Vegetative propagation of Solidago canadensis – do fragment size and burial depth matter?

Clonal species may benefit from human disturbance because their vegetative fragments may be distributed via soil. Solidago canadensis is an invasive rhizomatous perennial frequently found in ruderal environments. When creating new infrastructure, digging and cutting are two main factors that may influence the spread of S. canadensis into new areas. To have a better understanding of the invasive potential of S. canadensis, we investigated whether S. canadensis was able to survive and grow from stem cuttings as well as from rhizomes. Rhizomes and cuttings were collected from three populations in Eastern Norway. The rhizomes and cuttings were planted in a pot experiment to assess their vegetative ability to propagate. Rhizome fragments (5 and 10 cm long) were buried at 0.5, 10 and 30 cm depths. The cuttings were planted as 15 cm stems, with the bottom 5 cm pushed into the soil. The results showed that rhizome length did not have an effect on survival. Although some sprouting occurred at all burial depths, increasing depth had a negative effect on rhizome survival. In general, development of the cuttings was good, but there were differences between population performance and survival. These results imply that care must be taken when (i) constructing new sites, because digging and transport of soil masses may spread S. canadensis into new areas by rhizomes or cuttings, and (ii) mowing road verges and other ruderal areas to prevent the spread of stem cuttings from one area to another.


Introduction
Areas invaded by alien species are often ruderal habitats, characterised by high levels of human disturbance, such as cutting, digging and relocation of soil masses from infrastructure and expansion areas (Parendes & Jones, 2000). Disturbance affects plants and may lead to the development of various sizes of fragments of roots and shoots. Clonal species may benefit from such disturbance, depending on burial depth, fragment size and sprouting ability (Li et al., 2013). Fragments of plant species buried in soils at construction and waste dumping sites facilitate greater spread of invasive species into native habitats and gene flow between populations (e.g. Keller et al., 2011). Different populations of the same species may respond differently to the same treatment, due to genetic variation, environmental conditions, age or time of year. Many invasive alien plants are spread by clonal fragments. Baker (1974) identified the ability to form clones and vegetative propagation combined with strong lateral growth as common traits among alien invasive species, even if clonal plants seem to be under-represented among this group of plants (Dietz et al., 2002). To be able to manage the threat of vegetatively propagated alien species, we need a better understanding of their potential to establish and spread.
Solidago canadensis L. (Canadian goldenrod) is a rhizomatous perennial that originates from North America and has been introduced in Europe and temperate Asia (Weber & Schmid, 1998;Xu et al., 2014); in Norway, it has been classified to the highest level of alien species impact ('severe') (Artsdatabanken, 2018), due to its rapid invasion and spread and negative effect on native species, habitats and ecosystems (Elven et al., 2018). This species flowers in the second growth season (Bender et al., 2000), and the small, wind-dispersed seeds ripen about 6 weeks after inflorescence (Pavek, 2011). It is often found at disturbed sites, such as industrial areas, alongside pavements and crop fields, as well as in transitional areas between forests and cultivated land, and in southern Norway, it is one of the most common alien species along road verges and on cultivated land, displacing native species and altering the cultural landscape (Elven et al., 2018). The invasion success of S. canadensis is due to its rapid growth and prolific reproduction (both sexual and asexual) (Kabuce & Priede, 2010), rhizome survival in harsh conditions through the sharing of nutrients between ramets (Hartnett & Bazzaz, 1985) and allopathy that outcompetes other plants (Dong et al., 2006b); the formation of large colonies of S. canadensis may enhance nutrient cycling rates (e.g. Vanderhoeven et al., 2006), inhibit microbiological activities (Deng et al., 2015) and negatively affect biodiversity (Fenesi et al., 2015). Environmental impacts of S. canadensis depend on local climate and recipient communities (Dong et al., 2015b).
Management of S. canadensis in Norway includes application of herbicide during the early growth stage and mowing prior to flowering (cut material may lie on the ground, but not in contact with running water) (Fløistad, 2010). When creating new infrastructure, digging and cutting are two main factors that may lead to the spread of S. canadensis into new areas; however, Weber (2011) found that establishment of S. canadensis rhizomes was erratic. To obtain a greater understanding of the invasive potential of S. canadensis, we investigated population variation in rhizome and stem cutting establishment under contrasting burial regimes that mimicked disturbance conditions during construction work.
The research questions were as follows: (i) Do different rhizome lengths and burial depths affect the regrowth of S. canadensis? (ii) Do stem cuttings of S. canadensis propagate? (iii) Do different populations of S. canadensis show different regeneration capacities?

Site selection
Three sites in eastern Norway ( As and Røyken in Akershus County, and Drammen in Buskerud County) were selected for this study in which S. canadensis was the predominant species. All three sites are dispersed on both sides of the Oslo Fjord. Among the sites, annual normal temperatures are greatest at Drammen, and lowest temperatures and greatest amounts of precipitation are at Røyken; lowest levels of precipitation are at As (Table 1).

Sampling
Plant material from each population was collected on a single day (17, 18 and 19 June 2015), where five root clumps were removed from visually healthy plants that had large numbers of stems, and individual rhizomes were cut into 5 and 10 cm long pieces of approximately the same thickness (rhizome thickness was not measured). With existing roots still attached, five replicate single rhizomes (by size and population) were planted in square 25-L pots that were buried at one of three depths (0.5, 10 or 30 cm); the 90 rhizomes were randomly arranged in a grid of 6 9 15 pots ( Fig. 1). Three replicate stem cuttings (hereafter referred to as cuttings) were taken above the first pair of leaves on the same plants from which rhizomes were sampled and cut into 15 cm long pieces (N = 45). The cuttings were individually potted by inserting the lower 5 cm into the soil in the centre of each 2 L pot (Fig. 1).
The soil contained 85% sphagnum peat, 10% sand and 5% granulated clay, with 1 kg of multimix fertiliser (http://www.horticoop.dk/index.php/godning/81multimix-12-6-20-02-mo-npk-12-6-20-mg-mikro) and 4.5 kg of chalk dolomite added per 20 kg of soil. The pH values were 5.5-6.5. All pots were randomised and placed outside to mimic a natural environment. To minimise edge effects from sun exposure, an extra row of pots filled with soil were placed along the border of the pots on the western side for the rhizome experiment. For the cutting experiment, pots filled with soil were placed around the outer edge of all experimental pots. All pots were watered as a group, usually about 15 min each day or as needed, using an overhead irrigation system with spray nozzles.

Measurements
For both the rhizome experiment and the cutting experiment, the height of the tallest shoot was recorded from the first sprouting, and subsequently, on average, every third day from 24 June until 12 August 2015. The termination of both experiments lasted 3 days, from 12 to 14 August 2015. All plants were carefully removed from the pots, keeping roots as intact as possible. The development of the root system was judged visually on two different scales, one for the rhizome experiment and one for the cutting experiment. Both scales ranged from 0 to 6, where 0 is dead and 6 is excellent root development ( Table 2, Figs 2 and 3). All root systems were photographed upon termination.
In the rhizome experiment, the number of aerial shoots was recorded for all depths, and for the rhizomes buried at 10 and 30 cm, the number of shoots that did not reach the surface, their length (cm), and the extent of branching were also recorded. Shoots with inflorescence were also counted. For the cuttings, the number of shoots was counted, and the length of the longest shoot was recorded (cm).

Statistical analysis
We applied a Bayesian hierarchical model that allows experimental design with both effects associated with treatment and population-specific effects. The posterior distributions for the effects of interest were estimated using Integrated Nested Laplace Approximations  ). For all analyses, the mother plant was included as an independent and identically distributed random contribution, and population and treatments were considered as fixed effects. Root development and flowering (present or absent) assumed a binomial distribution, number of shoots assumed a negative binomial distribution and shoot length assumed a gamma distribution. Treatment effects on root development were analysed for living roots, where stages 1, 2 and 3 were classified as some root development and stages 4, 5 and 6 were classified as well-developed root system. The probability of inflorescence development in shoots was described as a function of aerial shoot length. Statistical analyses were performed using R v. 3.2.3 (R Core Team, 2018).

Survival of rhizomes from
As was lower than those from Røyken and Drammen (Table 3); there was >50% mortality of rhizomes from As by the end of the experiment and rhizome survival was greatest from populations at Røyken (Table 4). There was no effect on survival of rhizome length for any of the populations (Table 3). There were minor differences in rhizome survival among the populations at 0.5 cm burial depth (Table 4); however, increasing burial depth had a negative effect on survival (Table 3).

Rhizome development
Rhizome development decreased with increasing burial depth for all populations, but this relationship appeared weaker for the Røyken and Drammen populations than the As population (Fig. 4). There were no effects of burial at 10 cm on root system development, although effects of burial at 30 cm were varied (Table 5). Of rhizomes buried at 30 cm, 2 from the As population, 8 from the Røyken population and 2 from the Drammen population, developed shoots that did not reach the surface, with an average length of 16, 22 and 29 cm respectively.
Aerial shoots developed from 21 rhizomes from Røyken and Drammen and 11 from As, where there were fewer aerial shoots from rhizomes buried at 10 and 30 cm than from the other two populations. Mean length of the longest aerial shoot (500 mm) and number of aerial shoots per rhizome (4) for the As and Røyken populations were similar, and lower for the Drammen population (400 mm long; three shoots). Reaching the surface and starting to produce aerial shoots is a main factor on root system development (Fig. 5). The presence of two aerial shoots resulted in a 50% probability of a well-developed root system, while four aerial shoots increased this probability to >95%.

Inflorescence
None of the rhizome pieces buried at 30 cm developed inflorescence. However, the As population had three rhizomes with inflorescence, all from 0.5 cm burial depth. The Røyken population had 11 rhizomes with inflorescence and three from 10 cm burial depth. The Drammen population had nine rhizomes with inflorescence, eight from the 0.5 cm depth and only one from 10 cm burial depth. Estimated probability of inflorescence increased gradually with shoot height (Fig. 6). The largest increase in mean probability of development of inflorescence occurred when aerial shoot length increased from 500 to 600 mm (20-90%); at 660 mm, the mean probability of developing inflorescence was 98%.

Stem cutting development
Overall, survival and development of cuttings was high; mortality was observed only in 2 cuttings from Cutting is alive, with developed callus One to two new roots, but no root branching 3 One to two roots, without branching More than two new roots, with root branching 4 More than two roots, with branching Considerable root development with many new roots and root branching 5 Considerable root development, with many roots and significant branching Excellent root development with the roots filling most of the pot, and some new rhizomes 6 Excellent root development, numerous long roots with significant branching Excellent root development with the roots filling most of the pot, and development of thick rhizomes the Røyken population. There were no differences in root development (Fig. 4, Table 6), or number of shoots (Table 6) among the three populations; however, shoots were longer in the population from As than those from Røyken and Drammen (Table 6). None of the cuttings developed inflorescences, and we found that several leaves from the As population developed roots (14 August 2015; Fig. 7), unlike those of the other populations.

Discussion
Increasing burial depth had a negative effect on rhizome survival for all three populations. Decreasing survival or vigour of shoot growth from rhizomes with increasing burial depth is common for rhizomatous perennials, as their energy reserves are expended while growing towards the surface (Klimes et al., 1993; Dalbato et al., 2014). There was almost no difference in  survival between the populations at 0.5 cm burial depth, most likely due to the rhizomes containing enough energy to grow new shoots and produce assimilates (Price et al., 2001). However, at increasing burial depth, the number of dead rhizomes and differences in survival between the populations became evident; most of the rhizomes from As were dead, while most of the rhizomes from Røyken survived. Even though more stored energy is required to produce aerial shoots from a deeper burial, none of the burial treatments was fully effective in hindering sprouting for any of the populations. Rhizome length did not have a positive effect on survival. Still, the size of the rhizomes could have an effect, as the thickness of the rhizomes was not recorded for this experiment, and the biomass of the rhizomes might have given a more reliable result. Huang et al. (2015) found that thicker rhizomes of Mikania micrantha had more stored energy, but stolon thickness had a very limited role in the fragment survival. When looking at the underground shoot length of the rhizomes buried at 30 cm depth, there seems to be a threshold level of stored energy needed for rhizomes to reach the surface and survive. Beyond this burial depth threshold, new sprouts may fail to emerge because the carbohydrate reserves in the storage organs become completely depleted (Klimes et al., 1993). This threshold level may also vary between seasons. In this study, the rhizomes were harvested in June and buried down to 30 cm. These rhizomes had a higher survival rate than the rhizomes in a study by  . 4 Average root development for different populations (Drammen, Røyken and As), burial depths (0 cm,10 cm and 30 cm) and rhizome piece lengths (S = 5 cm and L = 10 cm), and for stem cuttings (C), where 0 is dead and 6 is excellent root development (Table 2). Each category is an average from 5 pots for the rhizomes and average from 15 pots for the cuttings. Standard error bars are included. Weber (2011), in which the rhizomes were harvested in early April and most of the rhizomes died at 5 cm burial depth. This may suggest that there are seasonal variations in stored energy of S. canadensis rhizomes. This is consistent with Bradbury and Hofstra (1977) who found lower carbohydrate levels in the rhizomes early in the growth cycle of S. canadensis. This means that the time of year for digging and burying may affect resprouting ability and the survival of the rhizomes, which is also found in other species (Nkurunziza & Streibig, 2011;Verwijst et al., 2018).
To attain a well-developed root system, development of above-ground shoots is important; the probability of attaining a well-developed root system is over 95% when the rhizome has four aerial shoots. Above-ground shoots give the photosynthetic capacity to support respiration and growth and develop root systems (Luo & Zhao, 2015). Solidago canadensis shoots grow rapidly and produce a great number of leaves early in their life cycle (Schmid et al., 1988). This suggests that once a rhizome has shoot growth that reaches the soil surface, it performs well. However, at 30 cm burial depth, this effect is not as obvious, as most of the rhizomes from Røyken and Drammen had not developed aerial shoots even though the rhizomes were still alive. Nonetheless, the rhizomes from Røyken and Drammen had underground shoots that were growing towards the surface. If the experiment had been terminated later, underground shoots may have reached the surface, and these rhizomes may also have attained well-  developed root systems. For Mischantus sacchariflorus, a rhizomatous grass that has ramet growth like S. canadensis, fragmentation and burial at 20 cm was shown to delay sprouting and subsequent growth, although it still managed to sprout (Chen et al., 2015). Similarly, burial depth may delay sprouting of S. canadensis rhizomes, which could lower their competitive ability as they have less time to develop shoots and store assimilates. The probability of inflorescence increases gradually between 400 mm and 650 mm, indicating that the height of inflorescence is not absolute. Schmid and Weiner (1993) and Schmid et al. (1995) did find that S. canadensis flowers, after reaching a certain size. Solidago canadensis normally flowers in its second year of growth when propagated from seeds (Bender et al., 2000). Many of the plants grown from rhizomes in this experiment flowered in their first growth season, indicating that the rhizomes were more than 2 years old.
The As population had rhizomes with the lowest survival, but developed longer shoots on the cuttings than the two other populations. The difference in survival and growth between the rhizomes and cuttings from Røyken, Drammen and As is associated with inherent differences among the populations, as the soil used in the experiment was homogenous and similar for all pots. Solidago canadensis is known to be a species with large genetic variance between populations in Europe (Weber, 1997). The difference in height and longer shoot lengths of the cuttings as Table 6 Cuttings analysed with respect to root development. Roots were considered good (classes 4-6) or bad (classes 1-3), number of shoots protruding, and the length of the shoots protruding. The population at As was the intercept well as the poor performance of the rhizomes of the As population could indicate a different subspecies, although it is uncertain how much genetic variance there is between the different populations in this experiment. There might also be trade-offs between above-ground and below-ground biomass allocation in this experiment. The As population may have a trade-off where the aerial shoots store more energy and have a stronger regenerative ability than the rhizomes, while the opposite may be the case for the Røyken and Drammen populations. This is not in accordance with Schmid and Weiner (1993) and Schmid et al. (1995) who described a linear relationship between allocation of resources to sexual and vegetative growth in Solidago species. However, Werner et al. (1980) found that S. canadensis tends to invest more biomass in sexual reproduction than vegetative reproduction as the plant ages. The As population is probably older than the two other populations and might therefore allocate more energy for inflorescence while storing less in its rhizomes. The younger populations in Røyken and Drammen, in contrast, may have stored more energy in their rhizomes. Also, plants growing on nutrient-rich soil often allocate more energy to above-ground growth (Aerts et al., 1991). The As population likely grows on a slightly more nutrient-rich soil than the other two populations which favours resource allocation towards more stored energy in above-ground biomass. This could help explain why among the cuttings the As population was the tallest, had the longest shoots and was the only population to develop vegetative roots from leaves. Solidago canadensis do affect abiotic and biotic soil composition (Dong et al., 2015a(Dong et al., , 2017; however, neither soil nutrients nor the nutrients in the plants were analysed in this study. Differences in nutrient content in both soil and tissue might have explained some of the differences in establishment between populations.

Implications for management
Although the mortality was higher with increasing burial depth, none of the burial depths were sufficient to completely stop the rhizomes from sprouting. If they manage to sprout, the rhizomes have a high likelihood of re-establishing themselves. A 30 cm burial depth may delay growth and stop development of inflorescence similar to the effect of mowing (Fløistad, 2010), although this is uncertain given the experiment's short timeframe. However, care should be taken at construction sites. Digging and transport of soil masses may spread S. canadensis into new areas by rhizomes or cuttings, as both have shown a capacity to re-establish themselves. Burying does not seem to prevent sprouting. Although our results indicate burying early in the season and at more than 30 cm may delay or hinder sprouting from rhizomes, this must be investigated further. When mowing road verges and other areas, care must be taken to ensure that cuttings are not moved from one area to another. Finally, it is important to manage the spread of S. canadensis by seeds, as sexual reproduction also may enable establishment of new populations (Dong et al., 2006a). Because S. canadensis is a common invasive species in road verges in Norway (Elven et al., 2018), the undertow from passing cars might also contribute to wind dispersal of its seeds.