Evidence of Hybridization Between Lythrum salicaria (Purple Loosestrife) and L. alatum (Winged Loosestrife) in North America

Abstract

• Background and Aims Although Lythrum salicaria (purple loosestrife) was introduced to North America from Europe in the early 1800s, it did not become invasive until the 1930s. Whether hybridization with L. alatum (winged loosestrife) could have played a role in its ultimate spread was tested.

• Methods Six diagnostic morphological traits (flower number per axil, leaf placement, calyx pubescence, style type, plant height and leaf shape) were surveyed in 30 populations of Lythrum across eastern North America. Patterns of AFLP variation were also evaluated using five primer pairs in a ‘global screen’ of the same North American populations of L. salicaria and L. alatum described above, in L. salicaria from 11 European populations located in Germany, England, Ireland, Austria and Finland, and in six L. salicaria cultivars.

• Key Results All of the North American L. salicaria populations had individuals with alternate leaf placement and 1–2 flowers per leaf axil, which have not been described in Eurasian L. salicaria but predominate in North American L. alatum. In addition, two L. salicaria populations were intermediate in height and leaf ratio between the typical L. salicaria and L. alatum populations in their native fields and when grown in a common greenhouse. In screens of variation patterns using 279 AFLPs, only two fragments were found that clearly supported introgression from L. alatum to L. salicaria.

• Conclusions The evidence indicates that L. salicaria may have hybridized with L. alatum, but if so, only a small fraction of L. alatum genes have been retained in the genome of L. salicaria. This is unlikely to have led to a dramatic adaptive shift unless the introgression of a few key genes into L. salicaria stimulated a genomic reorganization. It is more likely that crossing among genotypes of L. salicaria from multiple introductions provided the necessary variability for new adaptations to arise.

INTRODUCTION

A time lag is often observed in the development of invasive species, which remain close to their point of establishment for decades before range expansion (Cousens and Mortimer, 1995; Kowarik, 1995; Mack et al., 2000). Whether an exotic species becomes invasive depends to a large extent on evolutionary processes (Sakai et al., 2001). In some cases, population growth may be delayed until the species evolves new adaptations through re-assortment of existing genetic variability (Brown and Marshall, 1981; Clegg and Brown, 1983). Multiple introductions may be important in providing sufficient genetic variability to allow this differentiation to occur. In other cases, inter- and intra-specific hybridization may be the nucleating event that ends the lag phase by providing new sources of adapted genes (Abbott, 1992; Ellstrand and Schierenbeck, 2000).

Hybridizations between native and introduced species have often led to the development of new taxa and have even been implicated in the evolution of a number of new invasive species. Abbott (1992) estimated that 45 % of the British flora was alien, and 7 % of those introduced species were involved in the production of hybrids now prominent in the native flora. One of the most widespread examples is Senecio vulgaris var. hibernicus, a hybrid of native S. vulgarus var. vulgaris and introduced S. squalidus, which escaped from the Oxford Botanical Garden in 1794 (Abbott et al., 1992). Highly invasive thistles from Europe have widely hybridized in Australia (O'Hanlon et al., 1999). Ellstrand and Schierenbeck (2000) found 28 examples ‘where invasiveness was preceeded by hybridisation’ and at least half of these hybrid lineages were the product of native × non-native hybridizations.

Lythrum salicaria (purple loosestrife) has followed the classic pattern of establishment and range expansion of invasive plants (Batra et al., 1986). It was probably introduced into North America from Europe in the early 1800s through ship ballast and seed sales (Mack, 1991), but it was not recognized as invasive until the 1930s, when it began to form monospecific stands in the floodplain pastures of the St Lawrence River in Quebec (Louis-Marie, 1944). Since then, it has followed a distinct pattern of invasion across the United States. Typically, it remains unobtrusive for a long period (at least 20 years) followed by a brief period (of less than 3 years) in which it becomes dominant at multiple locations across a region (Stuckey, 1980). Although purple loosestrife has an affinity for moist soil zones of wetlands, lakes and rivers in Europe and North America, its dominance in North America, where it makes up more than 50 % of the biomass of emerged vegetation in many areas, is higher than in Europe (Thompson et al., 1987).

The greater dominance of purple loosestrife in North American wetlands compared with Eurasia may be due to additional genetic differentiation after it arrived in North America, unveiling previously untapped invasive potential. The alternative hypothesis is that adaptively significant genes could have introgressed from a close North American relative. If hybridization played a role in its becoming invasive, the most likely candidate is Lythrum alatum (winged loosestrife), a widespread diploid species in North America that has considerable habitat overlap with tetraploid purple loosestrife, but is typically found in drier areas. Winged loosestrife is a shorter, less showy species than purple loosestrife (Blackwell, 1970), and grows in wet meadows as a sub-dominant (Cody, 1978).

There is evidence that the genomes of winged and purple loosestrife are compatible even though their ploidies differ, as cultivars of purple loosestrife have been generated by hybridizing the two species (Anderson and Ascher, 1993). Presumably, the diploid winged loosestrife parents produced unreduced gametes that could fertilize those of the tetraploid L. salicaria. In a screen of 100 pollen grains from 20 randomly selected winged loosestrife plants, an overall average of 0·35 % of gametes were found to be unusually large and of probable unreduced origin (Houghton-Thompson, 2000). Most of the inter-species cultivars are self-sterile, but several studies have shown that they are fully fertile when crossed with the wild species of purple loosestrife (Anderson and Ascher, 1993; Lindgren and Clay, 1993; Ottenbreit and Staniforth, 1994). Although no direct evidence of hybridization in the wild has been observed between winged and purple loosestrife, bees and butterflies are known to move between these species in sympatric populations (Levin, 1970), and morphological characters (Table 1) have been observed in purple loosestrife populations in Minnesota that are not found in Europe (Anderson and Ascher, 1993).

Table 1.

Taxonomic characteristics that separate North American winged loosestrife (Lythrum alatum) and Eurasian purple loosestrife (Lythrum salicaria) (Graham, 1975)

Lythrum alatum	Lythrum salicaria
1–2 flowers per leaf axil	4 or more flowers per leaf axil
Leaves alternate	Leaves opposite or whorled
Calyx glabrous	Calyx pubescent
Distylous	Tristylous
Plant height up to 90 cm	Plant height at least 120 cm
Leaves oblong-ovate to linear-lanceolate	Leaves lanceolate

This study was designed to determine whether North American purple loosestrife has hybridized with winged loosestrife in North America and has acquired genes via introgression. To do this, patterns of AFLP and morphological variation were examined in European and North American populations of the two species. It was found that a limited amount of introgression may have occurred between L. alatum and L. salicaria.

MATERIALS AND METHODS

Morphological patterns of variability

Six diagnostic morphological traits (Table 1) were surveyed in 30 populations of Lythrum across eastern North America (Table 2). Twenty-five to 50 plants were sampled in transects through the widest diameter of each population area. Plants of L. salicaria were sampled at a distance of 10 m from each other, whereas the smaller L. alatum clones were sampled at a minimum 1 m apart. In most cases, individual clones were well separated by open water with no obvious rhizomatous connections. The Michigan populations were surveyed in August 1997, the Ohio populations in July 1998, and the Massachusetts and Wisconsin populations in July 1999. On all these dates, the populations were in full bloom. The average number of flowers per leaf axil and mean style length were calculated on one random shoot of each genotype, and the placement of leaves along that shoot was noted as either alternate, opposite or whorled. Another random shoot was used to rate the calyxes as either primarily glabrous or pubescent. The height of the tallest shoot of each clone was also measured, along with the leaf length and width of 2–3 randomly selected leaves on each plant.

Table 2.

Location of study sites in North America and the species present

State	Site	Species	Abbreviation	Latitude	Longitude
Massachusetts	Field Farm	L. salicaria	FFA	42°42′43″N	73°12′15″W
	Sheffield	L. alatum	SHE	42°06′37″N	73°21′20″W
	West Pittsfield	L. salicaria	WPI	42°25′51″N	73°18′37″W
Michigan	Crow Island St. Game Area	L. salicaria	CIA	43°28′11″N	83°54′14″W
	Harsen's Island	L. alatum	HIA	42°35′22″N	82°35′19″W
	Harsen's Island	L. salicaria	HIS	42°35′22″N	82°35′19″W
	Lake Lansing	L. salicaria	LLA	42°44′59″N	84°24′02″W
	Quanicassee Wildlife Area A	L. salicaria	QWA	43°35′00″N	83°40′51″W
	Quanicassee Wildlife Area B	L. salicaria	QWB	43°35′00″N	83°40′51″W
	Rose Island Railroad	L. alatum	RIR	43°46′58″N	83°25′53″W
	Sheep Farm	L. alatum	SFA	43°39′13″N	83°27′58″W
	Sheep Farm	L. salicaria	SFS	43°39′13″N	83°27′58″W
	Shiawassee R. St. Game Area	L. salicaria	SRA	43°23′13″N	83°57′58″W
	Wildfowl Bay	L. alatum	WFB	43°53′00″N	83°22′00″W
Ohio	Kildeer	Mixed	KIW	41°02′39″N	83°39′00″W
	Kitty Todd A	L. alatum	KTA	41°34′46″N	83°37′02″W
	Kitty Todd B	L. alatum	KTB	41°34′46″N	83°37′02″W
	Ottawa Natl Wildlife Refuge	Mixed	ONW	41°36′56″N	83°12′58″W
Wisconsin	Bark River	L. salicaria	BRI	43°04′50″N	88°15′40″W
	Duck Creek	L. salicaria	DCR	44°33′43″N	88°04′09″W
	Herbarium Preserve	L. alatum	HPR	43°04′53″N	88°54′42″W
	Janesville	L. alatum	JAN	42°40′58″N	89°01′07″W
	Nature Conservancy	L. alatum	NCO	42°30′44″N	87°48′33″W
	Senior Citizen Center	L. salicaria	SCC	42°54′38″N	87°51′38″W
	Tichigan Lake	Mixed	TLW	42°49′44″N	88°11′51″W
	Wee Know School	L. salicaria	WKS	43°06′18″N	88°20′31″W

Seed was collected from four Michigan populations of L. salicaria and grown in a common greenhouse in late September 2000. Two of these populations were allopatric and contained no L. alatum (LLA and QWB), whereas the other two populations were sympatric and contained both species (HIS and SFA). The sympatric populations of L. salicaria had been shown in our field screen to have unusually short plant heights and leaf lengths, whereas the allopatric populations were of more typical size for the species.

Seed of each population was collected from ten flowers on a randomly selected shoot from 20 clones spanning the width of each population area. The seeds were spread onto moistened soil in a covered tray and allowed to germinate under artificial light. After germination, seedlings were transplanted to 10-cm pots and grown in a single greenhouse at Michigan State University, East Lansing, under 12-h day lengths. Ten plants from each population were arranged on three benches in a completely randomized block design and allowed to grow until flowering (about 8 weeks). At anthesis, the height of the tallest shoot in each plant was measured. The number of flowers was counted for 10–20 random axils and the most common number was recorded. The lengths of the styles on a random shoot were rated as long, mid or short. Leaf length and width were also measured on a randomly selected mature leaf from each plant.

Molecular patterns of variability

Patterns of AFLP variation were evaluated using five primer pairs (M-CAG/E-ACT, M-CAG/E-AGG, M-CAG/E-AAG, M-CAG/E-ACG, M-CAG/E-ACC) in a ‘global screen’ of the same North American populations of L. salicaria and L. alatum described above, as well as L. salicaria from 11 European populations located in Germany (FEH—Fehmarn, KIR—Kirchwerder, LEM—Lembruch, NIE—Niederwetter, OFF—Offenburg and WET—Wetter), England (FAR—Farnham), Ireland (GG—Glengarriff and DER—Derrynane), Austria (ILL—Illmitz) and Finland (KUO—Kuopio), and the following L. salicaria cultivars: HAP—Happy, MG—Morden's Gleam, MP—Morden's Pink, PS—Purple Spires, ROB—Robert and RS—Roseum Superbaum. Morden's Glean and Morden's Rose were purportedly developed by the hybridization of Morton Pink with select forms of native L. alatum (Harp, 1957). The European samples were provided by Dr Bernd Blossey at Cornell University who grew them from pooled, open-pollinated seed in a field at Ithaca, New York. The cultivars were obtained commercially and grown in a greenhouse at Michigan State University. An additional 18 primer pairs were also evaluated in the eight populations of Michigan Lythrum grown in the common garden experiment (M-CAC/E-AAG, M-CAC/E-AAC, M-CAA/E-AAG, M-CAA/E-AAC, M-CAC/E-AGG, M-CAT/E-AAG, M-CTC/E-ACT, M-CTC/E-ACC, M-CTC/E-AGC, M-CTG/E-ACC, M-CTG/E-AGC, M-CTT/E-AGC, M-CTA/E-AGG, M-CTA/E-ACT, M-CAT/E-AAC, M-CAT/E-AGG, M-CTA/E-AAG and M-CTA/E-AAC).

Approximately ten young green leaves were collected from each clone in early to late June before full growth and before flowering had occurred. The leaves were placed in a ziploc bag with enough silica gel to cover the leaf tissue completely. Bags were then labelled as to population and individual, sealed and stored at room temperature for several months until the DNA was extracted. DNA was extracted according to the method of Doyle and Doyle (1990). AFLP analyses were performed following procedures similar to that of Vos et al. (1995) using Gibco BRL-Life Technology (Rockville, MD) reagents for all digestions, ligations, and other AFLP preparations and experiments. A 5 % acrylamide gel was run at 85 W with a variable voltage (1500–1700 V) for approximately 3 h. The amplification products were visualized by autoradiography and scored manually. The individual AFLP fragments (characters) were named based on their primer pair and molecular weight.

Statistical analysis

Model variance components were estimated using the SAS procedures (Cary, NC). Variance was partitioned into location (state), species, population and error. The percentage data were log transformed. To evaluate patterns of AFLP variation, the data set was put into a Nexus file format for analysis using PAUP* 4.0b10 (Swofford, 2002) and MacClade 4 (Maddison and Maddison, 2000). Two sets of AFLP data were evaluated. The first represented the global screen and consisted of 338 accessions and 64 binary AFLP characters from native North American L. alatum and L. salicaria, European L. salicaria and cultivars of L. salicaria. The second data set consisted of 279 AFLP markers analysed in five individuals from each of four Michigan L. alatum and L. salicaria populations.

Data from the 338 accession data set were analysed by Neighbor-joining (NJ) (Saitou and Nei, 1987) in PAUP*. From the initial NJ dendrogram, redundant taxa were identified and removed from the data set for subsequent analyses. An NJ tree was produced from the resulting data set of 71 non-redundant taxa and displayed as an unrooted phylogram.

Evidence of gene flow from L. alatum to L. salicaria was investigated by examining the evolution of each of the 64 characters in the data set using the ‘Trace Character’ function of MacClade 4 to trace each character onto the NJ tree of the 71 accession global data set. AFLP characters were classified into one of four categories based on their evolutionary pattern in the NJ tree: (1) present in L. alatum but absent in L. salicaria, (2) present in L. alatum and some North American L. salicaria but absent in European L. salicaria, (3) present in L. alatum and both European and North American L. salicaria, and (4) absent in L. alatum but present in L. salicaria. In this analysis, only characters in category 2 were considered as unambiguous evidence of hybridization and introgression.

Evidence of gene flow was also looked for in the Michigan data set of 279 AFLP characters in sympatric and allopatric populations. Characters were again placed into four categories based on the patterns observed: (1) present in L. alatum but absent in L. salicaria, (2) present in some L. alatum and some L. salicaria, (3) present in all L. alatum and L. salicaria, and (4) absent in L. alatum but present in L. salicaria. Characters in category 2 that were present in L. alatum and only sympatric populations of L. salicaria were considered to be evidence of introgression. However, these characters were also checked in the global screen for their presence in European L. salicaria; if they were present in Europe, they were deemed as possibly introduced.

RESULTS

Morphological data

All the L. salicaria populations had individuals which carried the L. alatum traits of fewer than four flowers per axil and alternate leaf placement (Table 3), although the calyx and style traits were always identical for L. salicaria. The field populations of L. salicaria were generally taller and had longer, narrower leaves than those of L. alatum; however, four sympatric L. salicaria populations had mean heights significantly closer to typical L. alatum than typical L. salicaria (HIS, ONP, KIL and SFA), and leaf length in two of the sympatric L. salicaria populations (HIS and SFA) were much closer to the typical mean leaf length of L. alatum. In the greenhouse comparisons, HIS and SFA maintained their significantly shorter heights and smaller leaf sizes than the other two L. salicaria populations (Fig. 1). However, only a few L. salicaria plants displayed the other two L. alatum traits found in the field (data not shown).

Table 3.

Location and mean plant height, leaf length and leaf ratio of populations of Lythrum alatum and L. salicaria sampled in North America

State	Species	Population	Height (cm)	Leaf length (mm)	Leaf ratio	Percent with <4 flowers	% alternate leaves
Massachusetts	L. salicaria	FFA	121·8			17	17
	L. alatum	SHE	51·3			100	100
		WPI	128·2			7	36
Michigan	L. alatum	ASP	68·1	11·4	3·97	100	100
		HIA	65·0	8·0	3·40	100	100
		RIR	67·0	12·6	3·93	100	100
		SFA	60·0	14·0	3·84	100	100
		WFB	55·6	11·0	3·88	100	100
	L. salicaria	CIA	146·4	52·6	4·49	20	17
		HIS	107·5	26·4	3·26	27	33
		LLA	173·6	48·0	5·47	22	18
		QWA	146·3	75·7	4·85	26	16
		QWB	126·3	62·1	5·43	30	12
		SFS	96·7	41·4	4·71	9	7
		SRA	183·0	78·9	5·59	30	8
Ohio	L. alatum	KIW	53·8			100	100
		KTA	50·4			100	100
		KTB	53·3			100	100
		ONW	64·9			100	100
	L. salicaria	KIL	74·8		3·59	0	0
		ONP	109·5			16	14
Wisconsin	L. alatum	JAN	55·3	22·4	3·51	100	100
		HPR	53·3	15·1	5·31	100	100
		NCO	53·8	21·0	3·77	100	100
	L. salicaria	BRI	137·0	54·8	5·06	10	13
		DCR	137·5	69·3	5·49	8	15
		SCC	129·9	68·7	5·63	0	5
		TLP	131·4	67·2		26	13
		WKS	136·7	58·2	5·49	8	16
	Significance (P)	Location	0·001	0·410	0·461	0·001	0·001
		Species	0·001	0·001	0·001	0·001	0·001
		Population (Species)	0·001	0·001	0·001	–	–

Fig. 1.

Mean plant height and leaf width/length ratio in four populations of Lythrum salicaria grown from seed in a common greenhouse at Michigan State University, East Lansing. Two of the populations were sympatric with L. alatum [Harsens's Island (HIS) and Sheep Farm B (SFB)], and two were allopatric [Lake Lansing (LLA) and Quanicassee B (QWB)]. Different letters at the top of each column indicate significant differences at P < 0·05 (LSD test).

Molecular data

The unrooted NJ dendrogram for the 71 accessions in the global data set is shown in Fig. 2. L. salicaria and L. alatum form distinct and well-separated clusters. Within the L. salicaria cluster, the introduced North American L. salicaria form a distinct terminal cluster that appears to be derived from within the larger cluster containing the cultivars and the European L. salicaria.

Fig. 2.

The unrooted neighbour-joining dendogram for 71 accessions in a global survey of Lythrum salicaria and L. alatum in North America and Europe. Variation in 64 AFLPs was analysed. Numbers along branches indicate bootstrap support for that branch (1000 replicates). Note that L. salicaria and L. alatum form well-separated clusters, and the North American L. salicaria are distinct from the cultivars and European L. salicaria.

In the individual examination of each of the 64 AFLP characters in the global data, 16 AFLP markers were found in only L. salicaria. These characters could not have been acquired by L. salicaria through introgression with L. alatum because they are not in L. alatum. Twenty-seven characters were shared by L. alatum and L. salicaria in a broad sense, with bands both in L. alatum and in either European populations or both the European and the native North American populations. Some of these characters could have been transferred to North American L. salicaria via introgression with L. alatum, but because they were found in European L. salicaria, it is also possible that they were introduced from Europe.

Ten AFLP characters were identified in L. alatum but not in European L. salicaria. Of these, eight were found only in L. alatum in North America and as a result had not been transferred to L. salicaria via hybridization. Two characters (M-CAG/E-AAG: 325 bp and M-CAG/E-AGG: 350 bp) were present in L. alatum and a few plants of L. salicaria in Massachusetts and Wisconsin. This is the expected pattern for a locus at which introgression occurred from L. alatum to L. salicaria. One other character (M-CAG/E-AAG: 650 bp) was found in L. alatum and two of the L. salicaria cultivars (Morden's Gleam and Happy). This suggests that breeders have hybridized the two species in their cultivar development efforts, although the reported hybrid Morton Rose did not carry this fragment. In the study of sympatric and allopatric populations in Michigan, these three characters were found in all L. alatum and at least some individuals of all L. salicaria populations (Table 4).

Table 4.

AFLP characters shared by Michigan populations of L. alatum and L. salicaria

		Presence in Michigan populations
		Allopatric			Sympatric
Primer pair	Fragment size (bp)	L. alatum	L. salicaria	L. alatum	L. salicaria
M-CAA/E-AAG	475	Polymorphic	Fixed	Fixed	Fixed
	472	Polymorphic	Polymorphic	Polymorphic	Polymorphic
	425	Polymorphic	Fixed	Fixed	Fixed
M-CTC/E-ACC	425	Absent	Fixed	Polymorphic	Fixed
M-CAG/E-ACT	350	Polymorphic	Fixed	Polymorphic	Fixed
M-CAG/E-AGG	500	Fixed	Polymorphic	Fixed	Fixed
	480a	Polymorphic	Fixed	Absent	Fixed
	480b	Polymorphic	Fixed	Absent	Fixed
	480d	Polymorphic	Fixed	Absent	Fixed
	450a	Polymorphic	Fixed	Absent	Fixed
	450b	Polymorphic	Fixed	Absent	Fixed
	400a	Polymorphic	Fixed	Fixed	Fixed
	400b	Polymorphic	Fixed	Polymorphic	Fixed
	325a	Polymorphic	Fixed	Polymorphic	Fixed
	300ab	Fixed	Absent	Fixed	Polymorphic
M-CAG/E-ACG	490	Fixed	Polymorphic	Polymorphic	Fixed
	350a	Absent	Fixed	Rare	Fixed
	185a	Fixed	Polymorphic	Fixed	Polymorphic
M-CAG/E-ACC	330	Fixed	Polymorphic	Polymorphic	Fixed

Lythrum salicaria and L. alatum also formed distinct and well-separated clusters in the unrooted NJ dendogram using the Michigan AFLP data set (Fig. 3). Two hundred and seventy-nine characters could be assorted into four categories based on their evolutionary pattern in the NJ tree. Sixty-six characters were fixed in both species, so could not be used to evaluate whether introgression had occurred. Another 71 characters were found in only L. salicaria, and as a result could not have been derived from introgression. One hundred and twenty-three characters were found in L. alatum but not in L. salicaria, which also does not support introgression.

Fig. 3.

The unrooted neighbour-joining dendogram for 40 accessions in a survey of four sympatric and allopatric populations of Lythrum salicaria and L. alatum in Michigan. Variation in 279 AFLPs was evaluated. Numbers along branches indicate bootstrap support for that branch (1000 replicates). Note that L. salicaria (below) and L. alatum (above) form distinct clusters.

Nineteen AFLP characters were found in Michigan populations of both L. salicaria and L. alatum (Fig. 4A); however, only one character (M-CAG/AGG: 300 bp) was found to be in L. alatum and only sympatric populations of L. salicaria (Fig. 4B). This pattern supports introgression, but the character was also found in European L. salicaria and therefore could have been introduced.

Fig. 4.

Individual patterns of AFLP variation in the Michigan survey of allopatric and sympatric populations of North American Lythrum salicaria and L. alatum. (A) M-CAA/E-AAG: 472 bp, which is found in L. alatum and both allopatric and sympatric L. salicaria, and (B) M-CAG/E-AGG: 300 bp, which is found in L. alatum and only sympatric L. salicaria.

DISCUSSION

The characters evaluated here are probably good representatives of the various taxa examined, as they clearly distinguished them in the phylogenetic analysis. North American L. salicaria formed its own distinct cluster within the larger L. salicaria group, with the European and cultivar samples clustering separately from the North American group. Likewise, L. salicaria and L. alatum were well differentiated within the sympatric populations. No L. salicaria from Asia or North Africa were examined.

The molecular data indicate that introgression may have occurred between the two North American Lythrum species, although the number of genes incorporated into the genome of L. salicaria appears to be limited. Of 115 diagnostic characters identified in L. alatum, only two were clearly shown to have introgressed into L. salicaria. In the global screen used, there were ten fragments that were found in L. alatum and not in European L. salicaria, but only two of them were observed in North American L. salicaria. In the Michigan screen of sympatric and allopatric populations, there were 123 characters found to be unique to L. alatum and 19 that were shared by some individuals of the two species. Only one of these was restricted to sympatric populations of L. salicaria and thus could have introgressed; however, it was also found in European populations of L. salicaria and as a result may have been introduced. It is possible that all the Michigan L. salicaria were fixed for the other 19 traits because of an ancient introgression resulting in a single Michigan founder, but this is deemed unlikely because L. salicaria carries so much genetic variability in Michigan.

All of the cultivars, including the apparent introgressants Morden's Gleam and Happy, grouped more closely with European L. salicaria than North American L. salicaria or L. alatum. This indicates that these cultivars have not extensively hybridized with North American L. salicaria, and the integrity of the cultivars in nursery stock remains. One character was found in North American L. alatum and in two of the cultivars, supporting a hybrid ancestry for them. Again, only a few L. alatum genes must have been retained in the L. salicaria cultivar background, as only this one marker was consistent with introgression, and no visual morphological differences were observed between the cultivars and wild purple loosestrife populations.

The morphological data also supported introgression between L. alatum and L. salicaria, as most of the North American L. salicaria populations had individuals that carried the L. alatum traits alternate leaf placement and 1–2 flowers per leaf axil. However, it is possible that the genes responsible were introduced from an unsampled part of the range of L. salicaria. There also appears to be a strong environmental component to these characteristics, as most of the L. salicaria plants grown in the common greenhouse had leaf placement and flower numbers typical of L. salicaria, even though plants of both typical L. salicaria and typical L. alatum were collected in the field for use in this study.

Two L. salicaria populations (Harsen's Island and Sheep Farm) were intermediate in height and leaf ratio between the typical L. salicaria and L. alatum populations in both the native field and the common greenhouse. The morphological intermediacy of these populations is consistent with their being hybrid swarms, but the molecular evidence does not support this conclusion, as the sympatric populations of L. alatum and L. salicaria did not carry any more of the unique L. alatum characters than did the allopatric populations. Harsen's Island and Sheep Farm appeared to be more xeric than most other L. salicaria habitats, so it is possible that L. salicaria is evolving a more xeric ecotype, by re-assortment and selection of genes already available in its genome.

It appears that L. alatum contributed some unique genes to L. salicaria through introgression, and this could have played a role in L. salicaria becoming more invasive in North America than in Eurasia. However, the number of L. alatum genes retained in L. salicaria is limited, suggesting that much of the adaptive switch in North American L. salicaria more likely came from the re-assortment and selection of genes within its own genome. Crossing among genotypes from multiple introductions may have played an important role in the invasion of L. salicaria, through segregation of previously unassociated genes. The autopolyploid nature of L. salicaria may have facilitated its adaptation to new habitats in North America, as the increased levels of heterozygosity generally observed in polyploids may have pre-adapted L. salicaria with sufficient plasticity to fill many habitats, and these high levels of allelic diversity were available for re-assortment after hybridization via tetrasomic inheritance.