Biological Invasions

, Volume 14, Issue 3, pp 693–699

Levels of novel hybridization in the saltcedar invasion compared over seven decades


    • USDA Agricultural Research Service
  • Adam S. Birken
    • Department of Forest, Rangeland, and Watershed StewardshipColorado State University
    • Graduate Degree Program in EcologyColorado State University
    • USDA Natural Resources Conservation Service
  • David J. Cooper
    • Department of Forest, Rangeland, and Watershed StewardshipColorado State University
    • Graduate Degree Program in EcologyColorado State University
Original Paper

DOI: 10.1007/s10530-011-0110-z

Cite this article as:
Gaskin, J.F., Birken, A.S. & Cooper, D.J. Biol Invasions (2012) 14: 693. doi:10.1007/s10530-011-0110-z


Hybridization is proposed as one process that can enhance a plant species’ invasive ability. We quantified the levels of hybridization of 180 saltcedar plants (Tamarix spp.) of varying ages that span the history of an invasion along the Green River, Utah, USA. Plants ranging in establishment dates from 1930s to 2004 were analyzed using Amplified Fragment Length Polymorphisms. All plants sampled, even those established before the Green River saltcedars were numerous, were assigned as hybrids, not as parental types that are still found in more extreme southern and northern latitudes in the USA. Our collections either did not capture the earliest parental types, parental types have failed to persist, or the first introductions to the Green River were already hybrids. In any case, it appears that hybrids have been a dominant part of this local invasion history, from establishment through invasion spread stages.




Hybrids and novel genotypes are involved in many plant invasions (e.g., Ainouche et al. 2009; Sloop et al. 2009; Moody and Les 2002; Blair and Hufbauer 2009; Williams et al. 2005; Zalapa et al. 2010; also see review by Schierenbeck and Ellstrand 2009). Creation of novel hybrids may make possible increased genetic variation (Anderson 1949), new interactions between genes (Templeton 1981), and the transfer of favorable genetic traits such as cold tolerance or disease/herbivore resistance (Milne and Abbott 2000; Abbott et al. 2003; Whitney et al. 2006; Rieseberg et al. 2007), all of which may enhance a plant’s invasive ability (Ellstrand and Schierenbeck 2000; Vila et al. 2000; Lee 2002; Sakai et al. 2001). If detected and quantified, especially over time, hybridization may help us understand the influence of introgression on population structure (e.g., Tung et al. 2008) and shifts in invasion dynamics from establishment to spread stages (Crooks 2005; Theoharides and Dukes 2007). Introgression of hybrids toward one parental type or the other, or the breakdown of heterosis (Lee 2002) can also be investigated, helping to target adapted evolutionary states for control efforts.

Detecting genetic history in an invasion is subject to the persistence of individuals over time, or well-preserved collections, both of which are rare for long time spans (e.g., Saltonstall 2002; Lelong et al. 2007). In this study, we use a genetic marker system to determine if levels of hybridization have changed during approximately 70 years of an invasion in a long-lived perennial shrub.

Saltcedar (Tamarix spp.; family Tamaricaceae) invasions in the USA predominantly contain a mix of diploid sexual parental genotypes and hybrids of Tamarix ramosissima Ledeb. and T. chinensis Lour. (Gaskin and Schaal 2002), with the former appearing to be dominant in northern latitudes and the latter in southern latitudes (Friedman et al. 2008; Gaskin and Kazmer 2009). At mid-latitudes (e.g., Utah), novel hybrids of these two parental Asian species appear to be most common (Gaskin and Kazmer 2009). Molecular data (Simple Sequence Repeats; SSRs) suggest that hybrid and parental types are diploid (Friedman et al. 2008), and both parental and hybrid genotypes produce high percentages of fertile seeds (Gaskin and Kazmer, unpublished). Together, these taxa form one of North America’s more infamous plant invasions (Cleverly et al. 1997; Di Tomaso 1998; Zavaleta 2000; Bailey et al. 2001; Friedman et al. 2005).

The history of the saltcedar invasion has been partially documented in the inter-mountain western USA. Escape of saltcedars from cultivation was first reported in the 1870s (Brotherson and Field 1987) and continued to occur slowly until the 1920s. Christensen (1962) performed an in-depth review of the timing of saltcedar naturalization in Utah, finding that rapid spread occurred in the state from 1925 to 1960. On the Green River in Utah saltcedar was noted, but not as an important species, as of 1931–1935 (Graham 1937), but was by 1957 considered abundant along the southern portion of this river (Hayward et al. 1958). Saltcedar recruitment along the lower Green River appears to have been caused by large annual floods that were followed by years with lower peak flows and may have greatly influenced floodplain development and riparian vegetation composition since the early twentieth century (Birken and Cooper 2006).

Materials and methods

Plant collection

Saltcedar plants from Asia (43 T. chinensis from China, 58 T. ramosissima from Asia: China, Azerbaijan, Republic of Georgia, Turkmenistan, Iran, and Kazakstan), were used as known parental types. These were identified morphologically and from DNA sequence data (phosphoenolpyruvate carboxylase) by Gaskin and Schaal (2002). Silica dried leaf material from 30 saltcedars from 6 age classes (180 plants total) was collected in 2004 in Utah, USA at six locations along a 45 air-mile section of the Green River (Table 1).
Table 1

Age classes and genetic diversity of saltcedar collections on the Green River in Utah, USA

Age class



Standard error (Hj)

No. of plants collected from each age class at each location (30 total per age class)a
















1950s–early 1960s









Late 1960s–1970s



























2003–2004 seedlings










Proportion of polymorphic loci. Values did not vary significantly between age classes in pairwise Mann–Whitney tests (all P > 0.4295, two-tailed)

Gene diversity (Hj) did not vary significantly in t tests comparisons between age classes (all P > 0.416)

aCollection locations: SCR Stone Cabin Rapids (N 39.138056, W 110.099167), BRP Butler Rapids Pool (N 39.159444, W 110.112778), STR Short Canyon Rapids (N 39.132222, W 110.108056), MBE Mineral Bottom Excavation Trench (N 38.538889, W 110.003611), MBB Mineral Bottom Boat Ramp, (N 38.525833, W 109.992778), SKG Sand Knolls Rapids (N 39.148056, W 110.112222)

Mixes of age classes were found at each location. Four of these sampling sites were located within Gray Canyon, approximately 15 river-miles north of the town of Green River, Utah, USA. The remaining two sampling sites were located within Labyrinth Canyon, approximately 50 river-miles south of the town of Green River. Age classes include plants that established in the 1930s–1940s, 1950s-early 1960s, late 1960s–1970s, 1980s, 1990s–2002 and 1st and 2nd year seedlings from 2003 to 2004. Previous dendrogeomorphic research (Birken and Cooper 2006) from these two stretches of the Green River provided spatial distribution data of the six saltcedar age classes we investigated. The six age classes also encompass the chronology of saltcedar invasion (starting with initial recruitment during the 1930s, followed by proliferation during the 1950s and 1960s, and concluding with relatively moderate to minimal expansion by the late 1960s to the present) and regulation of the Green River’s flow regime with the completion of Flaming Gorge Dam in 1963.

AFLPs (amplified fragment length polymorphisms)

Genomic DNA was extracted from approximately 20 mg of silica-dried material using a modified CTAB method (Hillis et al. 1996). The AFLP method followed Vos et al. (1995) with these modifications: restriction and ligation were performed during a single step in an 11 μL reaction containing 500 ng genomic DNA, 2 U MseI (New England Biolabs [NEB], Ipswich, MA, USA), 1 U EcoRI (NEB), 1 × T4 DNA ligase buffer (NEB), 0.45 U T4 DNA ligase (NEB), 0.05 M NaCl, 0.5 × BSA, 4.5 μM MseI adaptor, 0.45 μM EcoRI adaptor, and H2O. The restriction-ligation was incubated at room temperature overnight, then 5.5 μL of the product was diluted to 100 μL in TE (15 mM Tris and 0.1 mM EDTA). A pre-selective Polymerase Chain Reaction (PCR) was performed in a 20 μL reaction containing 4 μL of the diluted, restricted-ligated product, 1 × PCR buffer (Bioline, Taunton, MA, USA), 1.5 mM MgCl2, 0.2 mM each dNTP, 0.2 μM of each pre-selective amplification primer (MseI + C and EcoRI + A), 0.5 U Taq polymerase (Bioline) and H2O. The pre-selective PCR consisted of 20 cycles of: 30 s at 94°C, 60 s at 56°C, and 60 s at 72°C. 10 μL of the pre-selective amplification product was diluted to 200 μL in TE (15 mM Tris and 0.1 mM EDTA). The selective amplification was performed in a 20 μL reaction containing 3 μL of the diluted pre-selective amplification product, 1 × PCR buffer, 1.5 mM MgCl2, 0.2 mM each dNTP, 0.1 μM MseI selective primer, 0.05 μM EcoRI selective primer dye-tagged with 6-FAM (Integrated DNA Technologies, Coralville, IA, USA), 0.5 U of Taq polymerase and H2O. The selective PCR was 120 s at 94°C; 10 cycles of: 20 s at 94°C, 30 s at 66°C (decreasing by 1°C each cycle), 120 s of 72°C; 25 cycles of: 20 s at 94°C, 30 s at 56°C, 120 s at 72°C. 0.5 μL of each selective PCR product was combined with 0.25 μL of 600 base pair (bp) size standard and 9.25 μL of de-ionized formamide and loaded into an Applied Biosystems (Foster City, CA, USA) 3130 Genetic Analyzer. Loci were initially scored by the fragment analyzer software GeneMapper v 4.0 (Applied Biosystems). These bins were then manually screened, making this a semi-automatic scoring method, as suggested by Papa et al. (2005). All selective primer combinations of MseI + CAA, CAC, CAT, CTA, or CTA and EcoRI + AAG, ACC, or ACT were pre-screened for 8 samples in Gaskin and Kazmer (2009), and the two most polymorphic primer pairs were chosen (MseI + CAT/EcoRI + ACC and MseI + CTA/EcoRI + ACC).

Analysis of AFLPs

Clustering and assignment tests were performed using the software Structure v 2.3.3 (Pritchard et al. 2000; Falush et al. 2003; Falush et al. 2007). To determine the number of clusters (K), or species, represented in the native Asian plants, AFLP results were diploidized (Falush et al. 2007), no population information was included, admixture was assumed as possible, allele frequencies were considered to be independent in each population, and a 10,000 run burn-in (α stabilized at approximately 1,000 runs) and 10,000 run length were used. We tested for number of clusters (K = 1–8) with 10 repetitions for each K. Selection of K from this output data was done by two methods: (1) using suggestions in the software documentation (which looks at changes in slope of plotted values of log P(X/K)) and (2) with a more formal criterion (ΔK) suggested by Evanno et al. (2005). Average estimated admixture coefficient (Q, or assignment value) of each plant to T. chinensis and T. ramosissima was determined using Structure v 2.3.3. Prior population information was included for the native Asian plants (but not the USA plants), admixture was assumed to be possible, allele frequency was considered to be independent in each population and 50,000 burn-in and 50,000 run lengths were used. The software also calculated 95% probability intervals around the average assignment value of each plant using the 50,000 runs performed after burn-in. Average assignment values for each plant were plotted using Distruct v1.1 (Rosenberg 2004). Changes in levels of hybridization over time were determined with Kolmogorov–Smirnov (K–S test) pairwise comparisons of average assignment values of entire age classes (D statistic and P value of significance of <0.05).

Proportion of Polymorphic loci (PLP) at the 5% level (i.e. with allele frequencies between 0.05 and 0.95), gene diversity (Hj), and estimated allele frequencies at all 117 AFLP loci were calculated using the program AFLP-SURV (Vekemans et al. 2002). For estimation of allele frequencies, we used the option: “Bayesian method with non-uniform prior distribution of allele frequencies” as in Zhivotovsky (1999), and assumed Hardy–Weinberg equilibrium for this outcrossing species. Differences in PLP among age classes (determined to be non-normal data) were investigated using the Mann–Whitney test (non-parametric), and differences in gene diversity (Hj) among age classes (determined to be normal data) were investigated with t tests. To determine if individual loci were trending over time toward fixation (estimated allele frequencies of 1 or 0) or polymorphism (estimated allele frequency of 0.5), we used the absolute value that the estimated allele frequency varied from 0.5 and compared this value between the oldest and youngest age classes.


Initial output from the fragment analyzer software listed 318 and 276 fragments for MseI + CAT/EcoRI + ACC and MseI + CTA/EcoRI + ACC, respectively. After manual screening we scored 43 and 74 (total 117) unambiguous, polymorphic loci. Clustering analyses indicated that the native T. chinensis and T. ramosissima formed K = 2 clusters (Fig. 1). The ΔK method of Evanno et al. (2005) cannot be calculated for K = 1; however, the method used in the Structure software indicates that K = 1 is not an option for the dataset. The average assignment values of each individual plant to the two parental species are shown in Fig. 2, assuming two (K = 2) ancestral source populations in each case. The range of average assignment values for individual USA plants was 0.262–0.717 for assignment to T. chinensis (and 1 minus those values for T. ramosissima). 95% probability intervals around the average assignment value of each plant ranged from 0.129 to 0.871. Hybridization is indicated when an average assignment value of less than 0.9 toward both parental species is reported for an individual (Pritchard et al. 2000; Blair and Hufbauer 2009), and all of our USA plants scored as hybrids (highest average assignment over 50,000 runs for a USA individual = 0.717), even when considering the extremes of their 95% probability intervals (highest for any USA plant = 0.871).
Fig. 1

Graph of average mean posterior probabilities (black line) of 101 native Tamarix chinensis and T. ramosissima data for 10 runs of K = 1–8 (number of clusters) as determined by Bayesian clustering software Structure v 2.2 (Pritchard et al. 2000; Falush et al. 2003, 2007) and ΔK values (grey line) for those same 10 runs as determined by method of Evanno et al. (2005)
Fig. 2

Assignment values for 43 native Tamarix chinensis 58 native T. ramosissima from Asia, and 180 invasive Tamarix spp. from the Green River of Utah, USA as determined by Bayesian clustering software Structure v 2.2 (Pritchard et al. 2000; Falush et al. 2003, 2007)

Even though all saltcedars sampled on the Green River were assigned as hybrids, there were some differences in assignment to parental lineages over time. Plants in the oldest age class were assigned more to T. chinensis than all younger age classes (K–S test: D = 0.37–0.47; P = 0.002–0.026, data not shown). All other comparisons between younger age classes were non-significant (K–S test: D = 0.10–0.27; P = 0.200–0.997).

The proportion of polymorphic loci (PLP) did not vary significantly between any age classes in Mann–Whitney tests (all P > 0.4295, two-tailed). Gene diversity (Hj) also did not significantly vary in t tests comparisons between any age classes (all P > 0.416) (Table 1). Of the 117 loci, 17.1% trended toward fixation by changing estimated allele frequencies more than 5%, 56.4% did not significantly change estimated allele frequencies, and 26.5% trended toward polymorphism (data not shown).


Our original intent was to attempt to document a change from parental to hybrid types, perhaps correlating with a break in lag time of an invasion. We were surprised to find no parental types (non-hybrids) in 180 plants of the Green River, even in the oldest plants we sampled (established in 1930s). Parental types still exist in more northern and southern latitudes of the USA invasion (Gaskin and Kazmer 2009). The invasion in some southern latitudes is 15–37 years older than that of the Green River, and the invasion in the north is relatively young, with establishment on major rivers in Montana not being recorded until the 1960s (Robinson 1965). Our collections either did not capture existing parental types, parental types that colonized the region did not persist, or the first invaders in the Green River were already hybridized. While it is still possible that the increased spread of saltcedar after the apparent lag time (1930s–1950s) was driven by the formation of hybrids, the continuous presence of hybrids throughout the lag period suggests that other mechanisms may be responsible, such as increased propagule pressure, human disturbance (Everitt 1998), adaptation of existing hybrids, or environmental change such as peak flow events followed by years with smaller peak flows (Birken and Cooper 2006), etc. Our collections represent the oldest living saltcedars on the Green River (Cooper et al. 2003, Birken and Cooper 2006) and the lack of significant change in assignment to parental species since the 1950s indicates stabilization in genetic identity of this invasion, though future biotic and abiotic changes could certainly require further adaptation, perhaps detected as varying levels of introgression toward one or the other parental type. It is possible that evolution of highly fit hybrid genotypes could still be a relevant factor in the success of saltcedar in the Green River, as hybrid vigor may be determined disproportionately by a small number of key loci. We did find that a higher percentage of loci (26% vs. 17%) were trending away from fixation over time in the invasion, and the majority of loci were not changing frequency significantly. However, AFLPs, which are mostly selectively neutral, may not be well-suited for detecting the effects of natural selection, and further studies would be needed to correlate loci with adaptive phenotypic characters. The existence of >70 year old hybrids on the Green River suggests that for the saltcedar invasion in general, hybridization may have occurred early in its introduction into the western USA (>1850s; Horton 1964, Brotherson and Winkle 1986), or perhaps even earlier when it was first introduced on the east coast of the USA in the early 1800s as a horticultural plant, though T. chinensis is not listed among species available for sale in early catalogs (Horton 1964).

Saltcedar is a current target of biological control, and at this point it is not clear if the hybrids and parental species differ in their levels of resistance or tolerance to the released beetle agents (Diorhabda spp.). Information regarding the dominant and stable presence of hybrids in this region, as compared to lower levels of hybrids in more extreme USA latitudes, may be of use to help to explain establishment and efficacy of this biological control agent, or eventually guide the use of biological control options.


We thank GeorgeYatskievych, and others for donating Asian plant samples, and Kim Mann and Jeannie Lassey for AFLP analysis in the laboratory. This research was made possible through funding from the National Geographic Society Committee for Research and Exploration Grant # 6663-99, the Bureau of Land Management (Utah, Montana, South and North Dakota) and U.S. Department of Agriculture National Research Initiative Competitive Grants Program, Cooperative State Research, Education, and Extension Service Grant 2000-00836. Funding for the collection of Green River plant samples was provided by the U.S. Bureau of Reclamation, Upper Colorado River Regional Office.

Copyright information

© Springer Science+Business Media B.V.(outside the USA) 2011