role of Singapore’s recolonizing native wild pigs (Sus scrofa) in facilitating an invasive weed Miconia crenata into tropical rain forests, which are normally highly resistant to invasion. We conducted line-transect surveys in 11 Singaporean rain forests and used generalized linear mixed models to consider the contribution of pigs' soil disturbances, human forest paths, and other environmental covariates, on the density of M. crenata. We found that M. crenata was more abundant at forest edges and invasion into forest interior was facilitated by pigs, paths, and canopy gaps, but that these effects were all additive, not synergistic (i.e. not multiplicative). These results highlight how modern invasions are driven by multiple disturbances as well as propagule pressure (e.g. urban birds dispersing seeds at forest edges where they establish in pig soil disturbances). Singapore’s extensive native forest restoration efforts may have provided plentiful edge and secondary forests that are well suited to pigs and M. crenata, which in turn undermine the aims of fostering later-successional native plant communities. To prevent negative externalities, we suggest that plant restoration and rewilding projects consider the potential role of wildlife in facilitating non-native plants, and couple these actions with preliminary screening of unintended consequences and continued monitoring, as well as limiting human-mediated weed invasion to minimize propagule sources.

Site 

We conducted fieldwork in 11 secondary rain forest patches in Singapore (1°21’07.6”N, 103°49’11.3”E) from February 2019 to January 2020, including (1) the Dairy Farm and (2) Hindhede area of the Bukit Timah Nature Reserve (BTNR), (3) the MacRitchie in the Central Catchment Nature Reserve (CCNR), (4) Bukit Batok Nature Park, (5) East Coast Park, (6) Mandai Park, (7) the Fort Siloso area of Sentosa, (8) Telok Blangah, (9) Kent Ridge Park, (10) Clementi Forest and (11) Pulau Ubin Island (Table S1; Figure. 2). We conducted vegetation transects (see more below) on and off human paths in all forests and at the time of sampling, pigs had not recolonized (were absent from) Sentosa, Telok Blangah, and East Coast Park. 

Vegetation Transects

We established 2-5 linear “off path” transects spaced >200 m apart randomly in each forest, extending 30-200 m from the forest edge to the centre, depending on the size of the forest (Figure. 2b; Table S1). We observed M. crenata and covariates along 10 m-long contiguous segments along these transects, within a 2 m observation window on both sides, forming 40 m² plots (Figure. 2b). We surveyed 1-2 human paths in each forest, by establishing non-linear “on path” transects following human paths, extending 60 to 220 m along the paths (Figure. 2b; Table S1). We observed M. crenata and covariates along 10 m-long contiguous segments, within a 2 m observation window perpendicular to the path edges (left or right side, occasionally both sides), forming 20 m² plots on either side and grouped these into 40 m2 plots. Our unit of analysis was density per 40 m² in all cases. Figure. 2b; Table S1).

We recorded the following variables in each plot along the line transects: (1) number of M. crenata plants 40 m^-2, (2) percentage of canopy openness (% of the plot area), (3) canopy height (m), (4) pig soil disturbance (% of the plot area), (5) exposed soil without leaf litter that was not induced by pigs (hereafter, bare soil, measured as % of the plot area), and (6) distance to a forest edge (m), which was recorded at the starting point of each 10-m continguous segment for “off path” transects (Table S2). To ensure correct identification based on characteristic leaves, only M. crenata stems 10 cm or taller were counted, and each stem emerging from the soil was considered an individual plant, regardless of its potential connectedness to other stems underground. In choosing our sampling approach, we faced a trade-off between sampling feasibility versus the ideal metric for estimating invasion severity (e.g. abundance versus biomass). We opted to use the abundance of stems and acknowledge this may not scale linearly with M. crenata biomass or area coverage. This is one limitation of the study design.

We collected data on leaf litter because it has been previously shown to inhibit the germination of small-seeded shade-tolerant plants including M. crenata (Metcalfe & Turner, 1998; Samarasinghe et al., 2022; Teo et al., 2003). Prior work suggested Singaporean secondary forests have a thick leaf litter layer formed from slowly decomposing leaves from dominant trees and shrubs (Chua et al., 2013). 

Pig soil disturbances result in distinctive overturned soil, which is easily identifiable (e.g. Figure. 1c and 1d).The environmental covariates were estimated to the nearest 5% by dividing the 20 m² plots on either side of the transect into 1 m² sections and counting the proportion of affected. The canopy height was estimated at 2.5 m increments.

Microsoft Excel and R are needed to open the files.