Invasion of a Horticultural Plant into Forests: Lamium galeobdolon argentatum Affects Native Above-Ground Vegetation and Soil Properties

Horticultural trade is considered the most important pathway for the introduction of non-native plant species. Numerous horticultural plants are spreading from private gardens and public green space into natural habitats and have the potential to alter native biodiversity and ecosystem functioning. We assessed the invasiveness of the horticultural plant Lamium galeobdolon subsp. argentatum. We documented its spread in semi-natural habitats in the surroundings of Basel, Switzerland, over the past decades. We compared the performance of L. g. argentatum with that of the native subspecies Lamium galeobdolon galeobdolon based on surveys in forests and a pot experiment under standardized conditions. We also assessed whether the two subspecies differentially affect native forest vegetation and various physical, chemical and biological soil properties. The horticultural L. g. argentatum has tripled its occurrence in forests in the region of Basel in the last four decades. Lamium g. argentatum had both a higher growth rate and regeneration capacity than the native subspecies. Furthermore, L. g. argentatum reduced native plant species richness and changed the species composition of the ground vegetation, in addition to altering several soil properties in deciduous forests. Lamium g. argentatum should therefore be considered an invasive taxon.


Introduction
Numerous plant species are transported around the world for commercial and horticultural purposes [1,2]. As a side effect, the global horticultural trade became the strongest driver for the introduction of non-native plants [3]. This is because a significant proportion of horticultural plants become naturalized and some even become invasive [4,5]. In Europe, for example, more than 50% of the naturalized non-native plant species have been imported for horticultural purposes [6,7].
Invasive non-native plant species are considered a major threat to native biodiversity [8,9]. They can affect ecosystems by changing native plant diversity and species composition, and in some cases lead to the extinction of native species [10]. Non-native plant species also have the potential to change soil properties including pH and nutrient levels (e.g., phosphorus, nitrogen), and to alter the abundance and composition of soil microbial organisms [11,12]. Furthermore, non-native plants can disrupt symbiotic associations between soil fungi and host plants, which in turn affects biodiversity [13][14][15].
Successful plant invaders are often characterized by a rapid growth rate (including clonal expansion), large seed production, rapid seedling emergence, or high vegetative regeneration capacity after disturbance [16,17]. Any one of these traits, or a combination of them, allows rapid colonization of different types of natural habitats [18]. A comparison of regeneration capacity after disturbance [16,17]. Any one of these traits, or a com of them, allows rapid colonization of different types of natural habitats [18]. A c son of these traits, which are often summarized as performance, with the corres traits of closely-related native species allows an assessment of the invasivenes non-native species [6,19].
In our study, we compared the performance of the horticultural non-native galeobdolon subsp. argentatum (Variegated yellow archangel; Figure 1) escaped in with that of the closely-related native Lamium galeobdolon subsp. galeobdolon (L.) archangel; Figure 1) found in the same habitat. The horticultural non-native gentatum (Smejkal) J. Duvign (synonyms: Galeobdolon argentatum Smejkal, Lam argentatum (Smejkal) H. Melzer, Lamium montanum var. florentinum (Silva Tar.) B Schippmann, Lamiastrum galeobdolon subsp. argentatum (Smejkal) Stace, argentatum (Smejkal) Henker ex G. H. Loos, Galeobdolon luteum var. florentinum Si is a stolonifereous herb belonging to the family Lamiaceae. Lamium g. argentatu stolons up to 1 m long, consisting of multiple nodes with opposite unexpande and primordial roots at each node, connected by incompletely expanded interno leaves are silvery-colored and mostly with purple-colored spots. The yellow flowers are arranged in 2-10 whorls in a terminal, spike-like inflorescence. The are hermaphroditic and chasmogamous, pollinated by bees and bumblebees. S dispersed by ants. Lamium g. argentatum often forms a dense vegetation layer (F Lamium g. argentatum has the potential to hybridize with L. g. galeobdolon [20], a Lamium galeobdolon montanum ((Pers.) Hayek) [21], which also occurs at low abun the study region.   The closely-related native L. g. galeobdolon (Figure 1) has the same growth form and very similar plant traits as the horticultural L. g. argentatum, but the leaves are green in color and the yellow flowers are slightly smaller [22]. Its distribution area includes Central Europe, Great Britain and Ireland, western parts of Russia, and small areas in Turkey and Northern Iran [28]. Both subspecies are often found together in urban and suburban nutrient-rich, moist deciduous and coniferous forests [23]. In some European countries, L. g. argentatum is now more common in forests than the native L. g. galeobdolon [23].
There is a gap in knowledge about the effects of the horticultural L. g. argentatum on native vegetation and soil properties. To assess the invasiveness of L. g. argentatum, we conducted a series of comparative field studies and a pot experiment under standardized conditions, and documented the spread of the subspecies in semi-natural habitats in the surroundings of Basel, Switzerland, in the past decades. In particular, we examined the performance of L. g. argentatum and L. g. galeobdolon in 12 forests by comparing their patch sizes, plant cover, the number of both stolons and inflorescences, the number of nodes per stolon and stolon length. We also assessed whether the two subspecies have different effects on native forest vegetation as well as on various physical, chemical and biological soil properties. Furthermore, we investigated whether distinct patches of the two subspecies differ in their growth over 7 years. Using stolon fragments, we conducted a pot experiment under standardized conditions to examine whether differences in local environmental conditions could explain the observed differences in performance between the two subspecies.
In our study, we tested the following hypotheses: (1) Illegal garden waste dumping is a major source of non-native plants in suburban and urban forests [29,30], with an increasing frequency in the past decades [31]. We therefore expected that the number of L. g. argentatum records at the landscape level has also increased in recent decades. (2) Nehring et al. [23] reported that L. g. argentatum is now more common in suburban and urban forests than the native subspecies. This could be due to better performance of the horticultural L. g. argentatum compared to the native subspecies. We therefore expected that L. g. argentatum has a higher establishment rate of juvenile plants derived from clonal propagation, a greater number of stolons and longer stolons than native L. g. galeobdolon. Furthermore, we hypothesized that the horticultural subspecies has a greater patch growth rate and forms larger plant patches than the native subspecies. (3) We also hypothesized that compared to the native L. g. galeobdolon, the non-native L.
g. argentatum reduces the richness of native plants in the ground vegetation, changes the composition of native plant species and soil properties, and alters the abundance and composition of soil microbial organisms.

Spread of L. g. argentatum in the Region of Basel
The number of mapping areas in which L. g. argentatum was found tripled between the two surveys ( Figure 3a,b). In 1980-1996, individuals of L. g. argentatum were recorded in 18 mapping areas, in 2010-2020 in 54 out of 92 mapping areas (Figure 3a,b). In addition, the frequency of L. g. argentatum occurrences within mapping area increased between the two surveys (Chi 2 = 36.5, df = 3, p < 0.0001; Figure 4). In the 2010-2020 survey, a higher proportion of mapping areas with two or more records of L. g. argentatum was found ( Figure 4). (1) Illegal garden waste dumping is a major source of non-native plants in suburban and urban forests [29,30], with an increasing frequency in the past decades [31]. We therefore expected that the number of L. g. argentatum records at the landscape level has also increased in recent decades. (2) Nehring et al. [23] reported that L. g. argentatum is now more common in suburban and urban forests than the native subspecies. This could be due to better performance of the horticultural L. g. argentatum compared to the native subspecies. We therefore expected that L. g. argentatum has a higher establishment rate of juvenile plants derived from clonal propagation, a greater number of stolons and longer stolons than native L. g. galeobdolon. Furthermore, we hypothesized that the horticultural subspecies has a greater patch growth rate and forms larger plant patches than the native subspecies. (3) We also hypothesized that compared to the native L. g. galeobdolon, the non-native L.
g. argentatum reduces the richness of native plants in the ground vegetation, changes the composition of native plant species and soil properties, and alters the abundance and composition of soil microbial organisms.

Spread of L. g. argentatum in the Region of Basel
The number of mapping areas in which L. g. argentatum was found tripled between the two surveys ( Figure 3a,b). In 1980-1996, individuals of L. g. argentatum were recorded in 18 mapping areas, in 2010-2020 in 54 out of 92 mapping areas (Figure 3a,b). In addition, the frequency of L. g. argentatum occurrences within mapping area increased between the two surveys (Chi 2 = 36.5, df = 3, p < 0.0001; Figure 4). In the 2010-2020 survey, a higher proportion of mapping areas with two or more records of L. g. argentatum was found ( Figure 4).
2.3. Growth of L. g. argentatum and L. g. galeobdolon Patches over 7 Years Patches of the two subspecies differed in growth over 7 years (F 1,6 = 26.16, p = 0.002). Patches of L. g. argentatum increased from 1.5 m 2 to 4.6 m 2 (mean area covered), while those of L. g. galeobdolon increased from 1.2 m 2 to 1.8 m 2 ( Figure 5). The patch increase per year was 0.51 m 2 for L. g. argentatum and 0.10 m 2 for L. g. galeobdolon.

Performance under Standardized Conditions
The establishment rate of juvenile plants (percentage of juvenile plants that emerged from original stolon fragments per month) in the experimental pots was 41% higher in the horticultural L. g. argentatum than in the native L. g. galeobdolon ( Figure 6; F1,92 = 53.28, p < 0.0001), and was affected by the forest of origin of the stolons (F11,92 = 53.28, p = 0.0410).

Performance under Standardized Conditions
The establishment rate of juvenile plants (percentage of juvenile plants that emerged from original stolon fragments per month) in the experimental pots was 41% higher in the horticultural L. g. argentatum than in the native L. g. galeobdolon ( Figure 6; F 1,92 = 53.28, p < 0.0001), and was affected by the forest of origin of the stolons (F 11,92 = 53.28, p = 0.0410). The establishment rate of juvenile plants increased with the time elapsed in the experiment (F 5,590 = 334.55, p < 0.0001; Table S1). In addition, there was an interaction between subspecies and origin ( Figure 6; F 11,92 = 5.75, p < 0.0001), as the establishment rate of L. g. argentatum was higher than that of L. g. galeobdolon in eight forests of origin, equal in three forests of origin, and even lower in one forest of origin. Finally, differences in the establishment rate of juvenile plants of the two subspecies between March and April caused the significant interaction between subspecies and time ( Figure 6; F 5,590 = 9.47, p < 0.0001).

Performance under Standardized Conditions
The establishment rate of juvenile plants (percentage of juvenile plants that emerge from original stolon fragments per month) in the experimental pots was 41% higher i the horticultural L. g. argentatum than in the native L. g. galeobdolon ( Figure 6; F1,92 = 53.28, < 0.0001), and was affected by the forest of origin of the stolons (F11,92 = 53.28, p = 0.0410 The establishment rate of juvenile plants increased with the time elapsed in the exper ment (F5,590 = 334.55, p < 0.0001; Table S1). In addition, there was an interaction betwee subspecies and origin ( Figure 6; F11,92 = 5.75, p < 0.0001), as the establishment rate of L. argentatum was higher than that of L. g. galeobdolon in eight forests of origin, equal in thre forests of origin, and even lower in one forest of origin. Finally, differences in the estab lishment rate of juvenile plants of the two subspecies between March and April cause the significant interaction between subspecies and time ( Figure 6; F5,590 = 9.47, p < 0.0001 and Lamium galeobdolon subsp. galeobdolon (green line) from January to June. Asterisks indicate significant differences between the two subspecies in a particular month (** p < 0.01, **** p < 0.0001, n.s. not significant). Differences were tested using Chi 2 tests. Mean values of all populations ± s.e. are shown; n = 12.
Stolons emerging from juvenile plants were only recorded between June and July towards the end of the experiment (Table S2). The two subspecies did not differ in stolon emergence (percentage of stolons that emerged from juvenile plants; F 1,92 = 0.12, p = 0.722). However, the percentage of stolons that emerged from juvenile plants was affected by node position in the original stolon (F 4,92 = 2.50, p = 0.048), and the forest of origin (F 11,92 = 2.74, p = 0.004). In both subspecies, the percentage of stolons that emerged from juvenile plants was highest in fragments originated from the third node (17.5%), and lowest in fragments from the first node (9.2%). The percentage of emerged stolons of both subspecies combined varied from 6% (Zurich) to 26% (Seprais). The significant interaction between percentage of emerged stolons and forest of origin resulted from a lower percentage of emerged stolons in L. g. argentatum in three forests than that of L. g. galeobdolon.
The biomass of juvenile L. g. argentatum was 12% higher than that of L. g. galeobdolon (Table S3; F 1,91 = 4.90, p = 0.029). Furthermore, the biomass of juvenile plants was influenced by node position (F 4,91 = 5.43, p < 0.001), and forest of origin (F 11,91 = 3.63, p < 0.001). In both subspecies, the biomass of juvenile plants originated from the first node was 40% lower than that from the other four nodes. In addition, there was an interaction between subspecies and forest of origin (F 11,91 = 3.79, p < 0.001), as the biomass of juvenile L. g. argentatum was larger in nine forests of origin than that of L. g. galeobdolon, while the opposite was found in three forests of origin.

Native Ground Vegetation
Of a total of 61 plant species recorded in the ground vegetation, 34 species (55.7%) were found in L. g. argentatum plots, 49 (80.3%) in L. g. galeobdolon plots, 40 (65.7%) in L. g. argentatum control plots, and 38 (62.2%) in L. g. galeobdolon control plots (Table S4). Plant species richness (number of species per 3 m 2 ) was significantly reduced in L. g. argentatum plots compared to that in L. g. galeobdolon plots ( Figure 7, Table 2). This reduction in species richness was more pronounced in spring (62%) than in autumn (45%). The significant interaction between subspecies and season was due to the seasonal differences in plant species richness in plots of the two subspecies ( Figure 7, Table 2). Plant species richness was also positively related to the ground vegetation cover (those of the Lamium subspecies excluded) in all plot types in spring and autumn (in all cases p < 0.001).
lower than that from the other four nodes. In addition, there was an interaction betw subspecies and forest of origin (F11,91 = 3.79, p < 0.001), as the biomass of juvenile L. gentatum was larger in nine forests of origin than that of L. g. galeobdolon, while th posite was found in three forests of origin.

Native Ground Vegetation
Of a total of 61 plant species recorded in the ground vegetation, 34 species (55 were found in L. g. argentatum plots, 49 (80.3%) in L. g. galeobdolon plots, 40 (65.7%) in argentatum control plots, and 38 (62.2%) in L. g. galeobdolon control plots (Table S4). species richness (number of species per 3 m 2 ) was significantly reduced in L. g. argent plots compared to that in L. g. galeobdolon plots (Figure 7, Table 2). This reduction in cies richness was more pronounced in spring (62%) than in autumn (45%). The signifi interaction between subspecies and season was due to the seasonal differences in p species richness in plots of the two subspecies ( Figure 7, Table 2). Plant species rich was also positively related to the ground vegetation cover (those of the Lamium sub cies excluded) in all plot types in spring and autumn (in all cases p < 0.001).  argentatum and Lamium galeobdolon subsp. galeobdolon plots, and the corresponding control plots in both spring and autumn. Mean values ± s.e. are shown, n = 18. Differences were tested using paired t-tests; **** p < 0.0001, n.s. not significant).

Physical and Chemical Soil Properties
Plant-available phosphorus content was higher in L. g. argentatum plots than in L. g. galeobdolon plots (Figure 8, Table 2). There was a tendency for a slightly larger difference in plant-available phosphorus content between plots of the two subspecies in spring than in autumn ( Figure 8, Table 2). Soil moisture was not affected by the two subspecies (Table 2). However, the significant interaction between subspecies and season was the result of a slightly higher soil moisture in L. g. galeobdolon plots than in L. g. argentatum plots in spring, while the opposite was found in autumn. Table 2. Summary of linear mixed effects models testing the effects of subspecies (L. g. argentatum vs. L. g. galedobdolon), forests (three sites), season (spring, autumn) on plant species richness (number of species per 3 m 2 ) and various soil properties. In all analyses, we considered the difference in a variable between plots containing one of the Lamium subspecies and their corresponding control plots. galeobdolon plots (Figure 8, Table 2). There was a tendency for a slightly larger diffe in plant-available phosphorus content between plots of the two subspecies in spring in autumn ( Figure 8, Table 2). Soil moisture was not affected by the two subspecies ( 2). However, the significant interaction between subspecies and season was the resu slightly higher soil moisture in L. g. galeobdolon plots than in L. g. argentatum plo spring, while the opposite was found in autumn.

Soil Enzyme Activity
Acid phosphatase activity (nmol MUB released per g soil and h) was signific lower in spring than in autumn (Figure 9). In spring, acid phosphatase activity di differ between plots with either L. g. argentatum or L. g. galeobdolon and control ( Figure 9). In autumn, however, acid phosphatase activity was higher in L. g. argen plots than in L. g. galeobdolon plots (Figure 9, Table 2). Mean values ± s.e. are shown, n = 18. Differences were tested using paired t-tests; *** p < 0.001, **** p < 0.0001, n.s. not significant).

Soil Enzyme Activity
Acid phosphatase activity (nmol MUB released per g soil and h) was significantly lower in spring than in autumn (Figure 9). In spring, acid phosphatase activity did not differ between plots with either L. g. argentatum or L. g. galeobdolon and control plots ( Figure 9). In autumn, however, acid phosphatase activity was higher in L. g. argentatum plots than in L. g. galeobdolon plots (Figure 9, Table 2).
in autumn (Figure 8, Table 2). Soil moisture was not affected by the two subspecies ( 2). However, the significant interaction between subspecies and season was the resu slightly higher soil moisture in L. g. galeobdolon plots than in L. g. argentatum plo spring, while the opposite was found in autumn.

Soil Enzyme Activity
Acid phosphatase activity (nmol MUB released per g soil and h) was signific lower in spring than in autumn (Figure 9). In spring, acid phosphatase activity di differ between plots with either L. g. argentatum or L. g. galeobdolon and control ( Figure 9). In autumn, however, acid phosphatase activity was higher in L. g. argen plots than in L. g. galeobdolon plots (Figure 9, Table 2). **** *** Figure 9. Acid phosphatase activity (nmol MUB released per g soil and h) in Lamium galeobdolon subsp. argentatum plots, Lamium galeobdolon subsp. galeobdolon plots and corresponding control plots in both spring and autumn. Mean values ± s.e. are shown, n = 18. Differences were tested using paired t-tests; ** p < 0.01, *** p < 0.001, n.s. not significant).

Soil Fungal and Bacterial Community Profiles
The number of bacterial operational taxonomic units (OTUs) was significantly lower in L. g. argentatum plots than in L. g. galeobdolon plots ( Figure 10, Table 2). The significant interaction between subspecies and forest was due to a differential reduction in the number of bacterial OTUs in L. g. argentatum plots in the three forests ( Table 2). The significant interaction between subspecies, forest and season indicates a high seasonal variation in the number of bacterial OTUs across different plot types and forests ( Table 2).
The number of bacterial operational taxonomic units (OTUs) was significantly low er in L. g. argentatum plots than in L. g. galeobdolon plots ( Figure 10, Table 2). The signifi cant interaction between subspecies and forest was due to a differential reduction in th number of bacterial OTUs in L. g. argentatum plots in the three forests ( Table 2). The si nificant interaction between subspecies, forest and season indicates a high seasonal var ation in the number of bacterial OTUs across different plot types and forests (Table 2).

Physiological Profiles of Soil Bacteria
Metabolic activity of the soil microbial community indicated by the average we color development (AWCD) was not influenced by the two Lamium subspecies (Table 2 ANOSIM analysis based on AWCD data also showed that the CLPP patterns did no differ between L. g. argentatum and L. g. galeobdolon plots in spring or autumn in any fo est (in all cases, p > 0.101). In contrast, substrate richness was significantly higher in L. argentatum plots than in L. g. galeobdolon plots in both spring and autumn ( Figure 11, T ble 2).  Mean values ± s.e. are shown, n = 18. Differences were tested using paired t-tests; * p < 0.05, *** p < 0.001, n.s. not significant).

Physiological Profiles of Soil Bacteria
Metabolic activity of the soil microbial community indicated by the average well color development (AWCD) was not influenced by the two Lamium subspecies ( Table 2). ANOSIM analysis based on AWCD data also showed that the CLPP patterns did not differ between L. g. argentatum and L. g. galeobdolon plots in spring or autumn in any forest (in all cases, p > 0.101). In contrast, substrate richness was significantly higher in L. g. argentatum plots than in L. g. galeobdolon plots in both spring and autumn ( Figure 11, Table 2). cant interaction between subspecies and forest was due to a differential reduction in th number of bacterial OTUs in L. g. argentatum plots in the three forests ( Table 2). The sig nificant interaction between subspecies, forest and season indicates a high seasonal var ation in the number of bacterial OTUs across different plot types and forests ( Table 2).

Physiological Profiles of Soil Bacteria
Metabolic activity of the soil microbial community indicated by the average we color development (AWCD) was not influenced by the two Lamium subspecies (Table 2 ANOSIM analysis based on AWCD data also showed that the CLPP patterns did no differ between L. g. argentatum and L. g. galeobdolon plots in spring or autumn in any fo est (in all cases, p > 0.101). In contrast, substrate richness was significantly higher in L. argentatum plots than in L. g. galeobdolon plots in both spring and autumn ( Figure 11, Ta ble 2).  Mean values ± s.e. are shown, n = 18. Differences were tested using paired t-tests; *** p < 0.001, n.s. not significant).

Spread of L. g. argentatum in the Region of Basel
During recent decades, the trade in horticultural plants has increased as more geographic regions have been explored as new sources for horticultural plants [33,34]. The spread of horticultural plants (e.g., Sedum stoloniferum, Cotoneaster horizontalis) from private gardens and/or public green spaces in semi-natural habitats is well documented [29,30,35]. The three-fold increase in the occurrence of L. g. argentatum recorded in our study in the Basel region between the two surveys is similar to those found in the southern Swedish province of Scania between 1980 and 2020 (250%; [36]) and in Britain and Ireland in the same period (260%; [37]). The large increase in the occurrence of L. g. argentatum in the wild could be the result of several factors. In the 1960s to 1980s, L. g. argentatum was a popular horticultural species for decorating private gardens and public green spaces [38]. After that, L. g. argentatum has been replaced by new, more attractive horticultural plant species. As a result, L. g. argentatum was often dumped together with other garden waste in seminatural and natural habitats [31]. Illegal dumping of garden waste into forests adjacent to settlement areas are known sources of the spread of non-native plant species [29][30][31]. However, the establishment and growth of new populations take time, and this can result in delayed awareness of the presence of new plant species in a given habitat. In the case of L. g. argentatum, changes in taxonomy and difficulties in distinguishing this subspecies from different native Lamium subspecies could lead to misidentification of the ornamental plant, and thus further delay the awareness of its spread. For example, L. g. argentatum was considered as a native subspecies of Lamium galeodolon in the past [22].
Lamium g. argentatum has the potential to hybridize with the native L. g. galeobdolon [20]. Wegmüller [39] reported that the hybrid L. g. galeobdolon × L. g. flavidum produced few seeds that did not germinate and crosses between L. g. galeobdolon × L. g. montanum were not successful. So far, however, there is no study that has quantitatively examined this aspect.

Performance of Lamium subsp. in Swiss Forests
Assessing whether certain plant traits determine the invasiveness of non-native plant species is a prerequisite both for predicting which non-native species will become invasive and for managing already established non-native species [40]. One approach used to achieve this goal is to compare traits of non-native species with native species in a given environment [6,16]. In our study, the non-native L. g. argentatum had greater numbers and longer stolons, and a higher growth rate than the native L. g. galeobdolon. These differences in performance may account for the proven successful invasion of the non-native L. g. argentatum in the wild. A pilot study showed that apart from the difference in the number of inflorescences, the two subspecies did not differ in seed set, seed weight and germination rate [41]. This finding rules out the possibility that differences in reproductive performance could be another factor affecting the invasion success of L. g. argentatum. Our finding that growth rate is an important factor in the invasiveness of the non-native horticultural L. g. argentatum coincides with the results of a global meta-analysis [16]. Furthermore, the similarity of the non-native plant species to the native one, as in the present study, is assumed to be another factor favoring the invasiveness of non-native species [42].

Performance under Standardized Conditions
Many clonal invasive plants form dense monocultures by outcompeting native species, which can alter the structure and function of invaded ecosystems [43]. Clonal growth includes physiological integration and clonal storage organs [44,45]. Stolon storage resources become available for clonal fragment regeneration [45]. The higher regeneration capacity (expressed as a percentage of successfully established juvenile plants) of the non-native L. g. argentatum than the native L. g. galeobdolon found in our study agrees with the results of He et al. [46], who showed that a high regeneration capacity of clonal fragments can contribute to the invasiveness of non-native plants. In contrast, Song et al. [47] reported that regeneration capacity of clonal fragments did not differ between the invasive plant Alternanthera philoxeroides and its congeneric native Alternanthera sessilis. Furthermore, the regeneration capacity of 14 introduced stoloniferous species was not higher than that of 25 native stoloniferous species in China [48]. As reported in other studies, the higher biomass of the established juvenile plants of L. g. argentatum found in our study could be related to the higher regeneration capacity of the non-native plants [46,47]. Thus, the effect of clonal fragment regeneration on invasiveness of non-native plants is variable and seems to depend on the species. However, the regeneration capacity of clonal fragments is important for population dynamics and can contribute to the spread and colonization of non-native clonal plants [49].

Impact on Native Ground Vegetation and Soil Properties
In forests, non-native plants have the potential to alter vegetation structure by reducing the richness of native plant species and altering plant species composition [30,50,51]. The extent of reduction in plant species richness in ground vegetation recorded in our study (45-62%) is in the range observed for other non-woody invasive species including Lupinus polyphyllus (21% [52]), Solidago canadensis (66%; [53]), Reynoutria japonica (50% [54]), and for woody invasive species such as Robinia pseudoacacia (42-54% [55]), Lonicera maackii (50% [56]), and Prunus laurocerasus (40% [30]). Furthermore, the L. g. argentatum-induced changes in species composition parallel the results of several studies showing that nonnative plants have the potential to alter native species composition [30,51,57]. In our study, the change in plant species composition was more pronounced in spring than in autumn. This can be explained by the general reduction in both the abundance and richness of vernal geophytes in L. g. argentatum plots (e.g., Anemone nemorosa, Ranucultus ficaria, Arum maculatum). Here, we have to mention that the native L. g. galeobodolon did not affect the richness or composition of native plant species.
Plant-induced changes in physical, chemical and biochemical soil properties, as well as the activity and composition of microbial communities, can play a key role in the invasion success of non-native plant species [58][59][60]. However, a wide range of impacts on nonnative plants on given soil properties have been identified [12]. Our finding that soil moisture was not affected by L. g. argentatum contrasts with the results recorded in other invasive plants [15,61,62]. Furthermore, we found that the presence of L. g. argentatum increased plant-available phosphorus in the soil. This increase could be due to the high activity of acid phosphatase and the substrate richness, both of which can increase the mineralization rate of phosphorus in the soil. Another possibility could be that the two subspecies colonize forest sites with different levels of phosphorus in the soil. However, our small-scale study design and similar plant available phosphorus levels in the control plots and L. g. galeobdolon plots exclude this possibility. Similar changes in plant-available phosphorus have been reported for Ambrosia artemisiifolia [63] and Ageratina adenophora [64]. In contrast, the invasion of Solidago canadenis [53] and Chromolaena oderata [64] did not affect plant-available phosphorus.
Soil enzymes mediate and catalyze a number of biochemical processes in soil that are essential for the provisioning and recycling of soil nutrients [65]. The high activity of phosphatase and that of the inter-correlated β-glucosidase in soil invaded by L. g. argentatum is in line with the results of a global meta-analysis showing that invasive plants increase these soil enzymes [66]. It is therefore assumed that the higher activity of these soil enzymes after plant invasion is due to invasive plants with high nutrient requirements competing with soil microbial communities for available nutrients [67], and that effective limitation might stimulate enzyme production [68].
The soil microbial community is an essential component of the soil ecosystem and influences processes such as soil formation and fertility, nutrient turnover and carbon storage [69]. In our study, the presence of L. g. argentatum reduced soil bacterial richness (number of bacterial OTUs). Similarly, the invasive plant Rhus typhina reduced soil bacterial richness [70]. Other studies, however, reported no effect of invasive plants on soil bacterial richness (e.g., Solidago canadensis [71], Alliaria petiolata [72]). A meta-analysis revealed that besides a high variation in the impact of invasive plant species on soil bacterial communities, only allelopathic invasive plants increased soil bacterial diversity [73]. Soil bacterial richness was also positively affected by the magnitude of cover of the invasive plant Bromus inermis [74]. In addition, under conditions of reduced native plant species diversity, soil microbial diversity was lower [75,76]. Thus, the reduced bacterial richness recorded in our study could be due to the reduced plant species richness recorded in L. g. argentatum plots. However, this was not the case in our study because no relationship between the number of bacterial OTUs and plant species richness was found. The unexpected result could be a result of contrasting effects of changed soil properties and of both altered plant species richness and composition on soil bacterial richness.
The carbon utilization pattern (AWCD) represents a suitable indicator for the overall metabolic activity of soil microbial communities [63,77]. In our study, the presence of L. g. argentatum did not change the activity of the carbon sources. A similar lack of impact was noted in other invasive plants including Reynoutria japonica [78] and Bidens pilosa [79]. In contrast, various other studies demonstrated that non-native plants increase the metabolic activity of soil microbial organisms [30,63,80]. We also recorded an increase in substrate richness in soils invaded by L. g. argentatum. This result is supported by various other studies [63,80,81].
In our study, the presence of L. g. argentatum did not alter the carbon utilization pattern. This indicates that L. g. argentatum had no effect on the catabolic potential and functional diversity of the soil microbial community. In general, the different impacts on soil properties could be caused by differences in plant traits of the non-native species involved [11], differences in habitat and soil types [53], and by the extent of the invasion and its history [82].

Spread of L. g. argentatum in the Region of Basel
Brodtbeck et al. [32] provided detailed information on the occurrence and frequency of L. g. argentatum in the surroundings of Basel, Switzerland, in the years 1980-1996. Their study area covered approx. 765 km 2 , including border areas of France (Alsace) and Germany, divided into 92 units (mapping areas) ranging in size from 3 to 20 km 2 [32]. For each mapping area, the frequency of L. g. argentatum occurrence was presented using the following classes: 0, no record; 1, a single record; 2, 2-3 records; and 3, 4 and more records (Figure 3a). Using the same methods, we repeated the survey in the years 2010-2020 (Figure 3b).

Performance of Lamium subsp. in Swiss Forests
We used the distribution map of L. g. argentatum [83] to select 12 forests with occurrences of this subspecies ( Table 1). Eleven of the forests were evenly distributed over the Swiss Plateau and one forest was in Germany near the Swiss border (Table 1). In each of the 12 forests, we chose a study area of 50 m × 100 m containing both patches of L. g. argentatum and patches of native L. g. galeobdolon. We randomly selected six patches (>1 m 2 ) from both taxa. We measured the area of each patch (to the nearest 0.5 m 2 ) and visually estimated its cover by the respective taxa (in %). To examine whether the two Lamium subsp. differ in performance, we randomly selected three plant individuals from each patch. As measures of performance, we recorded the numbers of stolons and inflorescences, the number of nodes per stolon, and measured the length of the stolons for each plant (in cm).

Growth of L. g. argentatum and L. g. galeobdolon Patches over 7 Years
To examine whether patches of the two Lamium subspecies differ in growth, we selected six patches of similar size (0.8-1.2 m 2 ) from both taxa in the Allschwil forest near Basel. We marked the 12 patches and measured their area to the nearest 0.1 m 2 in September 2010. We repeated the measurement for each patch every September in the years 2011-2017.

Performance under Standardized Conditions
"Pot experiments" and "common garden experiments", as complements to field measurements, allow the investigation of plants under standardized conditions without distracting effects of heterogeneous environmental factors. We conducted a pot experiment to investigate whether small-scale differences in environmental conditions at the sites of origin (e.g., soil properties) could explain differences in patch growth (a measure of performance) between the two taxa. We randomly sampled 10 stolons with each five nodes from both subspecies in the 12 forests (hereafter origin) described in Section 4.1 (Table 1) in September 2016. From each stolon we made fragments of the five nodes with primordial roots and two leaves and a proximal and distal internode length of 1.5 cm. We planted 1200 stolon fragments in pots (diameter: 24 cm; height: 24 cm) filled with standard garden soil (Belflor, Ricoter, Aarberg, Switzerland) in October 2016. In each pot, we planted the five fragments of the same stolon 1 cm below the soil layer, resulting in a total of 240 pots. The pots were placed on a flat roof in the city of Basel. To avoid differences in light exposure, the positions of the pots were changed each week. To assess the establishment of juvenile plants, we checked the pots weekly for leaf sprouting and stolon shoots. At the end of the experiment in July 2017, we harvested the juvenile plants and determined their above-ground and below-ground biomass by drying the plants at 80 • C for 48 h. To assess the impact of L. g. argentatum on both native forest vegetation and soil properties, we selected 18 sites, each containing patches of the garden-escaped horticultural plant and nearby-situated patches of native L. g. galeobdolon. The study sites were evenly distributed among three deciduous forests in the suburbs of Basel, Switzerland. The vegetation type of these forests belongs to the Galio oderati-Fagetum association [84]. The most abundant tree species are European beech (Fagus sylvatica) and sycamore (Acer pseudoplatanus). The ground vegetation is rich in vernal geophytes, including Ranunculus ficaria, Anemone nemorosa and Alium ursinum. The distances between study sites within a forest ranged from 50 to 1280 m and the three forests were 1.7-7.0 km apart.
We installed three adjoining 1-m 2 plots in the center of one similar-sized patch of both taxa. At a distance of 2-3 m from each patch, we set-up three plots of the same size and same spatial arrangement, but without either Lamium subspecies (hereafter referred to as control plots). This small-scale arrangement of three types of plots minimized potential differences in site characteristics.
In a plot chosen at random from the three plots in the patches and from the three control plots, we determined all plants in the ground vegetation (herbs and woody plants up to a height of 40 cm) to the species and visually estimated their cover using the Domin scale [85]. To complete the species list of this patch, we carefully searched for additional species in the other two plots. Thus, the abundance of plant species was based on one plot of 1 m 2 , while for the records of plant species richness, three plots (3 m 2 ) were considered. Plant species were identified and classified as native or non-native to Switzerland according to Lauber et al. [86]. Plant surveys were carried out between April and September 2019, once in spring and once in fall.

Physical and Chemical Soil Properties
We collected five soil samples in one randomly selected plot from the three plots in all patches and the corresponding control plots using a soil corer (depth 5 cm, diameter 5.05 cm, volume 100 cm 3 ) in both late April and early October 2019. The five soil samples from each patch were pooled and sieved (mesh size 2 mm) for further analyses. The same procedure was used for the five soil samples from the control plots. A subsample from each patch and control plot was stored at -80 • C for the assessments of enzyme activity (see Section 4.5.3), soil fungal and bacterial community profile (T-RFLP; see Section 4.5.4), and physiological profiles of soil bacteria (Ecoplates; see Section 4.5.5). We dried the other part of each soil sample at 50 • C for 48 h. We determined soil moisture (%) using the fresh weight to dry weight ratio. Soil pH was assessed in distilled water (1:2.5 soil:water [87]. Total soil organic matter content (SOM, %) was determined as loss-on-ignition of oven-dried soil at 750 • C for 16 h [87]. We assessed plant-available phosphorus by extracting 2 g of soil with 50 mL of acetic acid-sodium acetate buffer (pH 4.8) and shaking for 16 h. After centrifugation at 4000 rpm for 15 min, we determined the phosphorus content (µg PO4 3− /g) of an aliquot using the molybdenum blue method [88].

Soil Enzyme Activity
We determined the activities of β-1,4-glucosidase (EC 3.2.1.21) and acid phosphatase (EC 3.1.3.2) of soil from each 18 patches of the two taxa and the corresponding control plots following the protocol of Saiya-Cork et al. [89]. Details of the method are presented in Appendix A.

Soil Fungal and Bacterial Community Profiles
We extracted DNA of fungi and bacteria in soil samples from each of the 18 patches of both taxa and the corresponding control plots to examine whether L. g. argentatum changes soil fungal and bacterial diversity. We used the T-RFLP method (terminal restriction fragment length polymorphism [90]) to examine potential differences in soil fungal and bacterial community profiles. Microbial diversity activity was expressed as average well color development (AWCD) corrected by the diffusion of the wells (absorbance at 750 nm) and the color development in the control well. Substrate richness was calculated as the number of different substrates used following Zak et al. [91]. The procedure is presented in detail in Appendix A.

Physiological Profiles of Soil Bacteria
Ecoplates assess physiological profiles of bacterial communities. We evaluated the potential influence of L. g. argentatum on the activity and metabolic diversity of soil bacterial communities using Biolog Ecoplates TM (Biolog Inc., Hayward, CA, USA). We examined soil bacterial communities from 18 L. g. argentatum patches, 18 nearby L. galedobdolon patches, and from each of the 18 control plots of both taxa, resulting in a total of 72 samples. Details of the procedure are presented in Appendix A.

Statistical Analyses
All statistical analyses were performed in R [92]. We applied a Contingency test to assess whether the frequencies of L. g. argentatum occurrence (expressed in four classes) in the 93 mapping areas differed between the two surveys (1980-1996 vs. 2010-2020). We used general linear models (GLM) to determine potential differences in performance measures (mean values of the following response variables: patch size, plant cover, number of inflorescences, number of stolons, stolon length per plant, number of nodes; in all variables n = 12) between the two Lamium subspecies. Subspecies and site (n = 12 forests) were included in the GLM models with gamma-distributed errors considering the distribution of the data.
To avoid temporal pseudo-replication, we used linear mixed models (LME) in the nlme package [93] to analyze the effects of the two Lamium subspecies, origin of the stolon fragments (n = 12 forests), and elapsed time on the response variable percentage of juvenile plants that emerged from stolon fragments per month (arcsine square-root transformed; n = 720). Subspecies and origin were included as fixed factor and elapsed time as random factor. We used analyses of variance (ANOVA) to assess the effects of the two Lamium subspecies, origin (12 forests), and node position in the original stolon (four positions) on the response variable percentage of stolon emerged (arcsine square root-transformed; n = 120), and on the response variable total biomass of the emerged juvenile plants (n = 120).
We used linear mixed models (LME) to analyze the effects of the two Lamium subspecies, forest (three forests), and season (spring, autumn) on the response variables above-ground plant species richness (number of plant species/3 m 2 ) and soil properties. Subspecies, forest and season were included in the model as fixed factors, and plot nested in season as a random factor. In all analyses, we considered the difference in a variable between plots containing one of the Lamium subspecies and their corresponding control plots (n = 72). Preliminary analyses revealed inter-correlations between several chemical and physical characteristics of the soil (soil pH vs. SOM: r s = 0.50, n = 144, p < 0.001; SOM vs. plant-available phosphorus: r s = 0.26, n = 144, p < 0.001). In addition, β-glucosidase activity and acid phosphatase activity correlated (r s = 0.60, n = 144, p < 0.001) as did the number of bacterial and fungal OTUs (r s = 0.24, n = 144, p = 0.004). In the final analysis, we therefore considered soil moisture, plant-available phosphorus, acid phosphatase activity, numbers of bacterial OTUs, AWCD (average well color development, 168 h) and substrate richness (168 h).
We used analysis of similarity (ANOSIM) in the vegan package [94] to examine differences in plant species composition and bacterial CLPP patterns between plot pairs (plots containing either of the two subspecies and their corresponding control plots; in both analyses: n = 144). ANOSIM is a nonparametric permutation procedure that allows comparison of between-group and within-group dissimilarities [95]. The procedure calculates R statistics ranging from −1 to 1. R = 0 indicates completely random grouping, while R = 1 when all replicates within groups are more similar than all replicates between groups.

Conclusions
Our study showed that the presence of Lamium g. argentatum reduced native plant species richness, changed the species composition and altered several soil properties in suburban deciduous forests. Our results demonstrate that L. g. argentatum should be considered an invasive taxon like Impatiens glandulifera [15] and Prunus laurocerasus [30]. We suggest that L. g. argentatum should be added to the Black List of Invasive Species in Switzerland as well as in other countries. The occurrence of L. g. argentatum can often be associated with the presence of nearby garden waste dumping sites. Dumping of garden waste should be avoided in forests. Prevention of invasion of non-native plants can further be improved by raising awareness of both garden owners and the general public about negative effects of non-native species that have become invasive [96]. McKinney [97] stressed the need to develop an ecologically better-informed public, as very often people cannot tell whether a species is native or not.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/plants12071527/s1, Table S1: Establishment of juvenile plants; Table S2: Emergence of stolons from juvenile plants; Table S3: Biomass of emerged juvenile plants; Table S4: Vegetation and soil characteristics of the plots. Data Availability Statement: Data presented in this study are available in the article and its Supporting Information Files (Tables S1-S4).