Liriope spicata is a low-growing, grass-like perennial that is native to Southeast Asia. This and several similar Liriope species have been widely introduced into the United States for ornamental and groundcover purposes. Liriope spicata can spread aggressively via seed dispersal and creeping rhizomes and form dense, monotypic patches. Little information is currently available in the literature concerning creeping lilyturf control. We tested seven herbicides in 2011 and 2012 using container grown creeping lilyturf that had been established for approximately one year prior to treatment. Visual evaluations at 30 and 60 days after treatment indicated that no treatment provided rapid control of the foliage. However, above- and belowground biomass harvested at 90 and 180 days after treatment (DAT) indicated that metsulfuron and imazapyr were highly effective in controlling L. spicata. Both herbicides reduced belowground biomass by greater than 97%. Glyphosate and imazapic, which have been recommended for creeping lilyturf control, only reduced belowground biomass by 43 and 45%, respectively, at 180 DAT. The herbicides 2,4-D and dicamba, which have known efficacy on other members of the family Liliaceae, did not effectively control L. spicata. These results indicate that metsulfuron and imazapyr are more effective treatment options for creeping lilyturf control than currently recommended glyphosate and imazapic treatments. However, in hardwood forest systems where creeping lilyturf is primarily invasive, their utility may be limited due to potential nontarget damage.
INTRODUCTION
Liriope spicata Lour. is a low-growing, grass-like perennial plant in the family Liliaceae that is native to southeastern Asia. It and several other Liriope species have been introduced into North America as ornamentals where they are typically used as ground covers, foundation plants, edging along walkways, and borders around landscaped areas (Fantz 1993; Nesom 2010). Liriope species have been popular in the southern United States due to their evergreen attractiveness, low maintenance requirements, and lack of serious pest issues (Hume 1961; Dirr 1983; Adams 1989).
Liriope spicata leaves are green, fleshy, 2–4 mm wide, and 10–26 mm long with serrulate margins and an erect or weakly arching growth habit. Inflorescences consist of somewhat hidden spikes of pale lavender to white flowers that bloom in the summer. Fruits are round, berry-like drupes, 6–8 mm in width that ripen to a blackish purple color, contain a single seed, and may persist into the winter. Fruits may move with water (Spaulding et al. 2010) but little is known concerning dispersal by other means. Roots are fibrous with small fleshy peanut shaped corms that may produce new plants. Plants are strongly rhizomatous and form dense, radially expanding patches (Fantz 2008; Miller, Chambliss, and Loewenstein 2010).
Liriope spicata, and closely related Liriope and Ophiopogon species, have been observed to escape cultivation and become invasive in natural areas (Spaulding et al. 2010). They also may become weedy where they spread from ornamental plantings into surrounding turf or other landscape settings. However, taxonomic uncertainty, interchangeably used common names (including lilyturf, mondograss, and monkey grass), and the availability of over 100 Liriope cultivars (Fantz 1993) have somewhat muddled the invasive issue. This type of confusion has also been observed with other horticultural plants including Nandina spp. (Knox and Wilson 2006) and Ligustrum spp. (Maddox et al. 2010).
Despite questions concerning the invasive potential of other Liriope or closely related species, it is clear that L. spicata has naturalized and become invasive across the southeastern United States (sensu Richardson et al. 2000). It is also one of the two Liriope species most frequently encountered in natural areas in the southeastern United States, the other being L. muscari (Miller, Chambliss, and Loewenstein 2010; Newsom 2010; Spaulding et al. 2010). Liriope spicata is reported to establish in both undisturbed and disturbed areas on woodland slopes, floodplains, and riparian habitats (Spaulding et al. 2010). It tolerates filtered sun to full shade (Fantz 2008) and has been reported to grow under dense stands of Chinese privet (Ligustrum sinense Lour.) (Spaulding et al. 2010) where few other species persist. It currently appears to be most abundant in urban forests where it likely escaped from nearby horticultural plantings (Miller, Chambliss, and Loewenstein 2010; Spaulding et al. 2010).
Impacts of L. spicata invasion are lacking documentation. Miller, Chambliss, and Loewenstein (2010) reported that dense stands displace native plants by forming a ground layer monoculture. Its utilization or avoidance by other species is currently unknown. Studies examining these aspects would be extremely useful to better understand its true impact.
In terms of control, Miller, Manning, and Enloe (2010) recommended physical or chemical approaches. These included frequent mowing, digging, and removing all rhizomes and corms, and multiple applications of glyphosate, imazapyr, or glyphosate plus imazapic. However, no research establishing the efficacy of these or other treatment approaches has been published. Spaulding (2010) stated that once established, the species was nearly impossible to remove. Given that excavating entire rhizome systems of creeping perennials is extremely labor intensive, herbicide treatment is the most likely management technique that would be utilized in most natural areas where L. spicata is found.
Table 1.
Control of container grown Liriope spicata with selected postemergence applied herbicides.
The overall objective of this research was to identify herbicide treatment(s) for effective L. spicata control. We quantified both above- and belowground responses to treatment with seven herbicides encompassing three specific herbicide modes of action. We hypothesized that L. spicata would respond to herbicides with known efficacy on other weedy species in the family Liliaceae and that herbicide rate would influence both above- and belowground responses.
METHODS
Since no previous research on L. Spicata control has been published, herbicide treatment selection was based upon current recommendations (Miller, Manning, and Enloe 2010), documented efficacy on other members of the Liliaceae, and by herbicides registered for use in natural areas, forestry, and noncrop settings where L. spicata infestations occur. Seven herbicides were selected for evaluation (Table 1). Dicamba and 2,4-D are synthetic auxin-mimics labeled for use in multiple crop and noncrop settings. While these herbicides are generally more effective on dicot species than monocot species, 2,4-D has been reported to be phytotoxic to both wild garlic (Allium vineale L.) (Klingman and Ahlgren 1951; Hardcastle 1976) and wild onion (Allium canadense L.) (Hardcastle 1976). Dicamba also controls both wild garlic and wild onion and is generally considered to be more effective on perennial species than 2,4-D (Zimdahl 1999). Glyphosate is also recommended for L. spicata control (Miller, Manning, and Enloe 2010). Glyphosate is well noted for its wide spectrum of weeds controlled and excellent translocation, which makes it effective in controlling many perennial species (Cobb and Reade 2010). Glyphosate inhibits the synthesis of the aromatic amino acids (Jaworski 1972; Steinrücken and Amrhein 1980). Imazapyr and imazapic are imidazolinone herbicides. Imazapyr is considered to be the most broad spectrum of this group in that it has minimal crop selectivity (Zimdahl 1999), but it is used in aquatic systems, natural areas, and forestry for the control of both grass and broadleaf weeds. Imazapic is labeled for use in natural areas, roadsides, range and pasture, and also controls wild onion and wild garlic. Metsulfuron is a sulfonylurea herbicide labeled for use in forestry, grasslands, and improved turf, and controls other Liliaceae species including wild onion and wild garlic. Sulfometuron is labeled for use in forestry and unimproved turf and provides broad spectrum control of several grass and broadleaf species (Anonymous 2010). Both the imidazolinones and sulfonylureas share the same mode of action; that is, inhibition of branch chained amino acid synthesis.
Experiments were conducted at the Paterson Greenhouse Complex of Auburn University, Auburn, Alabama. Divisions of L. spicata were collected in July 2011 and 2012 from a naturalized stand near the greenhouse complex and potted into 10-cm2 plastic pots, filled with a 6:1 (v: v) pine bark-sand substrate. This substrate was amended with a controlled-release granular fertilizer (Polyon® 17N-6P-12K, Harrell's Fertilizer, Inc., 203 West 4th Street, Sylacauga, AL 35105), dolomitic limestone, and a micronutrient fertilizer (Micromax®, O. M. Scott Corp., 14111 Scotts Lawn Road, Marysville, OH 43401) at 8.3, 3.0 and 0.9 kg m-3, respectively. Approximately 1 mo after planting, established divisions were transplanted into 3.8 L containers using the same media. Plants were maintained in an outdoor area under black (40%) shade cloth and received approximately 1.25 cm of overhead irrigation daily through the remainder of the growing season. Plants were covered with plastic film for protection during the following winter and used for experimentation the following summer. Thus, plants used in the experiment conducted in the summer of 2012 had been propagated and established in 2011, and plants used in the 2013 experiment had been propagated and established in 2012.
Experimental Procedures
The herbicide products used in this study were as follows: 2,4-D was applied as the dimethylamine salt, (2,4-D Amine Weed Killer, Universal Crop Protection Alliance LLC, 1300 Corporate Center Curve, Eagan, MN 55121), dicamba was applied as the dimethylamine salt (Banvel® , Arysta LifeScience North America LLC, 15401 Weston Parkway, Suite 150, Cary, NC 27513), glyphosate was applied as the isopropylamine salt (Roundup® Pro Monsanto, 800 N Lindbergh Blvd., St. Louis, MO 63167), imazapic was applied as the ammonium salt (Plateau®, BASF Corp., 26 Davis Dr., Research Triangle Park., NC 27709), imazapyr was applied as the isopropylamine salt (Arsenal®, also BASF), metsulfuron was applied as Escort® XP (E. I. Dupont de Nemours, 1007 Market St., Wilmington, DE 19898), and sulfometuron was applied as Oust® XP (also Dupont). Each of the seven herbicides was applied at two rates (Table 1). Rate structures were specifically designed with 1x and 2x rates that fell within the labeled range for each product, as used in many natural area or forest environments. All treatments also included a nonionic surfactant to improve foliar absorption (Agri-Dex®, Helena Chemical Company, 225 Schilling Blvd., Suite 300, Collierville, TN 38017) at 0.25% v/v. Treatments were applied during the first week of May using an enclosed-cabinet sprayer, calibrated to deliver 280 L ha-1 at 193 kPa. Treatments were applied to ten, single-pot replicates. Plants were returned to the same outdoor growing area after treatment. Treated plants were then protected from irrigation or rainfall for 72 h after treatment application. The experiment was repeated in two growing seasons, 2012 and 2013.
Data Collection and Statistical Analyses
Treated plants were visually evaluated at 30 and 60 DAT, using a scale of 0 (no effect) to 10 (complete shoot death). After the second rating at 60 DAT, foliage from all 10 replicates of each treatment was harvested by clipping at approximately 1 cm above the media surface. All shoots were separated into living or dead, and live foliage was dried at 70 °C for 24 h and weighed. At 90 DAT (or 30 days after the foliage harvest), shoots that had regrown (if any) were harvested and dry weights determined for five of the 10 treated replicates. Also, after foliage harvest, these plants were removed from their pots and all belowground living tissue including roots, rhizomes, and corms were harvested, oven dried, and weighed. At 180 DAT (or 120 day after initial foliage harvest), the foliage and belowground tissue dry weights were determined for the remaining five replicates as previously described. This longer term evaluation was important to collect as herbicides such as imazapyr and glyphosate may continue to work in susceptible species for several months. Tissue dry weight data were then converted to percent control values by expressing the harvested foliage and belowground root weights as a percent reduction relative to the appropriate nontreated control. Therefore, 100% control indicates no live tissue remaining (i.e., death), 0% control indicates tissue weight equivalent to the nontreated control (no effect), and a negative control indicates a weight greater than the control (i.e., a possible stimulatory effect).
The experimental design employed in this study was a completely randomized design with two experimental runs. All data were analyzed by SAS version 9.2 (SAS Institute, Cary, NC) using an analysis of variance (PROC GLM) with herbicide and herbicide rate as the primary treatment variables and with experimental run as the blocking factor. An herbicide by herbicide rate interaction factor was also included in the model statement. Treatment effects on visual control ratings at 30 and 60 DAT and root and shoot biomass at 90 and 180 days after herbicide application were considered significant at α = 0.05. If the interaction factor was significant, an analysis of variance was used to test the effect of herbicide within each herbicide rate and experimental run was used as a blocking factor. Model residuals did not violate the assumptions of normality and equality of variance so no transformations were required. Post hoc comparisons among herbicide types and among herbicide rates were performed using a Tukey's honestly significant difference test.
RESULTS
At 30 DAT, there was a significant herbicide by rate interaction for visual control ratings. The interaction was largely driven by the metsulfuron rate response, which went from 2 at the 1x rate to 4.2 at the 2x rate. At the 1x rate, visual control ratings were very low and ranged from 0 to 2 across all herbicide treatments (Table 1). Very few visual symptoms were apparent from any treatment. At the 2x rate, metsulfuron was significantly better than all other treatments. However, visual control ratings were still very low across treatments. Metsulfuron symptomology included shoot yellowing and some stunting, which is characteristic of herbicides that inhibit amino acid biosynthesis.
At 60 DAT, the herbicide by rate interaction was again significant. For this evaluation, the interaction was largely driven by imazapyr and metsulfuron. At the 1x rates, only imazapyr and metsulfuron resulted in visual control ratings above two and the same was also true for the 2x rates (Table 1). Herbicide injury from these two herbicides was apparent and included chlorosis of the new growth, some necrosis, and stunting, all at a slightly higher degree than at 30 DAT. Plants in almost all other herbicide treatments exhibited very little visually detectable damage.
At 90 DAT, herbicide treatment significantly influenced L. spicata above- and belowground biomass (Figure 1A and B). However, neither herbicide rate nor the herbicide by rate interaction was significant for either response variable. Shoot response varied by experimental run; herbicide treatments performed better in the first run compared to the second (data not shown). For aboveground (shoot) biomass, untreated plants averaged 7.4 g per pot. Metsulfuron and imazapyr reduced creeping lilyturf shoot regrowth by 74 to 79% and were not different (P = 0.9; Figure 1A). All other herbicide treatments resulted in a negative reduction in shoot biomass (i.e., a possible stimulatory effect).
Figure 1.
Liriope spicata above- (A) and belowground (B) biomass response 90 days after herbicide treatment. Negative values indicate a possible stimulatory effect. Means with the same lower case letters are not different at the 0.05 level of significance.
Reductions in belowground biomass at 90 DAT did not follow a similar pattern as reductions in aboveground biomass. Untreated creeping lilyturf belowground biomass averaged 29.6 g per pot, which accounted for 80% of the total biomass. Metsulfuron, imazapyr, sulfometuron, and glyphosate reduced belowground biomass by 41 to 56% and were not different from each other (Figure 1B). Metsulfuron reduced belowground biomass to a greater extent than imazapic, dicamba, and 2,4-D. However, there were few clear differences between most other treatments.
At 180 DAT, the herbicide by rate interaction was significant for creeping lilyturf aboveground biomass (Figure 2A and B). Untreated plants averaged 26.1 g of shoot biomass per pot. For the 2x herbicide rate, imazapyr and metsulfuron reduced creeping lilyturf shoot biomass by 99% and 85% respectively and were different from all other treatments. All other herbicide treatments reduced shoot biomass by less than 18% and were not different from each other (Figure 2A). At the 1x herbicide rate, imazapyr reduced shoot biomass by 85% and was different from all treatments except metsulfuron (Figure 2B). The significant herbicide by rate interaction for the shoot biomass reduction was driven by metsulfuron, which increased from 42% at the 1x rate to 85% at the 2x rate. No other herbicide treatment was influenced by increasing the rate from 1x to 2x. This is very noteworthy as it indicates considerable tolerance of L. spicata to several key herbicides, even when applied at relatively high rates.
For belowground biomass at 180 DAT, the herbicide by rate interaction was highly significant (Figure 3A and B). The untreated plants averaged 45.3 g of belowground biomass per pot, which accounted for 63% of the total biomass. The interaction was driven by the creeping lilyturf response to 2,4-D, glyphosate, imazapic, and imazapyr. As the rate doubled for each of these herbicides, belowground biomass reductions increased by 21, 50, 17, and 23%, respectively (Figure 3A and 3B). At the 2x rate, metsulfuron and imazapyr reduced belowground biomass by greater than 97% and were not different from each other (P = 0.48). Glyphosate, imazapic, and sulfometuron reduced belowground biomass by 42 to 45% and were not different from 2,4-D. Dicamba had very little effect on belowground biomass (Figure 3A). At the 1x rate, metsulfuron and imazapyr reduced belowground biomass by 74 and 97% and were different from all other herbicide treatments (Figure 3B). Sulfometuron and imazapic reduced belowground biomass by 26 to 43% and were not different from each other. Dicamba, 2,4-D, and glyphosate reduced belowground biomass by -6 to 4%, which was an essentially negligible effect.
DISCUSSION
Both metsulfuron and imazapyr were far more phytotoxic to L. spicata than the other five herbicides at the rates tested with a single application. Both were very slow to work within the target species, as visual ratings at 30 and 60 DAT were generally low (Table 1). These visual ratings at 60 DAT would be unacceptable if rapid control was desired and it would be easy to conclude that treatments were only marginally effective. However, at 90 DAT, quantitative above and belowground biomass data indicated that the herbicides were continuing to work (Figure 1). At 180 DAT, almost complete control of belowground biomass, which included rhizomes and corms, had occurred for the 2x rates of imazapyr and metsulfuron (Figure 3A). This degree of root system kill for an aggressive, creeping perennial is highly desirable to invasive plant managers. However, it often requires multiple herbicide applications over years to achieve. For example, a recent study found that cogongrass required up to six repeated glyphosate treatments before complete elimination of the rhizome layer (Aulakh et al. 2014). Glyphosate, imazapic, and sulfometuron all reduced belowground biomass by approximately 41 to 45% at 180 DAT. However, vigorous, new shoot growth was apparent in these three treatments and reductions in shoot biomass were much lower. This would suggest that followup treatments with these herbicides could provide good control. However, additional studies are needed to verify this.
Although metsulfuron and imazapyr provided excellent control of L. spicata, one of the major concerns with these herbicides is the potential for nontarget damage through soil residual activity, especially in hardwoods (Lawrie and Clay 1994; Kochenderfer et al. 2001). With the exception of Liriope control in turfgrass with metsulfuron or in pine plantations with metsulfuron or imazapyr, both herbicides would require very careful use in many natural areas where creeping lilyturf is commonly found (Miller, Chambliss, and Loewenstein 2010; Spaulding et al. 2010).
In terms of our specific hypotheses, we found that only certain herbicides with known activity on members of the family Liliaceae controlled creeping lilyturf. These included metsulfuron, imazapyr, and glyphosate to some extent. However, the herbicides 2,4-D and dicamba did not control creeping lilyturf. We also found significant herbicide by rate interactions for almost all variables tested, indicating the rate response was specific to certain herbicides. Metsulfuron and imazapyr were frequently responsible for the interaction and additional dose response studies should be conducted with the herbicides. Additionally, future research should examine sequential timings and tank mixes of the herbicides that were effective with the end goal of complete kill of all roots, rhizomes, and corms.
CONCLUSIONS
These data indicate that L. spicata can be effectively controlled. However, there are still limited options for effective control, especially given the potential for nontarget damage with metsulfuron and imazapyr. Several of the herbicides tested were not effective at rates commonly used for control of many other invasive plants and weedy members of the family Liliaceae. Further studies including field experiments should be conducted to evaluate additional control options and refine herbicide treatments that may lead to patch eradication of this and other aggressive Liriope species.
