Northern Prairie Wildlife Research Center
Until the 1980's, the resistance of purple loosestrife to available herbicides gave wetland managers a limited list of compounds for controlling this hardy exotic. Although not approved for aquatic use, dicamba (Banvel: 3,6-dichloro-o-anisic acid) was used on experimental plots at Parker River NWR with modest success (Parker River NWR, 1972 unpublished field report). A mixture containing equal amounts of dicamba and 2,4-D was applied to the infestation near Meridian, Idaho. Reed canarygrass was the dominant forage at this site; the infested area was accessible to power spraying equipment. Despite treatments during early bloom in 2 successive years, a scattering of surviving purple loosestrife plants in August of 1980 (D. Q. Thompson, unpublished field notes) indicated that the site was still threatened with a resurgence of the infestation. Our earlier reference to the work of Gilbert and Lee (1980) suggests that summer dormancy of a small percentage of L. salicaria root crowns could account for some of the difficulty of controlling this weed with chemicals.
Prospects for chemical control changed sharply with the advent of glyphosate (Roundup: N-[phosphonomethyl] glycine). First described by Baird et al. (1971) as a postemergence spray for the control of perennial herbaceous weeds, this broad-spectrum herbicide has many attributes that support its use on agricultural lands. It is a systemic compound that is absorbed by leaf surfaces and translocated to the entire plant, including roots and underground stems. It is effective at relatively low concentrations and has a low potential for bioaccumulation. The half-life of its biological activity in soil and water is fairly short—about 2 months (Rueppel et al. 1977). It is of relatively low toxicity to birds (Batt et al. 1980), mammals (Monsanto, mimeo. report), and aquatic invertebrates and fish (Folmar et al. 1979). The effectiveness of glyphosate in agriculture soon led to applications on perennial weeds on rangeland (Gottrup et al. 1976), ornamentals and Christmas tree plantings (Ahrens 1974), and in general forestry (Sutton 1978). The first report of use in an aquatic habitat is from Welker and Riemer (1982), who used glyphosate to control an infestation of fragrant waterlily (Nymphaea odorata) in New Jersey with application rates of 2 and 4 pounds per acre in successive growing seasons. The second year, spraying was needed to control seedlings germinating from seeds in bottom sediment and resulted in eradication for at least 3 more years. Rawinski (1982) and Malecki and Rawinski (1985) reported on L. salicaria tests at the Montezuma NWR in central New York; they used replicated plots to test three rates of glyphosate application (1.7, 3.4, and 6.7 kg/ha) against three stages of growth (vegetative, 13 June; early flowering, 13 July; late flowering, 11 August). Results showed no significant difference in application rates but a highly significant difference in the timing of applications. The late flowering application was the most effective with nearly 100% shoot reduction. In the following growing season, Rawinski found that the timing of application also affected the establishment of purple loosestrife seedlings—the plots sprayed in June became reinfested with seedlings, whereas the plots sprayed in July and August were free of purple loosestrife seedlings. Furthermore, the infested plots were colonized by several species of desirable waterfowl food plants. Despite the promise of these results, Rawinski (1982) cautioned against the widespread use of glyphosate without adequate knowledge of the effects of this new chemical on marsh ecosystems.
Christy et al. (1981) studied the effects of glyphosate on the growth rate of Chlorella sorokiniana. They reported the glyphosate concentration threshold for zero growth to lie between 17.7 and 23.7 × 10-6 M. In limited tests, they also found supporting evidence for soil inactivation of glyphosate with Kaolinite. They concluded with a warning (similar to that of Rawinski) that careless use of glyphosate in aquatic systems "could result in elimination of beneficial species, proliferation of noxious species, and disruption of food chains and/or nutrient cycles." The criticism of use of chemical controls on natural habitats hinges on the percentage of habitat treated and the care with which the treatment is delivered. The delivery system should be as gentle and as target specific as possible.
In 1982, a new formulation of glyphosate (Rodeo-EPA Reg. No. 524-343) was approved for use over water (J. B. Elder, USFWS memo, 14 February 1983). The new formulation includes the same active ingredient (glyphosate), but uses a new surfactant (Ortho X-77). Several field tests with the new formulation have shown that glyphosate is very effective in reducing large monotypes of reed (Phragmites australis) in eastern wetlands (G. W. Gavutis, USFWS memo). Similar work was conducted on purple loosestrife in marshes in northeastern Ohio (Balogh 1986).
The response of various growth stages of purple loosestrife to water levels is not well known. Without specifying water depths or timing, McKeon (1959) reported that water manipulation attempts on Travers Marsh (5 ha) were "completely unsuccessful." However, G. Cole (personal communication) reported that raising the water level about 60 cm in the 1960's resulted in the retreat of a monotype of purple loosestrife to the rim of this marsh leaving the deeper (120-150 cm) portions of the impoundment free of the plants. R. H. Smith (1964), doubtless referring to mature plants, stated that purple loosestrife will "survive for many years even when submerged to depths of 2 or 3 feet … " He did not attempt to use higher impoundment levels to control loosestrife and, indeed, could not raise the water on his experimental plots to more than 60 cm depth. Rawinski (1982) did not attempt to use flooding as an experimental procedure, but several of his field studies offered insights into the relation of water levels to purple loosestrife vigor and abundance. One set of plots compared the number of loosestrife and cattail shoots in a nearly monospecific stand of L. salicaria with an adjacent mixed-species stand (55% loosestrife: 44% cattail). Water levels were maintained at 40 cm depth throughout 3 years of study (1978, 1979, 1980); during this time the densities of loosestrife shoots decreased sharply in both sets of plots, whereas cattail increased threefold in the loosestrife plots and decreased to about two-thirds of its original densities in the mixed-species plots. Rawinski attributed the decline in loosestrife as "probably due to the combined effects of high water levels, damage by carp, and suppression by cattail." The effect of autotoxic feedback on Typha latifolia (McNaughton 1968) was not mentioned by Rawinski—but may have been responsible for some of the decrease observed in cattail shoots in the mixed-species stands.
The use of replacement control in native wetland communities has limited value in that the cure (discing and seeding) is potentially as disastrous to the infested community as the invading weeds. In some circumstances, however, this may be the habitat manager's best choice. To control purple loosestrife in badly infested agricultural fields, Muenscher (1955) recommended plowing and planting an intertilled crop. Novak (USFWS Field report, Region 5, 1967) described the control of a heavy purple loosestrife infestation at Great Meadows NWR near Concord, Massachusetts; the field was mowed repeatedly, plowed, and planted to reed canarygrass. Rawinski (1982) tested seven plant replacements for purple loosestrife and found that Japanese millet (Echinochloa frumentacea) and reed canarygrass were promising, in the order mentioned. Thereafter, a large-scale field trial was conducted to test the ability of Japanese millet to suppress naturally occurring loosestrife seedlings resulting from an early July drawdown. The millet seeding was especially successful on deep-muck sites. In addition to suppressing loosestrife seedlings, the mature emergent millet stands were used by mallards and black ducks (Anas rubripes). This was a minimum disturbance replacement in that Rawinski used a hand-cranked machine to seed the newly exposed bottom. Furthermore, Rawinski indicated that Japanese millet is not likely to become a weedy intruder where used in natural habitats.
When summer drawdown of infested wetlands cannot be avoided, seeding exposed muck flats with Japanese millet is a good temporary measure to forestall the establishment of purple loosestrife seedlings. This would be particularly useful on, say, an intensively used waterfowl display pool where the exposed flats were accessible for hand seeding. It would be less useful on large impoundments with scattered emergent stands and many remote flats.
The growth form and phenology of purple loosestrife indicate that it will probably not be susceptible to control with fire. As an herbaceous perennial, its overwintering growing points on the crown of the rootstock lie about 2 cm below the soil surface, where they are fairly well insulated from the heat of a surface fire. Furthermore, L. salicaria begins spring growth about a week or 10 days after broad-leaved cattail and reed canarygrass, thereby reducing the chances that a spring burn would decrease the vigor of the weed. Also, fuel energy tends to remain concentrated in the erect overwintering stems rather than accumulating as fine-textured fuels at the soil surface. This shortcoming was demonstrated by an unsuccessful prescribed burn on a 1.6 km-long dike at the Montezuma NWR in the spring of 1971 (D. Q. Thompson, unpublished field notes). The dike cover had been seriously overgrazed by Canada geese in preceding growing seasons, leaving a patchy distribution of mixed grasses and forbs that had been invaded by small, scattered clumps of purple loosestrife. Although a slow, backing fire would have caused maximum heat penetration to the loosestrife root crowns, the distribution of surface fuels lacked enough continuity to maintain a moving line of fire. We were forced to use a head fire, with many patches needing separate ignition to finish the task. In the following year, mortality to loosestrife root crowns was less than 10%; the remaining crowns made vigorous growth and new seedlings continued to expand the dominance of the weed in the dike cover.