Research Article
Print
Research Article
Weed co-occurrence patterns in revegetation plantings: Implications for managing multiple weed species
expand article infoDiana M. F. Borse‡, Angela J. Brandt§, Nick Waipara|, Margaret C. Stanley‡
‡ Waipapa Taumata Rau – University of Auckland, Auckland, New Zealand
§ Bioeconomy Science Institute – Maiangi Taiao, Dunedin, New Zealand
| Bioeconomy Science Institute – Maiangi Taiao, Auckland, New Zealand
Open Access

Abstract

Most weed research has focused on single species, whereas managers of conservation habitats often deal with multiple co-occurring species. Effective multi-weed management in revegetation plantings is crucial to prevent unintended responses to the removal of single species, which can change the successional trajectory of the planting project. Predicting positive and negative weed associations can help managers prioritise weed removal to prevent unintended responses from co-occurring weeds. Seedling plots were used to identify species co-occurring under each of three target woody weeds in revegetation plantings in Auckland, Aotearoa – New Zealand (Solanum mauritianum, Paraserianthes lophantha, and Ligustrum lucidum) and compare them with seedlings growing under native plants. Twenty-four pairs of weed species were significantly correlated with each other’s occurrence, either positively (often found together) or negatively (rarely found together). Mature, shade-tolerant L. lucidum was significantly correlated (both positively and negatively) with the most weed species (n = 9). Nitrogen-fixing P. lophantha had one positive and one negative pairwise association with other weed species, while allelopathic S. mauritianum had two negative associations. Shared bird dispersal and shade tolerance were associated with an increased probability of co-occurrence for seedlings. While bird-dispersed seedlings arrive at the site together, their associations are subsequently shaped by interspecific interactions, such as competition and facilitation. Higher housing density, resident native species richness, and vascular vegetation cover were all associated with higher weed richness, whereas higher human population density was associated with lower weed richness. Woody weeds, particularly shade-tolerant and bird-dispersed species, should be prioritised by managers of urban revegetation projects.

Key words:

Co-occurrence, environmental weeds, invasion, management, New Zealand, restoration, species interactions

Introduction

Environmental weeds (hereafter referred to as ‘weeds’) are invasive non-native plants that spread rapidly and have negative impacts on native species biodiversity and ecosystem functions (Falk-Petersen et al. 2006; Sax and Gaines 2008; Gallien and Carboni 2017). While management is essential for maintaining the integrity of high-value conservation habitats, investment in ecological restoration at the landscape level, particularly through revegetation, is critical for enhancing biodiversity (Bradshaw 1983; Reay and Norton 1999; Abella and Chiquoine 2019). However, these efforts are increasingly challenged by weed invasion, which can alter successional trajectories, reduce native species establishment, and compromise long-term restoration outcomes (Caut et al. 2009; Collins et al. 2009; Cummings et al. 2012; Wallace et al. 2017). An emerging issue for weed management is that weed invasion research tends to focus on single-species invasions and therefore has been unable to address continual introductions of multiple co-occurring weeds (Brandt et al. 2023). More than 65% of conservation habitats (valuable sites that are managed to reduce weeds) contain multiple invasive plant species, but only 33% of studies investigate co-occurring weeds (Kuebbing et al. 2013).

Conservation management decisions are currently made based on assessments of individual weed species’ impacts due to a lack of understanding of the interactions among co-occurring weeds and how these interactions shape the impacts these weeds have on native species (Rauschert and Shea 2012; Kuebbing et al. 2013; Brandt et al. 2023). The interactions among weeds may be positive, as in the invasional meltdown hypothesis, whereby the presence of non-native species facilitates the spread of other non-native species, accelerating the impact of the accumulated species (Simberloff and Von Holle 1999). Empirical evidence for invasional meltdown has not been conclusive (Gallien and Carboni 2017; Ricciardi and Simberloff 2025), though there is mounting evidence for the importance of positive interactions among invasive species. For weeds, positive interactions such as facilitation, whereby one weed directly shelters or indirectly facilitates another weed by displacing native competitors (Flory and Bauer 2014), may increase invasion success and impacts (Cavieres 2021). However, there remains a bias towards studying negative interactions (e.g. competition) among weeds (Gallien and Carboni 2017; Elton 2020). This bias assumes that competitive relationships among invasive species drive community composition more than positive relationships (Gallien and Carboni 2017). This competition among weeds could contribute to biotic resistance, or the ability of a resident community to prevent the establishment of newly arriving invasive species (Henriksson et al. 2016; Kuebbing and Nuñez 2016). Therefore, examining how the richness of resident, mature weeds correlates with weed seedling richness could indicate the relative contribution of weeds to biotic resistance (negative relationships) or facilitation (positive relationships) in weed establishment. The types of interactions among weed species are likely to shape overall impacts on native species and outcomes of restoration programmes. Beyond this, positive and negative interactions among weeds can influence the outcomes of weed control programmes. Particularly problematic for managers, single-species removal can lead to surprise effects, where a co-occurring weed species is released from competition (Caut et al. 2009; Kuebbing et al. 2013) and the ecosystem impacts increase rather than decrease as a consequence of management (Bergstrom et al. 2009). Therefore, improving understanding of weed species relationships will enable managers to optimise prioritisation of weed targets and increase the efficacy of control efforts.

Revegetation plantings are composed of newly planted vegetation that has not formed a dense canopy and is therefore vulnerable to weed invasion. These plantings are particularly vulnerable where they are adjacent to human settlements. Urban areas can be biodiversity hotspots, particularly in regions with high endemism (Ives et al. 2016), and habitat patches within cities can provide multiple benefits to the environment and to people (Stanley and Galbraith 2024). However, urban areas also contain more residential gardens that can be a source of weeds (Sullivan et al. 2004). Understanding the relative importance of anthropogenic influences, such as urban gardens, and weed interactions can inform management prioritisation of both the weed species combinations and the most vulnerable habitats to target.

Co-occurrence patterns among weeds in urban revegetation plantings were investigated by comparing weeds co-occurring in plots centred around a target weed, one of three highly invasive woody species of relevance to conservation managers (woolly nightshade, Solanum mauritianum; tree privet, Ligustrum lucidum; and brush wattle, Paraserianthes lophantha), paired with a plot centred around a native tree to assess whether differences were simply a function of the site. These weeds were selected because they are widespread, under regional regulatory management in Aotearoa (Auckland Council 2021), and have distinct traits that are likely to affect how they interact with other non-native and native plants. Aotearoa has high levels of both endemism and weed species, and Auckland itself is extremely weedy, making this a highly suitable location to examine both conservation and weed management (Esler 1991, 2006; Myers et al. 2000; Simberloff 2009). The most common, highly impactful weed species that were more or less likely to co-occur with each other within these revegetated communities were identified, indicating positive or negative relationships among mature plants and seedlings. Species attributes such as shade tolerance, dispersal mode, and habitat preferences were predicted to influence whether certain species tend to co-occur or establish in the understorey. Bird-dispersed species were predicted to co-occur more often than species that use other dispersal modes due to seeds of different species arriving together in bird faeces (Loyn and French 1991; Stanley and Lill 2002; Wotton and McAlpine 2015. The relative contribution of plot factors (e.g. species identity of the central individual, canopy cover, and slope) and anthropogenic factors (e.g. human population and housing density) to weed seedling species richness in these urban sites was also investigated. Given the known importance of gardens as a source of invasive plants (Sullivan et al. 2004) and the potential for weeds to facilitate additional weeds (Simberloff and Von Holle 1999), resident weed species richness and housing density were expected to be strong predictors of the number of weed species per plot.

Materials and methods

Study sites

Survey plots were located in revegetation plantings on public land and in native species plantings on smaller blocks of private land in the Auckland region of Aotearoa. Auckland is a coastal city positioned on an isthmus, with a subtropical climate and 1.797 million residents (Statistics 2023). The city extends over an area of 560 km2 that was covered primarily by temperate forest prior to human settlement (McArthur 2017). Plots were located at 20 sites ranging from 0.08 hectares to 121 hectares (Suppl. material 1: table S1). Each site contained species native to Aotearoa (primarily early successional species) planted by private owners or local government as part of revegetation programmes and one or more of the three target weeds (S. mauritianum, L. lucidum, and P. lophantha). Revegetation plantings are a common land use type in Auckland that is often invaded by non-native plant species, including declared weeds legally subject to regulation and management as listed in the Auckland Regional Pest Management Plan (Auckland Council 2021). The three target weeds are common, problematic weeds for regional managers, and each has distinctive traits that likely influence its interactions with other weed species. Solanum mauritianum (Solanaceae) is native to South America and has allelopathic traits, producing chemicals that are toxic to other plants (McAlpine et al. 2021). Solanum mauritianum reaches maturity within 6 months, has high annual seed production, and has a rapid growth rate (Van den Bosch et al. 2004). Ligustrum lucidum (Oleaceae) is native to southern China and has been described as one of the most invasive tree species globally; it has high physiological tolerance to a range of temperature and soil conditions and is shade-tolerant (Fernandez et al. 2020). Paraserianthes lophantha is a common species of Fabaceae native to western Australia that is known to form associations with soil bacteria that fix atmospheric nitrogen into bioavailable forms (Impson et al. 2011).

Plot surveys

To assess the differences in plant assemblages growing with each target weed, a total of 162 plots were sampled, with 27 centred around each of the three target weed species (n = 81 weed-centred plots), each paired with a native tree-centred plot to assess whether differences were simply a function of the site (Suppl. material 1: table S2). Several native species were represented in the native-centred plots: Coprosma repens, Coprosma robusta, Cordyline australis, Cyathea dealbata, Hedycarya arborea, Kunzea ericoides, Leptospermum scoparium, Melicytus ramiflorus, Metrosideros excelsa, Myrsine australis, Piper excelsum, Pittosporum crassifolium, Pseudopanax lessonii, Sophora microphylla, and Vitex lucens. These native species are commonly used in revegetation plantings. Native species-centred plots were within 10 m of the weed-centred plot and contained a central individual of a similar height to the weed species in the paired plot. All central individuals were at least 150 cm tall to enable measurement of diameter at breast height (DBH at 135 cm). Each plot was at least 3 m apart from any other plot pair (weed-centred and paired native-centred plot), but the plots were unevenly distributed across the 20 sites surveyed (Suppl. material 1: table S1). Solanum mauritianum plots were sampled at eight sites (mean = 4 plots per site), L. lucidum plots at 10 sites (mean = 3 plots per site), and P. lophantha plots at eight sites (mean = 3 plots per site). In total, 81 plot pairs were sampled from December 2022 to November 2024, primarily during the austral summer months (December–February).

Temporary 4 m2 circular plots were established by placing a peg with a 1.13 m string attached at the base of the central individual (weed or native) and circumscribing an area of 4 m2. Attribute data were collected for each plot, including canopy and ground cover, slope, height and DBH of the central individual, and presence of disturbance (e.g. footpaths, fences, dumped refuse) within the plot. Three forest canopy density measurements taken from three points equidistant from the central individual, with a random starting point, were averaged using a spherical densitometer at 30 cm above ground level to capture cover at seedling height. Vascular plant cover was estimated as the percentage of the ground area within the plot covered by vascular plants. Housing density within 250 m was determined by counting houses using Google Earth with satellite imagery from 31 December 2020. Human population density for the statistical reporting areas (defined by the government census in Aotearoa) around each site in 1996 and 2023 and the number of dwellings in 2023 were sourced from StatsNZ (Statistics 1996, 2023). Statistical reporting areas vary in size, but this was accounted for by calculating population density from the area (in km2) of each statistical reporting area.

All individual plants taller than 5 cm within each plot were identified and categorised by height tier to determine the number of individuals per species in each height category. Height tiers of < 15 cm, 15–50 cm, 50–150 cm, and > 150 cm were used, following standard vegetation survey methods in Aotearoa (Allen 1993). Species were classed as native or non-native, and a subset of non-native species was classed as ‘environmental weeds’, defined as those included in McAlpine and Howell’s (2024) list of environmental weeds in Aotearoa. Environmental weeds were the focus of the analyses, rather than low-priority, non-native plants with lesser or ephemeral impacts on native plant restoration. The analysis primarily focused on seedlings categorised by their growth habit (according to the classification in the NZ National Vegetation Survey databank) (Landcare Research 2002). Forbs, vines, and grasses were included as seedlings if they were < 15 cm tall (the first height tier), and trees and shrubs were considered seedlings if they were < 50 cm tall (the first two height tiers) to allow for a temporal delay between the establishment of the central individual and the seedlings. The remaining plants not considered seedlings were then defined as ‘resident’ in the plot.

Statistical analysis

All analyses were performed using the R programming language (R Core Team 2025). To understand how weed species assemblages varied among plots, compositional analysis of weeds in the three weed-centred plot types was undertaken using functions in the vegan R package (Oksanen et al. 2025). Compositional differences were first visualised with non-metric multidimensional scaling (NMDS) based on Bray–Curtis distances (Clarke 1993). To balance the contributions of dominant and rare seedlings, a square-root transformation was performed on the species abundance data prior to analysis (Clarke 1993). The compositional differences among plot types (S. mauritianum, L. lucidum, and P. lophantha) were then analysed using PERMANOVA (Anderson 2001). Native-centred plots had greater variation than weed plots; therefore, the native-centred plots were removed from the compositional analysis to allow for visualisation of differences among weed plots. Beta dispersion was mapped, and ANOVA was used (Suppl. material 1: fig. S1) to ensure that PERMANOVA results reflected differences in the assemblages and not the shapes of the distributions. Similarity percentage (SIMPER) analysis was employed to determine the species with the greatest contribution (up to 80%) to the compositional differences among the three weed-centred plot types (Clarke 1993).

The probability of co-occurrence and positive and negative relationships between the 15 most common weeds in all plots (S. mauritianum, L. lucidum, P. lophantha, and native-centred plots) were analysed using Veech’s probabilistic model (Veech 2013), implemented using the cooccur R package (Griffith et al. 2016). A threshold of p < 0.05 was set for significant positive or negative relationships between species (Veech 2013). To test whether seedlings of species dispersed by birds or those that were shade-tolerant had a greater probability of co-occurrence, a two-part hurdle model was fitted, with a binomial GLM to model whether species ever co-occurred and a gamma GLMM to model the magnitude of co-occurrence among pairs that co-occurred, using the glmmTMB R package (Brooks et al. 2017; McGillycuddy et al. 2025).

To understand which factors most influenced environmental weed species richness, a model selection process (Burnham and Anderson 1998) was performed using the following predictors in a Poisson generalised linear model (GLM), without any interactions between predictors: height of the central individual, central individual DBH, slope, percentage canopy cover, site (as a random effect), percentage vascular plant cover, disturbance in the plot, whether the plot was centred around a weed or native species, richness of resident native species, richness of resident weed species, number of houses within 250 m of the site, human population density of the statistical reporting area in 1996 and 2023, and number of dwellings in the statistical reporting area in 2023. Models containing site as a random effect and models without site were compared using ANOVA to determine whether site influenced weed seedling richness. Data from all plots (weed- and native-centred) were included in the modelling. Determining the relationship between weed seedling richness and resident native and weed species richness enabled the examination of evidence for biotic resistance to invasion or facilitation conferred by resident species (Gioria et al. 2023). Multicollinearity among variables was determined using variance inflation factors (VIF; car R package; Fox and Weisberg 2019), and variables with scores greater than five were removed. Human population density and the number of dwellings from the 2023 census data (Statistics 2023) were positively correlated. Given that the number of houses within 250 m of the site was already included in the model and was at a finer resolution than the census data, the number of dwellings was omitted, and human population density was retained in the model. The remaining variables were used in model selection using Akaike’s information criterion corrected for small sample sizes (AICc) with the dredge function in the MuMIn R package (Bartoń 2025). No evidence of overdispersion was found using tests within the performance R package (Lüdecke et al. 2021); thus, a Poisson GLM was used for the top model identified by AICc using the MASS R package (Venables and Ripley 2002). The effects R package (Fox 2003; Fox and Weisberg 2018, 2019) was used to visualise the effects of the factors included in the top model selected with the lowest AICc score.

Results

Overall, the proportion of native species in the seedling stage did not vary between weed- and native-centred plots (ANOVA: F = 1.8, p > 0.1, Fig. 1). The top 10 most common seedlings found in each plot type included 23 species in total, of which seven were native, 16 were non-native, and 12 were environmental weeds (Table 1; McAlpine and Howell 2024). Just over two-thirds of these common species have fleshy-fruited, bird-dispersed seeds. The most common native seedlings across all plots were three commonly planted revegetation species (Coprosma robusta, Pittosporum crassifolium, and Myrsine australis) and Lophomyrtus obcordata, a declining species in Aotearoa planted in high numbers at one site specifically to assist its conservation (Table 1). Coprosma robusta seedlings were also the most common species in all plot types except S. mauritianum-centred plots. Plots centred around P. lophantha had the highest proportion of native species among the most common seedlings, followed by L. lucidum- and native-centred plots, while S. mauritianum-centred plots had the lowest proportion of common native species (Table 1). There were 12 common weeds across all plot types (four trees, four vines, three forbs, and one shrub), half of which have fleshy-fruited, bird-dispersed seeds. Conspecific weed seedlings were common under each of the weed-centred plots. Ligustrum lucidum was ubiquitous, being common in native species-centred plots (n = 14), but also in P. lophantha-centred plots (n = 5) and L. lucidum-centred plots (n = 9). A congener, L. sinense, was common in native-centred (n = 20 plots) and L. lucidum-centred plots (n = 12). Tradescantia fluminensis, an invasive forb, was also common in many plots (native-centred = 15; L. lucidum-centred = 8; S. mauritianum-centred = 6).

Figure 1.

The proportion of seedlings that are native species in plots centred around native species and three target weeds. Data were collected in revegetation plantings in the Auckland region using 4 m2 plots centred around either a native species, Solanum mauritianum, Ligustrum lucidum, or Paraserianthes lophantha (n = 81 native-centred plots, n = 27 plots centred around each target weed). Horizontal lines represent the median values, the bounds of the boxes represent the interquartile range (IQR), notches represent the 95% confidence intervals, and the whiskers represent the largest and smallest data points, respectively, within 1.5× the IQR.

Table 1.

The 10 most common species found as seedlings, ordered from most (top) to least (bottom) common within each plot type, found in plots centred around native trees, Solanum mauritianum, Ligustrum lucidum, and Paraserianthes lophantha in 4 m2 plots in revegetation plantings in Auckland. Native species are in bold, and species included on the Aotearoa environmental weeds list (McAlpine and Howell 2024) are marked with an asterisk. Growth form is listed as tree (T), shrub (S), vine (V), or forb (F), followed by the percentage frequency.

Native species S. mauritianum L. lucidum P. lophantha
Coprosma robusta (T) 8.3% Solanum mauritianum *(T) 9.7% Coprosma robusta (T) 9.5% Coprosma robusta (T) 11.9%
Ligustrum sinense * (T) 5.1% Coprosma robusta (T) 8.0% Ligustrum sinense *(T) 8.1% Paraserianthes lophantha *(T) 5.9%
Tradescantia fluminensis *(F) 3.8% Ranunculus repens (F) 5.3% Ligustrum lucidum *(T) 6.1% Ligustrum lucidum *(T) 5.0%
Ligustrum lucidum *(T) 3.5% Tradescantia fluminensis *(F) 5.3% Tradescantia fluminensis *(F) 5.4% Lophomyrtus obcordate (S) 5.0%
Zantedeschia aethiopica *(F) 2.3% Calystegia silvatica *(V) 2.7% Lonicera japonica *(V) 4.1% Pittosporum crassifolium (T) 5.0%
Asparagus scandens *(V) 2.0 Melicytus ramiflorus (T) 2.65% Podocarpus totara (T) 4.1% Solanum mauritianum *(T) 4.0%
Myrsine australis (T) 2.5% Smilax asparagoides *(V) 1.8% Zantedeschia aethiopica *(F) Tradescantia fluminensis *(F) 4.0%
Ranunculus repens (F) 2.5% Hedychium gardnerianum *(F) 1.8% Alectryon excelsus (T) 3.4% Alectryon excelsus (T) 3.0%
Solanum mauritianum *(T) 2.5% Onopordum acanthium (F) 1.8% Asparagus scandens *(V) 3.4% Araujia sericifera *(V)
Corynocarpus laevigatus (T) 2.3% Araujia sericifera *(V) 1.8% Galium aparine (F) 3.4% Asparagus scandens *(V) 3.0%

Weed composition within target weed-centred plots

Weed seedling composition differed significantly between S. mauritianum- and L. lucidum-centred plots, with S. mauritianum plots having higher values in dimension 2 (Fig. 2; PERMANOVA, pseudo-F = 1.80; P = 0.02). Weed seedling composition of P. lophantha-centred plots did not differ as greatly from that of the other two weed-centred plot types.

Figure 2.

NMDS of weed seedling composition data collected in revegetation plantings in the Auckland region using 4 m2 plots centred around either Solanum mauritianum, Ligustrum lucidum, or Paraserianthes lophantha (n = 81 plots).

SIMPER analysis showed that nine to 11 species in each combination of plot types contributed to up to 80% of the compositional differences (Table 2). Conspecific seedlings contributed significantly (between 7.7 and 12.8%) to the compositional differences between each plot type. Ligustrum lucidum-centred plots also had more Ligustrum sinense and Lonicera japonica seedlings than either S. mauritianum- or P. lophantha-centred plots (Table 2). While L. japonica was often found in L. lucidum plots, other vines such as Smilax asparagoides, Calystegia sylvatica, and Araujia sericifera were more commonly found in plots centred around S. mauritianum and P. lophantha (Table 2). For S. mauritianum-centred plots, Tradescantia fluminensis was an important contributor to compositional differences and was more common in S. mauritianum plots than in the other two plot types (Table 2).

Table 2.

SIMPER results showing the top 12 species contributing to up to 80% of the pairwise compositional differences between Solanum mauritianum-, Ligustrum lucidum-, and Paraserianthes lophantha-centred plots in terms of the mean number of seedlings per plot centred around individuals of each species.

Species Mean number of seedlings
S. mauritianum plots L. lucidum plots P. lophantha plots
Ligustrum sinense 0.38 1.25 0.47
Tradescantia fluminensis 0.96 0.67 0.39
Ligustrum lucidum 0.11 0.91 0.39
Solanum mauritianum 0.96 0.00 0.21
Lonicera japonica 0.00 0.72 0.21
Paraserianthes lophantha 0.00 0.00 0.72
Zantedeschia aethiopica 0.07 0.21 0.00
Araujia sericifera 0.07 0.20 0.26
Smilax asparagoides 0.30 0.04 0.00
Calystegia silvatica 0.16 0.17 0.61
Asparagus scandens 0.74 0.08 0.35
Hedychium gardnerianum 0.15 0.18 0.00

Weed co-occurrence patterns in revegetation plantings

Analysis of probabilistic co-occurrence relationships between weeds showed that weeds have distinct co-occurrence patterns (Fig. 3). Most of these patterns changed when the focus was shifted from all plants in a plot to only seedlings (either < 15 or < 50 cm tall, depending on growth form), but overall, the same species were common as resident weeds (Fig. 3A) and seedlings (Fig. 3B). Among the 225 weed species pairs with all heights considered, there were 24 significant associations (10.6%), with nine of the associations being positive correlations (37.5%) and 15 being negative correlations (62.5%) (Fig. 3A). Ligustrum lucidum had the greatest number of significantly correlated species of any weed, with five negatively correlated species and four positively correlated species (Fig. 3A). Most of the positive correlations were between shade-tolerant L. lucidum and scrambling or climbing vines, such as Smilax asparagoides, Calystegia sylvatica, and Araujia sericifera. More than 70% of the 15 most common weed species are bird-dispersed, and they have primarily negative correlations between species pairs (Fig. 3). However, two-thirds of the species that had positive correlations were bird-dispersed. There were several positive correlations between species with similar growth forms and habitat preferences, e.g. Lonicera japonica and Asparagus scandens, two scrambling, light-demanding species. However, certain species pairs that share similar growth forms and habitat preferences were negatively correlated, such as the shade-tolerant forbs Tradescantia fluminensis and Zantedeschia aethiopica.

Figure 3.

The probability of co-occurrence among the 15 most common weeds of all heights (A) and among seedlings (B) found in 4 m2 vegetation plots centred around Solanum mauritianum, Ligustrum lucidum, Paraserianthes lophantha, and various native species in revegetation plantings across the Auckland region. The central individuals of each plot were not included in the correlation matrix. Pairings with positive probabilities of co-occurrence (p < 0.05) are shown in teal, pairings with negative probabilities of co-occurrence (p < 0.05) are shown in magenta, and non-significant relationship probabilities are shown in grey. Species that have bird-dispersed seeds are shown in bold.

For seedlings, there were seven significant associations among the 225 weed species pairs (0.3%), with two positive correlations (28.5%) and five negative correlations (71.5%; Fig. 3B). The positive correlations among seedlings included S. asparagoides and T. fluminensis, and Araujia sericifera and T. fluminensis, but these relationships were restricted to the seedling stage. The dispersal methods for these species vary, with S. asparagoides being predominantly bird-dispersed, T. fluminensis dispersing via stem fragments, and A. sericifera by wind. The negative associations from the seedling stage were largely retained when all height tiers were included, apart from those with H. helix, as this species was not common at taller heights. Ligustrum sinense was negatively associated with L. japonica as seedlings and resident plants, but L. sinense was only negatively associated with Z. aethiopica in the seedling stage.

For both native and non-native seedlings, pairs that included bird-dispersed species were more likely to co-occur than not, and this probability appeared to increase as the number of bird-dispersed species in the pair increased from zero to one and from one to two (Fig. 4A, β = 0.65, p < 0.0001). However, the gamma GLMM for the species that did co-occur showed no difference in the magnitude of co-occurrence probability (Fig. 4B, p > 0.5 for linear and quadratic contrasts). Similarly, species pairs with shade-tolerant seedlings were more likely to co-occur than not co-occur, and the probability of any co-occurrence increased as the number of shade-tolerant species in the pair increased from zero to one and from one to two (Fig. 4C, β = 0.43, p < 0.0001). Among the species that did co-occur, the magnitude of the probability of co-occurrence did not vary as the number of shade-tolerant species in the pair increased (Fig. 4D, gamma GLMM, p > 0.9).

Figure 4.

The associations between the probability of co-occurrence and the number of species in a pair that were bird-dispersed (A, B) and shade-tolerant (C, D) for native and non-native seedlings in revegetation plantings in the Auckland region using 4 m2 vegetation plots centred around either native species, Solanum mauritianum, Ligustrum lucidum, or Paraserianthes lophantha. The probability of any co-occurrence increased as the number of bird-dispersed species in the pair increased from neither to both species being bird-dispersed (A), and the magnitude of co-occurrence probability among species that did co-occur did not differ between pairs with neither, one, or both bird-dispersed species (B). The probability of any co-occurrence increased as the number of shade-tolerant species in the pair increased from neither to both species being shade-tolerant (C), and the magnitude of co-occurrence probability among species that did co-occur did not differ between pairs with neither, one, or both shade-tolerant species (D).

Factors influencing weed seedling richness in revegetation plantings

Model selection yielded a model for predicting weed seedling richness in the revegetation plots that included the number of houses within 250 m of the site, vascular plant cover, resident native species richness, and site (AICc = 701.5, Tables 3, Suppl. material 1: table S3). The site was a non-significant predictor of weed seedling richness (p > 0.1 from ANOVA comparing models with and without site as a random effect). Whether the plot was centred around a weed or a native species did not appear as a significant predictor in the top models. The number of houses within 250 m of the site, the percentage of the plot covered by vascular plants, and the richness of resident native plants in the plot were positively correlated with weed seedling richness in the plots (Fig. 5A–C, p < 0.01). The effect of housing increased weed richness from 3.36 species (95% CI [2.82, 4.01]) at zero houses to 8.84 species (95% CI [6.45, 12.11]) when there were 300 houses within 250 m. Similarly, the effect of vascular plant cover increased weed richness from 3.86 species (95% CI [3.31, 4.50]) at 2% cover to 6.00 species (95% CI [4.76, 7.58]) when vascular plant cover increased to 100%. The effect of the richness of resident native species also increased weed richness as the number of native species increased, from 3.55 weed species (95% CI [3.03, 4.17]) where there were zero resident native species to 6.54 weed species (95% CI [5.33, 8.02]) where there were six resident native species. The human population density of the statistical reporting area around the site was negatively correlated with weed richness (Fig. 5D; p < 0.001). This effect showed the highest weed richness (95% CI [4.25, 5.75]) where the population density was 40 people per km2 (4.94 species), and weed richness decreased as the population density increased to 6000 people per km2 (3.05 species; 95% CI [2.01, 4.63]).

Figure 5.

Effects plots of model-selected factors influencing weed seedling richness in revegetation plantings in the Auckland region using 4 m2 vegetation plots centred around either native species, Solanum mauritianum, Ligustrum lucidum, or Paraserianthes lophantha. The number of houses within 250 m of a site (A), the richness of resident native species in the plot (B), the percentage of vascular plant cover in the plot (C), and the human population density of the area in 1996 (D) all predicted weed seedling richness.

Table 3.

Model selection results with the following factors: height of the central individual, central individual DBH, slope and canopy cover of the plot, site (as a random factor), the percentage of the plot covered by vascular plants, presence of disturbance in the plot, the richness of resident native species in the plot, the richness of resident weed species in the plot, the number of houses within 250 m of the site, and the human population density of the statistical reporting area in 1996 and 2023. Factors were reduced by determining multicollinearity among variables using variance inflation factors. The model outputs with ΔAICc < 2 are included in the table below.

Dependent variable:
Response variable
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Model 10 Model 11 Model 12
Height of central individual –0.014
(0.023)
Disturbance -0.042
(0.083)
Canopy cover 0.002
(0.004)
Houses within 250 m 0.003*** 0.003*** 0.003*** 0.003*** 0.002*** 0.003*** 0.002*** 0.003*** 0.003*** 0.003*** 0.003*** 0.003***
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001)
Population density 2023 0.00004 0.00005
(0.0001) (0.00005)
Population density 1996 –0.0001** –0.0001** –0.0001* –0.0001** –0.0001** –0.0001** –0.0001** –0.0001* –0.0001** –0.0001**
(0.00004) (0.00004) (0.00004) (0.00004) (0.00004) (0.00004) (0.00004) (0.00004) (0.00004) (0.00004)
Richness of resident natives 0.102*** 0.099*** 0.101*** 0.097*** 0.104*** 0.103*** 0.103*** 0.101*** 0.105*** 0.102*** 0.100*** 0.102***
(0.026) (0.025) (0.026) (0.025) (0.026) (0.026) (0.026) (0.026) (0.026) (0.026) (0.026) (0.026)
Richness of resident weeds 0.040 0.045 0.043
(0.032) (0.032) (0.032)
Slope 0.002
(0.003)
Vascular plant cover 0.005*** 0.005*** 0.005*** 0.005*** 0.004** 0.004*** 0.004** 0.005*** 0.005*** 0.004*** 0.004*** 0.005***
(0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002)
Weed or native status central individual -0.090 -0.102 -0.091
(0.073) (0.074) (0.073)
Constant 0.932*** 0.890*** 1.064*** 1.035*** 0.909*** 0.907*** 1.041*** 0.910*** 0.974*** 0.951*** 0.861*** 0.783**
(0.114) (0.117) (0.156) (0.157) (0.122) (0.118) (0.161) (0.119) (0.131) (0.120) (0.122) (0.343)
AICc 701.061 701.657 701.748 701.978 702.228 702.583 702.852 702.866 702.898 702.993 703.032 703.036
ΔAICc 0 0.596 0.687 0.917 1.167 1.522 1.792 1.805 1.837 1.932 1.971 1.975
Weight 0.116 0.086 0.082 0.073 0.065 0.054 0.047 0.047 0.046 0.044 0.043 0.043
Observations 160 160 160 160 160 160 160 160 160 160 160 160
Log likelihood –344.256 –343.460 –343.506 –342.512 –345.919 –343.923 –345.152 –344.064 –344.081 –344.128 –343.039 –344.150
Akaike inf. crit. 700.512 700.920 701.011 701.024 701.838 701.846 702.303 702.129 702.161 702.256 702.078 702.299
Bayesian inf. crit. 718.963 722.446 722.537 725.626 717.214 723.373 720.754 723.655 723.687 723.783 726.680 723.825
Note: *p**p***p < 0.01

Discussion

Overall, the study found high weed richness, varying from four to 12 weed species in each plot across the sites and beneath different target weed and native species. This is a high number of weeds considering the small size of the sampled areas and that each additional weed species can substantially impact native plantings. Ligustrum lucidum was a key weed species with numerous significant associations, including several positive associations with high-impact environmental weeds, particularly vines, such as Araujia sericifera and Smilax asparagoides, suggesting that this weed should be a high priority for managers (Collins et al. 2009). This is especially apparent given that conspecific seedlings were commonly found under mature plants, so management will be necessary to prevent continued replacement of mature L. lucidum with conspecific seedlings (Esler 1991; McAlpine et al. 2021). Conversely, Paraserianthes lophantha-centred plots had more native species among the most common seedlings than all other plots. Although this may indicate that P. lophantha is a lower risk in facilitating weeds relative to L. lucidum and Solanum mauritianum, it should remain a managed species given that there was no difference in the overall proportion of native seedlings in P. lophantha-centred plots compared with plots centred around the other two weeds or native species. In addition, P. lophantha has several weedy qualities, such as a high growth rate, seed bank longevity, and changes to the soil that can lead to legacy effects (Impson et al. 2011; Kuebbing et al. 2016). Multiple anthropogenic factors were important determinants of weed seedling richness. In contrast to housing density, higher human population density in the past was associated with a decrease in weed richness, which may indicate that the number of houses and their associated gardens is more important than the total number of people in those houses in terms of weed propagule pressure. Overall, patterns of weed co-occurrence indicate a complex network of interacting factors that influence weed dynamics and can inform management decisions.

Weed seedling composition varies under different mature woody weed species

The most common seedling species in all plot types were weeds. The plots that had the highest number of common native seedlings were those centred around P. lophantha. Based on the nitrogen-fixing capacity of P. lophantha, its presence was expected to be associated with an increased abundance and impact of weed species rather than early successional native species adapted to low nutrient levels (Simberloff and Von Holle 1999; Craine et al. 2006; Sheppard 2013; Kuebbing and Nuñez 2016; Gioria et al. 2023). Ferenc and Sheppard (2020) also found in their study on co-occurring species that legumes had positive effects on native species more often than other types of species (e.g. grasses) had positive effects. Further research could confirm whether this association with native species is replicated in other woody legumes and whether growing alongside a nitrogen-fixing weed consistently increases native plant establishment, growth, and survival.

The composition of weed seedlings differed among the three target weed plot types, especially between S. mauritianum- and L. lucidum-centred plots. Conspecific seedlings were major contributors to the differences between each of the weed plot types, potentially indicating either an intraspecific nurse plant effect (Flory and Bauer 2014) or that the seedlings are common due to the large numbers of seeds falling from the central weed. Relatedness may be another important factor for species establishing together. This was evidenced in the plots by a closely related species, Ligustrum sinense, being more common in plots centred around its congener L. lucidum. This finding aligns with previous work conducted in French grasslands, where smaller phylogenetic distances between resident and newly introduced invasive plants corresponded with greater establishment success (Sheppard et al. 2018). The understorey seedling composition is concerning, as it indicates that weeds will not only be replaced by these conspecifics and close relatives but also by other weed species. The understorey is an important tool for predicting the future direction of restoration areas (McAlpine et al. 2021), and the understorey composition observed at revegetation sites in Auckland suggests that weed species will continue to be common in the future. Therefore, effective management of resident weeds is required to accomplish restoration goals (Esler 1991).

Do weeds have more positive or negative relationships with other weeds?

Co-occurrence associations among weeds were mostly negative, but approximately a third were positive, suggesting that common weeds may be capable of both facilitating and inhibiting the establishment of other weeds (Kuebbing and Nuñez 2016). Given the bias in the literature towards weed competition, a redress is required to more fully explore positive associations and their implications for invasion and management (Simberloff and Von Holle 1999; Gallien and Carboni 2017). The presence of positive associations among highly impactful weeds such as L. lucidum, S. asparagoides, A. sericifera, and S. mauritianum indicates that positive relationships between even highly competitive weeds exist and may shape the seedling assemblages at a site depending on the ecological context.

To understand positive and negative relationships, it is necessary to determine whether species co-occur more often than expected by chance after accounting for environmental filtering (Gallien and Carboni 2017). Dispersal is one such filter, and most of the common weeds in this study are dispersed by birds. Species that have bird-dispersed seeds had higher probabilities of co-occurrence than other pairings, indicating that bird dispersal can lead to increased co-occurrence. This pattern requires further investigation. Among weeds that could have arrived together in faeces, most of the relationships between these bird-dispersed weeds (e.g. L. japonica and L. sinense) were negative, suggesting that they are likely to compete. Habitat preferences are another influential filter. Some species with similar habitat preferences were negatively correlated, indicating competitive dynamics, while others were positively correlated. For example, Tradescantia fluminensis and Z. aethiopica were negatively correlated despite sharing a tendency towards growing in moist, shaded habitats, suggesting that these two weeds likely have a competitive relationship. Based on their growth forms, it is likely that Z. aethiopica shades out T. fluminensis when these species grow in close proximity. Conversely, several vines that have somewhat similar growth forms and tolerate shaded environments, such as Calystegia sylvatica and Asparagus scandens, were positively associated with each other. This relationship could be due to the combination of being dispersed together (by birds) and being able to tolerate shade, thus avoiding competition if light is the main limiting resource.

Several vines were positively associated with shade-tolerant L. lucidum seedlings, but most were not found commonly in L. lucidum-centred plots. Asparagus scandens, Araujia sericifera, C. sylvatica, and Lonicera japonica were all positively associated with L. lucidum. It is not clear from this study, with few shade-tolerant woody weed species recorded, whether this is a relationship specific to L. lucidum or whether other shade-tolerant woody weeds (e.g. Prunus campanulata or Syzygium smithii) have the same type of association with vines. This relationship may be due to vines’ ability to climb and outcompete plants with other growth habits in low-light environments. However, if this were the case, more of these vine species would be expected to be common in L. lucidum-centred plots, rather than only L. japonica. A more plausible explanation may be that, as a shade-tolerant weed, L. lucidum can successfully grow beneath resident vines more often than other woody species that are shaded out, such as S. mauritianum and P. lophantha (Fernandez et al. 2020). Furthermore, this may provide a competitive advantage to L. lucidum by reducing the success of shade-intolerant competitors (Flory and Bauer 2014). Given that most of the positively correlated vines were found infrequently in L. lucidum-centred plots, it appears that once L. lucidum trees mature, they may shade out the vines. These findings contrast with a study in the Czech Republic that found that several vines were more positively associated with areas of denser cover than with disturbed, high-light areas (Bitomský et al. 2019). Therefore, if the vines are being outcompeted for light by larger, more established L. lucidum, there would need to be significant shading from these trees. Based on the results, the dynamics between vines and shade-tolerant L. lucidum may shift depending on the life stages of each of the plants involved (Callaway and Walker 1997). The high number of significant correlations observed between L. lucidum and other high-impact weeds warrants further research given the well-established global weed status of L. lucidum (Fernandez et al. 2020), making it a candidate model species for studying weed co-occurrence.

Understanding the role of plant traits in interspecific associations could help optimise management priorities by accounting for facilitation between environmental weeds. In this study, bird-dispersed seeds and shade tolerance appeared to be associated with an increased probability of co-occurrence. Allelopathy, on the other hand, was not associated with either positive or negative associations. Nitrogen-fixing ability has been predicted to be associated with the facilitation of weeds (Craine et al. 2006; Gioria et al. 2023); however, the target nitrogen-fixing weed in this study, P. lophantha, was associated with more native species than the other target weeds. This indicates that assumptions about the impacts of certain traits on facilitation and competition between invasive species, compared with effects on native species, need to be tested and verified.

Factors associated with high weed species richness

Weed seedling richness did not differ between native-centred and weed-centred plots but increased with higher numbers of resident native plants in the plot and greater vascular plant cover. Therefore, it appears that the identity of a single central individual did not affect weed species establishment as much as the overall richness of resident native species in the plot. While resident weed richness did not predict weed seedling richness, many weed seedlings were commonly found under each of the three target weeds. Conspecifics made up a large proportion of these seedlings, indicating that these target weeds are more likely to replace themselves than be replaced by native species (McAlpine et al. 2021). The number of weed species in the seedling stage also increased as vascular plant cover increased. Due to the tendency for monocultures to dominate larger areas, vascular plant cover was expected to correlate with low weed species richness in the seedling stage. However, it appears that vascular plant cover was higher where more weed species co-occurred in a plot. This may indicate that the more co-occurring weed seedlings there are in a plot, the greater the impacts will be on native species (Beaury et al. 2023), assuming that native seedlings will be less able to germinate through higher weed vegetation cover. Neither native species nor weeds appeared to offer significant biotic resistance to the establishment of more weed species as seedlings in a plot (Kuebbing and Nuñez 2016). Rather, the results suggest that restoration actions aimed at benefiting native plant biodiversity may inadvertently offer weeds an advantage as well (Abella and Chiquoine 2019).

Weed seedling richness did not vary based on the plot’s central target weed species. One possible explanation is that the plot’s environmental context masked any effect of the central weed species. For example, if P. lophantha tends to establish where available nitrogen is low, then its presence may raise nitrogen levels to those in plots centred around other target species, or nitrogen levels may have been high in all plots due to high runoff in this urban area, such that nitrogen fixation by P. lophantha would not have a biologically significant impact. However, the most parsimonious explanation is that high weed propagule pressure from factors such as high housing density overwhelmed these finer-scale factors that could influence seedling establishment. Most of the plant species observed are dispersed by birds, consistent with previous work assessing revegetation plantings in Aotearoa (Reay and Norton 1999). Birds are dominant vertebrate dispersers for many weeds in Aotearoa, where there are few native mammalian dispersers and many non-native birds show a preference for non-native fruit in their diet (Wotton and McAlpine 2015). The high number of weeds in Auckland and the presence of frugivorous birds are therefore both likely to contribute to high propagule pressure in these revegetation sites.

There was also a relationship between weed seedling richness and housing density and 1996 human population density. Weed species richness increased with an increase in the number of houses around the site. Due to the gardens associated with residential houses, there is greater potential for propagule pressure from ornamental weeds (Hulme 2020). Increased human population density was also expected to be associated with increased numbers of weed species, based on Sullivan et al. (2004), who found that past population statistics were strong predictors of current high levels of weediness; however, the opposite pattern was found in this study. It appears that higher-density housing solutions such as multigenerational or multifamily homes, which are becoming more common in Auckland and which may quickly increase the density of the human population in an area, would not offer as many opportunities for garden escapees because these developments do not include as many gardens per resident as lower-density housing (Murphy 2008; Carroll et al. 2011). Therefore, the potential for new weeds is high where lower-density housing is being built in the less central parts of the city and lower where apartments and units are being built that house more people with fewer gardens. Management of the increasing number of weed species in the outer parts of the city where housing with gardens is being built will quickly become increasingly complicated and urgent, especially where these developments border revegetation plantings. Prevention is key to weed management, including encouraging responsible gardening among residents near conservation habitats, following guidance for planting future-proof native species suited to the area, and careful planning for shared green spaces (Stanley and Bassett 2015; Farrell et al. 2022; Tabassum et al. 2023).

Conclusion

The results indicate that active management of revegetation plantings is important in areas with higher native species richness due to the increased pressure from multiple weed species. Rather than offering biotic resistance, highly species-rich native plantings in urban environments appear to be favoured by birds, which can disperse predominantly weed seeds to these sites. To ensure that restoration goals are met, removal of shade-tolerant woody weeds and vines should be prioritised to give native species a competitive advantage.

Finding several positive associations was unexpected given the bias in the literature towards studying competition among weeds. While considerable variation was found in pairwise associations between species, certain species might be especially likely to form positive associations. Based on the results, Ligustrum lucidum is one such species, and this might be due to its key trait (shade tolerance), which makes it highly tolerant of competition from other species, or it may facilitate the growth of surrounding weeds, although other factors may be involved. This is a pattern that warrants further investigation – for example, whether it is a general trend for shade-tolerant species and whether L. lucidum is the driver or passenger in the association. Ligustrum lucidum already presents a high risk in terms of its impact but is an especially high-risk species given that it may be associated with an increase in co-occurring weed species. This risk is especially great given that an increased number of weed species was associated with greater vascular plant cover, which can suppress native species regeneration. As a widespread species in Aotearoa, site-led control of L. lucidum to prevent the establishment of both L. lucidum seedlings and co-occurring high-risk weeds in revegetation plantings may be prudent. Site-led control can be accompanied by investment in biocontrol measures such as the work currently underway on the biocontrol of its congener Ligustrum sinense using Leptoypha hospita (lace bug) (Environmental Protection Authority 2015; Shaw et al. 2018; Manaaki Whenua – Landcare Research 2025). Further research on shade-tolerant woody weeds would be useful in identifying solutions that target multiple weed species and therefore lead to better outcomes for restoration projects.

Higher-density housing solutions in urban areas such as Auckland may limit the spread of new weed species from increased human settlement. Unlike housing developments in the past, which included personal gardens, newer builds with apartment-style or multi-unit buildings introduce fewer opportunities for ornamental species to escape and establish as weeds in the surrounding areas. However, prevention, such as encouraging the planting of native or non-weedy non-native species, is vital for improving the success of restoring native species cover in urban areas.

Acknowledgements

The authors would like to thank the property owners and Auckland Council staff who gave permission to access and conduct this research on their land. Thanks to Bella Perham, Ella Speers, Arwen Barendregt, Sandra Anderson, Allan Wu, Julianne Gaviraghi Mussoi, and Kamya Patel for their assistance in conducting field surveys.

References

  • Abella SR, Chiquoine LP (2019) The good with the bad: when ecological restoration facilitates native and non-native species. Restoration Ecology 27: 343–351. https://doi.org/10.1111/rec.12874
  • Auckland Council (2021) Auckland Regional Pest Management Plan 2020–2030.
  • Beaury EM, Sofaer HR, Early R, Pearse IS, Blumenthal DM, Corbin JD, Diez J, Dukes JS, Barnett DT, Ibáñez I, Petri L, Vilà M, Bradley BA (2023) Macroscale analyses suggest invasive plant impacts depend more on the composition of invading plants than on environmental context. Global Ecology and Biogeography 32: 1964–1976. https://doi.org/10.1111/geb.13749
  • Bergstrom DM, Lucieer A, Kiefer K, Wasley J, Belbin L, Pedersen TK, Chown SL (2009) Indirect effects of invasive species removal devastate world heritage island. Journal of Applied Ecology 46: 73–81. https://doi.org/10.1111/j.1365-2664.2008.01601.x
  • Bitomský M, Mládková P, Cimalová Š, Mládek J (2019) Herbaceous climbers in herbaceous systems are shade-tolerant and magnesium-demanding. Journal of Vegetation Science 30: 799–808. https://doi.org/10.1111/jvs.12768
  • Bradshaw AD (1983) The reconstruction of ecosystems: Presidential address to the British ecological society, December 1982. Journal of Applied Ecology 20: 1–17. https://doi.org/10.2307/2403372
  • Brandt AJ, Png GK, Jo I, McGrannachan C, Allen K, Peltzer DA, D’Antonio C, Dickie IA, French K, Leishman MR, Ostertag R, Parker IM, Stanley MC, Suding KN, Bellingham PJ (2023) Managing multi-species plant invasions when interactions influence their impact. Frontiers in Ecology and the Environment 21: 370–379. https://doi.org/10.1002/fee.2658
  • Brooks ME, Kristensen K, van Benthem KJ, Magnusson A, Berg CW, Nielsen A, Skaug HJ, Maechler M, Bolker BM (2017) glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. The R Journal 9: 378–400. https://doi.org/10.32614/RJ-2017-066
  • Burnham KP, Anderson DR (1998) Practical use of the information-theoretic approach. In: Burnham KP, Anderson DR (Eds) Model Selection and Inference: A Practical Information-Theoretic Approach. Springer, New York, 75–117. https://doi.org/10.1007/978-1-4757-2917-7_3
  • Caut S, Angulo E, Courchamp F (2009) Avoiding surprise effects on Surprise Island: Alien species control in a multitrophic level perspective. Biological Invasions 11: 1689–1703. https://doi.org/10.1007/s10530-008-9397-9
  • Craine JM, Lee WG, Walker S (2006) The context of plant invasions in New Zealand: Evolutionary history and novel niches. In: Allen RB, Lee WG (Eds) Biological Invasions in New Zealand. Springer, Berlin, Heidelberg, 167–177. https://doi.org/10.1007/3-540-30023-6_11
  • Elton CS (2020) The Ecology of Invasions by Animals and Plants (2nd ed.). [with contributions by D. Simberloff&A. Ricciardi.] Springer Nature, Switzerland, 261 pp.
  • Farrell C, Livesley SJ, Arndt SK, Beaumont L, Burley H, Ellsworth D, Esperon-Rodriguez M, Fletcher TD, Gallagher R, Ossola A, Power SA, Marchin R, Rayner JP, Rymer PD, Staas L, Szota C, Williams NSG, Leishman M (2022) Can we integrate ecological approaches to improve plant selection for green infrastructure? Urban Forestry & Urban Greening 76: e127732. https://doi.org/10.1016/j.ufug.2022.127732
  • Ferenc V, Sheppard CS (2020) The stronger, the better – trait hierarchy is driving alien species interaction. Oikos 129: 1455–1467. https://doi.org/10.1111/oik.07338
  • Fernandez RD, Ceballos SJ, Aragón R, Malizia A, Montti L, Whitworth-Hulse JI, Castro-Díez P, Grau HR (2020) A global review of Ligustrum lucidum (OLEACEAE) invasion. The Botanical Review 86: 93–118. https://doi.org/10.1007/s12229-020-09228-w
  • Fox JA, Weisberg S (2018) Visualizing fit and lack of fit in complex regression models with predictor effect plots and partial residuals. Journal of Statistical Software 87: 1–27. https://doi.org/252010.18637/jss.v087.i09
  • Henriksson A, Wardle DA, Trygg J, Diehl S, Englund G (2016) Strong invaders are strong defenders – implications for the resistance of invaded communities. Ecology Letters 19: 487–494. https://doi.org/10.1111/ele.12586
  • Impson FAC, Kleinjan CA, Hoffmann JH, Post JA, Wood AR (2011) Biological control of Australian Acacia species and Paraserianthes lophantha (Willd.) Nielsen (Mimosaceae) in South Africa. African Entomology 19: 186–207. https://doi.org/10.4001/003.019.0210
  • Ives CD, Lentini PE, Threlfall CG, Ikin K, Shanahan DF, Garrard GE, Bekessy SA, Fuller RA, Mumaw L, Rayner L, Rowe R, Valentine LE, Kendal D (2016) Cities are hotspots for threatened species. Global Ecology and Biogeography 25: 117–126. https://doi.org/10.1111/geb.12404
  • Kuebbing SE, Nuñez MA, Simberloff D (2013) Current mismatch between research and conservation efforts: The need to study co-occurring invasive plant species. Biological Conservation 160: 121–129. https://doi.org/10.1016/j.biocon.2013.01.009
  • Kuebbing SE, Patterson CM, Classen AT, Simberloff D (2016) Co-occurring nonnative woody shrubs have additive and non-additive soil legacies. Ecological Applications 26: 1896–1906. https://doi.org/10.1890/15-1931.1
  • Landcare Research (2002) National Vegetation Survey Databank. Manaaki Whenua – Landcare Research, New Zealand.
  • Loyn RH, French K (1991) Birds and environmental weeds in south-eastern Australia. Plant Protection Quarterly 6(3): 137–149.
  • Lüdecke D, Ben-Shachar MS, Patil I, Waggoner P, Makowski D (2021) performance: An R package for assessment, comparison and testing of statistical models. Journal of Open Source Software 6: e3139. https://doi.org/10.21105/joss.03139
  • McAlpine KG, Lamoureaux SL, Timmins SM (2021) Understory vegetation provides clues to succession in woody weed stands. New Zealand Journal of Ecology 45: 1–10. https://doi.org/10.20417/nzjecol.45.4
  • McGillycuddy M, Warton DI, Popovic G, Bolker BM (2025) Parsimoniously fitting large multivariate random effects in glmmTMB. Journal of Statistical Software 112: 1–19. https://doi.org/10.18637/jss.v112.i01
  • Myers N, Mittermeier RA, Mittermeier CG, da Fonseca GAB, Kent J (2000) Biodiversity hotspots for conservation priorities. Nature 403: 853–858. https://doi.org/10.1038/35002501
  • Oksanen J, Simpson GL, Blanchet FG, Kindt R, Legendre P, Minchin PR, O’Hara RB, Solymos P, Stevens MHH, Szoecs E, Wagner H, Barbour M, Bedward M, Bolker B, Borcard D, Borman T, Carvalho G, Chirico M, Caceres MD, Durand S, Evangelista HBA, FitzJohn R, Friendly M, Furneaux B, Hannigan G, Hill MO, Lahti L, Martino C, McGlinn D, Ouellette M-H, Cunha ER, Smith T, Stier A, Braak CJFT, Weedon J (2025) vegan: Community Ecology Package. https://doi.org/10.32614/CRAN.package.vegan [January 19, 2026]
  • R Core Team (2025) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.r-project.org/ [January 17, 2026]
  • Rauschert ESJ, Shea K (2012) Invasional interference due to similar inter- and intraspecific competition between invaders may affect management. Ecological Applications 22: 1413–1420. https://doi.org/10.1890/11-2107.1
  • Ricciardi A, Simberloff D (2025) Revisiting invasional meltdown: Mechanisms and consequences of positive non-native species interactions. Biological Invasions 27: e183. https://doi.org/10.1007/s10530-025-03625-1
  • Sax DF, Gaines SD (2008) Species invasions and extinction: The future of native biodiversity on islands. Proceedings of the National Academy of Sciences 105: 11490–11497. https://doi.org/10.1073/pnas.0802290105
  • Shaw RH, Cock MJW, Evans HC (2018) The natural enemies of privets (Ligustrum: Oleaceae): A literature review, with particular reference to biological control. CABI Reviews 13: 1–24. https://doi.org/10.1079/PAVSNNR201813011
  • Sheppard CS, Carboni M, Essl F, Seebens H, Consortium D, Thuiller W (2018) It takes one to know one: Similarity to resident alien species increases establishment success of new invaders. Diversity and Distributions 24: 680–691. https://doi.org/10.1111/ddi.12708
  • Stanley MC, Bassett IE (2015) Environmental weeds in New Zealand: Impacts and management. In: Stow A, Maclean N, Holwell GI (Eds) Austral Ark: The State of Wildlife in Australia and New Zealand. Cambridge University Press, Cambridge, 134–161. https://doi.org/10.1017/CBO9781139519960.009
  • Stanley MC, Galbraith JA (2024) Connecting people with place-specific nature in cities reduces unintentional harm. Environmental Research: Ecology 3: e023001. https://doi.org/10.1088/2752-664X/ad3f22
  • Tabassum S, Beaumont LJ, Farzin S, Staas L, Griffiths G, Ossola A, Leishman MR (2023) Which plant where: A plant selection tool for changing urban climates. Arboriculture & Urban Forestry 49: 190–210. https://doi.org/10.48044/jauf.2023.014
  • Van den Bosch E, Ward BG, Clarkson B (2004) Woolly nighstshade (Solanum mauritianum) and its allelopathic effects on New Zealand native Hebe stricta seed germination. New Zealand Plant Protection 57: 98–101.
  • Wallace KJ, Laughlin DC, Clarkson BD (2017) Exotic weeds and fluctuating microclimate can constrain native plant regeneration in urban forest restoration. Ecological Applications 27: 1268–1279. https://doi.org/10.1002/eap.1520
  • Wotton DM, McAlpine KG (2015) Seed dispersal of fleshy-fruited environmental weeds in New Zealand. New Zealand Journal of Ecology 39: 155–169.

Additional information

Conflict of interest

The authors have declared that no competing interests exist.

Ethical statement

No ethical statement was reported.

Artificial Intelligence (AI) use

The authors accept full responsibility for the content of the manuscript, including the disclosure of any use of AI.

During programming, Microsoft Copilot was used to assist with generating R-code. The generated code was manually reviewed, optimized, and validated by the authors to ensure correct data execution and formatting. No scientific interpretations or claims were made by the AI software.

Funding

This work was supported by the University of Auckland and Bioprotection Aotearoa.

Author contributions

Conceptialisation, Methodology: All. Data curation, Formal analysis, Investigation, Project administration, Software, Visualisation, Writing – original draft; DMFB; Supervision, Validation, Writing – review & editing: MCS, AJB, NW; Funding aquisition, Resources: MCS.

Author ORCIDs

D.M.F. Borse https://orcid.org/0000-0003-0084-6930

A.J. Brandt https://orcid.org/0000-0001-5980-9128

N. Waipara https://orcid.org/0000-0002-8193-8302

M.C. Stanley https://orcid.org/0000-0003-0701-8935

Data availability

Data is available on Figshare: https://doi.org/10.17608/k6.auckland.30842993.

Supplementary material

Supplementary material 1 

Supplementary information

Diana M. F. Borse, Angela J. Brandt, Nick Waipara, Margaret C. Stanley

Data type: docx

Explanation note: table S1. Sites included in the study with area in hectares (Auckland Council Geomaps 2016). table S2. The frequency of plot pairs between native species and weed species as central individuals in order from the most common pairings to the least common. fig. S1. (A) Stress plot for the nMDS of the weed seedlings, stress = 0.081. (B) Goodness of fit for the nMDS of the weed seedlings. (C) The beta dispersion determining whether there is variance in the shape of the dispersions of the nMDS that would account for any differences found in the PERMANOVA.

This dataset is made available under the Open Database License (http://opendatacommons.org/licenses/odbl/1.0/). The Open Database License (ODbL) is a license agreement intended to allow users to freely share, modify, and use this Dataset while maintaining this same freedom for others, provided that the original source and author(s) are credited.
Download file (51.67 kb)
login to comment