I..-I .... !'I.~I Snm..r • ..J M .. "'Wmml 2008 1; 133-141

Invasion Dynamics and Genotypic Diversity of Cogongrass (lmperata cylindrica) at the Point of Introduction in the Southeastern

United States Ludovic J. A. Capo-chichi, Wilson H. Faircloth, A. G. Williamson, Michael G. Patterson, James H. M iller, and

Edzard van Santcn'

rJine sites of (:ogongrass were included in a study of genotypic divers ity and spread dynamics at Ihe point of introduction and its adjacent areas in the southern United States. Clones evaluated with tWO primer pairs yidded a IOtal of 137 amplified fr:1gmem length polymorphism (AF LP) loci of whi(:h 102 (74.4%) were polymorphic.

Genetic diversity was measured as the percentage of polymorphic, Shannon's information index, Nci's gene diversity, and panmictic heteroz),gosity. Nei's gene diversity (Hs) across all nine si res was estimated to be 0. 11 and

within site gene diversity ranged from 0.06 to 0.16. Bayesian esti mate of gene diversity and Shannon's information index were higher (0.1 7 and 0.1 7, respectively) . The samples from the point of introduClion (Pi) had the lowest genetic diversity for all types of estimates. Within site variance accounted for 56% of the total variation and among

site va riance 44% (P < 0.05). Differentiatio n among Sifes w:t$ asSe5SCt1 using FS 1. The grearest difference was found berween the I' i and the others. No relationship was found berween genetic and geographic distances. Principal

component analysis as well as cluster analysis separated individuals into three main clUSlers. The Pi fo rmed a separate subcluster. Gene flow (Nm), inferred from lP-statist ics describing the genetic differem iation berween pairs of sites

r:anged from 0.6 10 555 . The lack of significant relationship berween gene flow and geographic distance as well as

genetic and gt.'Ographic distances suggests that the invasion dynamics of cogongrass into the southern United States is primarily through anthropogen ic activi ries and to the lesser extent through natural forces.

NomenclalUre: cogongrass, Imperar'l cylindrica (L.) P. lkauv, I3r.lzil i:ul s:uintail, Impaara hrazifimsis Trin.; satint:lil , Impernta brevifolia Vasey; satsuma orange, Citrus reliculma L.; C rAB, cetyltrimcthyl ammonium bromide, DNA, deoxyribonucleic acid.

Key words: Invasive species, mQlecular makers, gene diversity, plant introduction.

DO[: IO.16 14/ 1I'SM-07-007. 1 ' Fillt author: POSI Doctoral Research Scientist, Dcpartmcm of

Crop and Soil Science;, University of Georgia, 1109 Experiment SI~t, Griffin, GA 30223- 1297; sc<ond author: Research Agrono­mist, National Peanut Research Lab, U.S. Department of Agriculture. P.O. Box 509, E. Dawson, GA, 39842, thi rd, fourt h. and six th authors: AssisGlnt, Extension Weal Scientisl and Professor, anti Professor, Department of Agronomy and Soils, Auburn Uni"ersity, 202 Funchess Hall, Auburn, AL 36849-5412; fiflh author: Research Ecologisl, U.s. Department of Agriculture Fordt Service, SOUlhern Research Station and Affiliate Profcsror, School of Foreslry and Wildlife Sciences, Auburn University, 520 Devall Drive, Auburn, AL 36849. Corresponding author's E-mail: capochi@uga.edu

The introduction of exotic species is one o f the most serious threats to native ecosystems worldwide (Simberloff and Von Holle 1999), Only a fraction of [he species ci lher purposefull y o r accidentally transported from one location [0 anOlher becomes established and invades [he new environment. The magnitude o f the problem is such that invasive species threaten most of the species listed in the U.S. Endangered Species Act (D 'Antonio ct al. 2001). ElTon has been spelH trying fa develop generalizations to determine which species are likely to become successfull y invasive (Ellsrrand and Schierenbeck 2000). Less freq uent­ly, possible genetic correlates have been sought to predic t which introduced species might become invasive (Gray 1986), Litlle attention has been givt:n to u nderstandin g the init ial spread dynam ics of an introduced species.

Capo-chichi et a!.: Cogongr.w: genetic diversity and spread 133

Interpretive Summary Invasive plantS are species that, after they have been transported

from their native habira! 10 a new location, have abiliey to

successfuHy disperse, colonize, spread on their own, and subsequently compele wi th and disrupt ecosystems. These characteristics rank cogongrass as one of ,he most nO)::ious weeds in tropical and subtropical regions around the globe. CogongtaSS is a ~rious threat to ecosystems worldwide covering 500 million ha. In sourhwest Alahama, where an estimated 200,000 ha are dominated by cogongras.s, infested arcas afC increasing rapidly. U;,mrolling lhis species through herbicides has proven (0 be expensive ($400 ha yr) requiring mllhiple applications pcr year and several years. While basic ecologic.l1 and herbicide-based comrol smdies have been conducted, there arc .Itill critica! information gaps that need ro be filled . This study employed amplified fragment length polymorphism (AFLP) analysis as the mok.::ular marker to examine genctic Volriation among sample sitC5 of oogongrass in the locale of Mobile, AL. The authors demonstrated chat the sample site 'IT the point or introduction is nO[ me only source or other infestations in the southern UniTed States. The spread dynamics afe greatly influenced by anthropogenic activities and natural factors to a much lesser cxtent. They have also shown that cogongrass exhibits high genetic diversity in the local environment of irs site of introduction. l'aCtors tlut may conuibule such high levels of genelic diversity include cross-compatibility between species allowing genetic exchange within and among sample sites and the rounding of the sample site.

The tesulrs of this study a~ essential for the development of appropriate comrol strategies by government agencics and landowners. The authors are confident that current and futurc research efforts will help to make the best·and COSt effective in the eradication program of cogongra5..'

Cogongrass [lmperata cylindrica (L) P. Beauv] was accidentally introduced into Alabama near Grand Bay about 1911 as seed in packing materials for Satsuma orange (Citrus rericuidta L.) rootstock from Japan (Dickens 1974; Tabor 1952). Cogongrass is a rhizomatous, perennial, invasive species that is generally considered a pernicious pest plant bec."1use of irs ability to successfully disperse and subse­quently compete with and displace desirable vegetation and disrupt ecosystems over a wide range of environmental conditions (Dozier er al. 1998; Holm et al. 1977). A few decades afrer the time of the innoduction near Grand Bay, Mobile County, AL., Tabor (1952) estimated that cogon­grass could be found on approximately 200 ha (494 ac) Later, estimates increased to nearly 4,000 ha for Mobile County (Dickens 1974). Current estimates exceeded 200,000 ha in sourhwest Alabama, with northward expan­sion ncar the Tennessee border (Faircloth et 'II. 2003).

The genetic composition of the founding population determines its ability to adapt to the new environment (Tsutsui and Case 2001) . However, dlis constitutes one, but not the only factor that might influence the ability of a species to become invasive. Effective size of an introduc­tion, genetic diversity of the source samples, number of

founding sources, and other characteristics may influence the genetic structure of a successful invasive population. Pappert et al. (2000) reported that populations that have been introduccd multiple times over an extended period ftom different origins should have higher genetic diversity dlan populations that were imroduccd once or only a few times. Characterizing the genetics of established sites of the aggressive species cogongrass is of interest because it may provide insights into the spread dynamics and the mode of local population cstablishmem.

In the United States cogongrass has twO close relatives: Brazilian satintail (lmperata brazitiensjs Trin.) and satintail (lmperata brevifotia Vasey) (Hitchcock 1950). Both Brazilian satimail and satintail arc described as native species by Hitchcock (1950) and are distinct in their ranges. Brazilian satintail is native to tbe pinelands and prairies of the Everglades, southern Florida, and Alabama, while satintail is confined to desert regions of western Texas to southern California. The nonnative cogongrass and the native Brazilian satintail have overlapping ranges, and discriminating these twO species can be difficult (Bryson and Carter 1993; Hall 1998). Frequent hybridization between them has been observed (Gabel 1982). Regardless, the introduction of cogongrass has resulted in the release of new genetic material into the U.S. genome (Gabel 1982; Hall 1998). The genetic diversity of cogongrass in the U.S. is unknown (Shilling et 'II. 1997). Worldwide, Hubbard et 'II. (I944) describes five varieties of cogongrass, mostly based on geographical distribmion, growth habit, and morphological characteristics_

Discrimination of individuals in clonal plants has been one of the methodical problems in plant demography (Ellstrand and Roose 1987) but recent advances in molecular marker techniques have enabled researchers to

overcome these limitations (Ellstrand and Roose 1987; Pornon et al. 2000). These markers provide means fOI· assessing genotypic diversity and phylogenetic relationship and have been extensively used to evaluate clonal structure of plants (For example, see Alben et a!. 2003; Arens et al. 2005; Douhovnikoff et at. 2004; Eckert et al. 2003i Kreher et a!. 2000). The AFLP technique, which is based on selective amplification of restriction fragments from a digest of total genomic DNA, has several advamages over other marker systems currently in usc (Vos et al. 1995). It does not require previous knowledge of the species genome, produces a large number of informative polymorphic markers per primer pair, is highly sensitive, and has proven to be robust and reliable (Mueller and Wolfenbarger 1999). Examination of the genotypic diversity of a. naturalized species in its new environment provides a first step for understanding the genetic consequences of its introductions and spread (Barrett and Husband 1990). This understanding is strengthened if points of introduc~ tion for the species of interest are known and potential

134 Invasive Plant Science and Management I, April-June 2008

Figure 1. Geographical locations ot the nine sample sites ot cogongras.s studied around Gl";lnd Bay, Mobile, Alabama.

source populations are idemified (Novak et al. 1993). In this pape r we examine the level of genotypic diversity and the invasion dynam ics of the imroduced cogongrass at the poim of imroduction and its adjacent areas. We exami ne the hypothesis that the population at the poi m ot introduction is a source of the other infestations in the southern United States.

Materials and Methods

Study Sites and Sampling. The point of introduction of cogongrass ncar Grand Bay in Mobile Counry, AL described by Tabor (1952) was located . We selected eight additional sites in the local Mobile, AL region on [he basis of information from records of the U.S. Deparrment of Agriculture Namral Resources Conservation Service (USDAfNRCS) (Figure I) . A site is defined as a single, contiguous population . Distances between collection sires ranged fro m 1.0 to 60.9 km (0.62 to 37.8 mi les). At each site we collected from 10 to 20 individuals along one or twO randomly placed transects depending on the size of the site. Samples were collected a mini mum of 0.3 and up

to 15 m (50 ft) apart across sites and latitudellongitude was obtained using C PS. A si ngle rhizome was taken from each sa mpling poim, stored in a plastic bag and transponed to the greenhouses at the Plam Science Research Center of the Alabama Agricultural Experiment Station (PSRC/AAES) located on the campus of Auburn Un iversity. Plants were grown in pOtS filled with standard gtowing medium di luted with coarse sand (I ; 1 v/v).

Genomic DNA was extracted from leaves according to a modified cetylrrimeth}~ ammonium bromide (crAB) method of Doyle and Doyle (1990). The presem smdy employed AFLI' analysis to examine genetic variation among (hese nine sites of cogongrJ.ss. We performed AFLP analysis according to Vos et al. (1995) with slight modifications with AFLP core reagent and starter primer ki ts purchased from Life Technology. Each AFLP marker was ucated as a unit character and scored as a binary code (I/O). The analysis of molecular variance (AM OVA) was performed as described by Excoffier et al. (1992) using Arlequin 2.000 software (Schneider et al. 2000). AMOVA produces estimates of variance components and fixation index statistics ((1» . {/>­

statistics are analogous to FST and describe genetic differentiation among sample sites. Gene Aow (Nm, number of migrants per generation) among sites was calculated based upon (/>-staristics using the method of Wright (1951 ). An alternative Bayesian approach (Holsingerer al. 2002) was also used to obtain independent estimates of FST. The original data matrix was im ported into HI CKORY, version 1.0 (Holsinger and Lewis 2003) and used for a full model, I = ° model, Theta-B (OB) = 0 modd, and I-free model. We conducted several runs with def.1ult sampling parameters (burn-in = 50,000, sample = 250,000, th in = 50) to ensure that the results were consistent. T he deviancc information criterion (DIC) (Spiegclhalter et al. 2002) was used to

estimate the fit between the data and a particular model and [Q choose among models (Holsinger and Wallace 2004). T he I-free model chooses I values ar random from its distribution while esti mating other parameters, resulting in estimates of OB, which incorporate all of the uncertainty in the prior of I (Holsinger and Lewis 2003). Recent studies have shown that estimates of I from dominant markers may be unreliable, particularly in data sets with small sample sizes (Holsinger and Wallace 2004), which is the situation in this study. Thus, the analysis below should be evaluated conservatively and be viewed as a complement to the other analyses.

Within-site genetic diversity was assessed for each site in several ways. We used m ultiple approaches, as each has a diffe rent SCt of assumptions, strengths, and weaknesses. Where they agree, we call be qui te con fidem in the analysis. The methods are as (I) percentage of polymorphic loci; (2) Nei's (1978) gene diversity (Hs), assuming Hardy-Wei nberg equi librium, as implemenred in the ARLEQU IN program; (3) Bayesia n gene diversity (Hd as implemented in the HICKORY program, which (loes not assume Hardy-

Capo·chichi et at.; Cogongrass: genetic diversity and spread 135

Table 1. Number of polymorphic fragments ~nd average gene diversiry u;l{ istics in cogongrass sample sites.

Number of Po lymorphic markers Average gene d ivcrsiry Number of allelcs'

Sam ple sites individuals Number % Nei's (Hs) Bayesian (H s) n, no

P I 10 59 43.1 0.1 2 0.1 9 1.3 1.2 0.16 1'2 10 55 40.1 0.15 0.21 1.4 1.2 0.21 1'3 10 68 49.6 0.19 0.2 5 1.5 1.3 0.25 1'4 18 58 42.3 0.12 0. 19 1.4 1.2 0.20 1'5 17 48 35.0 0.09 0. 16 1.3 1.2 0.15 1'6 19 59 43.1 0.10 0.17 1.4 1.2 0.16 1'7 15 44 32.1 0.12 0.1 8 1.3 1.2 0. 17 1'8 17 45 32.8 0.10 0.16 1.3 1.2 0.15 1'; 19 27 19.7 0.06 0.13 1.1 1.1 0.11 M~n 0. 11 0.1 7 1.3 1.2 0.1 7 Overdll 135 102 74.4

• Abbreviations: lIa, observed number of allelcs; ne, effcctive number of alleles.

Weinberg equilibrium within sites; and (4) Shannon 's information index (I) implemented in PO PGENE (Yeh and Boylc 1997) . We also conducred a homogeneity test in POPGEN E to evaluate whether allele fre<lucncil..'S were d ifferelH from each mher. Cluster analysis (CLA) was carried our to estimate the relationship based on the banding profile of indi,'iduals and assign these individuals to artificial clusters. Genetic sim ilariry coefficien ts bcfWccn individuals were calcul:lIed according to Nei and Li (J 979). Principal componcnr analysis (PCA) was performed to visualize the dispersion of the ind ividuals in relation 10 the first twO principal axes of variat ion. Com putatio ns were done usi ng rhe procedures in NTSYS-pc version 2.0 (Rohl f 1998). Mantel 's tests were calculated with a rejection zone from 100,000 random permutations.

Results and Discussions

AFLP and Gene Diversity. Two AFLP primer pairs identified a (Om] of 137 markers, 102 of wh ich (74%) were polymorphic between fWO or more ramcrs Crable I) . The

siz.c of the AFLP fragments, com pared to AFLP standard size markers, ranged from 50 to 600 base pai rs (bp). Polymorphic fragments were distributed across rhe enti re size range with the major proportion between 75 and 550 bp. Po lymorphic markers within sites varied from 27 (19.7%) 10 68 (49.6%). Nei's gene diversity (Hs) assuming Hardy-Weinberg cquilibrium ranged from 0.06 for Pi to 0.19 fo r P3, with a mean value of 0.1 1 (Table 1). Estimates of Bayesian gene diversiry (Hs) using the I-free model ranged from 0.13 for Pi TO 0.25 for P3, averaging 0. 17 (Table I). Shannon 's information index (1) varied from 0. 11 for Pi to

0.25 fo r 1'3, with a mean value of 0.17 (Table 1). However, these three estimates of genetic diversity were highly corre1an:d (r = 0.99, P < 0.00 1 fo r Hs vs. H II; r = 0.96, I' < 0.00 1 for Hs vs. I; and r = 0.97, I' < 0.001 for HB vs. /). Homogeneity tests of allele frequencies identified significant (I> < 0.001) differences across s,1mple sires for a large number of loci (75 OUI of 137). The genet"ic distance among sa mple sires ranged from 0.02 to 0.17 (Table 2).

The lowcst value of the gcnctic di versifY in Pi is an indication or low level ofhcrerozygosity among;individuals,

Table 2. Sample site pairwise Fsr (above (iiagona!), genetic dis tance (below d iagonal). and geographic distance (bold; km).

Sample si tes P I 1'2 1'3 1'4 1'5 1'6 1'7 1'8 1';

PI 0.12 0. 18 0.44 0.33 0.43 0.43 0.54 0.65 1'2 0.04 (1.0) 0.09 0.34 0.27 0.30 0.34 0.49 0.59 1'3 0.05 (7.9) 0.02 (6.9) 0.33 0.27 0.27 0.32 0.44 0.57 P4 0. 10 (1.5) 0.07 (1.4) 0.08 (6.9) 0.4 1 0.44 0.35 0.48 0.60 1'5 0.07 (32.1 ) 0.06 (3 1.3) am (25.6) 0.08 (60.9) 0.33 0.39 0.52 0.60 1'6 0.10 (3 1.6) 0.07 (30.8) 0.08 (25.4) 0. 11 (30.4) 0.06 (0.5) 0.03 0.54 0.62 1'7 0.11 (21.3) 0.09 (21.2) 0.11 (21.4) 0.09 (19.9) 0.09 (22.9) 0.07 (22.4) 0.18 0.48 1'8 0.16 (1.8) 0.1 4 (2.0) 0.14 (8.1) 0.12 (3.2) 0.12 (33. 1) 0.15 (32.6) 0.03 (23. 1) 0.43 1'; 0.1 7 (3.0) 0.15 (2.2) 0. 17 (5.2) 0.1 4 (3.3) 0.11 (30.4) 0.15 (30.0) 0.08 (22.1) 0.06 (2.9)

136 In V".lSive Planl Science and Management I, April-June 2008

Table: 3. Summary of {h~ analysis of mol ocular va riance (AM OVA) \\'i ~hin and amollg sample siu:s of cogongr.lSS. The s ignificanc~ (1))

of the (/>sr was b.1500 on Ihe procedure of 3,000 permutations."

Sum of Expocled mean Source of variat ion df S<)uaTeS squares

Among sample sites 8 753.6 n(J"! +~ Within sample sites 126 942.9 u; Toml 134 1,696.5 O"~.

• Abbrcvia~ions: nla, not applicable; df, degrees of freedom.

which suggestS high degree of clonality. In contrast, in sites PI th rough 1'8, genetic diversity is high. \Vle suggested four mechan isms that could expla.in such high genelic diversity. Fi rs riy, multiple d ifferent individuals from diverse sources may have established these populations. Secondly, popu· lalions may be founded from seedli ng recruitment. It is likely that long-distance dispersal is driven by seedling recruitment. Many of th e infestarions of cogongrass occur along righlS·of-way bord ering highways and railroads and open fields ar('3S where aerodynamic properties of wind dispersed seed fro m eogo ngrass may optimize d ispersal pOlential. Thirdly, si tes may be founded from the cross bcf'ovccn Brazilian satineail and cogongrass as reponed by G abel (1982). It has been shown that gen etic diversity at bolh population and species levels can increase due to natural hybrid ization (Vila el al. 2000). W e speculate tlm out-crossing involving the nonnative and native species may have an important impact on lhe levels of genetic d iversity, lead ing to a number of newly different genotypes.

Fourthly, the variation in gene diversity may al50 be due to small sample sizes or geno typic dynamics. The number of genets in a clonal population depe nds on ramel dynamics of existing genets and the recruitment of new genees fro m seeds (M clellan et al. 1997). If seed ling recru itment is infrequent in a clonal population and if increasing ramet density leads to the thinning of geners, genotypic diversi ty in a population is expected [0 be greater at the pioneer phase and to progressively di minish with population maturity and closure (Mclellan e( al. 1997). Population closure through ramets may lead to a dramatic loss o f genets and ro a selection o f h ighly genetically related genotypes (Pornon et al. 2000). As a result, population closure and maturation could both have reduced genotypic varia tion more in Pi , the oldest si te, tha n in Pl · P8. In general, one could argue that both seedli ng recruirmenr and clonal reproduction play an importanr role in genotypic dyna mics in the sites of clonal plants, which may innuencc the variation and level of genetic diversity. The mechanisms are nOt mutually exelus ive, and all fou r may contribute to

the high genorypic diversiry within samples.

AMOVA and C LA. T he analysis of variance indicated that 56% (I' < 0.05) of the total genetic variation was found

Variance Fixation Index

Coml>onenis Pereenrage 4>" Probability

5.8 43.8 0.44 < 0.05 7.5 56.2 ,I, ,I,

13.3 100

wi thin sites while the among-site component accou nted for 44% (P < 0.05; Table 3). CLA (Figure 2) using Un­weighted Pair G roup Method w ith Arithmetic M ean (UPGMA) and Nei and Li 's coeffi cient led to the separation of the individuals into three main dusters. C luster I grouped si tes PI , P2, P6, 9 ind i\' iduals OUI of the 10 sampled from 1'3, and 15 of the 17 from P5. Cl uste r II grou ped 17 of the 18 from P4 and 2 of the 17 from 1'5. ClUSter 111 captured all ind ividuals of Pi, 1'8, and 9 of 15 from P7. With in cl uster III , Pi formed a separate subcluster. This may indicate that these individuals arc highly clonal. As can be seen, an individual of P3 was d iscernible from the rest as well as one from P4 (F igu re 2). Few ind ividuals o f P5 and P6 could not be d ifferentiated (F igure 2) .

For the nine sitcs, a Bayesian esrimate of rST unde r a ran(lom-effects model of sampl ing (9t!) based on the J-free model from the Hickory program (Holsinger and Lewis

Figure 2. Dendrogram generaled by UPGMA clustering me~hod based on AFLP markers obf3ined with TWO primer pairs.

Capo.chichi CI al.: Cogongras.s: genetic diversity and spread 137

2003) genemed a mean of 0.39 :t 0.02 (95% confidence inten'als 0.35 and 0.43) wi th DIC values of 2,27 1 and Dbar of 1,839. Resuhs show that me I-free model yielded estimates of FIT lower than that of previously reported in the AM OVA. Again. this estimate should be viewed conservatively because of biases associated with too small sample sizes (Holsinger and Lewis 2003).

The value for the withi n-site variance componenr was sliglu ly higher. This high within-site estimate may be indicative of om-crossi ng species. Cogongrass has been described as an obl igate al1ogamous species (Gabel 1982). Viable seed is produced only when cross-poll inated (McDonald et al. 1996). In contrast, in self-poll inated species, one would predict a greater panitioning of diversiry among rather than with in sites in the absence of human­mediated gene flow among sites. With predictions based on previous evidence of inter- and intraspecific hybridization as a frequenr cause (Ellstrand and Schierenbcck 2000). cogongrass exhibits a high level of genetic variation and several genetically differentiated sites around the poim of introduction.

In the CU, the nine sites using UPGMA and Nci and Li 's coefficients showed three distinct main clusters. The resultS indicated that within clusters. there has been relative exchange of migrants among the sample si tes at different levels. For instance, in cluster I grouping sites PI , P2, PG, and mOSt of the individuals of P3 and P5. there is no clear subcl uster composed of all individuals of a given site as can be seen in cluster II I in which all individuals of Pi formed a separate subcluster. The sites P5 and PG fJ;om Dauphin Island arc grouped together with most of the sites near the poim of introduction , although these sites are farthest away fro m that poi m and separated by an environ mcncal barrier. We suggest that anthropogenic activities fo llowed by local migration of cogongrass could be an explanation.

PCA. The first rwo principal components account<.x! for 70% of the total variation. The scaner-plOt represemarion of the PCA analysis showed a dear-cut separation of the 135 individuals inco three main dusters in rdation to the first rwo principal axes of variation (Figure 3). Cluster I com prised all individuals o f PI , P2 , P3, PG, 5 our of 18 samples from (>4, IG our of 17 from P5, and 4 om of IS from P7. C luster II comprised 13 our of 18 from N. 2 OUt

of 15 from (>7, and I om of 17 from P5, while duster 111 grouped 9 our of 15 from 1'7 and all individuals of P8 and Pi. G enetic distances among sites ranged from 0.02 to 0.17 widl rhe highest values between Pi and the others. Geographic distances ranged from 1 to GO.9 km (Table 2) . A$ can be seen, most of the ind ividuals are grouped in the same d usters as the three of the previous CU. We suggest that all of rhe samplcd individuals from 1'8 and 9 of the individuals from 1'7 most probably descend from Pi o r arc closely related.

V5 0 0

° ° 80 <Po~

0 o~J~ Cluster I

0.15 <l> o~ 't>

;}S ° 00° ,

0 Cluster II

N -{).05 °l u ..

o PSb (PSb - lndMdual b In the sample slit S)

-{) .25

Clus ler III

-{).45 l-___ -2=~-_ 0.45 0.59 0.73 0.86 1.00

PC 1

Figure 3. Scatter plot of the first two principal componentS (PC I and PC 2) fo r 135 cogongrass ramers. Variation explained by rhe firs[ principal component is 63%; by the second principal component is 7%.

When comparing the dispersion pattern of the individ­uals within dusters as an indication of individual-level genetic variation, along me first principal axe, individuals of cl uster III arc less dispersed man mose of clusters II and 1. The most likely explanation is that dusters with more dispersion pattern may be the result that some sexual reproduction is occurring; thus, indicating me develop­ment of new genOtypes. Overall, the AF LP approach provided an effective means of exam ining relationship and genetic sim ilariry within and among sites of cogongrass in me locale of Mobile, AL.

Site Differentiation. The res ults of analysis of F-statistics as a measure of d ifferentiation among the si tes sampled arc presented in Table 2. Most pairwise estimates of FST (9 1% ) were significant (P < 0.05; Table 2). suggesling that each sample site is disti nc t. The FIT estimates between Pi and [he resc of the sites ranged from 0.43 to 0.G5. The data were examined for evidence of thc model of isolation-

138 InV2sivc Plant Science and Management 1, April-June 2008

by-distance (Slatkin 1993) . Dispersal is oft~'n distarlCt· dependant. such that populations n(:;lr many other populations receive a greater number of migranrs. whereas more isolated populations receive fewcr (I3rown and Kodric-Hrown 1977). The g.cographic distance between pairs of sites ranged !l"Otll \ to 60.9 kill, averaging 17.0 km. Estimates of gcn(: flow among sites rang('d from 0.5 to 5.0, averaging 1.0 . G~ncrally. if Nm < 1. [hen loctl differentiation of populations would result, ;llld if Nm < I, then [here will be litde differenti;uion among populations (McDermott and Iv!cDol1:lld 1993). The pattern of gent' flow was not rdated w geographic d istance (r = 0.09). Lower vailles (NIII < I) were obsefved in 27 uf 36 p;lirs of sites (data not $hown). The highest value was estimated to be 5.0 berween a pair of sample sin~s separated by 7 klll, in(l ic<l ting signiJic<ltlt geneti.:: ex.::hange. The lowesT gene flow value was estimated to be 0.26 betwccn Pi and PI d ist:lIlf by 3 km. Accordingly, There was nei rher significant corrdation between the matri..:es of gem~ flow and geographic distance (M:llltd 's h':St, r = - 0.13, t = - 0.70, l' = 0.24) nor between geneTic and geographiC" di.\t'lllces (Mantel's tcst, r = -0. 13, t = - 0.72, P = 0.23). Indeed, rhe loweST I~r value ohscTved bcrwccn sites P4 and 1'5 may sllggest th:1I they have cxch:lllged a greater number of migr:mls even though tI;ey are hlrther apart. However, these sites may also have received migranl~ frorn the same sources. Whidock ;lIld McCa uley ( 1999) reported rhat a p:lir of popuLttions dUT receivcs 1lligrant~ from the sal11e source would have a low F\T cven if they exchange no migr:lnrs at all.

Modd-hasetj res ts tor site d iOcrellt iation showed Ih:1{ the cogongra~s sample site~ arc genetically distinct. BOTh Bayesian :lIId AMOVA ("~t il1l:H eS of F'>T (0.39 and 0.44, respectively) are high for cogongrass, an om-crossing species. High fixaTion illl!cX was fou nd in the clona l Ottt­

crossing speci l"~ Pitm;mill gepkesii L. B. Smith in till' family Brollld iacL";te (Sanhou et al. 2(01). The I'~r cstimates for all sire pairs showed a high level of genetic differentiation. The subdivision index was lower I(>f tilt" pair P2-]>3 (I-SI = 0.09, P < 0.05) ,md higher for all other pairs wiTh Pi , "'ST ranging from 0.43 to 0.65. This may indica[(' dlat Pi has exch:tr:ged few migr'llIts with mhers. The high F~J values ohlailllx\ in the analysis may be becHise of genetic isolatioll of the sites alld the individuals wirh in rhese sires. This result is sll rpri~illg, given the geographical proximity of one sample site to another. Such differences among sites sugge~t th;H eit!K'r tll ey hal'e been genelic.llIy isolated at some lime during their spre;ld or humans have hrought together cross-compatible species thaT had becil n;lfltrally isohlTe.:l as adulOwle.:lglxl by And('rson and St(;bbing (1954) . TllU~. the lewis of among-si te diversi fY and The difficulty of explaining the observed paltCflIS of sitl' dif"l('reIHialion may be due 10

varying degn:es of inu:rspel' itlc hybridi1.ation ill some

sample sites and taxonomic misidemitlcltioll. In future work, it would be important to L'I'aluare the genetic str llcrure of cogongrass along the dispersal routeS due to

seed being blown by prev3iling winds from rhe coasral to

inland areas. Us ing molecular m3rkcrs. uur result., showed dwt

cogollgrJ5S exhibits high levels of genetic variation in the local environment of its sire ofi1l1 roduction in the southern Un it ed StaTes. ThaI nriation sholl'e(1 few panefJlS gcographi..:alJy, ,md the sile at the poin l of i1t1roduction contained the lowest genetic vari:ltioll. These p;merns lead to the co nclusion Ihar our (bra do nO[ suppOrt the hypothesis that lhe population at the point of inrroduction is a SOUfce of lbe other iule.,rations ill the southern Unite\1 States. With thc exccption uf 1'7 and !l8. the sites sampled showed linle evidence that Pi, the initial poi m of introduction of the species, pbycd a dominant rule ill fo unding till:: olher sit~s salllpkd.

The alllount of genetic I"ari:!rion within and bcrween sites was q ui te b rge given how recemly cogongtass w~!s itHroduced. Facwrs that Illay conrrihute sllch high genetic d iv('rsity include cross-compatibility between species al­lowing gencric cxch;Hlge wirhill and ~lmU!lg sites ;!IId the manner ill which {he sites wefe founded. \'(Ihat is dear frolll This assessment is That seed ling recruitment is far more important than has been previously assul1led. 'f he spread dynamics of cogongra:.s is greatly inlluenccd by anfhropo· genic activities such as so il disturbance and canopy rCl1loval, and narural factors intluence spread to a much lesser extent. \'(Ie do not claim thar these factors explain the mechanism of inva~il·ent~~' . They au to accelera\(~ the opportunities for bringing together cross-cumpatible species Th;n had been prCl'iously i501:lIed by n;lfltral evems such as ecology and geogr:lphy. This would have influenced the variation and \cvds o j" gcnetic diversity. The remits represen t an important first step ill rcvealing the mech:misms of invasion o r cogongr.lss an(1 cO lH ribllle to

the growing bndy of empirical studies of mechanisms and genetic co mequences of pblll invasions. h om a m:llIage­mem perspecrivl" the knowledge of generic di"ersity and popu\;'tion genetic StrUCTU re o f invasive pLt Ill.~ is essen tial for improving t.he eflicacy of control strategy (\,\/ang l't a1. 2007). Tlic more ali ke individllals are, the e:lsier Ihey should be to manage. The genetics should have a larger role in the developmen l of policy 10 man:lgc ;Illd conrrol invasive species th rough a hCHer llnd('rsrallding of till' risk thai p;lrticlllar genotypes pose. The conn"]ll could also be applied to other im-Hive donal species. ft would bc in teresting for fmure studies to inveslig:ne the patteTII of distr ihmion of the species along its dispersal route.

Sources of Materials

, SUrfer primer kils. I fc Ta;hnology. (;i[,.:;o BIZ!.. G.,i{\,,,r.,burg. MD. USA

Cap,)-chichi CI JI. : CO!!:OIl!!:r:I>S: ge'nedc dil"crsity and sprc;ld 139

Acknowledgments

This research was supported by the Alabama Agricultural Experiment Station through Project no. ALA080-0 18.

Literature Cited

Albert, T., 0. Raspe, and A. L. Jacqueman. 2003. Clonal Strucrure in Vaccinium myrtillus L. revealoo by RAPD and AFLP markers. 1m. J. Plam Sei. 164:649-655.

Anderson, E. and G. L. Stebbing Jr. 1954. Hybridization as an evolutionary stimulus. Evolution 8:378- 388.

Arens, P., C. j. Grashof-Bokdam, T. van def Sluis, and M. ]. M. Smulders. 2005. Clonal diversity and genetic differentiation of Mnjanrhnnum bi[ofium among forest fragments of different age. Plant Ecol. 179:16-180.

Barrell, S. C. H. and B. C. Husband. 199(). The genetics of plant rlligration and colonil.ation. Pages 254- 277 in A. H. D. Brown, M. T. Clegg, A. L Kahler, and B. S. Weir, oos. Plam Population Genetics, Breeding, and Genetic Resources. Sunderland, MA: Sinaeur Associates.

Brown, J. H. and A. Kodric-Brown. 1977. Turnover rates in insular biogeography: effe<:t of immigration on atincrion. Ecology 58: 445-449.

Bryson, C. T. and R. Carter. 1993. Cogongrass, Im/X'alll cylindrica, in the United Srates. Weed Technol. 7:1 005- 1009.

D'AlIIonio, R. , L Meyerson. and]. Denslow. 2001. Research priorities related to invasive exotic species. Pages 59-60 in M. Soule, G. Orians, and I>. Dee Boersma, cds. Conservation Biology: Research priorities for the coming decade. Covelo, CA: hland Press.

Dickens, R. 1974. Cogongrass in Alabama after sixty years. Weed Sci. 22:177- 179.

Douhovnikoff, V .. A. M. Cheng, and R. Dodd. 2004. Incidence, size and spatial s(rucrure of clones in second-growth srands of ooas( redwood, &quuia sempn1lirm s (Cupressaceae) .. Am. J. Bot. 91: 1140--114G.

Doyle, j . 1- and 1- L Doylc. 1990. Isolation of DNA from fresh tissue. Focus 12: 13-15.

Dotier, H.,j. F. Gaffney, S. K. McDonald, E. R. R. L.Johnson, and D. G. Shilling. 1998. CogongrJss in dIe Uni(ed Sratcs: hiswry, ecology, impacts, and management. Weed Technol. 12:737- 743.

Eckert, C. G., K. Lui, K. Bronson, P. Corradini, and A. Bruneau. 2003. Population gene(ics consequences of arreme variation in sexual and donal reproduction in an aquaTic plant. Mol. Ecol. 12:331- 344.

Ellstrand, N. C. and M. L. Roose. 1987. Patterns of genotypic diversity in clonal plant species. Am. J. Bot. 74:123-131.

Ellstrand, N. C. and K. A. Schierenb.::ck. 2000. Hybridi7..ation as a stimulus for the evolution of invasiveness in plants. Pages 7043-7050 in Proceedings of the National Academy of Sciences of the United States of America.

Exooffier, L. P .. E. Smouse, and J. M. Quattro. 1992. Analysis of molecular variance inferred from metric dis(ances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131:479-491.

Faircloth. W. H., M. G. Pafterwn, j . H. Miller, and D. H. Teem. 2003. Wanre<! Dead Not Alive: Cogongrass. Auburn, AL: Alabama Cooperative Ex(ension System Rep. AN R-124 I. 4 p.

Gabel, M. L. 1982. A biosystematic study of the genus Impera'a (Gramineae: Andropogoneae). Ph.D. dissertation. Allies, IA: Iowa Sra(e University. 94 p.

Gray, A. J. 1986. Do invading species have definable genetic­characteris(ies~ I'hilos. Trans. R. Soc. Lond. B BioI. Sci. 314:655-674.

Hall, D. W. 1998. Is Cogon grass really an exotic? Wildland Weeds 1: 14-15.

140 Invasive Plant Scicnce and Managcmcnt I, April-June 2008

Hitchcock, A. S. 1950. Manual of the Grasses of the Unite<! Stales. 2nd ed, revised by Chase A. Washington, DC: U.S. Department of Agriculture. Misc. Pub, No. 200. 1051 p.

Holm. L. G., P. Donald, j. V. Pancho, and J. P. Herberger, eds. 1977. The World's Worst Weeds: Distribution and Biology. Honolulu: T he University Press of Hawaii. 609 p.

Holsinger, K. E. and 1'. O. Lewis. 2003. H ICKORY: a package for analysis of population geneTic data Vl.0. http://www.ceb.uoonn.ootL.

Holsinger, K. E., 1'. O. Lewis, and K. D. Dipak. 2002. A Bayesian approach (0 inferring popula(ion S(fucture from dominant markers. Mol. Ecol. 11:1157- 1164.

Holsinger, P. E. and L. A. Wallace. 2004. Bayerian approaches for The analysis of population generic structure: an example from P1manrhertl !eucupbtlta. Mol. F...ca!. 13:887- 894.

Hubbard, C. E" R. O. Whyte, D. Brown, and A. P. Gray. 1944. Imperara cyli"dn'm: taxonomy, distribution, eoonomic significance and control. Imperial Agricultural Bureaux Joint Publication 7: 1-63.

Kreher, S. A., S. A. Fore, and B. S. Collins. 2000. Genetic variation within and among pa(ches of the clonal species, Vaccinium sramintum L. Mol. Eool. 9:1247-1252.

McDermon, J. M. and B. A. McDonald. 1993. Gene flow in plant pathosystcms. Annu. Rev. Phy(opa(ho!. 31 :353-373.

McDonald, S. K., D. G. Shilling, C. A. N. Okoli, T. A. Bewick, D. Gordon, D. Hall, and R. Smith. 1996. Populalion dynamics of cogongrass, Impmua cylilldrica. Page 156 in Proceedings of the Southern Weed Science Society of America.

MclA:lIan, A. ]., D. Prali, O. Kaltz, and B. Schmid. 1997. Structure and analysis of phenotypic and genetic varia(;on in donal plants. Pages 185-210 in H. de Kroon and J. van Groenendael, cds. The Ecology and Evolution of Clonal Plana;. Lciden, The Netherlands: Backbuys.

Mueller, U. G. and L. L. Wolfenbarger. 1999. AFLP genotyping and fingerprinting. Trends Ecol. Evol. 14:389- 394.

Nei, M. 1978. Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89:583-590.

Nei, M. and W. Li. 1979. Ma(hematical model for studying genetic varrarron in terms of restriction endonudeascs. l'ag<!':s 526\1- 5273 in Proceedings of the National AC3d~my of Sciences of (he USA.

Novack, S. J., R. N. Mack, and 1'. S. Soltis. 1993. Genetic variation in Bromm ltc/orum (Poaceae): in(roduc(ion dynamics in North America. Can.J. Bol. 71:1441-1448.

l'appcTI, R. A" J. L. Hamrick, and L. A. Donovan. 2000. Genetic variation in Purran'a lobmtl (Fahaceae), an inlTociuced, clonal. invasive plant of the southeastern Unite<! St;ues. Am. J. Bol. 87:1240-1245 .

Pornon, A., N. Escaravage, P. Thomas, and P. Taberlel. 2000. Dynamics of genotypic strUClUre in clonal Rhododelldron ferrugin~um (Ericaceae) populations. Mol. Ecol. 9:1099-1111.

Rolf, F. J. 1998. NTSys.pc, version 2.0: Numerical ra~nomy and multivariate analysis sysr<!':m. hrrp:llexeterwfrware.com/cat/ntsyspd ntsyspc.

Sanhou, c., S. Sarnadi, and M. C. Boisselitr-Dubayle. 2001. Genetic strucrure of the saxicole Pi/cairnia gryslmii (Bromcliaceae) on insdbergs in French Guina. Am. J. Bot. 88:861-868.

Schneider, S., D. Roessli, and L. Excoffier. 2000. Arlequin, Vctsion 2000: Software for Population Genetics Dara Analysis. http:// anthropologie.unige.cn/arlequin/.

Shilling, D. G., T. A. Beckwick, J. F. Gaffney, S. K. McDonald, C. A. Chase. and E. R. L. Johnson. 1997. Ecology, Physiology, and Management of Cogongrass (/mpera/a cylindrica). Bartow, FL: Florida InstilUte of Phosphate Research Rep. 03-1070· 140. 144 p.

SimberlofT, D. and B. Von Holle. 1999. Positive in(eractions of nonindigenous species: invasional mel(down~ BioI. Invasions. 1: 21-32.

Slatkin, M. 1993. Isolation by distance in equilibrium and non­equilibrium popularions. EvolUTion 47:264-279.

Spkgdhalt~r, D. j., N. G. Iksr, B. P. D rlin , and A. van det Lind~. 2002. Bayesian mrasurcs of modd complaity and fir. j ournal of th~ Royal S(;IristiCiI Society, series B 64:583--639.

Tabor, P. 1952. Cogongf'25$ in Mobi le County, AbI;Qm;I. Agron.j. 44:50. T$Ul$ui, N. D. and T. j. Case. 2001. Population genC"lics .and colony

SlIUClure of rhe Argenril"\(' anr (Linl"pilhnna humi/t-) in ill n;uivt and inrroduad ranges. Evolulion 55:976-985.

V"do., M., E. Weber. and M. D·Antonio. 2000. ComelY." ':'" i,,,I'];""('U,,, of invasion by plant hybridization, BioI. Invasions 2:207- 217.

Y,n, P., R. Hogers, M. Reijam, T. van de Lo:c, M. I-iornes, A. Frijlen;, J. 1'01, J. Pcleman. M. K" ipu, and M. Z:..be ... u. 1995. AFLI'; a new technique for DNA fingerprinring. Nucleic Acids Res. 23: 4407-4414.

Wang, T., Y. Su, and G. Chen. 2007. Population gen~tic \':iliation and structure of Ih(' invasive ~ Milrania mirranthil in South~rn China: consequences of r:lpid r:lnge expansion. journ ... ] of Heredity. DOl 10. 1093Ijherffilesm080.

Whidock, M. C. and D. E. McCauley. 1999. Indirect measur(' of g('ne Rowand migrarion: FST " 11(4Nm + I). Heredity 82:[17-125.

Wright. S. 195 I . The genelical slrucnm of populations. Ann. Eugen. 15:323-354 .

Yeh, F. C. and T. J. B. Boyle. 1997. Porgene. the user.friendly sha«:ware for POI)ulation genel ic analysis. hnr:llwww.nall~rl a.cal - fYch/index.htm.

R«rilld funl" 5, 2007. and approll(d NOlI(mwr 21. 2()()7.

U!KI-chichi <:t al.: Cogongrass: g('netic diversity and spr~d 14 1