Dry weight minimum in the underground storage and proliferation organs of six creeping perennial weeds

Many herbaceous perennial plant species gain significant competitive advantages from their underground creeping storage and proliferation organs (CR), making them more likely to become successful weeds or invasive plants. To develop efficient control methods against such invasive or weedy creeping perennial plants, it is necessary to identify when the dry weight minimum of their CR (CR DW min ) occurs. Moreover, it is of interest to determine how the timing of CR DW min differs in species with different light requirements at different light levels. The CR DW min of Aegopodium podagraria , Elymus repens and Sonchus arvensis were studied in climate chambers under two light levels (100 and 250 μ mol m −2 s −1 ), and Reynoutria japonica , R. sachaliensis and R. × bohemica under one light level (250 μ mol m −2 s −1 ). Under 250 μ mol m −2 s −1 , the CR DW min occurred before one fully developed leaf in R. sachaliensis , around 1– 2 leaves in A. podagraria and E. repens and around four leaves in S. arvensis , R . japonica and R. × bohemica . In addition to reducing growth in all species, less light resulted in a higher shoot mass fraction in E. repens and S. arvensis , but not A. podagraria ; and it delayed the CR DW min in E. repens , but not S. arvensis . Only 65% of planted A. podag-ragra rhizomes produced shoots. Beyond the CR DW min , Reynoutria spp. reinvested in their old CR, while the other species primarily produced new CR. We conclude that A. podagraria , R. sachaliensis and E. repens are vulnerable to control efforts at an earlier developmental stage than S. arvensis ,

ornamental plant that has become a problematic garden weed in Scandinavia and invasive in North America (D′Hertefeldt et al., 2014). The tall dicotyledonous Reynoutria japonica Houtt.
Creeping perennials derive a major competitive advantage from their underground storage and proliferation organs: rhizomes in E.
repens, A. podagraria and Reynoutria spp.; and thickened roots in S. arvensis, hereafter collectively named CR (creeping rhizomes or (thickened) roots). Firstly, CR enable asexual reproduction by creating clonal plants as they grow away from the mother plant. Secondly, as long as the clonal network is intact, it can increase the competitive ability of the clones through the sharing of resources and information (Liu et al., 2016). Thirdly, CR function as exploration organs for finding unexploited resources (Kleijn and Van Groenendael, 1999).
Fourthly, CR store a large proportion of the energy and nutrients captured by the plant and they can, therefore, produce new shoots when the old ones die (e.g. due to winter, ploughing etc.).
Because of the vast energy resources in their CR, it is generally not sufficient to simply destroy the aboveground biomass of perennial weeds (van Evert et al., 2020). Once these species are established, regular herbicide spraying or intensive tillage/cutting is often required to manage them (Håkansson, 2003). Systemic herbicides (e.g. glyphosate) is the most common chemical control method against perennial weeds as they can be transferred down to the CR and consequently has the potential to kill the whole plant. In the absence of herbicides, a common control method in agriculture is to use repeated tillage operations to force the CR to re-sprout over and over, hence starving them of resources (Brandsaeter et al., 2017;Ringselle et al., 2016). The CR can also be starved by repeatedly cutting away the aboveground biomass, but the efficacy of this method vary greatly between perennial weed species (Thomsen et al., 2015). For instance, it is not considered effective against established stands of R. japonica (Jones et al., 2020). To make control measure as effective and resource efficient as possible, it is essential to determine when these weedy plants are most susceptible to different control measures.
Perennial weeds are considered to be at their most vulnerable to disturbance when their CR are at their dry weight minimum (CR DW min ).
This occurs after the loss of their shoot biomass (e.g. by winter or mowing). Until the plants return to the compensation point (i.e. when the photosynthetic production is equal to the respiration loss), respiration and root exudation losses will expend CR resources (Verwijst et al., 2018). The overall plant dry weight will continuously decrease until the plants have reached the compensation point and then continuously increase. In comparison, the dry weight of the storage organs may increase, stabilise or continue to decrease after the compensation point depending on whether the plant prioritises growth or storage.
Mechanical control measures (e.g. tillage or cutting) are usually recommended to be performed no later than at the CR DW min to maximise the starvation effect and prevent the build-up of resources in the storage organs (Håkansson, 2003). In comparison, systemic herbicides are recommended to be applied once the plant is passed CR DW min , since at earlier stages resources are not being transferred to the storage organs, and consequently herbicides are not either (Hunter, 1995).
To be useful to end users (e.g. farmers, gardeners and landscapers), studies on arable weeds have generally sought to discover correlations between CR DW min and easily identifiable developmental stages, such as the number of leaves of the main shoot during the early vegetative phase (cf. the BBCH scale by Lancashire et al., 1991). For instance, CR DW min have been estimated to be just before 3-4 leaves in E. repens (Håkansson, 1967), at 5-7 leaves in S. arvensis (Håkansson, 1969a) and at 4-7 (Gustavsson, 1997)

or 8 leaves in
Cirsium arvense (L.) Scop. (Creeping thistle) (Nkurunziza and Streibig, 2011). However, more recent studies have placed CR DW min for S. arvensis at four leaves (Tavaziva, 2012) and C. arvense at 3-4 leaves (Verwijst et al., 2018). The discrepancy between different estimates illustrates the importance of both revisiting old CR DW min estimates and determining CR DW min of unstudied species.
One reason CR DW min estimates differ is that the allocation of resources is influenced by environment, in particular resource availability (Poorter et al., 2012), temperature (Tørresen et al., 2020) and biotic factors, such as competition. For example, Håkansson (1969b) found that a lower light level delayed CR DW min for E. repens, and Tavaziva (2012) found that inter-specific competition delayed CR DW min for S. arvensis. Thus, light availability plays a significant role in deciding at which developmental stage CR DW min occurs. However, relatively little is known about how the same change in light availability affect CR DW min of different creeping perennials, in particular species with different light requirements such as light-demanding agricultural weeds compared to more shade-tolerant species.
The concept of resource sinks and sources can explain many aspects of plant growth in different plant species (White et al., 2016).
Exploiting phenological changes in CR source-sink relationships increases the efficacy of control treatments (Jones et al., 2018).
Beyond the CR DW min , creeping perennials can use excess resources to refill their old CR or form new ones, that is let their old CR become resource sinks rather than stay as resource sources. Which strategy prevails appears to differ greatly between species. For instance, E. repens tend to invest in new rhizomes as its rhizomes are relatively short-lived (usually not surviving longer than one to three years) and older rhizomes are more vulnerable to disturbance, for example tillage (Majek et al., 1984). Sonchus arvensis thickened roots are similarly short-lived as E. repens (Håkansson, 1969a). However, environmental factors may also play a part. Ringselle et al., (2017) observed that E. repens slightly increased its biomass allocation to old rhizomes over new rhizomes when grown under lower nutrient availability. In species with relatively long-lived CR like Reynoutria spp. (Price et al., 2002) and A. podagraria (Meyer and Hellwig, 1997), the old CR are more likely to become sinks beyond the CR DW min .
Our aim was to study how CR DW min differs in terms of developmental stage between different creeping perennial plants, some that have been studied before (E. repens and S. arvensis) and some that have not (A. podagraria, R. japonica, R. sachalinensis and their hybrid R. × bohemica). Moreover, we wished to study how light levels affect when the CR DW min of different species occur. The following hypotheses were tested: (1) under adequate light levels, CR DW min will occur at a later developmental stage in the agricultural weeds (E. repens and S. arvensis) than in the shade-tolerant invasive species (A. podagraria and Reynoutria spp.); (2) a reduction in light supply will cause the CR DW min to occur at a later developmental stage in E. repens and S. arvensis, but not in A. podagraria (since it is relatively shade-tolerant); and (3) the species with long-lived CR (A. podagraria and Reynoutria spp.) are expected to primarily refill their old CR beyond the CR DW min while the species with short-lived CR (E. repens and S. arvensis) will primarily create new CR.

| Experimental setup
CR were collected just before the soil was frozen in autumn, late-October to late-November before the experimental years, in a nearby cereal field (E. repens and S. arvensis) or along roadsides and building sites (A. podagraria, R. japonica, R. sachalinensis and R. × bohemica). After collection, the CR were stored in buckets with soil in a cooling chamber at 2-4℃ until the start of the experiments. The buckets were irrigated as needed.
The day before starting the experiments, rhizomes were cut so they had two nodes, and regenerative roots so they were 5 cm long. One CR piece was planted per pot in all experiments except in E2017. There, two pieces were placed in each pot and one removed if both produced aerial shoots. Prior to planting, the CR pieces were weighed. To estimate their water content and initial dry weight, 10 additional CR pieces of each species were weighed before and after being dried for 72 hr at 60°C. In E2013, a sandpeat mixture (33% sand and 66% peat) was used, and a 100% peat-soil in the other experiments. The peat type was a limed peat enriched with nutrients [Tjerbo Torvfabrikk 'Plantejord', containing 80% (volume per cent) sphagnum peat, 10% composted bark and 10% fine sand. Each unit (50 litre) enriched with limestone flour (6 g) and 2 kg fertiliser (NPK 12-4-18), pH 5.5-6.5 and density 360 kg/m 3 (applied volume)].
The pots used were plastic and square-shaped (10 × 10 cm) and TA B L E 1 Number of pots in each experiment (E2013, E2015, E2016 and E2017), in total and how many were excluded from analyses either due to the rhizome/root not producing any shoots by harvest (dead) or because they advanced to a leaf stage that had too few representatives and was thus not relevant enough to include in the analysis by grouping leaf stages (outliers). High light is 250 µmol m −2 s −1 and low light 100 µmol m −2 s −1 Pots were watered from below (i.e. water was poured onto the trays). During the first 3 weeks, the pots were irrigated with only tap water and afterwards fertilised with a complete nutrient solution.

| Statistical analysis
Pots where the CR had not produced shoots were removed prior to analyses (Table 1). The remaining pots were divided into groups based on the number of leaves on the largest shoot in the pot. If there were fewer than three plants/pots in the group, they were generally combined into a larger group for the analyses of that species ( Figure 1). If only one or two plants of the same species had produced a higher number of leaves than most other plants, they were considered outliers and omitted from the analyses (Table 1).
For example, one plant with four leaves and one with five leaves (both in the 250 µmol m −2 s −1 treatment) were omitted from the A.
podagraria analyses, as all other plants had ≤3 leaves.
The shoot mass fraction (SMF) was calculated by dividing the total shoot biomass (above + belowground shoots) with the total plant biomass. Planted CR fresh weight was used as a covariate as this was a likely source of variation. Tukey-Kramer groupings at α = 0.05 were used for determining significant differences between treatments.

| Elymus repens
The light level significantly affected E. repens biomass production, and the difference increased with increasing leaf stage (  There was no interaction between light level and leaf stage for E. repens old rhizome weight ( Table 2). The old rhizome biomass was significantly lower at all leaf stages compared to plants at leaf stage 0 (i.e. before the plant has even produced one leaf with lamina ≥4 cm), except leaf stage 4 and 10-11 ( Figure 1, Table 2). New rhizome biomass gradually increased with increasing leaf stage, but the increase was lower in plants grown at 100 µmol m −2 s −1 (Figure 1,  3). When grown in 100 µmol m −2 s −1 , plants had their lowest total rhizome biomass at leaf stage 3 and it was not until leaf stages 10-11 that total rhizome biomass was significantly higher than at leaf stage 1-5 (Figures 1 and 3). There was no interaction between light level and leaf stage for total biomass (Table 2). On average across both light levels, plants at leaf stage 2 had already amassed a significantly larger total biomass than those at leaf stages 0 and 1.

| Sonchus arvensis
The light level significantly affected S. arvensis total biomass production, but not all biomass fractions (    The difference in new regenerative roots or aboveground shoots between light levels was not significant ( Table 2). The  (Table 2) as the SMF appeared similar at leaf stages 4 and 10-12 ( Figure 2).
There was a significant interaction between leaf stage and light level for old regenerative roots (Table 2). However, the old regenerative root weight was still lowest at leaf stage 4, at both light levels (Figure 1), where it was significantly lower than plants  (Figure 1). Since old regenerative root biomass was lowest at leaf stage 4 and new regenerative root biomass did not truly begin being produced until leaf stage ≥5, the total regenerative root weight (old + new) was significantly higher for plants at leaf stage ≥5 than at leaf stage 4, under both light levels (Figures 1 and 3). Similarly, the total biomass was significantly higher in plants at leaf stage ≥5 than those at leaf stage ≤4 (Figures 1 and 3).
On average across all leaf stages, A. podagraria plants grown at 100 µmol m −2 s −1 had 29% lower total biomass than those at 250 µmol m −2 s −1 (Figure 1; Table 2), but the SMF was not significantly affected by the light level ( Figure 2). There were significant or almost significant interactions between leaf stage and light level for old rhizomes, roots and aboveground shoots (Table 2).
For old rhizomes, this was due to a lower old rhizome biomass at leaf stage 2 compared to leaf stage 3 at 250 µmol m −2 s −1 , but not at 100 µmol m −2 s −1 (Figure 1). At 250 µmol m −2 s −1 , root and TA B L E 2 ANOVA-table showing the F-values and significance level of the analyses of total plant biomass, total belowground (BG) biomass, aboveground (AG) and BG shoots, non-regenerative roots and old and new rhizomes/regenerative roots (CR), shoot mass fraction (SMF) and CR relative reduction of Elymus repens, Sonchus arvensis, Aegopodium podagraria, grown at two light levels (100 or 250 μmol m −2 s −1 ) and three Reynoutria spp. grown at one light level (250 μmol m −2 s −1 ), and harvested at different leaf stages.
At leaf stage 3, eleven out of eighteen (61%) plants grown at 250 µmol m −2 s −1 and eight out of fifteen (53%) plants grown at 100 µmol m −2 s −1 had produced new rhizomes, compared to only one plant of each at leaf stage ≤2. As a result, the total rhizome biomass was significantly higher at leaf stage 3 than at leaf stage 2 (Figure 3).

| Reynoutria spp
The Reynoutria spp. were only grown at 250 µmol m −2 s −1 . There were significant interactions between the three Reynoutria species and leaf stage for total biomass, root biomass, aboveground shoot biomass and SMF (Table 2). These interactions are likely because R. sachalinensis had a large relative change in rhizome weight early on ( Figure 3) and rapidly expanded its root and shoot biomass, particularly around leaf stage 4 and 5. However, the increase was relatively high in the shoot biomass as the SMF of R. sachalinensis was

F I G U R E 2
The LS means of the shoot mass fraction (SMF) of Elymus repens, Sonchus arvensis, Aegopodium podagraria and three Reynoutria spp. at different leaf stages. Error bars show the standard error. Please note the difference in the x-and y-scale between species. To improve visibility, the means and error bars have been jittered SMF Number of leaves on the largest shoot at harvest significantly higher than the other Reynoutria species; on average 0.7 compared to 0.56 in both R. japonica and the hybrid (Figure 2).
At leaf stage 7, and only in R. japonica, one-third of the plants started producing new rhizomes, but the produced biomass was low.
Thus, old rhizome weight and total rhizome weight were essentially the same. In R. sachalinensis, the total rhizome biomass increased with increasing leaf stage. In R. japonica and the hybrid, the total rhizome biomass was lowest at leaf stage 4 and increased from leaf stage 5 (Figures 1 and 3). In comparison, the total regenerative root biomass had significantly increased by leaf stage 5. Thus, the results of our study support the findings of Tavaziva (2012) that CR DW min occurs at leaf stage 4 in S.

| Developmental stage for CR DW min of different creeping perennials
arvensis. In E. repens, the old and total rhizome biomass were both at their lowest point at leaf stage 1-2, with new rhizome biomass starting to appear at leaf stage 2. Thus, both the CR DW min and the initiation of new rhizome production occurred at an earlier stage than in Håkansson (1967). However, the difference between leaf stages 1-2 and 'just before leaf stage 3-4' is not large and could be due to differences in experimental setups (e.g. Håkansson, 1967's experiment was conducted in pots outside), leaf stage determination or clonal variation (e.g. Neuteboom, 1981;Westra and Wyse, 1981).
The shade-tolerant species did not have a consistently lower CR DW min than the agricultural weeds. Under adequate light conditions (250 µmol m -2 s -1 ), A. podagraria had its lowest total rhizome biomass at around leaf stage 1-2, but there was a strong interaction with light level (refer to section 4.2). The three Reynoutria species differed in that R. sachalinensis had a strong early reduction in rhizome biomass (Figure 3) accompanied by a quick expansion in shoot and root biomass (Figure 1), leading to a higher SMF than the other Reynoutria species (Figure 2). The total rhizome weight of R. sachalinensis then consistently increased with increasing leaf stage. Reynoutria japonica and R. × bohemica had a comparatively small initial reduction in rhizome weight, followed by a relatively stable old rhizome weight with a slight dip at leaf stage 4 before starting to increase. One could therefore argue that the CR DW min of R. sachalinensis occurred very early, perhaps even at leaf stage 0 (i.e. before the plant has even produced one leaf with lamina

| Effect of light level
Elymus repens, S. arvensis and A. podagraria all had a slower and more gradual biomass build-up when grown in less light. In particular, the refilling of their old CR and/or creating new CR were greatly delayed by a reduction in light level (Figure 3). One would expect the SMF to increase with less light (Poorter et al., 2012;Ringselle et al., 2017), and this was the case for E. repens and S.
arvensis, but not for A. podagraria (Figure 2). This can be contrasted with Elemans (2004) who found that a reduction in light caused a reduction in SMF in A. podagraria. However, Elemans (2004) were comparing relatively low light levels (2 and 8% vs. 66% of full light in a greenhouse).
The second hypothesis stated that a reduction in light level would cause the CR DW min to occur at a later leaf stage in E.
repens and S. arvensis, but not in A. podagraria. This hypothesis was not supported as reduced light delayed the CR DW min in E.
repens podagraria was at leaf stage 2, it would be expected that plants in both light treatments would lose old rhizome weight until at least that stage. As this did not occur in the 100 µmol treatment, it seems more likely that the higher light level caused A.
podagraria to invest more of its reserves into creating roots and shoots. This explanation is in line with A. podagraria's strategy of fast expansion in spring (Meyer and Hellwig, 1997 One factor that has not been considered is the influence of the quality, age and origin of the sampling material. First, the age of the CR were not determined before the study. Age can reduce CR viability (e.g. Majek et al., 1984). The age may also af-

| Implications for management
Mechanical control is arguably most effective around the CR DW min as it maximises the loss of stored energy (Håkansson, 2003). Based on the current study, the CR DW min occurred around leaf stage 0 (i.e. before the plant has even produced one leaf with lamina ≥4 cm) in is that the experiments showed that a higher light level can lead to a reduction in stored resources, most likely due to an induced growth spurt. However, larger plant size can also reduce the efficacy of some herbicides.
Reynoutria sachaliensis is considered a less invasive plant than R. japonica and R. × bohemica (Herpigny et al., 2012). The larger inititial depletion of rhizome resources to invest in roots and shoots should in theory give it a larger competitive advantage than the other Reynoutria species, but also make it more vulner-  (Jones et al., 2018). As a result, Jones et al., (2020) strongly warns against mowing and other mechanical control of R. japonica as it requires too many resources/treatment times and risks spreading the infestation further. In conclusion, while applying control efforts correctly in relation to the CR DW min may potentially increase the control efficacy against Reynoutria spp., only field trials testing this hypothesis could determine if the effect is significantly enough to recommend mechanical control against these species.
That so few A. podagraria rhizome pieces produced shoots illustrates how vulnerable the species is to tillage. podagraria as it prevented it from initiating a growth spurt at the cost of additional CR resources.

| CON CLUS IONS
3. Beyond CR DW min , the old CR became resource sinks in the three Reynoutria species, while production of new CR was seemingly prevalent in the other species.
4. Aegopodium podagraria and R. sachaliensis are likely to be vulnerable to control methods very early during sprouting compared to the other species, followed by E. repens.

ACK N OWLED G EM ENTS
The authors acknowledge the contributions of Marit Helgheim, Anne-Kari Holm, Knut Asbjørn Solhaug and PLV340 NMBU students. Parts of this work was funded by Regionale Forskningsfond Oslofjordfondet project no 245824.

CO N FLI C T O F I NTE R E S T
There is no conflict of interest.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/wre.12476.