Competition between the invasive Impatiens glandulifera and UK native species: the role of soil conditioning and pre-existing resident communities

Himalayan balsam (Impatiens glandulifera) is a highly invasive annual herb that has become extremely prevalent in riparian zones across the UK. The competitive ability of I. glandulifera, both in terms of resource exploitation and allelopathy (i.e., the release of biochemicals that may be toxic to neighbouring plants), is considered a key determinant of its success. Little is known, however, about the effects of the resident community on the establishment and growth of I. glandulifera. Here, we aim to increase our understanding of the competitive ability of this highly invasive plant by investigating the effects of soil conditioning on the performance of four co-occurring native species (Tanacetum vulgare, Urtica dioica, Chelidonium majus and Arabidopsis thaliana). In addition, we also aim to investigate the effect that the pre-existing species composition have on the performance of I. glandulifera seedlings by establishing artificial communities (monocultures and mixtures of four UK native species, including U. dioica). We found negative effects of soil conditioning by I. glandulifera in all four species, either by reducing above-ground biomass, chlorophyll content or both. Monocultures of U. dioica were the only artificial communities that reduced growth of I. glandulifera, and we did not find any support for the idea that a more diverse community may be more resistant to invasion. Our results confirm the high competitive ability of I. glandulifera and highlight how the identity of the natives in the resident community may be key to limit its success.


Introduction
Biological invasions are considered a key driver of global change that can pose a major threat to biodiversity (Vitousek et al. 1997;Pejchar and Mooney 2009;Franklin et al. 2016), but see also Cook-Patton and Agrawal (2014) for evidence of positive effects. Non-native species can negatively affect invaded ecosystems by altering species diversity, community structure and interactions between organisms, which in some cases leads to local extinction of native species (Vilà et al. 2011).
Although not yet fully understood, the success of plant invasions seems to depend to a large extent on their interaction with organisms native to the invaded system. For example, invasive plants often exhibit a superior competitive ability in resource use and acquisition than natives-even in low resource environments (Funk and Vitousek 2007)-that may explain their dominance (see Gioria and Osborne 2014 and references therein). Superior competitive abilities of invasives over natives may not only relate to resource exploitation, but also to other types of competition, e.g., allelopathy-the release of biochemicals toxic to neighbouring plants (Qin et al. 2013).
Invasion success may also depend on the resident community of plants, and the 'diversity-invasibility hypothesis' predicts that high native species richness will increase resistance of ecosystems to invasion (Levine and D'Antonio 1999;Jeschke 2014). In theory, in a species-rich community there is a more complete occupation of most potential niches and more complete and/or efficient uptake of available resources, which would prevent invasion (Elton 1958). However, empirical evidence lacks a consistent diversity-invasibility pattern and supports both direct and inverse relationships between native and exotic species richness (Jeschke 2014;Bjarnason et al. 2017). Contrasting evidence may relate partly to the different spatial scales considered in the studies, with large scales usually supporting a direct relationship and smaller scales an inverse relationship (Fridley et al. 2007). At large scales, environmental heterogeneity and disturbance may promote the coexistence of native and exotic species (Stohlgren et al. 2003(Stohlgren et al. , 2006. At small scales, species interactionsincluding the two above mechanisms (resource competition and allelopathy)-are likely the key drivers determining the invasibility-diversity pattern (Smith and Côté 2019). In addition, patterns of diversity and alien invasion may depend on other aspects-rather than or in addition to species richness-such as the identity of the dominant species in the native community (Funk et al. 2008;Perkins et al. 2011). Native community composition, in terms of presence and/or relative abundance of particular species, may therefore be key to understand patterns of invasion (Emery andGross 2006, 2007). In fact, significant effects of species identity in promoting or resisting plant invasion have been found (Thomsen and D'Antonio 2007;Emery and Gross 2007;Qin et al. 2013;Zheng et al. 2015), highlighting the relevance of incorporating this latter aspect in future work.
Here, we aim to increase our understanding of the role of competition and native community composition in the performance of a highly invasive plant in the UK, the Himalayan balsam (Impatiens glandulifera). In particular, in this study: (1) We aim to investigate the effect of soil conditioning by I. glandulifera on the performance of four co-occurring UK native species (Tanacetum vulgare L. (Asteraceae), Urtica dioica L.
(Urticacaeaea), Chelidonium majus L. (Papaveraceae) and Arabidopsis thaliana L. (Brassicaceae). In order to reduce the potential interference by allelopathic chemicals in the soil and allow differentiation between the negative effects of allelochemicals and resource competition, all species were planted with and without the addition of finely ground activated carbon to the substrate (see Fig. 1; Callaway and Aschehoug 2000). If allelopathic substances are present in the soil, native species that grow in soil previously used (conditioned) by I. glandulifera will have an enhanced performance-in terms of growth and chlorophyll content-when activated carbon is added to the soil (see Vivanco et al. 2004;Prati and Bossdorf 2004). However, if resource competition is stronger than allelopathy, no differences should be observed between growing in soil with and without activated carbon.
(2) In addition, we also aim to investigate the effect that the pre-existing species composition (monocultures and mixture of four UK native species, including U. dioica) has on the performance of I. glandulifera seedlings (see Fig. 2). Following the 'diversity-invasibility hypothesis' (Levine and D'Antonio 1999;Jeschke 2014) we may expect that I. glandulifera will have lower performance-in terms of growth and chlorophyll content-when growing in the mixture of species than in the monocultures.

Study species
Impatiens glandulifera Royle (Balsaminaceae), the Himalayan balsam, is a highly invasive annual herb, native to the Himalayas and introduced to UK in 1839 (Beerling and Perrins 1993). It is now widespread in  Monocultures were established by planting 4 individuals of the same species in each pot; mixture was established by planting 1 individual of each of the 4 species. One week after transplanting the native plants, one seedling of I. glandulifera was transplanted to each pot the UK and Europe, and shows high reproductive output (a single plant can produce 800-2500 seeds), rapid growth and high competitive ability (Beerling and Perrins 1993). There is contrasting evidence of the impact of I. glandulifera on the species richness and diversity of invaded communities, ranging from negative effects (Hulme and Bremner 2006;Rusterholz et al. 2017;Kiełtyk and Delimat 2019) to weak or even no effects (Hejda and Pyšek 2006;Hejda et al. 2009;Č uda et al. 2017). The allelopathic potential of this species, via naphtoquinones release, has been identified (Ruckli et al. 2014a), together with its ability to disrupt mycorrhizal networks with detrimental effects for native species (Ruckli et al. 2014b(Ruckli et al. , 2016.

Seed origin and germination
Seeds of I. glandulifera were collected from three large established populations along the riverbanks (River Taff, Cardiff) in Autumn 2017. A random bulk sample of these seeds was used for this experiment. Seeds were placed on moist filter paper in Petri-dishes and stored in a refrigerator (ca. 4°C) in darkness for 2-months prior to the start of the experiment in order to break their dormancy requirements.
To assess the effect of I. glandulifera on the growth and performance of neighbouring plants, and also to assess the effect of those plants on I. glandulifera, we chose three herbaceous perennials: Tanacetum vulgare L. (Asteraceae), Urtica dioica L. (Urticacaeaea) and Chelidonium majus L. (Papaveraceae) and one annual species: Arabidopsis thaliana L. (Brassicaceae). These species were found to co-occur with I. glandulifera during field observations in Cardiff, UK. Seeds of the native species were purchased from Nicky's Nursery Ltd (Kent, UK) and sown in germination trays in the glasshouse.

Experimental design
Experiment 1 Effect of soil conditioning by I. glandulifera on above-ground biomass and chlorophyll content of native species (see Fig. 1).
Seedlings of I. glandulifera were transplanted at the cotyledon stage (approximately 1-week old) to 1L pots (one seedling/pot) filled with a 3:1 soil (Verve multipurpose compost, UK) to sand (Westland horticultural sand, UK) and left to grow for 2 weeks under glasshouse conditions in Talybont (Cardiff University, Cardiff, UK). Prior to transplanting, activated carbon (AC), NORIT TM SA 2, ACROS Organics TM (Fisher, UK) was added to half of the pots at a rate of 20 ml/L (6 ml per pot). Therefore, there were two different AC treatments: pots without AC (-AC) and with AC (? AC). Following the 2 weeks, the above-ground biomass of I. glandulifera was harvested from a total of 96 pots, which were then used for the 'conditioned' soil treatment for assessing the growth of the native species. Our approach using artificial soil aims to avoid confounding factors of natural field-sampled soil, such as differences in soil characteristics (e.g., pH, texture, organic matter) (Pernilla Brinkman et al. 2010).
After germination, seedlings of the native species were transplanted into the pots conditioned by I. glandulifera, i.e., where I. glandulifera had been previously growing and also to 'control' pots, i.e., ? AC and -AC pots without previous growth of I. glandulifera. The experiment consisted of one species per pot, and there were 12 replicates per each species and AC (? AC, -AC) and soil treatment ('control', 'conditioned') combination (total N = 192) (see Fig. 1). Pots were randomly allocated to avoid position-bias within the glasshouse and plants left to grow for 3 weeks.
Three weeks after growing in the experimental conditions, plants of A. thaliana, T. vulgare, C. majus and U. dioica were harvested, and their above-ground biomass was oven-dried for 72 h at 65°C. Then, the above-ground dry mass was weighed with a precision balance in the laboratory (± 0.0001 g). Prior to harvest, we estimated chlorophyll content by measuring three fully developed young leaves with a handheld chlorophyll meter (SPAD-502, Minolta Camera Co., Osaka, Japan), which calculates an index based on absorbance at 650 and 940 nm. SPAD values are well correlated with the chlorophyll content of leaves (Wood et al. 1993;Markwell et al. 1995).
Experiment 2 Effect of species composition on above-ground biomass and chlorophyll content of I. glandulifera (see Fig. 2).
After germination, seedlings of the native species were transplanted into 1L pots filled with a 3:1 soil (Verve multipurpose compost) to sand (Westland horticultural sand) mixture. A total of 4 seedlings (either of a single species or a 'mixture' of each of the 4 species) were transplanted into each pot. The experiment consisted of establishing 6 different treatments of 10 replicates each: monoculture of 'A. thaliana', 'T. vulgare', 'C. majus', 'U. dioica', 'mixture', and 'none'. The treatment 'none' consisted of pots that were established without any native species. One week after transplanting the native plants, one seedling of I. glandulifera was transplanted to each pot (see Fig. 2). Pots were randomly allocated to avoid position-bias within the glasshouse.
Two weeks after growing in the experimental conditions, the above-ground biomass of I. glandulifera was harvested, oven-dried for 72 h at 65°C and weighed with a precision balance (± 0.0001 g). As above, prior to harvest, we estimated chlorophyll content.

Data Analysis
All data analyses were performed using R version 3.1.2 (R Development Core Team 2014). The effects of 'AC' and 'treatment' on the above-ground dry mass and chlorophyll content of the native plants were analysed using a two-way analysis of variance (ANOVA) by means of the 'aov' function. AC, treatment and their interaction were added as fixed factors to the model. We checked graphically for normality of residuals and homogeneity of variance. The above-ground dry mass for T. vulgare, C. majus, and U. dioica was log-transformed to meet the assumptions of the analysis of variance. The effects of the pre-existing 'species composition' on the above-ground and chlorophyll content of I. glandulifera were analysed using a one-way analysis of variance by means of the 'aov' function. The preexisting 'species composition' was added to the model as fixed factor. We obtained P-values using the 'Anova' function from the 'car package' (Fox and Weisberg 2011) with type III sum of squares. We tested post hoc comparisons between means using Tukey's HSD tests, with a \ 0.05.

Results
Soil conditioning by I. glandulifera: effects on native species The effects of soil conditioning by I. glandulifera on above-ground biomass and chlorophyll content of natives differed between species (see Table 1, Fig. 3). In particular, the above-ground biomass and chlorophyll content were both reduced in C. majus (Table 1, T: P = 0.007 and P = 0.004, respectively; Figs. 3c, 4c) and U. dioica (Table 1, T: P = 0.001 and P \ 0.001, respectively; Figs. 3d, 4d). For T. vulgare, soil conditioning by I. glandulifera reduced the aboveground biomass but did not affect the chlorophyll content (Table 1, T: P = 0.020 and P = 0.549, respectively; Figs. 3b, 4b). In contrast, the above-ground biomass of A. thaliana was not affected, but there was a reduction in chlorophyll content in response to soil conditioning by I. glandulifera (Table 1, T: P = 0.760 and P = 0.008, respectively; Figs. 3a, 4a).
The presence of activated carbon in the soil reduced the above-ground biomass of A. thaliana (Table 1, AC: P = 0.003, Fig. 3a) and the chlorophyll content of T. vulgare (Table 1, AC: P = 0.012, Fig. 4b). The interaction of AC and treatment was only significant for U. dioica (Table 1, AC 9 T: P = 0.009); there was no difference in above-ground biomass between control soil and soil conditioned by I. glandulifera when activated carbon was added (Fig. 3d). However, there was a significant reduction in the above-ground biomass of U. dioica in response to growing in soil conditioned by I. glandulifera when soil was not amended with activated carbon (control soil) (Fig. 3d).

Natives species composition: effects on I. glandulifera
The above-ground biomass and chlorophyll content of I. glandulifera was significantly affected by species composition (Table 2, P = 0.037 and P = 0.013, respectively). In particular, monocultures of U. dioica reduced the above-ground biomass of I. glandulifera (Fig. 5a). However, there was no effect on aboveground biomass of the invasive when growing in monocultures of A. thaliana, T. vulgare and C. majus or in the mixture of species (Fig. 5a). Similarly, the chlorophyll content of I. glandulifera was reduced when growing with U. dioica, and also A. thaliana, but no effect was found when growing with any of the other monocultures or in the mixture of species (Fig. 5b).

Discussion
We found negative effects, either in terms of reduced chlorophyll content, reduced above-ground biomass or both, exerted by the invasive on four different cooccurring native species.
The depletion of limiting resources by previous growth of I. glandulifera is undoubtedly a major contributing factor in generating these results (Schenk 2006). However, plants can alter subsequent growth by changing soil properties, including not only resource depletion, but more broadly its chemistry, structure and biota (Kulmatiski et al. 2008;van der Putten et al. 2013). So, to what extent these results can be explained beyond resource competition?
Our experimental design aimed at disentangling the effect of resource competition and allelopathy, by the addition of activated carbon. However, results from adding AC were not consistent across species. Although AC has been recommended and used extensively for testing for the effects of allelopathy (Inderjit 2003a, b), its presence has in some cases been found to affect-either increasing or reducing-the growth of plants by modifying nutrient availability (Lau et al. 2008). Here, we found negative effects of AC on A. thaliana and T. vulgare, which precludes us from obtaining conclusions about allelopathic influence of I. glandulifera on these species (Lau et al. 2008). No effects of AC were found for C. majus; however, for U. dioica, our results suggest a role for allelopathy in contributing to the competitive ability of I. glandulifera. Interestingly, the allelopathic effect of I. glandulifera has been demonstrated in a previous study with U. dioica as a neighbour (Gruntman et al. 2014).
U. dioica was the only species that reduced growth of I. glandulifera; however, we did not find any support for the idea that a more diverse community may be more resistant to invasion (Levine and D'Antonio 1999;Jeschke 2014). The negative effect of U. dioica over I. glandulifera is not surprising given that it is a highly competitive ruderal species, abundant and dominant in the areas invaded by I. glandulifera, where it often forms monospecific stands (Taylor 2009). In fact, negative effects on the growth of I. glandulifera have been reported before by young U. dioica plants (Bottollier-Curtet et al. 2013) and using U. dioica rhizome segments (Tickner et al. 2001). What is perhaps more surprising is that in the current study, individuals of U. dioica were at the seedling stage when 'artificial invasion' by I. glandulifera was imposed. This indicates that the balance between the competitive abilities of the invasive I.
glandulifera and U. dioica may be affected by the developmental stage of the competing individuals. From our study, it seems that once that I. glandulifera is established, this may have negative consequences on U. dioica. However, at the same time, our study also suggests that the presence of individuals of U. dioica, even at the seedlings stage, may hinder the invasion success of I. glandulifera. The lack of support of our results to the idea that species-rich communities are more resistant to invasion (Elton 1958;Levine and D'Antonio 1999) adds to the studies with contrasting results that found both positive and negative diversity-invasibility relationships (Jeschke 2014;Bjarnason et al. 2017). These studies include both artificial and natural communities, and it has been suggested that artificial Fig. 3 Above-ground biomass of a Arabidopsis thaliana, b Tanacetum vulgare, c Chelidonium majus and d Urtica dioica in response to growing in soil 'conditioned' or not ('control') by Impatiens glandulifera, and in presence or absence of activated carbon (? AC, -AC, respectively). Bars and error bars represent means and SE, respectively (N = 12). Symbols centred above the bars indicate whether the difference between 'conditioned' and 'control' soil is statistically not-significant (ns) or significant (*P \ 0.05, **P \ 0.01). In d different letters above bars indicate significant differences between groups for the treatment and soil interaction (P \ 0.05, Tukey's HSD test) communities may be generally young, quite species poor, and structurally simpler-with similar age plants-than natural communities (Zeiter and Stampfli 2012). In addition, the number of species may not always be the best predictor of how communities function, and several studies have argued about Fig. 4 Chlorophyll content (SPAD units) of a Arabidopsis thaliana, b Tanacetum vulgare, c Chelidonium majus and d Urtica dioica in response to growing in soil 'conditioned' or not ('control') by Impatiens glandulifera, and in presence or absence of activated carbon (? AC, -AC, respectively). Bars and error bars represent means and SE (N = 12), respectively. Symbols centred above the bars indicate whether the difference between 'conditioned' and 'control' soil is statistically notsignificant (ns) or significant (*P \ 0.05; **P \ 0.01; ***P \ 0.001)

Conclusions
Soil conditioned by I. glandulifera negatively affected all chosen native species (A. thaliana, T. vulgare, C. majus, U. dioica). Negative effects were most likely mediated through resource competition, however, our results for U. dioica also suggest a role for allelopathy in contributing to the competitive ability of I. glandulifera. In our artificial communities, monocultures of U. dioica were the only ones to reduce the growth of I. glandulifera, which may relate to the ruderal and highly competitive ability of this dominant native species. Our results do not support the idea that a more diverse community may be more resistant to invasion. Overall, our results indicate that growth of I. glandulifera has negative consequences for native species, and that the identity of the native species in the resident community may be key to limit establishment and therefore the success of I. glandulifera.
Acknowledgements We thank Mr Lyndon Tuck for his invaluable assistance in setting up and looking after the species. Bars and error bars represent means and SE, respectively (N = 12). Different letters above bars indicate significant differences between the different pre-existing species treatment (P \ 0.05, Tukey's HSD test) experiments at the plant growth facilities in Talybont. We also thank Dr. Roiloa, Prof Inderjit and two anonymous reviewers for their useful advice to an earlier draft of this manuscript.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.