m e genetic analysis of colony and population structure of three...

MOLECULAR ECOLOGY AND EVOLUTION

Genetic Analysis of Colony and Population Structure of ThreeIntroduced Populations of the Formosan Subterranean Termite(Isoptera: Rhinotermitidae) in the Continental United States

EDWARD L. VARGO,1, 2 CLAUDIA HUSSENEDER,3 DAVID WOODSON,4

MICHAEL G. WALDVOGEL,1 AND J. KENNETH GRACE5

Environ. Entomol. 35(1): 151Ð166 (2006)

ABSTRACT The Formosan subterranean termite, Coptotermes formosanus Shiraki, is a major inva-sive pest species in many parts of the world. We compared the colony breeding system and populationgenetic structure in three introduced populations in the continental United States: Charleston, SC;City Park, New Orleans, LA; and Rutherford County, NC. Based on worker genotypes at 12 micro-satellite loci, we found that colonies were mainly genetically distinct entities consisting of either simplefamilies headed by monogamous pairs of reproductives or extended families containing multipleneotenic (replacement) reproductives descended from simple families. Populations varied from 48%simple families in Charleston to 82% simple families in City Park. Extended family colonies in all threepopulations were likely headed by �10 neotenic reproductives. There was no signiÞcant isolation bydistance in any of the populations, suggesting that colonies reproduce by relatively long-range matingßights and/or human-mediated dispersal within each population. The Charleston population showedevidence of a recent genetic bottleneck and most likely was founded by very few colonies. Clusteranalysis indicated that the Charleston and City Park populations are quite genetically distant from eachother and most likely originated from different source populations. The more recently introducedRutherford County population was genetically most similar to City Park. These Þndings, together withresults from other infested sites, indicate considerable variation in the genetic structure and breedingsystem of introduced populations of this species, making it unlikely that there is a simple genetic orbehavioral explanation for the success of C. formosanus as an invasive species.

KEYWORDS microsatellite markers, breeding system, social organization, Formosan subterraneantermite, Coptotermes formosanus

INVASION OF BIOLOGICAL COMMUNITIES by exotic speciesis a serious and growing problem that is having enor-mous ecological and economic impacts around theworld (Pimentel et al. 2000, Mooney and Cleland2001). Invasive social insect species are especiallydamaging; several species of ants, wasps, and termiteshave severely disrupted ecological communitiesand/or caused signiÞcant economic damage in manyof their introduced ranges (Vinson 1986, Williams1994, Moller 1996). The Formosan subterranean ter-mite, Coptotermes formosanus Shiraki, is a highly in-vasive species that has become established in numer-ous parts of the world (Su and Tamashiro 1987, Su2003). A native of East Asia, most likely mainland

China (Kistner 1985), C. formosanus has invaded sev-eral other oriental locations (Taiwan, Japan, and SriLanka), the PaciÞc (Hawaii, Guam, Midway, and Mar-shall Islands), South Africa, and many areas of thecontinental United States. Its initial introduction intothe U.S. mainland is closely associated with militaryports receiving equipment and supplies from the Pa-ciÞc theater after World War II (La Fage 1987). Aspecimen collected from Charleston, SC, in 1957 is theÞrst record of C. formosanus on the U.S. mainland(Chambers et al. 1988). Other early reports of thisspecies on the U.S. mainland include Houston, TX, in1965 (Anonymous 1965) and New Orleans in 1966(Spink 1967). C. formosanus is now widespread butpatchily distributed on the U.S. mainland, occurring in10 states, including virtually all the southeastern andsouthcentral states and California (Woodson et al.2001).

In this study, we used microsatellite markers toinvestigate colony and population genetic structure inthree populations of C. formosanus from the south-eastern Unites States, including two populations thatwere among the Þrst sites known to be invadedÑ

1 Department of Entomology, Box 7613, North Carolina State Uni-versity, Raleigh, NC 27695Ð7613.

2 Corresponding author, e-mail: [email protected] Department of Entomology, Louisiana State University Agricul-

tural Center, Baton Rouge, LA 70803.4 U.S. Geological Service, Western Regional OfÞce, BRD, 909 First

Ave. Suite 800, Seattle, WA 98104.5 Department of Plant and Environmental Protection Sciences,

University of Hawaii at Manoa, 3050 Maile Way, Room 310, Honolulu,HI 96822.

0046-225X/06/0151Ð0166$04.00/0 � 2006 Entomological Society of America

Charleston, SC, and New Orleans, LAÑand a recentlydiscovered population from western North Carolina.The populations of Charleston and RutherfordCounty are relatively isolated (Chambers et al. 1988;unpublished data), whereas the City Park populationis embedded in a much more extensive population inand around New Orleans (Woodson et al. 2001). OurÞrst objective was to infer colony breeding structurein these three populations to determine whether therewere common features that might help account for theexceptional invasion success of this species. In otherhighly invasive social insects, primarily ants, reducedgenetic variability after introduction events can favorinvasiveness by lowering intercolonial aggression,leading to large, unicolonial societies that are ecolog-ically dominant (Holway et al. 2002, Tsutsui and Su-arez 2003, Payne et al. 2004). Thus, it is conceivablethat founder effects in C. formosanus could promote aparticular type of colony social organization underly-ing its invasion success.

Genetic markers are a powerful tool for inferringcolony breeding structure in social insects (Thorne etal. 1999, Ross 2001), and there has been an increasingnumber of genetic studies on colony social organiza-tion of termites (Atkinson and Adams 1997, Thompsonand Hebert 1998a, b, Goodisman and Crozier 2002),especially subterranean termites (Clement 1981, 1984,Reilly 1987, Kaib et al. 1996, Husseneder et al. 1997,1998, 1999, 2005, Jenkins et al. 1999; Bulmer et al. 2001,Clement et al. 2001, Husseneder and Grace 2001,Vargo 2003a, b, Vargo et al. 2003a, b, DeHeer andVargo 2004, Dronnet et al. 2005, DeHeer et al. 2005).These studies indicate that colonies of subterraneantermites are generally founded by pairs of unrelatedprimary (alate-derived) reproductives. As coloniesage, the primary reproductives are replaced or sup-plemented by neotenics (precocious nonwinged re-productive forms that develop within the colony) whoinbreed within the colony. In at least one species,Reticulitermes flavipes Kollar, there is evidence that asmallpercentageofcoloniescancontain threeormoreunrelated reproductives (Jenkins et al. 1999, Bulmer etal. 2001, DeHeer and Vargo 2004), and both laboratory(Fisher et al. 2004) and Þeld (DeHeer and Vargo2004) studies have shown that such genetically mixedgroups can arise through colony fusion. It has beenthought that, as colonies of subterranean termitesgrow and expand their foraging ranges, they fre-quently fragment into independent daughter colonies(Shellman-Reeve 1997, Myles 1999, Thorne et al.1999). Although there is some genetic evidencesuggesting that colony reproduction by buddingmay occur often in some species, e.g., the Africansubterranean termite Schedorhinotermes lamanianus(Sjostedt) (Husseneder et al. 1998), it is not universalbecause intensive studies of a number of populationsof R. flavipes have failed to Þnd evidence for it (Bul-mer et al. 2001, Vargo 2003a, DeHeer and Vargo 2004).

To date, we have studied the breeding structure oftwo introduced populations of C. formosanus, andthese show considerable variation. In two populationsfrom southern Japan, we found that 90% of the colo-

nies were simple families headed by closely relatedreproductives, and the remaining 10% consisted ofhighly inbred extended family colonies (Vargo et al.2003a). In contrast, we found that a little more thanone-half the colonies (57%) in a New Orleans, LA,population were comprised of simple families headedby outbred reproductives, and the remaining colonieswere only slightly inbred extended families (Hus-seneder et al. 2005). Comparative studies of additionalintroduced populations should help reveal whetherthere are features common to colony breeding struc-ture across introduced populations, and, if so, whatrole these may play in the invasion success of thisspecies. Inaddition,understanding thebreeding struc-ture of colonies may be important in management,especially in the case ofC. formosanus,where there areefforts in place to eliminate all colonies within largeareas. For example, if colonies frequently form spa-tially separated reproductive centers with limitedmovement of workers among them, this could restrictthe distribution of bait toxicants, interfering with theability of baits to eliminate entire colonies (Hus-seneder et al. 2003).

Our second objective was to study the genetic re-lationships among these three populations as well astwo other introduced populations that were previ-ously studied in Japan (Vargo et al. 2003a) and anotherpopulation in New Orleans, LA (Husseneder et al.2005). Analyses of the genetic relationships amongintroduced populations of invasive species can helpdetermine whether they arose from a single introduc-tion or multiple introductions, whether some intro-duced populations likely originated from the samesource population, and to infer patterns of dispersalwithin introduced ranges (Goodisman et al. 2001, Buc-zkowski et al. 2004, Johnson and Starks 2004, Kolbe etal. 2004). For example, it is of interest to know whetherthere was a single introduction ofC. formosanus to theU.S. mainland that spread to various locations throughhuman-aided dispersal or whether there were multi-ple introductions from different source populations.

Materials and Methods

Sample Collection. Samples were collected fromnatural wood or artiÞcial feeding stations in threepopulations: City Park, New Orleans, LA; Charleston,SC; and Rutherford County, NC. Figures 1Ð3 show therelative locations of the collection points in each pop-ulation. With few exceptions, samples from each col-lection point consisted of at least 20 individuals Ðwork-ers, soldiers and sometimes alates. Termites wereplaced directly into 95% ethanol and stored at �20�Cuntil DNA extraction.City Park, New Orleans, LA. A total of 19 samples

was collected between 19 March and 16 May 2001(Fig. 1). As part of a larger study of termites infestingtrees in City Park, bucket traps consisting of 15.25-cmcircular valve box covers Þlled with rolled corrugatedcardboard were installed in the ground within 1 m ofa tree in spring 2001. Only trees and shrubs �10 cm indiameter at ground level were used.C. formosanuswas

152 ENVIRONMENTAL ENTOMOLOGY Vol. 35, no. 1

most commonly found with southern live oak (Quer-cus virginiana Miller), followed by southern yellowpine (Pinus sp.), Chinese tallow (Sapium sebiferum L.Roxburgh), and bald cypress (Taxodium distichumL.). The location of each sample was determined usinga Trimble Pro XLR GPS (Trimble, Sunnyvale, CA)device.Charleston, SC. Thirty-six samples were collected

from Charleston, SC (Fig. 2). Samples of workers andsoldiers, and in some cases, alates, were collected fromthe trunks or the bases of 21 trees in Charleston, SC,on 3 June 2002. Most of the samples were collected inor near Hampton Park located adjacent to the Citadelnorthwest of downtown Charleston. The location ofeach sample was determined using a Trimble GeoEx-plorer 3 handheld GPS device. An additional 15 sam-ples were collected from trees and wood debris (treestumps and down limbs) in Charles Towne LandingState Historic Site on 13 and 19 May 2003. This park islocated �2 km from Hampton Park across the AshleyRiver. The location of samples was manually markedon a map, and the distance between collection pointswas measured on the ground using a tape measure.

Rutherford County, NC. Sixteen samples were col-lected from infested trees, buildings, and railroad tiesalong an �8-km stretch of U.S. Highway 221A runningthrough the towns of Rutherfordton, Spindale, andForest City in Rutherford County, NC, on 12 and 13August 2003 (Fig. 3). This was part of a cooperativeeffort of the Structural Pest Control Division, NorthCarolina Department of Agriculture and ConsumerServices, North Carolina Cooperative Extension Ser-vice, North Carolina State University, and local pestcontrol companies to delineate the boundaries of apopulation Þrst discovered in 1992 (M.G.W., unpub-lished data).Microsatellite Analysis. Genomic DNA was ex-

tracted from individual termite whole bodies using theDNeasy Tissue Kit (Qiagen, Valencia, CA). Twentyworkers from each collection point were genotyped at12 microsatellite loci (Table 1), 11 of which wereidentiÞed from C. formosanus (Vargo and Henderson2000) and 1 of which, Rf 6–1, was identiÞed fromReticulitermes flavipes (Vargo 2000). We followed theprotocols of Vargo (2000) and Vargo and Henderson(2000) for polymerase chain reaction (PCR) ampliÞ-cation and genotype scoring.Colony Affiliations. To determine if samples from

nearby collection points represented the same or dif-ferent colonies we used three criteria. First, all pairs ofcollection points within each population were testedfor genotypic differentiation by means of a permuta-tion test using the program FSTAT (Goudet 2001).Second, we compared the number of private alleles(the number of alleles present in one collection pointbut not the other) in each pair of collection points.Groups of workers with different alleles present wereconsidered to belong to different colonies. Third, wetook into account the distance between collectionpoints. Because the maximum linear distance reportedfor foragers of a single C. formosanus colony is 115 m(Su and Scheffrahn 1988), collection points separatedby much further (�500 m) were considered to belongto different colonies. In practice, collection pointsbelonging to different colonies were signiÞcantly dif-ferentiated and had many private alleles. More em-phasis was placed on analysis of private alleles in theCharlestonpopulationbecauseof the lowgeneticvari-ability present at this location.Classification of Colonies. Following the classiÞca-

tion of Vargo (2003a, b), Vargo et al. (2003a), andDeHeer and Vargo (2004), colonies were placed inone of the following two groups: simple families orextended families. Simple families were those presum-ably headed by a monogamous pair of reproductives.Colonies were considered simple families if the ge-notypes of the workers were consistent with thoseexpected for a single pair of parents and if the ratiosof the observed genotypes did not differ signiÞcantlyfrom the expected Mendelian ratios as determined bya G-test. For each colony, an overall G-value wasobtained by summing all the locus-speciÞc G-values.In contrast to simple families, extended families areheaded by multiple kings and/or queens. Extendedfamily colonies were those having genotypes that

Fig. 1. Locations of C. formosanus samples from the CityPark, New Orleans, LA, population. Dark shapes representbodies of water. Encircled collection points were part of thesame colony.

February 2006 VARGO ET AL.: GENETIC STRUCTURE OF C. formosanus COLONIES 153

were inconsistent with a single pair of reproductives(e.g., one or more loci with Þve or more genotypicclasses or three classes of homozygotes) or those inwhich the genotypes were consistent with the pres-ence of a single pair of reproductives but the observedfrequencies of the genotypes deviated signiÞcantlyfrom the expected values in simple families (P� 0.05,G-test).Colony and population genetic structure. Colony

and population genetic structure were assessed byestimating the coefÞcient of relatedness among nest-mate workers and by estimating F-statistics. The av-erage relatedness among colony mates was estimatedusing the program Relatedness v. 5.0.8 (Queller andGoodnight 1989) with colonies weighted equally. TheSEs were obtained by jackkniÞng over loci. F-statisticswere estimated using FSTAT (Goudet 2001). We fol-lowed the notation of Thorne et al. (1999) and Bulmeret al. (2001) in which each colony is treated as asubpopulation, and genetic variation is partitionedamong the following components: the individual (I),the colony (C), and total (T). According to this no-

tation, FIT is analogous to the standard inbreedingcoefÞcient, FIS, and is a measure of the level of in-breeding in individuals relative to the population. FCT

is analogous to FST and represents the genetic differ-entiation among colonies. FIC is the coefÞcient ofinbreeding in individuals relative to their colony andis particularly sensitive to the numbers of reproduc-tives present and their mating patterns within colo-nies. As shown by Thorne et al. (1999) and Bulmer etal. (2001), FIC is expected to be strongly negative insimple family colonies, to approach zero with increas-ing numbers of reproductives within colonies and tobecome positive with assortative mating among mul-tiple groups of reproductives within colonies or withmixing of individuals from different colonies. For eachof the F-components, SEs were estimated by jackknif-ing over loci. SigniÞcance of the F-values and therelatedness coefÞcients was determined by comparingthem to predicted single point values by means ofone-sample t-tests (P� 0.05). Pairs of empirical valueswere tested for signiÞcance using two-sample t-testsand not assuming equal variances. Pairwise genetic

Fig. 2. Locations ofC. formosanus samples from the Charleston, SC, population. Collection points that are encircled werepart of the same colony.


differentiation (FST) among populations was deter-mined using the program GDA (Lewis and Zaykin2000) by performing a hierarchical analysis incorpo-rating colony level structure simultaneously.

From the worker genotypes present in each of thesimple family colonies, the genotypes of the parentswere reconstructed, and F-statistics and the coefÞ-cient of relatedness among nest mate reproductiveswere estimated from the inferred genotypes. The like-lihood that the two reproductives in simple familycolonies were full siblings (nest mates) was tested

using the program Kinship (Goodnight and Queller1999). For the primary hypothesis, we used the aver-age degree of nestmate relatedness for each popula-tion, and this was tested against a null hypothesis ofzero. Log likelihood ratios were calculated based on1,000 iterations.IsolationbyDistanceAnalysis.To assess isolation by

distance, we Þrst obtained pairwise FCT-values for allpairs of colonies within each population. These valueswere converted to FCT/(1 � FCT), and the Pearsonproduct correlation coefÞcient between these values

Fig. 3. Locations ofC. formosanus samples from the Rutherford County, NC, population. Encircled collection points werepart of the same colony.


and the ln of geographic distance for all pairs of col-onies was calculated. The signiÞcance of the correla-tion coefÞcient was assessed by means of a Mantel testwith 1,000 iterations as implemented in Genepop onthe Web (Raymond and Rousset 1995).Tests for Genetic Bottleneck. Because C. formosa-

nus is an introduced species with a relatively shorthistory in the U.S. mainland, we expect that the studypopulations experienced a recent genetic bottleneck.To determine if there was evidence for a bottleneck ineach population, worker genotypes were tested forheterozygosity excess using two tests implemented intheprogramBottleneckv. 1.2.02(Piryet al. 1999).Onetest, developed by Cornuet and Luikart (1996), de-termines whether there is a signiÞcantly greater pro-portion of loci with heterozygosity excess than ex-pected for a population at mutation-drift equilibriumusing a sign test, whereas the other test detects sig-niÞcant heterozygosity excess on average across lociusing a Wilcoxon sign-rank test (Piry et al. 1999). Thelatter test is considered to be the more appropriate andpowerful of the two tests for �20 loci (Piry et al. 1999).The tests were conducted on each of the 20 resampleddata sets for each population using the two-phasedmodel of mutation, which includes a mixture of single-stepandmulti-stepmutations, as recommendedby theauthors for microsatellite loci (Piry et al. 1999). Wealso followed the authorsÕ suggestion in setting thesingle-step mutations to 95% and the variance amongmultiple steps to 12.

In addition, for each locus, we calculated M � k/r,the ratio of the number of alleles to the range of allelesizes, where k � the number of alleles and r � thenumber of possible alleles sizes between the smallestand the largest observed alleles (Garza and William-son 2001). We calculated mean M across all loci foreach population. This mean value was compared withM � 0.68, the threshold value below which a popu-lation can reasonably be assumed to have undergonea recent reduction in population size (Garza and Wil-liamson 2001).

Unweighted Pair-Group Method with ArithmeticAverage Clustering. To study the relationships of thethree populations to each other and to two otherintroduced populations previously studied in Japan(Vargo et al. 2003a), as well as a previously studiedpopulation from Louis Armstrong Park, New Orleans(Husseneder et al. 2005), we used the program Mega2(Kumar et al. 2001) to construct a tree using theunweighted pair group method with arithmetic meanfrom pairwise FST values in GDA. Twelve loci wereused in this analysis for each population, except theLouis Armstrong Park population, for which eight lociwere used.

Results

Genetic Variability. The three populations differedconsiderably in genetic variability. As shown in Table1, 11 of the 12 loci examined were variable in the CityPark population, with two to seven alleles per locus.Ten loci were variable in the Rutherford County pop-ulation, with two to six alleles per locus. Despite hav-ing the most samples and the most colonies present,the Charleston population was the least variable, withno more than three alleles per locus. Of the total of 19alleles present in the Charleston population, only 2were unique, both of which occurred at locus Cf4:1A2–4.Colony Affiliations. Based on permutation tests of

genetic differentiation, 13 of the 19 collection pointsin City Park were genetically distinct and were con-sidered separate colonies. Two pairs of collectionpoints located 13 and 55 m apart, respectively, werenot signiÞcantly differentiated and had identical al-leles and genotypes. Each of these pairs was groupedtogether into a colony (Fig. 1). There were two othercollection points that also were not signiÞcantly dif-ferentiated from each other, but there were two pri-vate allele and four genotypes unique to one of thesecollection points. Moreover, these were separated by900 m. Because of the presence of private alleles and

Table 1. Variability of microsatellite loci in the Charleston, SC; Rutherford County, NC; and CityPark, New Orleans, LA, populations

Locus

City Park, New Orleans, LA Rutherford Co. NC Charleston, SC

No.alleles

Freq. mostcommon allele

HeNo.

allelesFreq. most

common alleleHe

No.alleles

Freq. mostcommon allele

He

Cf 1–1 1 Ñ 1 Ñ 2 0.82 0.29Cf 4:1 A2–4 3 0.36 0.66 3 0.81 0.32 4 0.38 0.71Cf 4:1 A2–5 2 0.71 0.41 2 0.95 0.10 2 0.94 0.11Cf 4–4 2 0.57 0.49 5 0.58 0.57 7 0.33 0.76Cf 4–9A 2 0.54 0.50 4 0.37 0.69 3 0.41 0.66Cf 4–10 2 0.55 0.49 3 0.48 0.64 3 0.41 0.66Cf 8–4 2 0.77 0.35 6 0.33 0.74 4 0.37 0.71Cf 10–4 1 Ñ 3 0.72 0.42 3 0.76 0.39Cf 10–5 2 0.75 0.37 3 0.63 0.49 3 0.60 0.53Cf 11–1 1 Ñ 1 Ñ 1 ÑCf 12–4 2 0.87 0.23 4 0.50 0.59 3 0.39 0.66Rf 6–1 1 Ñ 2 0.76 0.36 2 0.80 0.40Mean � SDa 1.9 � 0.6 0.44 � 0.13 3.5 � 1.3 0.49 � 0.20 3.1 � 1.5 0.53 � 0.21

Allele frequencies were calculated by the program RELATEDNESS (Queller and Goodnight 1989) using all worker genotypes and coloniesweighted equally.a Polymorphic loci only.


the large distance between them, these two collectionpoints were considered different colonies. In Ruther-ford County, we grouped the 16 collection points intoeight genetically differentiated colonies, three ofwhich were present at multiple sites spanning from 46to 144 linear m (Fig. 3).

Because of the low variability of the microsatelliteloci in the Charleston population, assigning colonyafÞliations was not as straightforward as in the othertwo populations. Of the 315 pairwise tests of genotypicdifferentiation, 49 (16%) were not signiÞcant (P �0.05). However, we excluded 29 of these pairs as be-longing to the same colony because there were one ormore private alleles present. The other pairs wereconsidered part of the same colonies. Thus, one pair ofnearby collection points located 101 m apart in Hamp-ton Park was grouped together into the same colony(Fig. 2). Table 2 gives examples of two pairs of samplesthat were not signiÞcantly differentiated. CT1Ð17 andCT1Ð8 were not signiÞcantly differentiated (P � 0.8,exact test of genotypic differentiation) but were as-signed to different colonies because of the presence of

one private allele (185 at locus Cf1:4A2–4) and threegenotypes unique to CT1Ð8 (Cf 1:4A2–5:188/185 and185/185; Cf 10–5:281/281), as well as the long distance(874 m) between them. Samples CT1Ð13 and CT1Ð16were grouped into the same colony because they werenot signiÞcantly differentiated (P � 0.9, exact test ofgenotypic differentiation), they shared all the samealleles and genotypes, and were located relativelyclose to each other (101 m). The differences betweenCT1Ð13/1Ð16 and CT1Ð17 were more typical of dif-ferent colonies; they were signiÞcantly differentiated(P� 0.00001, exact test of genotypic differentiation),and they differed in having three private alleles andseven unique genotypes. The Charles Towne LandingState Historic Site had four multiple-site colonies, con-sisting of two to Þve collection points, the most ex-pansive of which spanned 133 linear m (Fig. 2). TheÞve collection points that comprised this expansivecolony all had identical genotypes, and this colony wasa simple family.Classification of Colonies. There was variation in

the proportion of simple and extended family coloniespresent in the three populations, ranging from aboutone-half simple families in Charleston to �80% simplefamilies in City Park (Table 3). Of the 13 extendedfamily colonies present in Charleston, 7 had genotypesinconsistent with the presence of a single pair of re-productives (e.g., too many homozygous classes),whereas 6 had Mendelian genotypes but the frequen-cies differed signiÞcantly from those expected (P �0.05, G-test summed across loci). Because of the lowgenetic variability in this population, we could notexclude the possibility that some of these “extendedfamilycolonies” wereactually “mixed familycolonies,”i.e., headed by two or more unrelated same sex re-productives (DeHeer and Vargo 2004). However,given the high degree of relatedness among workernestmates (r � 0.49) in this population, we can con-clude that if such mixed family colonies exist, they arenot common. In the Rutherford County and City Parkpopulations, the extended family colonies all had ge-notypes incompatible with a single pair of reproduc-tives. Despite greater genetic diversity in these twopopulations (seven alleles at one locus in City Parkand two loci with Þve or more alleles in RutherfordCounty), no more than four alleles per locus werepresent in any of the colonies, suggesting that repro-ductives within extended family colonies in these pop-

Table 2. Examples of genotypes of C. formosanus workersfrom pairs of collection points in which the genotypes within eachpair were not significantly differentiated

LocusPair 1a Pair 2a

CT1Ð17 CT1Ð8 CT1Ð13 CT1Ð16

Cf 4:1A2–4197/197 1 1197/188 11 10 7 9197/185 13 11188/188 7 3188/185 4185/185 2Cf 4:1A2–5163/163163/160 10 13160/160 10 6 20 19Cf 4–4248/248 11 13248/230 9 7 11 10230/230 9 10Cf 4–9A302/302 1 2 6 5302/287 14 11 8 10287/287 5 7 6 5Cf 4–10245/245 1 3 7 6245/236 12 10 7 9236/236 6 7 6 5Cf 8–4243/243 20 20 9 12243/234 11 8Cf 10–5296/296 10 12 10 9296/281 10 7 10 11281/281 1Cf 12–4191/191 17 12 20 20191/173 3 8

a Pair 1 was considered to belong to different colonies because ofthe presence of one allele and three genotypes present in CT1Ð8 butnot CT1Ð17 and the long distance between them (874 m); whereaspair 2 was considered part of the same colony because they shared allthe same alleles and were located relatively close to each other (101m).

Table 3. Comparison of the breeding structure of three intro-duced populations of the Formosan subterranean termite

PopulationNo.

colonies

Simple families No.reproductivesin extended

families

No.(%)

Inbred

City Park, New Orleans,LA

17 14 � 3Ð9(82%)

Charleston, SC 25 12 � 4Ð9(48%)

Rutherford Co., NC 8 6 � 3Ð9(67%)


ulations were descended from monogamous pairs ofreproductives.Colony and Population Genetic Structure. The F-

statistics and relatedness values estimated from theworker genotypes are shown in Table 4, along withvalues derived from computer simulations based onThorne et al. (1999) and Bulmer et al. (2001). Overall,workers in all three populations were signiÞcantlyinbred (FIT � 0.085Ð0.239; all P � 0.05, one-sampleapproximate t-test) and they were closely related toeach other (r � 0.49Ð0.64). The Rutherford Countypopulation was the most inbred overall (FIT � 0.239),but it differed signiÞcantly only from the City Parkpopulation (P� 0.03, two-sample approximate t-test).In all three populations, the extended family colonieswere signiÞcantly more inbred and had signiÞcantlyhigherFIC values than their corresponding simple fam-ily colonies (all P � 0.05, approximate t-test). In CityPark, FCT for extended families was signiÞcantlygreater than that for simple families (P � 0.03, two-sample approximate t-test). The F-statistics and relat-edness coefÞcient for workers in simple family colo-nies in the Charleston population were close to and

not signiÞcantly different from those predicted forcolonies headed by outbred pairs of monogamous re-productives, suggesting these colonies were headedby unrelated primary reproductives. However, work-ers in simple family colonies in Rutherford Countywere signiÞcantly more inbred than expected (FIT �0.132, P � 0.04, one-sample approximate t-test), aresult consistent with the elevated degree of related-ness between reproductives heading these colonies.Workers in the City Park simple families had signiÞ-cantly higher values of FCT, FIC, and r than predicted(all P � 0.03, one-sample approximate t-test), mostlikely because the reproductives in these coloniesoriginated from inbred colonies. As shown in Table 3,the main conclusions regarding the simple families inthe three populations are the following: those inCharleston and City Park tended to be headed byunrelated reproductives resulting in outbred workers,whereas workers in the Rutherford County coloniesare slightly inbred because nestmate reproductivesare more closely related in this population.

Considering the extended family colonies, the pres-ence of 10 or more neotenics of each sex (Table 4,

Table 4. F-statistics and relatedness coefficients for worker nestmates of C. formosanus from Charleston, SC; Rutherford County,NC; and City Park, New Orleans, LA, and values expected for some possible breeding systems of subterranean termites as derived fromcomputer simulations

FIT FCT FIC r

Empirical valuesCharleston, SC

All colonies (n � 25) 0.139 0.280 �0.196 0.492(SE) (0.046) (0.021) (0.036) (0.050)Simple family colonies (n � 12) 0.036 0.285 �0.349 0.584(SE) (0.070) (0.035) (0.041) (0.076)Extended family colonies (n � 13) 0.194 0.238 �0.058 0.420(SE) (0.055) (0.036) (0.057) (0.092)

Rutherford Co., NCAll colonies (n � 8) 0.239 0.379 �0.225 0.637(SE) (0.065) (0.047) (0.038) (0.111)Simple family colonies (n � 6) 0.132 0.323 �0.282 0.616(SE) (0.065) (0.046) (0.030) (0.107)Extended family colonies (n � 2) 0.434 0.499 �0.127 0.684(SE) (0.137) (0.123) (0.062) (0.229)

City Park, New Orleans, LAAll colonies (n � 17) 0.085 0.339 �0.384 0.609(SE) (0.022) (0.011) (0.029) (0.014)Simple family colonies (n � 14) 0.008 0.306 �0.428 0.596(SE) (0.039) (0.023) (0.025) (0.021)Extended family colonies (n � 3) 0.380 0.475 �0.176 0.649(SE) (0.085) (0.077) (0.083) (0.186)

Simulated breeding system(A) Simple family colonies headed by outbred reproductive pairsa 0.00 0.25 �0.33 0.50(B) Extended family colonies with inbreeding among neotenics(1) Nf � Nm � 1, X � 1a 0.33 0.42 �0.14 0.62(2) Nf � 2, Nm � 1, X � 1b 0.26 0.35 �0.14 0.55(3) Nf � 2, Nm � 1, X � 3a 0.52 0.59 �0.17 0.78(4) Nf � 5, Nm � 1, X � 1b 0.27 0.34 �0.11 0.53(5) Nf � Nm � 10, X � 1c 0.33 0.34 �0.01 0.51(6) Nf � Nm � 10, X � 3a 0.37 0.38 �0.02 0.56(7) Nf � 200, Nm � 100, X � 3a 0.34 0.34 0.00 0.71

Empirical values were based on eight microsatellite loci in the Charleston, SC, population, 10 loci in the Rutherford County, NC, population,and 11 loci in the City Park, New Orleans, LA, population. For the simulated breeding systems, X represents the no. of generations of productionof replacement reproductives within a colony; Nf and Nm represent the no. of replacement females and males, respectively, produced pergeneration.a From Thorne et al. (1999).b Results of simulations using the methods of Thorne et al. (1999).c From Bulmer et al. (2001).


cases B5 and 6) could be excluded for all three pop-ulations, because they had either FIC values that weresigniÞcantly lower than expected (Rutherford Countyand City Park, both P� 0.05, one-sample approximatet-test) or had signiÞcantly lower FIT and FCT values(Charleston,P� 0.03, one-sample approximate t-test).As discussed below, the relatively high value of FIC inthe Charleston extended family colonies was becauseof two colonies with highly positive values. Whenthese two colonies were excluded from the analysis,FIC for this class of colonies became more stronglynegative (�0.178), a result consistent with relativelyfew neotenics present. Thus, extended families in allthree populations appeared to have on average fewneotenics, from three to nine (Table 3).

A detailed view of the individual colony inbreedingcoefÞcients (FIC) shows a more complex picture forthe Charleston population (Fig. 4). The only colonieswith positive FIC-values were the two colonies fromCharleston at the far right side of the graph in Fig. 4,with values of 0.28 and 0.59. The elevated values ofthese two colonies account for the relatively highmeanFIC-value for the extended family colonies in thispopulation. In these colonies, all individuals at eitherone locus (in the colony with FIC � 0.28) or two (inthe colony with FIC � 0.59) were one of two homozy-gous genotypes with no heterozygotes present. In thelatter colony, there were only two homozygous classesat loci Cf 4–4 and Cf 4–10, and these were associatedsuch that individuals who were230/230 atCf4–4were245/245 at Cf 4–10, whereas 248/248 homozygotes atCf 4–4were 236/236 at Cf 4–10. These results suggestsome degree of assortative mating among the repro-ductives within these colonies.

Table 5 shows the values for the inbreeding coef-Þcient (FIS) for reproductives in simple family colo-nies and the relatedness coefÞcient between nestmatereproductives in simple family colonies based on theirgenotypes as inferred from their worker offspring. Thereproductives had high coefÞcients of relatedness inCharleston and Rutherford County (r � 0.209 and0.312, respectively), but in neither case was this sig-niÞcantly greater than zero (both P � 0.1, t-test).Compared with worker FIT-values, the equivalent ofFIS in the reproductives, functional reproductiveswithin simple family colonies were slightly less inbredthan workers in Charleston and Rutherford Countyand slightly more inbred than workers in City Park,but none of these differences were signiÞcant basedon overlapping 95% conÞdence intervals. Results ofthe analysis with the program Kinship showed similarproportions of colonies headed by putative sibÐsibpairs in all three populations: 21% (3 of 14) in CityPark, 33% (4 of 12) in Charleston, and 33% (2 of 6) inRutherford County.

Fig. 4. Frequency distribution of point estimates for individual colony inbreeding coefÞcients (FIC) of extended familycolonies for three introduced populations of C. formosanus.

Table 5. Coefficients of relatedness (r) between nestmatereproductives and of inbreeding (FIS) of reproductives in simplefamily colonies of C. formosanus

Populationr

(SE)FIS

(95% CI)

Charleston, SC 0.209 0.060(0.213) (�0.015Ð0.150)

Rutherford County, NC 0.312 0.104(0.240) (0.026Ð0.228)

City Park, New Orleans, LA 0.015 0.184(0.058) (0.119Ð0.254)


Isolation byDistance Analysis. There was no strongisolation by distance in any of the populations (Fig. 5).Although there was a positive correlation betweencolony pairwise FCT values and geographic distance inall three populations, the correlations were not sig-niÞcant in either City Park (r2 � 0.096, P � 0.119,Mantel test) or Rutherford County (r2 � 0.044, P �0.618, Mantel test). There was a weak but signiÞcantcorrelation in the Charleston population (r2 � 0.087,P� 0.02, Mantel test), but this was most likely causedby the slight differentiation between the HamptonPark colonies and the Charles Towne Landing StateHistoric Site colonies, which were separated by �1 kmof water, because this correlation disappeared when

considering each population separately (r2 � 0.050,P � 0.30 and r2 � �0.339, P � 0.75 for the HamptonPark and Charles Towne Landing colonies, respec-tively).Tests for Genetic Bottleneck. For the Charleston

population, there was strong evidence of a geneticbottleneck. All of the 20 resampled data sets showedsigniÞcant heterozygosity excess averaged across loci(P � 0.05, Wilcoxon sign-rank test). Similarly, a signtest for a greater proportion of loci with heterozygos-ity excess than expected was signiÞcant (P� 0.05, signtest) in all cases. The mean � SD ratio of the numberof alleles to the range in allele size (mean � 0.426 �0.276) was signiÞcantly lower than the threshold valueof 0.68 (P � 0.017, one-sample t-test), providing fur-ther support of a strong, recent genetic bottleneck.For the City Park population, there was signiÞcantheterozygosity excess (P � 0.05) in only 3 of the 20resampled data sets according to the Wilcoxon sign-rank test, and in no cases using the less powerful signtest. M � 0.631 � 0.303 was lower than the thresholdof 0.68 but not signiÞcantly different from it (P� 0.33,one-sample t-test). Also, there was no strong evidenceof a recent bottleneck in the Rutherford County pop-ulation; there was signiÞcant heterozygosity excess inonly 2 of the 20 resampled data sets according to theWilcoxon test and in only 1 of the data sets accordingto the sign test. The average ratio of the number ofalleles to the range of allele sizes was again below thethreshold value (mean � 0.631 � 0.305) but not sig-niÞcantly so (P � 0.31, one-sample t-test).Unweighted Pair-Group Method with ArithmeticAverage Analysis. Results of the unweighted pair-group method with arithmetic average clustering (Fig.6) suggest that the Rutherford County, NC; City Park,New Orleans; and Louis Armstrong Park, New Orleanspopulations are the most closely related, whereas theCharleston population is rather distant to these. TheJapanese populations cluster together into a branchthat is quite removed from the other four populations.

Discussion

A principle Þnding of this study was that the threepopulations of C. formosanus investigated here, each

Fig. 5. Relationship between genetic differentiation(FCT) estimated from microsatellite genotypes and physicaldistanceamongall pairsofC. formosanuscolonies in the studypopulations.

Fig. 6. Unweighted pair group method with arithmeticmean (UPGMA) tree showing the genetic relationshipsamong the three studied populations of C. formosanus andtwo previously studied populations in Japan (Vargo et al.2003a) and Louis Armstrong Park, New Orleans, LA (Hus-seneder et al. 2005). The tree was constructed from pairwiseFST values calculated from 12 loci in all populations except forthe Louis Armstrong Park population, in which 8 loci wereused.


with its own distinct introduction history, exhibitedconsiderable variation in colony breeding structures.Although all three populations were primarily com-prised of colonies consisting of genetically distinctfamily groups, they varied in the relative proportionsof simple and extended families and in the overalldegree of inbreeding. The two populations differingthe most in the proportions of the two family typeswere from the areas with the longest history of For-mosan termites on the U.S. mainland: Charleston, with�50% extended families, and City Park, New Orleans,with �20%. Rutherford County, a relatively recentlyintroduced population, was intermediate, with one-third extended families. Variation among the popula-tions in the overall degree of inbreeding among work-ers was complex and did not merely reßect theproportions of extended family colonies in each pop-ulation, as might be expected. For example, Ruther-ford County had the highest level of inbreeding (FIT� 0.24) even though this population had a lower per-centage of extended families than did Charleston.

Based on F-statistics and the coefÞcient of related-ness, simple families in Charleston provided a reason-able match to the values expected for colonies headedby outbred monogamous pairs. The simple family col-onies in Rutherford County and City Park variedsomewhat from expected but in different ways. In theformer population, workers were signiÞcantly moreinbred than expected, presumably because nestmatereproductives were more closely related in this pop-ulation than in the other two populations. In City Park,nestmate workers were more closely related to eachother and less inbred relative to their own colonies(more negative FIC) than expected, because func-tional reproductives in simple family colonies weresigniÞcantly more inbred than the population average.One possible explanation for this Þnding is that alatesin City Park issued predominately from extended fam-ily colonies, which produce individuals who are moreinbred than do simple family colonies. Investigatingwhether simple and extended family colonies differ intheir relative reproductive output would requirequantifying alate production by colonies of knownbreeding structure.

The average inbreeding coefÞcients for extendedfamily colonies in all three populations, especially thehighly negative FIC-values, were consistent with col-onies headed by relatively few closely related neo-tenic reproductives, on the order of three to nine, alldescended from monogamous pairs of founders.Higher numbers of reproductives can be excluded,because according to the predictions from computersimulations, this would have resulted in FIC-valuescloser to zero (Thorne et al. 1999, Bulmer et al. 2001).Moreover, the presence of multiple unrelated repro-ductives in most colonies, which may arise throughcolony fusion, can also be ruled out because suchcolonies are expected to have positive FIC-values. Thefact that no more than four alleles per locus werefound in extended family colonies in the City Park andRutherford County populations, despite having up toseven and six alleles per locus, respectively, also sup-

ports this conclusion. The possibility of multiple un-related reproductives in the Charleston extended fam-ilies is more difÞcult to exclude. First, the low geneticvariability in this population, with a maximum of threealleles per locus, prevented us from detecting thepresence of any multiple unrelated reproductiveswithin colonies. Second, the relatively small averageFCT-value obtained, indicating low genetic contrastsamong colonies, is consistent with some fusion amongcolonies. However 10 of the 13 colonies had negativeFIC-values, making it unlikely that colony fusion wascommon, if it occurred at all, in this population. None-theless, two of these colonies had strongly positiveFIC-values suggesting some degree of assortative mat-ing, one possible cause of which is fusion between twoor more colonies. Other mechanisms that could alsounderlie assortative mating are the presence of phys-ically separated reproductive centers within colonieswith no breeding between the different groups ofreproductives and selective mating within a singlegroup of reproductives, such as two cohabiting mo-nogamous pairs. Similarly high FIC-values were ob-tained by Vargo et al. (2003a) in an introduced pop-ulation of C. formosanus in Japan, suggesting somelevel of assortative mating within extended family col-onies in this population as well. In an expansive ex-tended family colony of C. formosanus in Louis Arm-strong Park, New Orleans, Husseneder et al. (2005)found slight but signiÞcant genetic differentiationamong spatially separated foraging sites, suggestingthepresenceof twoormore reproductivecenterswithunequal mixing of workers produced in each. Similarspatial and genetic substructure has been found in theAfrican subterranean termite Schedorhinotermes la-manianus (Husseneder et al. 1998). More detailedstudies are needed of individual colonies to determinewhether such genetic substructuring in C. formosanuscolonies is widespread.

The presence of multiple unrelated reproductivesin termite colonies has been shown using molecularmarkers in a few termites, such as the mastotermitidMastotermes darwiniensis Froggatt (Goodisman andCrozier 2002) and the termitid Nasutitermes corniger(Motschulsky) (Atkinson and Adams 1997). In sub-terranean termites, the co-existence of multiple un-related reproductives occurs at low frequency in R.flavipes(Jenkinset al. 1999,Bulmeret al. 2001,DeHeerand Vargo 2004). It was reported to be common in R.grasseiClement (Clement 1981), but new data for thisspecies (DeHeer et al. 2005) suggest that it may berare at best. Colony fusion has been observed in theÞeld and laboratory as a mechanism leading to theco-existence of offspring produced by multiple unre-lated reproductives (DeHeer and Vargo 2004, Fisheret al. 2004). Evidence for colony fusion in C. formo-sanus is sparse and indirect. Based on the location ofdyed termites and changes in worker size over time, Suand Scheffrahn (1988) reported the possible fusion oftwo C. formosanus colonies in Florida. However, thelack of clear genetic evidence of fused colonies in anumber of populations with sufÞcient allelic variabil-ity to detect it, including Oahu, HI (C.H., E.L.V., and


J.K.G., unpublished data), and Kyushu, Japan (Vargoet al. 2003a), from previous studies, and City Park,New Orleans, and Rutherford County in this study,suggests that, if colony fusion occurs in introducedpopulations of C. formosanus, it is rare.

The inferred number of neotenic reproductives inextended family colonies in the study populations wasrelatively low, more so than what might be expectedbased on Þeld collections. Although collections offunctional reproductives in colonies of C. formosanusare few, the numbers of neotenics observed withincolonies ranges from 6 to �100 (King and Spink 1969,Myles 1999). We inferred an average of six or fewerfunctional neotenics per colony for the study popu-lations, which is low compared with what has beenfound in the Þeld. Possible explanations for this dis-crepancy are that only exceptionally large and oldcolonies with abnormally high numbers of reproduc-tives were excavated, Þeld collections came from pop-ulations different from the study populations (China,Hawaii, South Africa, and Lake Charles, LA) whereneotenic numbers were higher, or there is signiÞcantdisparity among the reproductive output of cohabitingneotenics resulting in a lower effective number ofreproductives, as occurs in many social insects (Ross2001).

The variation in the family types observed hereamong the different populations is well within thewide range reported for other introduced populationsofC. formosanus.At one extreme, Vargo et al. (2003a)found that simple families made up 90% of 30 coloniesin two Japanese populations. At the other extreme,30% of 20 colonies were simple families in Oahu, HI(Vargo et al. 2003b). And between these extremes,Husseneder et al. (2005) found, in a detailed study ofcolonies in Louis Armstrong Park, another New Or-leans population, 57% of 14 colonies were simple fam-ilies. In contrast, a native population of 14 coloniesfrom Guangdong Province, China, consisted of all ex-tended families (C.H., E.L.V., and J.K.G., unpublisheddata). The reasons for this large variation are not clear,but could include age structure of the populations,local ecological conditions, or genetic make-up.

In this regard, it is of interest to compare theseÞndings for City Park to previous Þndings on LouisArmstrong Park (Husseneder et al. 2005), a populationlocated only 3 km away that was intensively studiedusing the same microsatellite loci employed in thisstudy. The results from these two populations, whichshare not only common ecological conditions but asimilar genetic composition, are alike. Both popula-tions were comprised of a majority of simple familycolonies. Although a higher proportion of simple fam-ily colonies was present in City Park (82%) than inLouis Armstrong Park (57%), this difference was notsigniÞcant (P � 0.23, Fisher exact test). The coefÞ-cients of inbreeding and relatedness in both popula-tions suggested that simple family colonies wereheaded by outbred pairs of primary reproductives andthat extended families were headed by relatively fewneotenics descended from the original founding pair.The degree of relatedness between nestmate repro-

ductives in simple families was low in the two popu-lations (r � 0.02 and 0.11 in City Park and LouisArmstrong Park, respectively) and not signiÞcantlydifferent from zero. Also, there was a similar propor-tion of cohabiting reproductives that were putativesiblings: 21% (3 of 14) in City Park and 17% (1 of 6)in Louis Armstrong Park. Moreover, there was nosigniÞcant isolation by distance in either population,consistent with a lack of strong population viscositywithin either population. These two populations arecertainly more similar to each other than they are toother introduced populations that have been studiedto date. Because City Park and Louis Armstrong Parkshare largely the same ecological conditions and aregenetically similar, it is not possible to tease apartwhich of these two factors may be more important inshaping colony breeding structures. Future studiesexamining other urban populations along the GulfCoast with distinct introduction histories or compar-ing nearby urban and natural populations may help inthis regard.

Our results showing variation in the breeding sys-tem in different introduced populations of C. formo-sanus are consistent with the results of studies onnativepopulationsof subterranean termites. Studiesofthe eastern subterranean termite, R. flavipes, haverevealed variation in colony breeding structure amonggeographically distant populations. Studies in centralNorth Carolina (Vargo 2003a, b, DeHeer and Vargo2004) show that about three-quarters of the coloniesare simple families headed by outbred reproductives,about one-quarter are extended families descendedfrom simple families and headed by relatively fewreproductives, and a very small number of colonies(2%) are mixed families formed by the fusion of twoor more colonies. In contrast, Bulmer and Traniello(2001) found that 27% of 22 colonies in Massachusettswere simple families headed by monogamous pairs ofoutbred reproductives, 59% were extended familiesheaded by numerous neotenics, and 14% were com-prised of mixed families. In a study of R. grassei insouthwestern France, DeHeer et al. (2005) found vari-ation among populations, ranging from all extendedfamilies with numerous neotenics in an area north ofBordeaux to nearly one-half simple families and one-half extended families in an area further south. ThusintraspeciÞc variation in colony breeding structuremay be common in subterranean termites, and a majorquestion to address in future studies concerns thecauses of this variation.

Our results are consistent with those of previousstudies showing expansive colonies of C. formosanus.Although our sampling scheme was not designed toexplicitly study the foraging areas of colonies, wefound collection sites from the same colony separatedby 144 m in Rutherford County, and one colony inCharleston spanned 133 linear m. These distances areclose to the reported record for this species of 185linear m observed by Su (1994) in Florida, where thisauthor also reported a colony with a foraging distanceof 100 m. Distances of �100 m also have been reportedin Florida by Su and Scheffrahn (1988), in Louisiana


by King and Spink (1969) and Messenger and Su(2005), and in Hawaii by Lai (1977). Thus, two col-onies in this study are among the most expansive C.formosanus colonies reported to date. Although themost spatially expansive colony we found was an ex-tended family, the Charleston colony that spanned133 m was a simple family. To the extent that foragingarea reßects colony population size, the second largestcolony among those studied here had only a singlequeen. These results conÞrm previous Þndings for C.formosanus in Louis Armstrong Park, New Orleans(Husseneder et al. 2005) and for R. flavipes in NorthCarolina (DeHeer and Vargo 2004), indicating thatthe presence of multiple neotenics is not necessarilyassociated with larger foraging areas and populationsizes as is sometimes assumed (e.g., Grube and For-schler 2004).

There was no signiÞcant isolation by distance in anyof the populations once the large water barrier sepa-rating the two groups of colonies in Charleston wastaken into account. In a study of two introduced Jap-anese populations on a slightly larger scale, Vargo etal. (2003a) also did not Þnd signiÞcant isolation bydistance. Isolation by distance is expected if dispersalis relatively limited over the spatial scale studied, ei-ther because of short range mating ßights by alates inthe absence of inbreeding avoidance and/or frequentcolony reproduction by budding. This lack of strongpopulation viscosity suggests that budding is not com-mon and that during mating ßights reproductives dis-perse relatively far over the spatial scale studied orthey actively engage in avoiding relatives when form-ing tandem pairs. Alternatively, no strong populationstructure is expected if human-mediated dispersalplays an important role in population expansion. Stud-ies of the spatial distribution of C. formosanus in NewOrleans (La Fage 1987) and Charleston (Chambers etal. 1988) indicate a rather patchy occurrence in theseareas, consistent with limited natural dispersal andfrequent spreading by human transport of infestedmaterials from one part of the city to another. Thus,the absence of strong local population structure inthese two older populations is not unexpected.

Results of the bottleneck tests provide strong evi-dence that the Charleston population underwent astrong genetic bottleneck on the introduction of C.formosanus to this area. The small number of allelespresent is consistent with this population originatingfrom very few colonies. Neither the tests for heterozy-gosity excess nor the mean ratio of the number ofalleles to the range in allele sizes detected signiÞcantbottlenecks in the City Park or Rutherford Countypopulations. The reason for the lack of detection inthese two recently introduced populations is unclear.It may be that the sample sizes were rather minimal forthese tests, especially in the Rutherford County pop-ulation, where the number of colonies (n � 8) wasbelow the minimum of 10 recommended for the het-erozygosity excess tests (Piry et al. 1999). Hussenederet al. (2005) recently detected a bottleneck in a pop-ulation of 14 C. formosanus colonies from Louis Arm-strong Park, New Orleans, using the test for heterozy-

gosity excess. However, a study of Japanesepopulations, where this species was introduced �300yr ago, did not detect a recent bottleneck (Vargo et al.2003a), presumably because there has been sufÞcienttime since the introduction to Japan to erase the het-erozygosity excess characteristic of recently bottle-necked populations (Piry et al. 1999).

The Charleston population, with a maximum ofthree alleles per polymorphic microsatellite locus andan average of fewer than two, is the least variablepopulation of C. formosanus so far studied. Even therelatively recent Rutherford County population withonly eight colonies had considerably more geneticvariability. Other populations that have been studiedinclude two Japanese populations with an average of6.3 and 2.7 alleles per locus, respectively (Vargo et al.2003a), Louis Armstrong Park, New Orleans, with 2.9alleles (Hussenederet al. 2005), andOahu,HI,with3.9alleles (C.H., E.L.V., and J.K.G., unpublished data). Tothe extent that allele number reßects the effective sizeof the founding population, Charleston seems to havebeen founded by the least number of colonies, possi-bly only a single colony. Alternatively, it could havebeen founded by several colonies introduced from apreviously bottlenecked population from Asia or else-where.

Another main Þnding of our study is the distinctintroduction history of the three studied populations.Results of the unweighted pair-group method witharithmetic average clustering indicated strong differ-ences between the Charleston and City Park popula-tions (FST � 0.30), Charleston and Louis ArmstrongPark population (FST � 0.24), and the Charleston andRutherford County populations (FST � 0.36), suggest-ing different introduction histories for Charleston andthe other three populations. These results conÞrmthose of an earlier study using allozymes suggesting atleast two separate introductions of C. formosanus tothe U.S. mainland (Korman and Pashley 1991, Wangand Grace 2000a, b). In contrast, City Park, LouisArmstrong Park, and Rutherford County were muchmore genetically similar (FST � 0.05Ð0.16), suggestinga close association between these three populations.However, by comparing the alleles present in each ofthese populations, we can exclude the possibility thatthe Rutherford County population could have origi-nated solely from the City Park population. This isbecause the alleles present in Rutherford Countywere not simply a subset of those in City Park. In fact,of the44 total allelespresent inbothpopulations, therewere 12 (27%) private alleles, 7 of which resided inRutherford County. Moreover, it is unlikely that theRutherford County population originated simply fromother populations in or around New Orleans, becausethe Louis Armstrong Park population closely resem-bles the City Park population. It is possible that theRutherford County population may have originatedfrom another nearby population from outside NewOrleans or may have resulted from two or more col-onies coming from different populations within theUnited States. Based on cytochrome oxidase II se-quence data, Jenkins et al. (2002) concluded that four


colonies from Atlanta, GA, likely originated from Lou-isiana, most probably having been transported by rail-road ties. A similar route of introduction was likely forthe Rutherford County population, although from anunknown location, because of the close associationbetween this small population and the abandonedrailroad line (Fig. 3).

The colonies studies here, as well as those in otherintroduced populations studied previously, are closefamily groups that retain their distinctness even in theface of high population densities. Thus, the invasionsuccess of C. formosanus cannot be attributed to abreakdown in nestmate recognition and an ensuingshift to unicoloniality in introduced populations, amajor factor associated with highly invasive ants (Hol-way et al. 2002, Tsutsui and Suarez 2003). Nor wouldit seem that the exceptional success of C. formosanusas an invader is caused by any one dominant form ofsocial organization, given the high variability in colonybreeding structure among introduced populations.Perhaps the invasion success of this species is relatedto the plasticity of the breeding structure allowingpopulations to adapt to local ecological conditions.However, it may be that this species has such a supe-rior competitive ability in many non-native areas be-cause of ecological release that it can invade success-fully despite the variability in breeding structure.Future studies investigating the breeding structure inadditional populations, particularly within the nativerange, and the nature of competitive interactions withrival species in both the native and introduced rangesshould shed more light on the factors underlying therepeated success of this important invader.

Acknowledgments

We thank T. Juba and M. Puente for valuable technicalassistance and C. DeHeer for both technical support andinsightful comments on the manuscript. E. Adams kindlyprovided the results of computer simulations of breedingsystems. J. McGuinn of the Rutherford County CooperativeExtension Center, C. Falco and the Structural Pest ControlDivision of the North Carolina Department of Agricultureand Consumer Services, D. Spillman of Goforth Termite andPest Control, and Thompson Brothers Exterminating helpedwith collecting samples in Rutherford County. We are grate-ful to the South Carolina Department of Parks, Recreation,and Tourism for permission to collect samples in CharlesTowne Landing State Historic Site. S. Fortson and S. Perry-man of Terminix assisted in collecting samples from Charles-ton. This work was funded by USDA T-STAR Program GrantsHAW00987Ð1011S and HAW00990Ð1013S to C.H., K.G., andE.L.V. and USDA NRICGP Grants 00-35301Ð9377 and 02-35302Ð12490) to E.L.V.

References Cited

Anonymous. 1965. Formosan subterranean termiteÑTexas.U.S. Dep. Agric. Coop. Econ. Insect Rep. 15: 907Ð908.

Atkinson, L., and E. S. Adams. 1997. The origins and relat-edness of multiple reproductives in colonies of the ter-mite Nasutitermes corniger. Proc. Roy. Soc. Lond. B. 264:1131Ð1136.

Buczkowski, G., E. L. Vargo, and J. Silverman. 2004. Thediminutive supercolony: the Argentine ants of the south-eastern United States. Mol. Ecol. 13: 2235Ð2242.

Bulmer,M. S., E. S. Adams, and J.F.A. Traniello. 2001. Vari-ation in colony structure in the subterranean termiteReticulitermes flavipes. Behav. Ecol. Sociobiol. 49: 236Ð243.

Chambers, D. M., P. A. Zungoli, and H.S.J. Hill. 1988. Dis-tribution and habitats of the Formosan subterranean ter-mite (Isoptera: Rhinotermitidae) in South Carolina. J.Econ. Entomol. 81: 1611Ð1619.

Clement, J.-L. 1981. Enzymatic polymorphism in the Euro-pean populations of various Reticulitermes species(Isoptera), pp. 49Ð62. In P. E. Howse and J.-L. Clement(eds.), Biosystematics of social insects. Academic, NewYork.

Clement, J.-L. 1984. Diagnostic alleles and systematics intermite species of the genus Reticulitermes in Europe.Experientia. 40: 283Ð285.

Clement, J.-L., A.-G. Bagneres, P. Uva, L. Wilfert, A. Quin-tana, J. Reinhard, and S. Dronnet. 2001. Biosystematicsof Reticulitermes termites in Europe: morphological,chemical and molecular data. Insectes Soc. 48: 202Ð215.

Cornuet, J.M., andG.Luikart. 1996. Description and poweranalysis of two tests for detecting recent population bot-tlenecks from allele frequency data. Genetics 144: 2001Ð2014.

DeHeer, C. J., and E. L. Vargo. 2004. Colony genetic orga-nization and colony fusion in the termite Reticulitermesflavipes as revealed by foraging patterns over time andspace. Mol. Ecol. 13: 431Ð441.

DeHeer, C. J., M. Kutnik, E. L. Vargo, and A.-G. Bagneres.2005. The breeding system and population structure ofthe termite Reticulitermes grassei in southern France.Heredity. 95: 408Ð415.

Dronnet, S., M. Chapuisat, E. L. Vargo, C. Louhou, andA.-G.Bagneres. 2005. Genetic analysis of the breeding systemof an invasive subterranean termite, Reticulitermes san-tonensis, in urban and natural habitats. Mol. Ecol. 14:1311Ð1320.

Fisher,M.L.,R.E.Gold,E.L.Vargo, andA. I.Cognato. 2004.Behavioral and genetic analysis of colony fusion in Re-ticulitermes flavipes (Isoptera: Rhinotermitidae). Socio-biology. 44: 565Ð576.

Garza, J. C., and E. G. Williamson. 2001. Detection of re-duction in population size using data from microsatelliteloci. Mol. Ecol. 10: 305Ð318.

Goodisman, M.A.D., and R. H. Crozier. 2002. Populationand colony genetic structure of the primitive termiteMastotermes darwiniensis. Evolution. 56: 70Ð83.

Goodisman, M.A.D., R. W. Matthews, and R. H. Crozier.2001. Hierarchical genetic structure of the introducedwaspVespula germanica in Australia. Mol. Ecol. 10: 1423Ð1432.

Goodnight, K. F., and D. C. Queller. 1999. Computer soft-ware for performing likelihood tests of pedigree relation-ship using genetic markers. Mol. Ecol. 8: 1231Ð1234.

Goudet, J. 2001. FSTAT, a program to estimate and test genediversities and Þxation indices (version 2.9.3). http://www.unil.ch/izea/softwares/fstat.html.

Grube, S., and B. T. Forschler. 2004. Census of monogyneand polygyne laboratory colonies illuminates dynamics ofpopulation growth in Reticulitermes flavipes (Isoptera:Rhinotermitidae). Ann. Entomol. Soc. Am. 97: 466Ð475.

Holway, D. A., L. Lach, A. V. Suarez, N. D. Tsutsui, and T. J.Case. 2002. The causes and consequences of ant inva-sions. Annu. Rev. Ecol. Syst. 33: 181Ð233.


Husseneder, C., and J. K. Grace. 2001. Similarity is relative:hierarchy of genetic similarities in the Formosan subter-ranean termite (Isoptera: Rhinotermitidae) in Hawaii.Environ. Entomol. 30: 262Ð266.

Husseneder, C., R. Brandl, C. Epplen, J. T. Epplen, and M.Kaib. 1998. Variation between and within colonies inthe termite: morphology, genomic DNA, and behaviour.Mol. Ecol. 7: 983Ð990.

Husseneder, C., M. Kaib, C. Epplen, J. T. Epplen, and R.Brandl. 1997. Small-scale population structure of thetermite Schedorhinotermes lamanianus: aggression mod-ulated by genetic and environmental factors. Mitt. Deut-sch. Ges. All. Ang. Entomol. 11: 183Ð187.

Husseneder, C., R. Brandl, C. Epplen, J. T. Epplen, and M.Kaib. 1999. Within-colony relatedness in a termite spe-cies: genetic roads to eusociality? Behaviour. 136: 1045Ð1063.

Husseneder, C., E. L. Vargo, and J. K. Grace. 2003. Molec-ular genetic methods: new approaches to termite biology,pp. 358Ð370. In B. Goodell, T. P. Schultz, and D. D.Nicholas (eds.), Wood deterioration and preservation:advances in our changing world. American Chemical So-ciety, Washington, DC.

Husseneder, C., M. T. Messenger, N.-Y. Su, J. K. Grace, andE. L. Vargo. 2005. Colony social organization and pop-ulation genetic structure of an introduced population ofthe Formosan subterranean termite from New Orleans,Louisiana. J. Econ. Entomol. 98: 1421Ð1434.

Jenkins, T.M., C. J. Basten, S. Kresovich, and B. T. Forschler.1999. Mitochondrial gene sequence questions Reticuli-termes sp. social structure (Isoptera: Rhinotermitidae).Sociobiology. 34: 161Ð172.

Jenkins, T. M., R. E. Dean, and B. T. Forschler. 2002. DNAtechnology, interstate commerce, and the likely origin ofFormosan subterranean termite (Isoptera: Rhinotermiti-dae) infestation in Atlanta, Georgia. J. Econ. Entomol. 95:381Ð389.

Johnson, R. N., and P. T. Starks. 2004. A surprising level ofgenetic diversity in an invasive wasp: Polistes dominulusin the northeastern United States. Ann. Entomol. Soc.Am. 97: 732Ð737.

Kaib, M., C. Husseneder, C. Epplen, J. T. Epplen, and R.Brandl. 1996. Kin-biased foraging in a termite. Proc.Roy. Soc. Lond. B. 263: 1527Ð1532.

King, E. G., andW. T. Spink. 1969. Foraging galleries of theFormosan subterranean termite,Coptotermes formosanus,in Louisiana. Ann. Entomol. Soc. Am. 62: 536Ð542.

Kistner, D. H. 1985. A new genus and species of termito-philous Aleocharinae from mainland China associatedwith Coptotermes formosanus and its zoogeographic sig-niÞcance (Coleoptera: Staphylinidae). Sociobiology. 10:93Ð104.

Kolbe, J. J., R. E. Glor, L. R. Schettino, A. C. Lara, A. Larson,and J. B. Losos. 2004. Genetic variation increases duringbiological invasion by a Cuban lizard. Nature (Lond.).431: 177Ð181.

Korman, A. K., and D. P. Pashley. 1991. Genetic compari-sons among U.S. populations of Formosan subterraneantermites. Sociobiology. 19: 41Ð50.

Kumar, S., K. Tamura, and I. B. Jakobsen. 2001. MEGA2:molecular evolutionary genetics analysis software. Bioin-formatics. 17: 1244Ð1245.

LaFage, J. P. 1987. Practical considerations of the Formosansubterranean termite in Louisiana: a 30-year-old problem,pp. 37Ð42. InM. Tamashiro and N.-Y. Su (eds.), Biologyand control of the formosan subterranean termite. Col-lege of Tropical Agriculture and Human Resources, Uni-versity of Hawaii, Honolulu, HI.

Lai, P. Y. 1977. Biology and ecology of the Formosan sub-terranean termite, Coptotermes formosanus, and its sus-ceptibility to the entomogenous fungi,BeauvariabassianaandMetarrhiziumanisopliae.Department of Entomology,University of Hawaii, Honolulu, HI.

Lewis, P. O., and D. Zaykin. 2000. Genetic data analysis:computer program for the analysis of allelic data. Version1.0 (d12). http://alleyn.eeb.uconn.edu/gda/.

Messenger,M. T., andN.-Y. Su. 2005. Colony characteristicsand seasonal activity of the Formosan subterranean ter-mite (Isoptera: Rhinotermitidae) in Louis ArmstrongPark, New Orleans, Louisiana. J. Entomol. Sci. 40: 268Ð279.

Moller, H. 1996. Lessons for invasion theory from socialinsects. Biol. Conserv. 78: 125Ð142.

Mooney, H. A., and E. E. Cleland. 2001. The evolutionaryimpact of invasive species. Proc. Natl. Acad. Sci. U.S.A. 98:5446Ð5451.

Myles, T. G. 1999. Review of secondary reproduction intermites (Insecta: Isoptera) with comments on its role intermite ecology and social evolution. Sociobiology. 33:1Ð87.

Payne, C. M., C. V. Tillberg, and A. V. Suarez. 2004. Rec-ognition systems and biological invasions. Ann. Zool.Fenn. 41: 843Ð858.

Pimentel, D., L. Lach, R. Zuniga, and D. Morrison. 2000.Environmental and economic costs of nonindigenousspecies in the United States. Bioscience. 50: 53Ð65.

Piry, S., G. Luikart, and J. M. Cornuet. 1999. BOTTLE-NECK: a computer program for detecting recent reduc-tions in the effective population size using allele fre-quency data. J. Hered. 90: 502Ð503.

Queller, D. C., and K. F. Goodnight. 1989. Estimating re-latedness using genetic markers. Evolution. 43: 258Ð275.

Raymond, M., and F. Rousset. 1995. GENEPOP (version1.2): population genetics software for exact tests andecumenicism. J. Hered. 86: 248Ð249.

Reilly,L.M. 1987. Measurements of inbreeding and averagerelatedness in a termite population. Am. Nat. 130: 339Ð349.

Ross, K. G. 2001. Molecular ecology of social behaviour:analyses of breeding systems and genetic structure. Mol.Ecol. 10: 265Ð284.

Shellman-Reeve, J. S. 1997. The spectrum of eusociality intermites, pp. 52Ð93. In J. C. Choe and B. J. Crespi (eds.),Social behavior in insects and arachnids. Cambridge Uni-versity Press, Cambridge, UK.

Spink, W. T. 1967. The Formosan subterranean termite inLouisiana. Louisiana State Univ. Circ. 89: 1Ð12.

Su, N.-Y. 1994. Field evaluation of a hexaßumuron bait forpopulation suppression of subterranean termites(Isoptera: Rhinotermitidae). J. Econ. Entomol. 87: 389Ð397.

Su, N.-Y. 2003. Overview of the global distribution and con-trol of the Formosan subterranean termite. Sociobiology.41: 7Ð16.

Su, N.-Y., and M. Tamashiro. 1987. An overview of the For-mosan subterranean termite (Isoptera: Rhinotermitidae)in the world, pp. 3Ð14. In M. Tamashiro and N.-Y. Su(eds.), Biology and control of the Formosan subterraneantermite. University of Hawaii, Honolulu, HI.

Su, N.-Y., and R. H. Scheffrahn. 1988. Foraging populationand territory of the Formosan subterranean termite(Isoptera: Rhinotermitidae) in an urban environment.Sociobiology. 14: 353Ð360.

Thompson,G. J., andP.D.N.Hebert. 1998a. Probing termitesocial systems through allozyme and mtDNA analysis: a


case study ofNasutitermes nigriceps andNasutitermes cos-talis (Isoptera, Termitidae). Insectes Soc. 45: 289Ð299.

Thompson, G. J., and P.D.N. Hebert. 1998b. Population ge-netic structure of the neotropical termite Nasutitermesnigriceps (Isoptera: Termitidae). Heredity. 80: 48Ð55.

Thorne, B. L., J.F.A. Traniello, E. S. Adams, and M. Bulmer.1999. Reproductive dynamics and colony structure ofsubterranean termites of the genus Reticulitermes(Isoptera Rhinotermitidae): a review of the evidencefrom behavioral, ecological and genetic studies. Ethol.Ecol. Evol. 11: 149Ð169.

Tsutsui, N. D., and A. V. Suarez. 2003. The colony structureand population biology of invasive ants. Conserv. Biol. 17:48Ð58.

Vargo, E. L. 2000. Polymorphism at trinucleotide microsat-ellite loci in the subterranean termite Reticulitermes fla-vipes. Mol. Ecol. 9: 817Ð820.

Vargo, E. L. 2003a. Hierarchical analysis of colony and pop-ulation genetic structure in the eastern subterranean ter-mite, Reticulitermes flavipes, using two classes of molec-ular markers. Evolution. 57: 2805Ð2818.

Vargo, E. L. 2003b. Genetic structure of Reticulitermes fla-vipes and R. virginicus (Isoptera: Rhinotermitidae) col-onies in an urban habitat and tracking of colonies fol-lowing treatment with hexaßumuron bait. Environ.Entomol. 32: 1271Ð1282.

Vargo,E.L., andG.Henderson. 2000. IdentiÞcation of poly-morphic microsatellite loci in the Formosan subterranean

termite Coptotermes formosanus Shiraki. Mol. Ecol. 9:1935Ð1938.

Vargo, E. L., C. Husseneder, and J. K. Grace. 2003a. Colonyand population genetic structure of the Formosan sub-terranean termite,Coptotermes formosanus, in Japan. Mol.Ecol. 12: 2599Ð2608.

Vargo, E. L., C. Husseneder, J. K. Grace, G. Henderson, andD. Ring. 2003b. Colony and population genetic struc-ture of Formosan subterranean termites from Hawaii andLouisiana. Sociobiology. 41: 67Ð69.

Vinson, S. B. 1986. Economic impact and control of socialinsects. Greenwood, Westport, CT.

Wang, J., and J. K.Grace. 2000a. Esterase differences amongCoptotermes formosanus (Isoptera: Rhinotermitidae)populations. Sociobiology. 36: 1Ð6.

Wang, J., and J. K. Grace. 2000b. Genetic relationship ofCoptotermes formosanus (Isoptera: Rhinotermitidae)populations from the United States and China. Sociobi-ology. 36: 7Ð19.

Williams, D. F. (ed.). 1994. Exotic ants. Biology, impact,and control of introduced species. Westview Press, Boul-der, CO.

Woodson, W. D., B. A. Wiltz, and A. R. Lax. 2001. Currentdistribution of the Formosan subterranean termite(Isoptera: Rhinotermitidae) in the United States. Socio-biology. 37: 661Ð671.

Received for publication 18May 2005; accepted 9 September2005.


m e genetic analysis of colony and population structure of three...

Documents