Regeneration and growth in crowns and rhizome fragments of Japanese knotweed (Reynoutria japonica) and desiccation as a potential control strategy

Fieldwork

Three sites in the North of England, United Kingdom (Fig. 2), two with two years of herbicide treatment and one with no recorded treatment history, were sampled in early April 2019 near the start of the growing season, which avoided concerns about winter dormancy impacting results. Both treated sites (Burnley in Lancashire and Rodley, Leeds in West Yorkshire) were located by the side of the Leeds–Liverpool Canal, surrounded by a riparian habitat, with similar treatment histories (two years of glyphosate-based herbicide treatment applied in October 2017 and 2018). Both are managed by the Canal and Rivers Trust. The untreated site, Huddersfield, West Yorkshire, was similar though on an “island” in the Colne River, with slightly sandier soil and managed by the Yorkshire Wildlife Trust. Collection of plant material was undertaken between 1st and 4th April 2019. Ten crowns and three sections of rhizome per crown were excavated from each site using spade, trowel and hand (to gently remove the ends of the more delicate rhizomes). Crowns were selected randomly once they had at least one live cane emerging (current year’s growth) and one dead cane (showing that the rhizome had produced viable above-ground material in a previous year. Rhizomes were separated from the crowns at the point where they emerge from the crown. Once excavated, crowns and rhizomes were sealed in plastic bags and transported to the lab. Permission to sample was obtained in advance by the Canal and Rivers Trust and the Yorkshire Wildlife Trust respectively.

Laboratory work

Once in the lab, crowns were fresh weighted and potted. Crowns varied in size as described in Supplemental Information. Plant fragments were potted in pots of an appropriate size based on fragment size. Rhizome fragments were potted in square sided pots of nine cm per side and crowns were potted in either medium-sized pots of 21 cm diameter or large-sized pots for crowns over 900 g fresh weight (five per site; Supplemental Information) of 30 cm diameter). Crowns were potted at surface level and rhizomes were covered with soil to a depth of approximately one cm. Plants were potted the day after collection and drying was prevented by storage overnight in a cold room. Since the composition of soil biota is a potential variable that could affect R. japonica growth (Parepa, Schaffner & Bossdorf, 2013), the soil mixture was standardised throughout the experiments, consisting of multi-purpose compost (John Innes Multi-Purpose compost) mixed with garden soil at a ratio of 1:4. The garden soil was sampled from an area in Wakefield, Leeds, that is known to not contain R. japonica and is a silty clay loam, similar to the sites where the samples were collected.

For the rhizome investigation, three segments of rhizome from each crown were collected, yielding a total of 30 segments from each site. One fragment with two nodes, one fragment with one node and one fragment with no nodes were sub-sampled from each rhizome segment resulting in three subsamples of each node classification from each crown, to reduce the possibility of one unhealthy rhizome distorting results. This resulted in the potting of 30 full crowns and 270 rhizome fragments in total.

All crowns and rhizomes were watered twice a week and the amount of water was normalised to the size of the pot (so that all samples received the same relative amount of water). Fertiliser (Baby Bio Original Plant Feed: standard NPK fertiliser with Nitrogen (10.6%); Phosphorus (1.9%), Potassium (1.4%) and organic matter) was added to the watering regime (10 ml per litre) on day 1, day 30 and day 50 of the 60-day experimental period to ensure the availability of nutrients throughout the experiment. Pots were randomly organised and placed on a laboratory bench by the window so that they were exposed to sunlight. They were subject to room temperature (with nightly minima above approximately 9 °C and daily maxima between 15–36 °C, with average daily temperature ranges of approximately 19–25 °C).

Both emergence (new growth) and regeneration were monitored throughout the experiment. Emergence was considered to have occurred when a new bud or stem was observed growing from the crown or rhizome fragment. Successful regeneration was considered to have occurred when a new bud emerged, burst and formed one leaf. Stem height was measured by recording the height from the surface of the soil to the tip of the stem. If stems did not show the traits mentioned, they were classed as an emergence but not regeneration.

Once the growth phase of the experiment ended (after 60 days), all underground biomass was left on plastic sheets to dry in the lab, which was generally between 19 and 25 °C during the day and around 9 °C minimum at night. Plant fragments were weighed at least once per week and this drying process continued until 10% or less of mass was lost between measurements and lasted for 38 days (by which time all plant fragments met these criteria including even the largest crowns (see Supplemental Information). The dried biomass was replanted under the same conditions as the first pot experiment and observed for an additional 60 day period. Treatment of the planted dry biomass replicated the initial 60 day growth experiment with fertiliser (Baby Bio) applied on day 1, 30 and 50 and water applied twice per week.

All field and laboratory work and the disposal of material at the end of the experiment followed strict biosecurity protocols in line with the University of Leeds health and safety standards.

Statistical analysis

Anderson-Darling normality tests were completed prior to any test for differences in regeneration and growth rates between six categories (ANOVA or Kruskal-Wallis testing). Subsequent post hoc tests included Fisher’s pairwise comparison, Mann–Whitney U test or T-test depending on their distributions. A range of statistical software packages were used to obtain the results including R, Minitab (version 18) and GPower (Faul et al., 2009; GPower 3.1.9.4). ImageJ (2019, v1.51c: documentation and downloads at website http://rsbweb.nih.gov/ij/, National Institutes of Health, Bethesda, Maryland, USA) was used to measure leaf area. All raw data are provided in the Supplemental Information.

Emergence and regeneration

The experiment ran for 60 days with 88.5% (293/331 stems) of stems emerging within the first 30 days (Fig. 3) and an additional 10 emerging during the final four days of the experiment. A total of 38 stems (11.48% of all regenerations) emerged between day 31 and day 60 (Fig. 3; Table 1), highlighting that 30 days in not long enough to track emergence (or regeneration) of R. japonica. All three sites had emergences during this period from both crowns and rhizomes.

Day 16 had the most new emergences with 57 new buds emerging above the surface of the soil. Figure 3 shows that emergences from rhizomes experience a longer lag time compared to full crowns with the first crown emergences occurring on day 3 and the first rhizome emergences occurring on day 7.

Successful regeneration occurred in both crowns and rhizomes with one or two nodes. No emergence (or regeneration) was recorded from any rhizome fragment with zero nodes, regardless of fragment weight or length. Table 1 shows that full crowns exhibited the highest success rate with the only failed regenerations occurring in Rodley (20% (two crowns) failed to regenerate) and that with the increasing size categories (zero to one to two nodes to full crowns) there was an increase in successful regeneration.

Growth

Most of the substantial increase in height was observed in the first three weeks, with maximum growth usually occurring in weeks 2–4 (Fig. 4), for all plant fragment types but maximum growth was recorded for two Rodley one-node fragments and two Huddersfield crowns in week 5 and for one Rodley one-node fragment in week 6 (Supplemental Information). Thereafter, growth decreased or ceased towards the latter half of the study period. The highest weekly growth increment was 111.8 cm per week (Supplemental Information), by a Huddersfield full crown in its second week of growth. The minimum growth rate recorded was 0 cm for at least one plant fragment of each type .There was no significant difference between crowns (Kruskal-Wallis: p = 0.262, H = 2.68; Figs. 4B(i) and Figs. 4C(i)) or one node fragments (p = 0.206, F = 1.7; Figs. 4B(iii) and 4C(iii)) across the three sites. Two node rhizome fragments from Rodley produced stems with significantly higher growth than Burnley but Huddersfield showed no significance between the two (p = 0.056, F = 3.27; Fig. 4C(ii)). There was also a significant difference between full crowns and rhizomes at all three sites (Burnley: ANOVA, p < 0.0001, F = 32.25; Rodley: Kruskal-Wallis: p < 0.0001, H = 15.86; Huddersfield: ANOVA, p < 0.0001, F = 48.88) but no difference between one and two nodes for average weekly growth (Fig. 4).

The maximum heights of stems regenerating from full crowns showed no significant difference between sites (ANOVA: p = 0.575, F = 0.57; Fig. 4B(i)). There was also no significant difference between either the two-node or one-node stems between the sites (two-nodes: ANOVA; p = 0.584, F = 0.54; one-node: ANOVA; p = 0.166; F = 1.86; Fig. 4B(ii)). Stems that regenerated from crowns were significantly taller than stems that regenerated from rhizomes across all three sites (Burnley: p < 0.0001, F = 48.32, Rodley: p < 0.0001, F = 191.33, Huddersfield: p < 0.0001, F = 68.49, Fig. 4). There was also no significant difference in the maximum height between stems that regenerated from rhizome fragments with one or two nodes across each of the three sites (ANOVA; p = 0.567; F = 0.78).

Statistical analysis of the diameters measured from each regenerated stem at the end of the 60 day experiment (Fig. 4) showed that Rodley had significantly larger diameters than Huddersfield and Burnley when considering one and two node rhizome fragments (ANOVA: 1 node: p = 0.006, F = 6.51, 2 nodes: p = 0.007, F = 6.23) but no difference between node categories within sites. Full crowns showed no significant difference between sites (p = 0.190, F = 177, Fig. 4). Stems that originated from full crowns had a significantly larger diameter than stems from any other category across all three sites (Fig. 4).

The maximum number of leaves that grew from a single stem was 174 from a stem that grew from a Burnley crown (Supplemental Information). The average number of leaves from crown regeneration was 28 from Burnley, 35 from Rodley and 16 from Huddersfield. Stems that regenerated from Huddersfield full crowns grew significantly fewer leaves than Rodley (ANOVA: p = 0.039, F = 3.72). Huddersfield also exhibited significantly fewer leaves (average = 3) from stems regenerating from one node fragments than the other sites (average = 5) (p = 0.002, F = 7.98), whilst two node fragments experienced no significant difference between all sites (p = 0.264, F = 1.41), with an average of five from Burnley and Rodley and four from Huddersfield (Fig. 5).

Leaf area analysis showed Burnley had significantly larger leaves than Rodley from full crowns (ANOVA: p = 0.107, F = 2.45; different Fisher’s groupings) and showed that Huddersfield leaves sampled from stems growing from 1 and 2 node fragments were significantly smaller than leaves from Burnley or Rodley (one node: Kruskal-Wallis, p = 0.014, H = 8.55; two nodes: ANOVA, p = 0.003, F = 7.63). When comparing the leaf area between the size of planted material within each site, full crowns always showed significantly larger leaf areas (Burnley: Kruskal-Wallis, p < 0.001, H = 13.11, Rodley: ANOVA, p < 0.001, F = 14.31, Huddersfield: p < 0.001, F = 57.86) than rhizome fragments (Fig. 5).

Leaf fresh weight analysis revealed no significant difference between sites for full crowns (Kruskal-Wallis: p = 0.115, H = 4.33) and Huddersfield leaves sampled from stems growing from one or two node fragments were significantly lighter than those from Burnley and Rodley (one node, p = 0.022, H = 7.64, two nodes: ANOVA, p = 0.003, F = 7.4). When considering individual sites and size categories, full crowns produced significantly heavier leaves than rhizome fragments (Burnley: Kruskal-Wallis, p = 0.001, H = 15.2, Rodley: Kruskal-Wallis, p < 0.001, H = 16.26, Huddersfield: ANOVA, p < 0.001, F = 55.63; Fig. 5). Only Rodley showed a significant difference between one and two node rhizome fragments regarding leaf weight with two node fragments producing heavier leaves than one node rhizome fragments (T-Test: p = 0.045, T = 2.22).

Rhizome growth

Figure 6 shows successful regeneration according to initial weights and lengths of the rhizome fragments from all node categories. There is an increase in successful regeneration with increasing number of nodes that is not apparent when considering weight. Although there is an increase in regeneration success with increasing length, this only applies when the longer fragment also had more nodes. This suggests that regeneration success is most associated with number of nodes in R. japonica rhizomes. It also shows that fragments with no nodes did not have any emergences or regenerations regardless of fragment weight or length.

Desiccation

Figure 7 shows the decrease in mass recorded after plant fragments were removed from the soil at the end of the 60-day experiment and air dried. By the 33rd day of air drying all crowns were at (or close to) their dry weight. No more than 10% of the remaining mass was lost between days 35 and 38, indicating that virtually all moisture was gone from the plant fragments. When exposed to air, plant material can potentially take up small amounts of moisture, so this was considered sufficient time to record how long it took the plant fragments to dry. For rhizome fragments with one or two nodes, there was little if any change in weight after day 11 and rhizome fragments with no nodes reached dry weight between day 11 and day 21.

All plant fragments were replanted after 38 days of air drying and treated the same way as throughout the growth experiment in relation to water and nutrient supplementation for 60 days. Regardless of size or plant fragment type, no growth was recorded from any crown or rhizome during the 60 days that they were monitored after being replanted post air drying.

Growth

Desiccation

Few plants can withstand desiccation and once a constant dry weight is reached, plant material is generally considered dead. There is no particular reason to assume that R. japonica would be any better able to recover from desiccation than other angiosperms; however we could not find an instance of where this has been tested. No new growths emerged from the planted dry biomass from either crowns or rhizomes after 60 days air drying, which suggests that the R. japonica material had been killed by the period of air drying and was no longer capable of emerging or regenerating. Smaller rhizomes and rhizomes in the zero node category reached dry mass faster than larger fragments and crowns, but 30 –38 days was sufficient for even the larger plant fragments to reach constant dry weight. The desiccation experiment was conducted via air drying, rather than use of an oven as in other desiccation experiments, e.g., Alshallash & Hatcher (2020), to purposely simulate what might be possible for home-owners with a small to moderate infestation. The experiment suggests that if carefully removed and left to air dry for approximately 38 days, then even medium sized crowns are likely to have died and be unable to produce new emergences or regenerate. Doubling this time to 76 days would provide an extremely conservative measure and likely enhanced protection from regeneration. It is worth noting that the laboratory was warm and dry, meaning that this is likely to work better during the warmer summer months and requires a warm, dry location to facilitate air drying to kill the plant material. This requires field trials, but strongly suggests a useful methodology for dealing with small R. japonica infestations, particularly in residential settings. Careful removal of crowns and larger pieces of rhizome will remove the material with the highest regenerative capabilities, which may simplify subsequent herbicide treatment by reducing the potential for regrowth and the quantity of herbicide required. This research demonstrates that such material can be desiccated relatively quickly and that once desiccated the material is no longer capable of growth. Accordingly, this opens up options for reuse of desiccated plant material in log piles or dead hedge (as refuges for wildlife) or returning the material to the soil to decompose.

Given concern around the use of herbicides in general (e.g., Hasanuzzaman et al., 2020; Thiour-Mauprivez et al., 2019) and glyphosate in particular (e.g., Farina et al., 2019; Sikorski et al., 2019) and that some studies suggest herbicides may become less effective under elevated CO₂ in the mid-term future (Waryzak et al., 2018), identifying methods of controlling invasive non-native plant species that do not rely solely on herbicides should be considered a priority. Although, it is not suggested that this would be a viable management strategy for landscape-scale infestations, desiccation could be a useful component of a holistic management strategy for smaller infestations that are more typical of residential settings.