USGS - science for a changing world

Northern Prairie Wildlife Research Center

  Home About NPWRC Our Science Staff Employment Contacts Common Questions About the Site

Spread, Impact, and Control of Purple Loosestrife (Lythrum salicaria) in North American Wetlands

Chronology and Mode of Spread

Early Spread in Northeast: Pre-1880

Although L. salicaria was well established along the New England seaboard by the 1830's, its spread into vast stretches of interior drainage basins (Fig. 5) could not have been easily accomplished without (a) profound disturbance and stress in native wetland plant communities and (b) a ready mode of dispersal that connected marine estuaries with interior wetlands. Community stress and perturbation had its beginnings in 1790 with the assignment of military lots in upstate New York to veterans of Sullivan's 1779 campaign to suppress the tribes of the Iroquois nation. In the rush of agricultural settlement that followed, wetland meadows were scythed, trampled, and grazed; upland watersheds were drastically altered as forests were cleared for cropland and pasture. A less obvious change was the cessation of aboriginal fires following early settlement. Day (1953) and Thompson and Smith (1970) documented the purposeful use of fire by Indians of the Iroquois Nation. By 1822, the effects of fire cessation were obvious to Dwight (1823), who astutely commented on the remarkable buildup of unburned surface litter in forest stands encountered along his stagecoach route through the upper Finger Lakes region. The effects of aboriginal fire on interior marshes is unknown, but the frequency of fire in these wetlands probably decreased sharply following settlement. We do not know enough of L. salicaria's response to fire to judge the importance of this likely change in fire regimen.


Early canals not only provided aquatic weeds access to the interior of the continent, they also added a new source of disturbance to native plant communities. In 1817, construction of the middle section of the Erie Canal began (U.S. Inland Waterways Commission [USIWC] 1972). By the following year more than 2,000 men and as many horses and oxen were building locks, excavating channels, and smoothing adjoining towpaths (Harlow 1926). In many stretches, the canal followed upland contours that led away from natural water routes. These stretches added new seepage habitat for plant immigrants. Moreover, each canal segment required "feeders" to replace the volume of water lost in normal lock operations. Surface water diversions, flumes, and feeder canals extended into the surrounding watershed to provide an adequate water supply; profound habitat disturbance thus preceded the completion of the Erie Canal in 1825. The enormous success of the new canal was foretold in more than a half-million dollars of tolls collected in its first year of operation (Harlow 1926). In the euphoria of expectations that followed the opening of the Erie Canal, more than 4,800 km of canals were constructed by 1840 in New England, New York, Pennsylvania, New Jersey, and Ohio (Tanner 1970). The most important of these water routes are shown on Fig. 5. Careful examination of the early L. salicaria plots shows that with the exception of the interior Pennsylvania canals, the spread of purple loosestrife was closely related to canal traffic moving inland from the northeastern marine shipping estuaries. Twenty-seven of the 30 collection localities were near seaports or along canals. Despite the abundance of early records along the Delaware and Raritan Canal, the invasion of Pennsylvania, Maryland, Virginia, and West Virginia was very slow compared with the invasion of upstate New York along the Erie. The greatest concentration of L. salicaria establishments was along the Delaware and Raritan Canal and the Erie Canal. Several factors may account for the sharp differences in spread. First, these canals were low profile waterways that offered direct connection between seaboard and interior. In 68 km, the D&R Canal needed only 14 locks and 35 m of lift to connect Raritan Bay with the Delaware River. The Erie Canal needed only 83 locks in 616 km to lift its barges from the Hudson River over a 192-m summit on their transit to Lake Erie. In contrast, canals leading into the interior of Pennsylvania were confronted by the mountainous northeast-southwest ridges of the Appalachian Region. The Morris Canal had to climb from sea level to a summit of 279 m before dropping 230 m into the upper Delaware River at Phillipsburg. Twenty-three inclined planes were needed to winch the barges overland from one watershed to the next, thus breaking the water connection between each segment. Second, in contrast to the broad, open glacial basins of upstate New York, the interrupted waterway segments in Pennsylvania moved along relatively narrow valleys with thin, early-drift mantles and well-shaded stream margins. All of these factors would have been deterrents to the spread of an emergent aquatic plant.

Other Routes of Early Spread

The National Road, various turnpikes, and regional railroad networks were also important routes for the westward expansion of the new republic. The National Road (Fig. 5) reached the Indiana line by 1840 (Ohio Writers' Project 1940); however, despite heavy wagon traffic over the route since its beginnings in 1811 at Cumberland, Maryland, it was of little importance as a means of L. salicaria spread. The single establishment of L. salicaria reported from the bottomland of the Ohio River at Wheeling, West Virginia, in 1878 might well have arrived with river traffic. The first railroad connection between Chicago and the eastern seaboard came into service in 1852 (Meyer 1969); railroad tonnage surpassed canal tonnage in New York and increased steadily to more than 3 times the tonnage of canals in 1880 (USIWC 1972). Nevertheless, an inspection of railroad routes and L. salicaria collection plots shows little correspondence. This is not surprising; railway rolling stock and cargos are a possible but not very likely mode of spread for an aquatic weed. Most railway trackage is well drained and crosses wetland habitats on raised embankments. Moreover, railroad rolling stock seldom comes in contact with purple loosestrife seed sources. L. salicaria rarely occurs in seed or grain cargos, although it could have gained access in marsh hay used as feed or bedding for livestock. Last, although livestock shipments on midwestern railroads began to reach significant tonnages after 1850, regional railroad networks were built out from centers that they served. Traffic flow into the Chicago hub came mostly from purple loosestrife-free areas in the adjoining midwestern States and the Great Plains.

In summary, all evidence suggests that the early phase of spread of L. salicaria into the interior of North America was by waterborne commerce into recently disturbed or stressed habitats. Finally, although some of the new establishments were probably escapes from cultivation, more than three fourths of the records indicated fortuitous or accidental spread from other sources.

Spread into Midwest: 1880-1900

Figure 6 includes the 30 collections shown in Fig. 5 along with the localities of 40 new collections made between 1881 and 1900. From 1800 to 1900, canal traffic declined and railroads not only took over all priority shipments but made severe inroads into the bulk cargo (coal and grain) that had once brought the principal revenues to canal carriers. Most of the regional and continental railroad networks were established in this period. Yet, despite the tremendous expansion of rail traffic, little additional spread in L. salicaria (Fig. 6) can be related to traffic on the new rail networks. Most of the new records of L. salicaria establishment occurred along marine commerce routes or at coastal or inland ports. From 1880 to 1900, maritime traffic was very active from Buffalo to ports on the Great Lakes. Immigrants, livestock, and cargo were transferred from the Erie Canal terminus at Buffalo to sailing schooners and steamers for transshipment into the midwestern frontier. The collection site in Michigan is at Muskegon Harbor—an early lumber port.

GIF-Spread of purple loosestrife as of 1900
Fig. 6. Spread of Lythrum salicaria as of 1900 (after Stuckey 1980).

Colonization of Midwest and Northwest Pacific Coast: 1901-1940

The continued spread of L. salicaria during the first four decades of the 20th century is shown in Fig. 7. New records were reported from maritime States and Provinces of the Northeast. Coastal sites in eastern Maine, western Nova Scotia, the Gaspé Peninsula, Anticosti Island, and Newfoundland were colonized by L. salicaria by 1940. Coastal schooner trade in lumber, coal, grain, and hay was very active early in this period and may have been responsible for the spread of L. salicaria from colonized areas to unoccupied estuaries. For the first time, establishments were reported from marine estuaries in the Pacific Northwest, suggesting again that marine commerce was the principal mode of spread.

GIF-Distribution of purple loosestrife as of 1940
Fig. 7. Distribution of Lythrum salicaria as of 1940 (after Stuckey 1980). Stipple pattern indicates extent of Wisconsin Age glacial drift.

The early 1900's saw the last struggles of a declining and obsolescent canal network. In New York, the shipment of grain shifted from 96% by canal in 1868 to 82% by railroad in 1906 (USIWC 1972). Two transcontinental railroad routes in Canada and six in the United States were in operation for most of this period, but contributed only a few new records of L. salicaria spread. The construction of our first State and Federal highway networks occurred during this period with seemingly little effect on the spread of purple loosestrife.

Spread into the Arid West: 1941-1985

L. salicaria's known distribution as of 1985 (Fig. 8) shows a greatly expanded range since 1941. In the Northeast, herbarium specimens were obtained for the first time from New Brunswick and Prince Edward Island. Eastward range extension was also obvious on both sides of the St. Lawrence estuary. Colonization of the northern Midwest was now nearly complete with range extensions occurring in western Minnesota. New locations also appeared in the upper Red River Valley along the Minnesota-North Dakota border and extended across the international border into Manitoba and the wetlands of Lakes Winnipeg and Manitoba. But the most dramatic colonizations since 1940 occurred in interior western North America where L. salicaria infestations appeared in an additional 12 States and 2 Canadian Provinces lying west of the 100th meridian—Powell's (1962) definition of the eastern boundary of the Arid West. With the exception of the Arkansas and Colorado rivers and the Rio Grande, all of the major watersheds within this vast expanse had now been colonized. Vigorous new colonies in California, Idaho, Nebraska, Wyoming, and Washington suggest that L. salicaria may become troublesome in the wetlands of the Arid West.

GIF-Distribution of purple loosestrife as of 1985
Fig. 8. Distribution of Lythrum salicaria as of 1985.

Two major public works programs contributed to massive habitat changes that favored purple loosestrife spread in western States. The idea of using the abundant water sources of the western cordilleras to irrigate outlying steppe or desert benchlands had its roots in the life and work of John Wesley Powell, These concepts took form as one of Theodore Roosevelt's legislative priorities and gave rise to the Federal Reclamation Act of 1902. At this time the total number of hectares under irrigation (Fig. 9) on Federal projects was about 8,000. By 1910, the efforts of the newly formed Reclamation Service had increased the total number of hectares under irrigation by more than 20-fold to 192,000. Rapid growth continued through the next 20 years to about 1,100,000 ha in 1930. Shortly thereafter, Reclamation received a vigorous boost from Franklin D. Roosevelt's New Deal, and new projects were established that eventually pushed the irrigated hectares through a threefold increase to about 3,800,000 by 1980. Although most irrigated land is in cultivated crops and therefore probably not susceptible to invasion by purple loosestrife, off-water swales that collect runoff from irrigated fields, and seepage slopes along canals and ditches, are likely places for L. salicaria encroachment. The recent infestations in California, Idaho (Fig. 10), Washington, and Wyoming (Fig. 11) are all in irrigation areas. Chemical control has already been necessary in irrigated rice in California (Howell and True 1966) and in a seepage pasture west of Boise in Ada County, Idaho (W. Hartman, personal communication).

GIF-Chart of FRP land amount under irrigation from 1906 to 1980 Fig. 9. Hectares (in millions) of land under irrigation on Federal Reclamation Projects from 1906 to 1980; the 1980 estimate is projected from indices of irrigated acreage on Federal Reclamation Projects 1956-1978 (U.S. Water and Power Resources Service 1979).

Fig. 10. A mature and robust purple loosestrife plant (height 2 m, width 1.5 m) grows in an "off-water" ditch in an irrigated hay meadow 10 km west of Boise, Idaho, 16 July 1980. jpg-Purple loosestrife in Idaho

jpg-Purple loosestrife in Wyoming Fig. 11. Wyoming's only reported Lythrum salicaria infestation has invaded a once-thriving stand of Typha latifolia, 6 km southwest of Powell, Wyoming, 27 August 1982.

Called the largest public works project in United States history (Leavitt 1970), the national superhighway network has resulted in 97,000 km of road construction at an average cost of $930,000 per km. Specifications for the interstate system called for well-drained crowns; the low gradients of even the steepest hills were achieved with massive excavations and backfills. The excavations often cut through local groundwater seams and created new habitat for emergent aquatic plants. These moist areas are usually first colonized by Typha spp. and other native aquatics. More recently, many of these seepages have been invaded by purple loosestrife, perhaps from seed transported by highway traffic.

H. Brumsted (letter, 22 August 1978) described the status of purple loosestrife along I-84 between Danbury, Connecticut, and the junction of State highway 17, near Middletown, New York. He noted, "as expected, Lythrum was present in all types of depressions associated with the highway … What I did not expect to find, however, were single clumps and scattered colonies of Lythrum amid areas of grass on the median, and also on fill slopes, that certainly were not ditches or wet depressions … [what] surprised me most … was [L. salicaria] growing at the left edge where pavement joined the median … during the return trip on the 21st, I had opportunity to observe the effect of a trailer truck's passing … one roadside clump of Lythrum … The truck, passing at about 65 mph at my left, caused a partial vacuum … that caused a clump of Lythrum at the median edge of the pavement to bend over the road at what appeared to be an angle of at least 45 degrees. Were this to occur when Lythrum's capsules were dehiscent, it would seem virtually certain that seed would be swept into the air currents and onto the highway. I see this … as a primary means by which the minute seeds of Lythrum could … be transported great distances." In addition to suggesting the role of superhighways in creating newly disturbed habitat that enlarges the opportunities for weeds, Brumsted pointed out that whereas earlier highways followed the ridges and valleys of the Appalachian region, the superhighways cut through them and thereby broke open the barriers that had previously blocked the spread of purple loosestrife into the next watershed.

Rate of Spread

Mack (1981) and Forcella and Harvey (1983) used herbarium specimen records and reports of field collections or observations as a means of estimating the rate of spread of rangeland weeds. Although plant collection records have many limitations as indicators of abundance, they are useful as records of occurrence. Several assumptions must be met before the value of these sources can be accepted: (a) a reasonably uniform level of field activity must carry through the span of the collections, (b) the field activity must occur in a more or less random pattern over all of the collection area, and (c) the plant must be easily seen and identified. Of the first assumption (a), by the early 1800's, botanical field collections were well under way in eastern North America; John Torrey and Asa Gray published the first volume of their projected A Flora of North America in 1840. Moreover, their efforts were vigorously supplemented by colleagues at other institutions and by private collectors—often medical doctors (as were Gray and Torrey), whose interest in botany may have stemmed from training in herbal medicine. A dozen or more academies and natural history or scientific societies published the records of field collections by this network of enthusiastic but highly competent amateur botanists. If the era of contributing amateur botanists diminished with the turn of the century, their collecting efforts were replaced by trained botanists at newly established private and public institutions. It is less certain that all of the collection range (b) has been evenly covered in a random pattern. This assumption seems to be reasonably met in early collections, but once a species became well represented in an herbarium, the level of collecting effort declined (Stuckey 1980, Table 1), making an estimate of the frequency of encounter less sensitive. Last (c), the tall and colorful inflorescences of purple loosestrife are dramatically obvious and identification is quite straightforward.

We divided four States (New York, Michigan, Wisconsin, and Minnesota) into 15' latilong blocks and scored the cumulative occurrences of distribution from 1860 to 1980 (Fig. 12). These States have similar climates and vegetative zones, a common history of Pleistocene glaciation, and an abundance of open glacial marshes—habitat that is highly susceptible to invasion by purple loosestrife. A latilong block has an area of 771.7 km²; 1,172 blocks make up the area of the four States. Several generalizations are obvious in the configuration of the data points. Purple loosestrife's rate of spread has been slow compared with the invasion of the western range by cheat grass, Bromus tectorum (Mack 1981); however, within 20 years of its first report in the southern Hudson Valley in New York, purple loosestrife had begun a pattern of geographic spread that indicated a high degree of preadaptation to North American habitats. The infestation of native habitat has been increasing exponentially since 1880, with a sharp upturn in the rate of spread beginning in 1940. From 1900 to 1940, the rate of spread was roughly 1 latilong block per year; from 1940 to 1980, the rate increased to 1.5 latilong blocks per year. It could be argued that the sharp increase in L. salicaria's rate of expansion is as much a measure of pervasive habitat disturbance as of the aggressiveness of the weed. The extent of habitat disturbance, including various enhancement plans for improving the productivity of wetlands, has been, and continues to be, parallel to the exponential increase of our human population. Plots of human impact on the North American environment, that is, energy consumption, wastes created, acres cultivated, volume of water impounded, highway mileage, chemicals and fertilizers used—all would approximate an exponential curve.

GIF-Graph of latilong blocks occupied by purple loosestrife
Fig. 12. Number of 15' × 15' latilong blocks occupied in westward spread of Lythrum salicaria through New York, Michigan, Wisconsin, and Minnesota from 1840 to 1980. N total blocks = 1,171; area each block = 771.7 km²

Previous Section -- Arrival and Establishment in North America
Return to Contents
Next Section -- Purposeful Introduction

Accessibility FOIA Privacy Policies and Notices

Take Pride in America logo logo U.S. Department of the Interior | U.S. Geological Survey
Page Contact Information: Webmaster
Page Last Modified: Saturday, 02-Feb-2013 06:29:55 EST
Menlo Park, CA [caww55]