Introduction
A recent review of the history of knotweed (Polygonum spp.) invasiveness (Drazan et al. Reference Drazan, Smith, Anderson, Becker and Clark2021) and an assessment of the impacts and costs of management (Hocking et al. Reference Hocking, Toop, Jones, Graham and Eastwood2023) provide a comprehensive background on the need to manage invasive Polygonum spp. and the challenges to doing so. Herbaceous perennials such as invasive Polygonum spp. must annually rebuild their aboveground photosynthetic architecture and do so by storing nonstructural carbohydrates in crowns, rhizomes, and roots between growing seasons (Child and Wade Reference Child and Wade2000). Seasonal patterns of nonstructural carbohydrates in crowns have long been used to optimize crop production in herbaceous perennials such as alfalfa (Medicago sativa L.) (Bélanger et al. Reference Bélanger, Kunelius, McKenzie, Papadopoulos, Thomas, McRae, Fillmore and Christie1999; Chatterton et al. Reference Chatterton, Carlson, Hart and Hungerford1974; Jung and Smith Reference Jung and Smith1962; Sheaffer et al. Reference Sheaffer, Lacefield and Marble1988; Sheaffer and Marten Reference Sheaffer and Marten1990; Smith Reference Smith1962). Conversely, depleting carbohydrates to manage aggressive herbaceous perennial invasive plants is also not a novel concept (Arny Reference Arny1932; Bakke et al. Reference Bakke, Gaessler and Loomis1939; Barr Reference Barr1936). Timmons and Bruns (Reference Timmons and Bruns1951) found that a shallow wheel hoeing at a 4-wk interval was optimal to eliminate hemp dogbane (Apocynum cannabinum L.) in 3 yr, with longer intervals reducing effectiveness, and shorter intervals increasing the number of operations necessary to achieve the same outcome. A similar study by Seely (Reference Seely1952) defined a 4-wk shoot-removal interval optimal to control Canada thistle [Cirsium arvense (L.) Scop.]. Although not stated, because C. arvense flowering is induced with long days, with buds developing into open flowers beginning the last week of June following summer solstice, the 4-wk interval coupled with the trial start time means the first shoot removal occurred in June in their four-cut system, the time of lowest storage energy reserves (Welton et al. Reference Welton, Morris and Hartzler1929). Similarly, the lowest reserves and reversal of seasonal carbohydrate flow occurred at initiation of flowering for A. cannabinum (Becker and Fawcett Reference Becker and Fawcett1998) and purple loosestrife (Lythrum salicaria L.) (Katovich et al. Reference Katovich, Becker, Sheaffer and Halgerson1998).
Control of Polygonum spp. has been inconsistent (Bailey et al. Reference Bailey, Bímová and Mandák2009; Bímová et al. Reference Bímová, Mandák and Pyšek2001), and the best management approaches are debated (Hocking et al. Reference Hocking, Toop, Jones, Graham and Eastwood2023). Connecting carbohydrate flow to management options via visual growth and development cues or, more concisely, growing degree day (GDD) models should improve herbicidal and mechanical management outcomes. Seiger and Merchant (Reference Seiger and Merchant1997) predicted that at least four cuttings per season would be required to result in a net depletion of Polygonum spp. belowground biomass based on same-season rhizome-transplanted container-grown plants. Sink-source interactions with herbicide transport and effectiveness were thoroughly reviewed by Bashtanova et al. (Reference Bashtanova, Beckett and Flowers2009) and broadened by Jones et al. (Reference Jones, Bruce, Fowler, Law-Cooper Graham, Abel, Street-Perrott and Eastwood2018). Timing of systemic herbicide application when net assimilate translocation is to belowground sinks should improve efficacy and consistency of control (Becker and Fawcett Reference Becker and Fawcett1998). Additionally, timing of mechanical or hand removal when carbohydrate reserves in rhizomes and crowns are at seasonal lows reduces the number of removal operations needed to achieve control (Seely Reference Seely1952; Timmons and Bruns Reference Timmons and Bruns1951), reducing disturbance events, time, cost, and non-target impacts.
Price et al. (Reference Price, Gamble, Williams and Marshall2002) documented assimilate partitioning to roots, shoots, and flowers of Japanese knotweed [Polygonum cuspidatum Siebold & Zucc.; syn.: Fallopia japonica (Houtt.) Ronse Decr.] at key months of the year in overwintered, container-grown plants using 14CO2 labeling as a surrogate for carbohydrate analysis. The work of Callaghan et al. (Reference Callaghan, Scott and Whittaker1981), sparked by the energy crisis of the late 1970s, characterized above- and belowground biomass, biomass recovery following harvest, elemental nutrients, and structural and nonstructural carbohydrates as exploratory work to find suitable candidates for plant biomass. Both provide considerable insight as to carbohydrate quantity and quality in Polygonum spp. An in situ developmental, seasonal assessment of Polygonum spp. carbohydrate partitioning is needed to optimize mechanical/hand-roguing methods of control and herbicide efficacy to bridge the gaps in Callaghan et al. (Reference Callaghan, Scott and Whittaker1981) and validate the container-grown 14C partitioning work of Price et al. (Reference Price, Gamble, Williams and Marshall2002). In addition, this would provide phenology-related nonstructural carbohydrate quality and quantity data to support the ideas put forth by Bashtanova et al. (Reference Bashtanova, Beckett and Flowers2009) and expanded by Jones et al. (Reference Jones, Bruce, Fowler, Law-Cooper Graham, Abel, Street-Perrott and Eastwood2018) and Kuehne and Heimer (Reference Kuehne and Heimer2023) to improve herbicide translocation efficiencies and therefore efficacy. The objective of our study was to characterize the seasonal variation of nonstructural carbohydrate quality and quantity in crowns of the hybrid Bohemian knotweed [Polygonum ×bohemicum (J. Chrtek & Chrtková) Zika & Jacobson [cuspidatum × sachalinense]; syn.: Fallopia ×bohemica (Chrtek and Chrtková) J.P. Bailey] and P. cuspidatum in the context of developmental phenology.
Materials and Methods
We monitored representative populations of the two most common Polygonum spp. in Minnesota, P. ×bohemicum and P. cuspidatum (Drazan Reference Drazan2022) to characterize nonstructural carbohydrates in the crowns from April through December and included a distribution sampling of lateral rhizomes on April 2, 2021. All populations were known to have been established for at least 7 yr and were undisturbed. The P. ×bohemicum samples were collected in Brooklyn Center, MN (45.073506°N, 93.331595°W) in 2020 and in Bloomington, MN (44.814635°N, 93.267628°W) in 2021. In 2021, a P. cuspidatum population was also sampled near Cottage Grove, MN (44.846527°N, 92.981658°W). The sampled patches were 7 to 10 m in diameter. We moved to the Bloomington site 30 km south-southeast of the Brooklyn Center site for the 2021 samples, because the Brooklyn Center site was inadvertently disturbed by park staff. The P. cuspidatum site was approximately 37 km southeast of the Brooklyn Center site and 23 km east-southeast of the Bloomington site. The 2020 P. ×bohemicum site was in a city park in the open without any nearby trees, the 2021 P. ×bohemicum site was at the edge of a city park partially overstoried by mature mixed hardwood trees that opened into a city right-of-way that was maintained as bluegrass (Poa pratensis L.) turf, and the P. cuspidatum 2021 site was partially overstoried by mature mixed hardwood trees in a right-of-way for a regional flood control earthen drainage causeway in Washington County. We targeted April 1 as the first sample date, as the ground was still frozen to the depth of sample removal in March. Due to historically mild fall temperatures for Minnesota, we were able to obtain samples through early December each year, as surficial freezing did not prevent sample removal. Samples were obtained ±4 d of the first day of each month. Phenology notes were taken at each sample date at each collection site. In addition, weekly phenology observations were made in a common garden established on the St Paul campus in 2019 that included both taxa plus a population of giant knotweed (Polygonum sachalinense F. Schmidt ex Maxim.; syn.: Fallopia sachalinensis (F. Schmidt ex Maxim.) Ronse Decr.]. Additionally, Montgomery et al. (Reference Montgomery, Carlson, Anderson, Govindan and Morey2024) have since released a GDD model linked to phenology based on citizen scientist observations during our study period that accurately predicted the phenology events of our sites (data not shown).
The crown was defined as the woody, enlarged area in the transition region between the root, rhizomes, and stems, often with numerous new shoots emerging among dead shoots from the previous season. For each month, three randomly selected crowns were sampled from each site, with each crown considered a replication. Stems were cut to 25 cm above the soil surface, and the crown and associated rhizomes were removed around each crown in a radius of 25 cm and to a depth of 30 cm. Excess soil was removed, and the samples were placed on ice for transport to the lab. For the April 2, 2021, samples, rhizomes attached to sampled crowns were excavated until either ending when another secondary crown was encountered or when physical obstructions prevented following the rhizome farther. For all other sample periods, rhizomes attached to the central crown were discarded. Anecdotally, the P. cuspidatum population had shorter inter-crown distances than P. ×bohemicum, and P. cuspidatum had a higher number of rhizomes proliferating from individual crowns. Smith et al. (Reference Smith, Ward, Child and Owen2007) measured rhizome networks in P. cuspidatum and found 0.27- to 1.60-m separations among 10 crowns along a rhizome. Anecdotally, in our study, inter-crown distances were 1.40 to 1.63 m for P. ×bohemicum and 0.76 to 1.27 m for P. cuspidatum, within the apical dominance determined distances noted by Smith et al. (Reference Smith, Ward, Child and Owen2007). Polygonum ×bohemicum rhizomes tended to be thicker than those of P. cuspidatum (1.9 to 3.2 vs. 1.0 to 2.5 cm in diameter, respectively). Rhizomes of both taxa tended to run laterally at similar depths ranging from 10 to 25 cm. The crowns of P. ×bohemicum were typically 5 to 10 cm in diameter versus 2.0 to 5.0 cm diameters for P. cuspidatum. Sampled crowns and rhizomes had three to six and two to five annual rings, respectively, visible in cross section.
Soil was removed with pressurized water at the lab. If samples could not be processed the day they were harvested, they were stored at in a deep freeze at −18 C until processing. Depending on the diameter of the crown, four to six 0.5-cm-thick subsample cross sections were cut from the 15-cm midsection of crowns with a band saw. Rhizomes from the April 2, 2021, sampling were subdivided into three to five representative sections depending on the length of rhizome obtained. Three to five representative subsamples for carbohydrate analysis were cut from the midsections of each rhizome subsection with hand-pruning shears. Rhizome and crown samples were then cut into small 2- to 7-mm chips with end-cutting nippers or pruning shears to expedite drying, weighed, and dried at 32 C until stable weights were obtained (usually 3 d). Dried samples were pre-ground to 6 mm using a Wiley Mill (Thomas Scientific, Swedesboro, NJ, USA) and halved in volume twice using a sample splitter; the resulting subsample (which averaged 27.2 g) was ground to 1 mm in a Cyclone mill (Foss, Hillerød, Denmark) for carbohydrate analysis.
Ground samples were analyzed for their total starch content using the Megazyme total starch assay kit according to the AACC method 76-13.01 (Cereals & Grains Association n.d.). Sugars were extracted and analyzed according to the method described by Boakye et al. (Reference Boakye, Kougblenou, Murai, Okyere, Anderson, Bajgain, Phillipp, LaPlante, Schlecht, Vogel, Carlson, Occhino, Stanislawski, Swan Ray and Annor2022) using a high-performance anion-exchange chromatography (HPAEC) system (Dionex ICS-5000+ HPAEC System, Dionex, Sunnyvale, CA, USA) equipped with a pulsed amperometric detector. The separation of saccharides was performed on a Thermo Scientific Dionex CarboPac PA200 analytical column (3 by 250 mm) protected by its corresponding guard column (3 by 50 mm). The eluents used were 225 mM sodium hydroxide (A), 500 mM sodium acetate (B), and ultrapure water (C). A composite of five crowns sampled from the Brooklyn Center site on October 1, 2019, was used as the lab standard for carbohydrate analysis.
For GDD calculations, daily temperature data were obtained for the study period from the nearest weather station with the most complete, quality weather data. Data for 2020 were obtained from NOAA Station MN087 (elevation 282.5 m, 45.0961°N, 93.4500°W). Data for 2021 were from NOAA Station ID AU649 (elevation 276.1 m, 44.8261°N, 92.9122°W). Daily GDD were estimated from daily temperature data using the equation from McMaster and Wilhelm (Reference McMaster and Wilhelm1997):
$${\rm{GDD}} = \left[ {\left( {{{{T_{{\rm{max}}}} + {T_{{\rm{min}}}}} \over 2}} \right) - 0} \right]$$
where
${T_{{\rm{max}}}}$
and
${T_{{\rm{min}}}}$
refer to daily maximum and minimum temperatures, respectively.
Cumulative GDD were calculated using the following equation:
$${\rm{Cumulative\ GDD}} = \sum\nolimits_{i = 1}^n {{\rm{GDD}}} $$
Unable to find research-based values for these taxa, we use the minimum and maximum of 0.0 and 30.0 C, respectively, to match the data used in Montgomery et al. (Reference Montgomery, Carlson, Anderson, Govindan and Morey2024) to develop their citizen science-based phenology research model for P. cuspidatum.
To investigate the variation in temporal and lateral distribution of nonstructural carbohydrates, we utilized generalized additive mixed-effects models (GAMM) using the mgcv package in R v. 4.4.0 (R Core Team 2024; Wood Reference Wood2017). The 3 site-years of data (2 for P. ×bohemicum and 1 for P. cuspidatum) were considered random variables and combined for analysis. Note that P. ×bohemicum is a hybrid of P. cuspidatum and P. sachalinense, and much of the germplasm in Minnesota is heavily backcrossed to P. cuspidatum and P. ×bohemicum rather than P. sachalinense, as P. sachalinense is rare in Minnesota (Drazan Reference Drazan2022), so we did not foresee a biological basis for the taxa to differ in carbohydrate quantity nor quality. Percent dry weight data for each of the carbohydrates were regressed over cumulative GDD using thin-plate regression spline, where the site-year was considered as a random effect (Wood Reference Wood2003). Similarly, percent dry weight data for carbohydrates were regressed over the lateral distance of crown and rhizomes from the central crown, with taxa considered as a random effect. Based on lower Akaike information criterion (AIC) values and suitability of beta distribution to model data values between 0 and 1, GAMM models were constructed using beta distribution and the logit link function in R. Furthermore, carbohydrates were classified into two groups. Group 1 included the major contributors to the carbohydrate pool starch and sucrose, and the sum of all nonstructural carbohydrates monitored termed total carbohydrates. Group 2 included the minor carbohydrates comprising glucose, fructose, raffinose, and stachyose. Separate models were fit for each group to account for the substantial difference in mean values between them and the visual clustering of residuals, which indicated the necessity of distinct variance structures for the two groups. To restrict the potential overfitting, the number of effective degrees of freedom was restricted to four (k = 5). Predicted mean carbohydrate percent values and 95% simultaneous confidence intervals were plotted. Additionally, cumulative GDD values corresponding to the lowest percent dry weight for each carbohydrate were estimated using the predict function in the mgcv package (see the GAMM parameters in Supplementary Material 1 for details).
Results and Discussion
Phenology and GDD
Montgomery et al. (Reference Montgomery, Carlson, Anderson, Govindan and Morey2024) developed phenology models based on cumulative GDD. Relating carbohydrate quantity and quality to cumulative GDD or to key phenology events allows managers to improve outcomes of management practices. In April through May in the St Paul campus common garden, shoot emergence and elongation was nonuniform and occurred over an extended period within each population. The attainment of full bloom in September and complete senescence in October was more uniform, occurring within a relatively short period of time. In the St Paul metropolitan area, soil frost-free dates historically occur in late March to early April. Crown buds had swollen and deepened to a reddish-pink color the first week of April, with initial shoot emergence occurring the second and third weeks of April, and widespread shoot emergence and elongation began the last 2 wk of April. Uneven shoot elongation continued through May, reaching maximum shoot height by the end of June. Continued canopy development via leaf expansion peaked mid- to late July. Freestanding shoot heights in July for the sampled P. ×bohemicum populations averaged 3.4 m in height, with a maximum shoot height of 3.7 m. The sampled P. cuspidatum population averaged 2.4 m in shoot height, reaching a maximum of 3 m.
In the common garden, early onset of first floral buds began August 9 in 2020 and August 18 in 2021, when a few developing flower buds were clearly visible in leaf axils as the racemes began to elongate (erect racemes to 10 mm in length composed of a compact cluster of developing panicles). Early development and elongation of individual male panicles were broadly visible by the end of the second and third weeks of August. Female panicles began to elongate the last week of August through the first week of September, approximately 10 to 14 d after male flower development. Plants reached full bloom the second week in September each season and by the end of September, male flowers were in sharp decline, mostly senesced, leaving bare racemes of erect panicle “skeletons.” Females had mature seed by the end of September to the first week in October, within the time frame observed by Bram and McNair (Reference Bram and McNair2004), with leaf senescence beginning at the same time. Change in tepal color began the end of September, transitioning from greenish-white to brownish-white coloration during October 15 through October 26 and was not uniform.
Regarding the window for absorption of foliar-applied herbicides, leaves began to senesce, developing a uniform, yellowish cast in early to mid-September. Some early leaf senescence began the third week of September and progressed to full leaf loss by early to mid-October, while stems remained green and moist. Stems of all Polygonum taxa appeared completely dry by the third week of October, enough so that movement of assimilates could no longer occur even within the stem. Note that total nonstructural carbohydrates continued to accumulate in crowns through the beginning of November and occasionally December (Figure 1) after complete leaf loss. Assimilate flow and canopy health indicate that early August through the third week of September would have been the ideal application window for systemic herbicides corresponding to approximately 2,400 to 3,400 GDD (Figure 1). Very high utilization by pollinators may push some managers to a narrow application window before bud to first bloom in the second to third week of August (approximately 2,300 to 2,700 GDD), or waiting until flowing ends in the third to fourth week of September (approximately 3,200 to 3,300 GDD) (see GDD phenology, Supplementary Material 2). The end of this application window depends on the canopy health to support herbicide absorption. Healthy leaf tissue to support herbicide absorption and translocation was no longer present by the end of September for P. cuspidatum and P. sachalinense and was no longer present approximately 1 wk later for P. ×bohemicum in our common garden. Key phenology stages estimated via GDD calculations with the research model of Montgomery et al. (Reference Montgomery, Carlson, Anderson, Govindan and Morey2024) and the GAMM estimates for seasonal lows for key storage carbohydrates are available in the GDD phenology supplement (Supplementary Material 2). The actual lab sample carbohydrate data are also available in the carbohydrate data supplement (Supplementary Material 3).
Figure 1. Estimated major nonstructual carbohydrates (CHO) by cumulative growing degree days (GDD) of Polygonum ×bohemicum in 2020 and P. ×bohemicum and Polygonum cuspidatum in 2021 sampled at the beginning of each month April through December. Annotations added to show observed phenology, recommended management windows, and approximate first of each month. Shaded areas denote 95% simultaneous confidence intervals.
Seasonal Carbohydrate Storage and Utilization
Estimated values for nonstructural carbohydrates on a dry weight basis are shown in Figures 1 to 4. Starch was the most common nonstructural carbohydrate present, followed by sucrose (Figure 1). At much lower concentrations, glucose and fructose were the next most common, and finally, of the carbohydrates monitored, raffinose and stachyose grouped as the lowest seasonal concentrations (Figure 2). The estimated GDD and nonstructural carbohydrate minimums are shown in Table 1. Inulins, the primary storage polysaccharides in Jerusalem artichoke (Helianthus tuberosus L.) (Rubel et al. Reference Rubel, Iraporda, Manrique, Genovese and Abraham2021) and chicory (Cichorium intybus L.) (Van Laere and Van Den Ende Reference Van Laere and Van Den Ende2002) were not a key nonstructural carbohydrate in either Polygonum taxon.
Figure 2. Estimated minor nonstructual carbohydrates (CHO) by cumulative growing degree days (GDD) of Polygonum ×bohemicum in 2020 and P. ×bohemicum and Polygonum cuspidatum in 2021 sampled at the beginning of each month April through December. Annotations added to show observed phenology, recommended management windows, and approximate first of each month. Shaded areas denote 95% simultaneous confidence intervals.
Table 1. Minimum and maximum nonstructual carbohydrate and corresponding estimated cumulative growing degree days (GDD) for Polygonum ×bohemicum in 2020 and for P. ×bohemicum and Polygonum cuspidatum in 2021 sampled at the beginning of each month April through December.
a GDD were calculated using the base and maximum temperatures of 0.0 and 30.0 C, respectively.
We found monthly average minimum to maximum starch levels on a dry matter basis ranged from 5.2% to 23.3%, a broader range than the 9.1% to 19.3% noted by Callaghan et al. (Reference Callaghan, Scott and Whittaker1981) (Table 2), and total nonstructural carbohydrate levels ranged from 13.9% to 41.3% versus their 19.2% to 38.5% range. The level of starch in Polygonum spp. was somewhat lower overall compared with the 7% to 38% starch range in A. cannabinum (Becker and Fawcett Reference Becker and Fawcett1998) and the 4% to 32% range in L. salicaria (Katovich et al. Reference Katovich, Becker, Sheaffer and Halgerson1998). In general, the highest total carbohydrate and starch levels occurred in the fall following the reproductive phase or in early spring before dormancy break and shoot elongation (Figure 1; Table 2).
Table 2. Three-crown average and individual crown high and low percent nonstructural carbohydrates among Polygonum ×bohemicum in 2020 and P. ×bohemicum and Polygonum cuspidatum in 2021 sampled at the beginning of each month April through December.
Glucose and fructose tended to fluctuate month to month in unison, as did raffinose with stachyose (Figure 2). Midseason fluctuations in levels of glucose and fructose approximated the trends for starch as intermediary building blocks of sucrose, the translocated carbohydrate. Even though average monthly sucrose levels (4.8% to 16.5%; Table 2) were higher compared with A. cannabinum (3.9% to 7.8%; Becker and Fawcett Reference Becker and Fawcett1998) and L. salicaria (6.6% to 12.8%; Katovich et al. Reference Katovich, Becker, Sheaffer and Halgerson1998), starch was still consistently the highest nonstructural carbohydrate present. Sucrose levels did not show a consistent inverse correlation with starch metabolism as the growing season progressed (Pearson’s r = −0.05 to 0.35; data not shown) (Figure 1). A negative correlation with starch levels would be expected to reflect metabolism of starch reserves to sucrose for transport to metabolic sinks during the season. The month-to-month fluctuations of glucose and fructose were positively correlated with each other (Pearson’s r = +0.92 to 0.99; data not shown), as were raffinose and stachyose (Pearson’s r = +0.88 to 0.96; data not shown). Of note, significant aboveground tissue loss due to frost damage occurred when shoots were to 0.3 to 1.5 m in height midway between the May and June sample collection both years. Additionally, Japanese beetle (Popillia japonica) feeding damage occurred mid-July through August 5 in both years with up to 50% leaf tissue loss in the upper one-half of the canopies. These defoliation events may account for some of the fluctuations in total nonstructural carbohydrates in the samples collected following those events.
Respiration losses during the winter have been shown for alfalfa and red clover (Trifolium pratense L.) (Jung and Smith Reference Jung and Smith1962), with as much as a 50% loss of nonstructural carbohydrates observed in alfalfa (Sheaffer et al. Reference Sheaffer, Lacefield and Marble1988). There were only minor losses of nonstructural carbohydrates over winter for A. cannabinum (Becker and Fawcett Reference Becker and Fawcett1998) and L. salicaria (Katovich et al. Reference Katovich, Becker, Sheaffer and Halgerson1998). Despite the inability to sample the same site one year to the next, our season-beginning carbohydrate levels were similar or slightly lower than season-end carbohydrate levels, supporting the findings of Price at al. (Reference Price, Gamble, Williams and Marshall2002) that Polygonum spp. use little to no carbohydrate for respiration during winter, indicating full dormancy. Additionally, in A. cannabinum, starch levels dropped in November each year from the seasonal high that occurred early September with a corresponding rise in ethanol-soluble reducing and nonreducing sugars (Becker and Fawcett Reference Becker and Fawcett1998). Barr (Reference Barr1936) observed a similar rise in soluble sugars in field bindweed (Convolvulus arvensis L.) as fall temperatures lowered, termed “hardening off.” We observed a similar rise in sucrose (Figure 1) and in glucose and fructose (Figure 2), possibly indicating physiological hardening-off for winter.
Lateral Distribution of Nonstructural Carbohydrates
The lateral distribution of carbohydrates was a onetime observation intended to inform the implications of sampling only crowns to characterize the broader underground carbohydrate storage potential. Most rhizomes at all sites ended with an ascending, more juvenile crown. There was a trend for increased starch and a corresponding total nonstructural carbohydrate in rhizomes compared with crowns (Figure 3). Lower levels of starch were also noted in crowns than in the roots of L. salicaria (14.9% vs. 19.5%, respectively; Katovich et al. Reference Katovich, Becker, Sheaffer and Halgerson1998), and in A. cannabinum crowns versus lateral roots (30.1% vs. a range of 38.5% to 52.3%, respectively) or crowns compared with ascending vertical roots (15.6% vs. a range of 18.6% to 39.4%, respectively) (Becker and Fawcett Reference Becker and Fawcett1998). Additionally, starch and total sugars were considerably lower in crowns than in roots of both alfalfa and red clover (Jung and Smith Reference Jung and Smith1962). The uneven distribution we observed may be typical for perennial herbaceous dicots in general.
Figure 3. Estimated lateral distribution of major nonstructural carbohydrates (CHO) for each of three crowns and associated rhizomes of Polygonum ×bohemicum and Polygonum cuspidatum sampled April 2, 2021. Shaded areas denote 95% simultaneous confidence intervals.
Figure 4. Estimated lateral distribution of minor nonstructural carbohydrates (CHO) for each of three crowns and associated rhizomes of Polygonum ×bohemicum and Polygonum cuspidatum sampled April 2, 2021. Shaded areas denote 95% simultaneous confidence intervals.
In Minnesota, the interval from frozen ground to crown bud dormancy break was relatively narrow, with the soil frost-free date centered around the first of April. Crown buds, though not elongating, were swollen and reddish-pink in color, indicating dormancy break was underway. First signs of shoot elongation began approximately 1 wk after the April 2 sampling date, with shoot elongation, although still very uneven, well underway by 2 to 3 wk after this sample date. Adachi et al. (Reference Adachi, Terashima and Takahashi1996) noted apical dominance of crown buds maintaining dormancy of lateral rhizome buds unless the crowns were disturbed. The lateral system was undisturbed at our two sites, so we suspect the first metabolically active site coming out of dormancy was the crown bud area. Thus, lower starch concentrations in the crowns may have been indicative of differential lateral carbohydrate metabolism in preparation for crown bud dormancy break. Adachi et al. (Reference Adachi, Terashima and Takahashi1996) also noted that lateral distribution of colonizing P. cuspidatum varied along metabolic gradients, with the main activity in their case in the outer periphery of an expanding population. Perhaps a similar reallocation of carbohydrates away from the established crowns was occurring in our study.
Implications of Seasonality
Our findings clarify the seasonal gaps in Callaghan et al. (Reference Callaghan, Scott and Whittaker1981). Although lacking a full growing season continuum without data for April or July through September, Callaghan et al. (Reference Callaghan, Scott and Whittaker1981) did note that seasonal lows occurred in June. Our work further defined that the seasonal low occurs at the end of June and extends through July, but transport reverses dramatically in August. The lowest levels of total carbohydrate reserves were estimated to occur between 1,200 to 1,800 GDD (Figure 1), with the estimated low in total carbohydrate reserves bottoming out at 1,454 GDD (Table 1). Callaghan et al. (Reference Callaghan, Scott and Whittaker1981) found that shoot nonstructural carbohydrate levels, although not tied to underground carbohydrate allocation, plummeted after their October sampling. Considering the longer growing season in the Callaghan et al. (Reference Callaghan, Scott and Whittaker1981) study, this supports our findings of accelerated accumulation of carbohydrates in the crowns in September through October in Minnesota. Additionally, our research confirms the assumption made by Price et al. (Reference Price, Gamble, Williams and Marshall2002), that their 14C-labeled assimilate utilization patterns in container-grown plants do indeed reflect nonstructural carbohydrate utilization patterns of established plants in the field.
Our findings indicate that removal of aboveground biomass from mid-June through mid-July after Polygonum spp. had reached maximum vegetative growth but before the initiation of flowering in August could maximize depletion of underground carbohydrate reserves and further deplete remaining underground carbohydrates to rebuild aboveground photosynthetic architecture (Figure 1). Subsequent removal of vegetation two more times that season targeting 30- to 40-d intervals until the growing season ends may provide the optimal depletion strategy, defined as the least amount of labor needed to maximally deplete carbohydrates and weaken a population. This approach would require diligent removal of all shoots in a stand two to three times a season for at least 3 yr, with return visits in years 4 and 5 to remove the few missed plants that may persist.
A single annual renewable energy biomass harvest of shoots of P. cuspidatum at various times of the year had little to no effect on rhizome biomass (Callaghan et al. Reference Callaghan, Scott and Whittaker1981). However, cutting intervals of 4 wk for container-grown Polygonum spp. did (Seiger and Merchant Reference Seiger and Merchant1997), which coincides with the optimal cutting intervals found for C. arvense (Seely Reference Seely1952) and A. cannabinum (Timmons and Bruns Reference Timmons and Bruns1951). Seiger and Merchant (Reference Seiger and Merchant1997) also noted that for Polygonum spp., shoot removal once in a season was equally effective in reducing belowground biomass when removed at least 7 wk before senescence. At our locations, this 7-wk time frame would have occurred in early August, when crown carbohydrate reserves consistently began to rebuild. Our observed time of carbohydrate reversal from acropetal to basipetal flow also corresponds with the noted shift from aboveground to belowground biomass accumulation by July 30 (Seiger Reference Seiger1993) and the observed maximum aboveground biomass accumulation (Fabiszewski and Brej Reference Fabiszewski and Brej2008; Horn Reference Horn1997). Baker (Reference Baker1988) controlled P. cuspidatum with mowing at 2-wk intervals for 2 yr or hand pulling once at maximum height in July for 3 yr. In light of our findings, their single hand-pulling operation was done when crown carbohydrate reserves were maximally depleted, and mowing at 2-wk intervals could have been less frequent and likely would have accomplished the same outcome. However, Baker (Reference Baker1988) did also find that attempting to pull shoots in a less-accessible streambank terrain for 10 yr did not control that population, highlighting the potential for inconsistent outcomes when using nonherbicidal control and the need to find and remove all shoots within a colony. A note of caution, repeated use of the same management timing may select for biotypes adapted to that practice, as was done to select alfalfa germplasm tolerant to increased cutting schedules (Chatterton et al. Reference Chatterton, Carlson, Hart and Hungerford1974).
We acknowledge the challenges of removing top growth to control Polygonum spp. (Jones and Eastwood Reference Jones and Eastwood2019; Jones et al. Reference Jones, Bruce, Fowler, Law-Cooper Graham, Abel, Street-Perrott and Eastwood2018; Jones et al. Reference Jones, Fowler, Hocking and Eastwood2020), and particularly the need to prevent vegetative spread through shoot or rhizome fragments. Yet this strategy may be useful for the many small, easily accessible patches where cut or dug material can be managed in place and herbicidal control may not be an option. Larger, more-inaccessible infestations would be best controlled by incorporating the use of herbicide.
Integrating carbohydrate allocation and herbicide movement to improve Polygonum spp. management outcomes has been discussed (Kuehne et al. Reference Kuehne and Heimer2023; Jones et al. Reference Jones, Bruce, Fowler, Law-Cooper Graham, Abel, Street-Perrott and Eastwood2018), although carbohydrate levels have yet to be monitored concomitant with an efficacy study or in a more comprehensive time of application study throughout the growing season. Our work indicates that herbicides that rely on passive translocation with assimilates to belowground reproductive structures would be most effective if applied in a relatively narrow window in late August through mid-September (Figure 1) if not previously disrupted earlier in the season with physical shoot removal or early-season herbicide applications (Boyd et al. Reference Boyd, White and Larsen2017). If disrupted earlier in the season, we speculate the window for downward assimilate movement may be pushed later in the season and the herbicide application window possibly narrowed to early to mid-September. However, with climate change, timing for fall application is becoming erratic in Minnesota and uncharacteristically extended through the last week of October for P. ×bohemicum in 2023 based on foliage conditions conducive for assimilate absorption and translocation. Preference for fall applications also assumes plants are actively growing. We have observed P. ×bohemicum, P. cuspidatum, and especially P. sachalinense in the common garden on the St Paul campus were very susceptible to moisture stress and to even short-duration mild frosts (−1.1 to 0 C) at spring emergence or in the fall.
Our findings clarify the quality, quantity, and seasonality of nonstructural carbohydrate utilization in P. cuspidatum and the hybrid P. ×bohemicum. Future work is needed to verify optimal depletion of carbohydrates to control Polygonum spp. and the nuances of using management to break apical dominance in rhizomes and lateral buds whether via physical disruptions such as tillage or mowing or improved use of herbicides informed by knowledge of herbicide modes of action, seasonality of transport, and behavior in the environment.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/wsc.2025.11
Acknowledgments
We are indebted to Ted Jeo for his expertise and assistance in carbohydrate analysis design and sample preparation. We thank Doug Miller for his technical and statistical analysis support. For access to sample sites, we thank the City of Brooklyn Center, David Hanson with the Bloomington Maintenance Division, and Washington County with access provided by Ron Cozad and family, Cottage Grove, MN.
Funding statement
This work was supported by the Minnesota Invasive Terrestrial Plants and Pests Center and the Minnesota Environment and Natural Resources Trust Fund as recommended by the Legislative-Citizen Commission on Minnesota Resources; University of Minnesota Extension; and the College of Food, Agricultural and Natural Resource Sciences; and facilitated by the Minnesota Department of Agriculture.
Competing interests
The authors declare no conflicts of interest.
Open access
