BIOLOGICAL CONTROLÐWEEDS Feeding and Oviposition Behavior of latipes Korotyaev (Coleoptera: ) and Its Predicted Effectiveness as a Biological Control Agent for Polygonum perfoliatum L. (Polygonales: Polygonaceae)

1 2 K. COLPETZER, J. HOUGH-GOLDSTEIN, K. R. HARKINS, AND M. T. SMITH

Delaware Agricultural Experiment Station, Department ofEntomology and WildlifeEcology, College ofAgriculture and Natural Resources, University ofDelaware, Newark, DE 19716Ð2160

Environ. Entomol. 33(4): 990Ð996 (2004) ABSTRACT Feeding and oviposition on different parts of mile-a-minute weed, Polygonum perfo- liatum L. (Polygonales: Polygonaceae), by Korotyaev (Coleoptera: Curcu- lionidae), a potential biological control agent for the weed, were studied in quarantine. An additional experiment was conducted to test the effects of different levels of simulated damage by R. latipes on P. perfoliatum growth, survival, and reproduction. Female weevils consumed more P. perfoliatum overall than males and selectively fed on capitula more than on ocreae or leaves, whereas males fed more on ocreae than on leaves or capitula. More eggs were also laid on capitula than on other plant parts. Female feeding preference is probably because of the high protein content of the capitula, because protein is required for continued egg production, whereas males may maximize their reproductive success by feeding low and close to P. perfoliatum stems to intercept females as they emerge from pupation in the soil and ascend the plants to feed. The feeding and oviposition preferences of female R. latipes for plant capitula suggest that host speciÞcity tests for this species should be conducted with plants that are ßowering. Damage that simulated the effect of R. latipes larval feeding caused plant mortality when it was initiated on small P. perfoliatum plants and reduced biomass and seed production when it was initiated on larger plants. Thus R. latipes could have a substantial impact on P. perfoliatum ifthe weevil is released into the weedÕs introduced range in North America.

KEY WORDS Rhinoncomimus latipes, mile-a-minute weed, Polygonum perfoliatum, feeding, simu- lated herbivory

PRACTITIONERS OF CLASSICAL BIOLOGICAL control of Study in quarantine ofthe insectÕs feedingand ovi- weeds are increasingly challenged to show the position behavior on the host plant can help predict its safety of the organisms they propose to import and potential impact on the plant. For example, Eustenopus their predicted effectiveness (DeClerck-Floate and villosus Boheman (Coleoptera: Curculionidae) feed Bourchier 2000, Kluge 2000, McEvoy and Coombs and oviposit on the capitulum (developing seed head) 2000, Louda et al. 2003). Although various methods for ofyellow starthistle ( Centaurea solstitialis L.; Forna- prerelease assessment have been proposed, Blossey sari et al. 1991). This is thought to be an effective (1995) concluded that such methods did not effec- control agent because its feeding and oviposition dis- tively allow for selection of the most successful agents torts C. solstitialis buds and reduces seed production before release, even for insect species that had been (Fornasari et al. 1991, Fornasari and Sobhian 1993, extensively studied in their European home range Connett et al. 2001). Study ofthe herbivoreÕs pre- before release in North America. For alien plants that ferred feeding and oviposition site can also help with originate in Asia, the lack ofeasily accessible literature the design ofan appropriate postrelease monitoring and research infrastructure has often hampered such strategy by suggesting where the insect is most likely extensive Þeld studies in the native range. Neverthe- to be found and may also help ensure the proper less, several approaches may help assess potential ef- evaluation ofhost speciÞcity. For example, ifan agent fectiveness of new agents before release in a new area. only oviposits on the ßowers ofits host or only ovi- posits when its host is ßowering, host speciÞcity tests should be conducted with plants that are ßowering. 1 E-mail: [email protected]. 2 United States Department ofAgricultureÐAgricultural Research Both egg maturation and oviposition behavior can Service BeneÞcial Insect Introduction Research Unit, Newark, DE vary substantially depending on host availability and 19716. quality (Papaj 2000).

0046-225X/04/0990Ð0996$04.00/0 2004 Entomological Society ofAmerica August 2004 COLPETZER ET AL.: R. latipes ON P. perfoliatum 991

Experimental simulation ofthe agentÕs observed plete development, exit the stem, and drop to the soil mode ofdamage to the host plant is another possible for pupation. Damage to the plant occurs primarily approach to assessing potential effectiveness. For ex- from larval feeding, which kills the stem from the exit ample, McEvoy et al. (1991) used simulated herbivory hole to the stem terminal (Price et al. 2003). In China, as part ofan assessment ofthe impact ofcinnabar moth adult weevils overwinter in litter and emerge in the (Tyria jacobaeae L., Lepidoptera: Arctiidae) feeding spring when P. perfoliatum is 30Ð50 cm tall. The weevil on tansy ragwort (Senecio jacobeae L.). Simulated is thought to be univoltine in northern China, but may damage can suggest the plantÕs sensitivity to the type have multiple generations further south (D. Jianqing, ofdamage inßicted by the proposed agent, although personal communication). the effects of mechanical damage may differ from actual herbivory (Baldwin 1990), and damage to in- Materials and Methods dividual plants may not translate directly to impacts on plant populations. TifÞn and Inouye (2000) compared Rhinoncomimus latipes Feeding and Oviposition estimates ofplant damage fromnatural versus artiÞcial Behavior. Rhinoncomimus latipes originating from or manipulated herbivory and concluded that esti- adults collected in Changsha, Hunan province of mates derived from simulated herbivory are likely to China in 2000 and 2001 were reared on P. perfoliatum be unbiased, but less precise, than those from natural in a quarantine room kept at 22 Ϯ 0.5ЊC at the USDA- herbivory. ARS BeneÞcial Introduction Research Unit in In this study, feeding and oviposition on different Newark, DE, using the methods ofPrice et al. (2003). parts ofmile-a-minute weed, Polygonum perfoliatum Voucher specimens of R. latipes have been deposited L. (Polygonales: Polygonaceae), by Rhinoncomimus in the University ofDelaware Department ofEnto- latipes Korotyaev (Coleoptera: Curculionidae) were mology and Wildlife Ecology Insect Reference Col- studied in quarantine. In addition, an experiment was lection. conducted to test the effect of different levels of sim- Polygonum perfoliatum plants were grown in a ulated damage by R. latipes on P. perfoliatum growth, greenhouse from seeds collected locally and germi- survival, and reproduction. nated using the methods ofColpetzer and Hough- Polygonum perfoliatum is an annual vine indigenous Goldstein (2004). Plants were supplied with drip ir- to temperate regions ofBhutan, China, India, Indo- rigation and supplemental light that provided a 16-h nesia, Japan, Korea, Nepal, the Philippines, Russia, and photophase and a minimum photosynthetic photon Vietnam (Wu and Raven 2003). It was introduced into ßux of200 ␮mol/m/s. They were fertilized every other a nursery in Stewartstown, PA, during the 1930s (Moul week with 250 ml ofa 0.4% solution ofMiracle-Gro 1948, Riefner 1982), and has since spread to Delaware, Professional (21% N; 5% P; 20% K). the District ofColumbia, Maryland, New Jersey, New Ten newly emerged male and 10 newly emerged York, Ohio, Virginia, and West Virginia (Oliver 1996), female R. latipes adults were placed individually in and more recently, Connecticut (Lamont and Fitzger- plastic cylindrical cages (15 cm diameter by 20 cm ald 2001). It has been listed as a noxious weed in high) with removable screen covers. The weevils were several states because it forms dense, prickly thickets sexed by assessing the shape ofthe metasternum, that displace native vegetation, interfere with forest which is more convex in the female than in the male regeneration, and reduce the recreational value of (Price et al. 2003). Once oviposition began, females natural areas (Mountain 1989, McCormick and were marked with white-out (Sailor Corporation of Hartwig 1995, Oliver 1996). America, Fayetteville, GA) on their elytra to facilitate The USDA Forest Service initiated a classical bio- sorting following mating. logical control program for P. perfoliatum in 1996. Each container was supplied with a freshly cut During 6 yr ofÞeld surveys in China, over 100 insect P. perfoliatum stem in a 125-ml Erlenmeyer ßask Þlled species were collected from P. perfoliatum (Ding et al. with water and stopped with cotton. All stems were 20 2004). One ofthe most promising, both in terms of cm long, with a single capitulum (Fig. 1), and were observed impact on the plant in China and host- taken from 1.5- to 2-mo-old plants. Stems were re- speciÞcity (Price et al. 2003, Colpetzer et al. 2004), placed every3dfor63d.When plant material was was Rhinoncomimus latipes Korotyaev. This weevil changed, all 20 R. latipes were placed in a 9-cm petri was initially misidentiÞed as Homorosoma chinensis dish and allowed to mate for Ϸ2h. (Wagner) (synonym: Homorosoma chinense Wagner). Each time plant material was changed, feeding dam- Specimens were later sent to B. A. Korotyaev ofthe age was measured on ocreae (saucer-shaped leafy Zoological Institute ofthe Russian Academy ofSci- structures that encircle the stem at each leafnode), ences, who identiÞed them as R. latipes Korotyaev leaves, and capitula (including the three terminal (Korotyaev 1997). Field release ofthis weevil in North ocreae and two terminal leaves associated with ßowers America has been recommended by the Technical or fruits: Fig. 1). Although some feeding by females Advisory Group for Biological Control Agents of was observed on ßowers and fruits, this was not quan- Weeds, and an application for a release permit is cur- tiÞed. Area consumed was measured by pressing the rently under consideration by the USDA APHIS-PPQ. individual plant parts, photocopying them onto trans- Adult R. latipes are Ϸ2 mm long, eat young leaves of parencies, and processing the transparencies through P. perfoliatum, and lay eggs on leaves and stems. After a portable leafarea meter (LI-3000A; LI-COR, Lin- hatching, larvae bore into the stem, where they com- coln, NE) twice before and twice after Þlling in the 992 ENVIRONMENTAL ENTOMOLOGY Vol. 33, no. 4

Simulated Damage to P. perfoliatum. Polygonum perfoliatum plants were grown from seed collected locally in September 2001 and germinated using the methods ofColpetzer and Hough-Goldstein (2004). Germinated seeds were planted in Promix in plastic pots on 12 April 2002, and placed in a greenhouse under drip irrigation as described above. On 7 May, 52 P. perfoliatum plants, 30Ð50 cm tall, were transplanted into a 12 by 39 m untilled Þeld (planted in corn the previous year) at the University ofDelaware Experiment Station Farm, Newark, DE. The Þeld was divided into 3 by 3-m plots, and a single plant was transplanted into the center ofeach plot. Plots were arranged in a randomized complete block Fig. 1. Aerial structures of P. perfoliatum. design with four replications and 13 treatments, in- cluding the control. Several seedlings failed to estab- area eaten using a permanent marker. Eggs found on lish and were replaced with plants from the green- capitula, stems, leaves, and ocreae were also counted. house before treatments were applied. To determine whether male R. latipes emerged from Treatments consisted ofclipping plants with dis- pupation before females, three male and three female section scissors, beginning when plants were small (15 R. latipes adults were put in each offourplastic cy- leaves), medium-sized (60Ð90 leaves), or large (240 lindrical cages containing 5 cm ofmoist vermiculite. leaves). For each size class, damage by one to four Each cage was supplied with a 30-cm length of P. female per plant was simulated. In the labo- perfoliatum stem, still attached to a potted plant. After ratory, R. latipes eggs are laid singly or in small clusters 24 h, and on 4 consecutive d, the weevils were moved at a rate of Ϸ3 eggs/d, and newly hatched larvae crawl to new cages with new stems. All containers were held down the stem and bore into the Þrst unoccupied without weevils to allow eggs that had been laid on the node. Larval feeding then causes death of the stem stems to hatch, larvae to develop, and pupation to from that node apically to the stem terminal (Price et occur in the vermiculite. After 20 d, P. perfoliatum al. 2003). Therefore, damage by one was sim- stems were removed from the containers, which were ulated by clipping three nodes from a stem terminal then monitored daily for emerging adults. The sex of each day, two beetles by clipping six nodes, etc. Only newly emerged adults was determined as described the shoot that had the greatest length from the base of above. the plant to the apex was clipped each day. Clipping The amount of P. perfoliatum consumed by R. latipes continued for each plant until that plant died or until was log (x ϩ 1) transformed and analyzed using a 6 August, when the experiment was terminated. All repeated measures two-way analysis ofvariance seeds produced before 6 August were collected into (ANOVA) by sex and plant part, with cages as the marked paper bags and left to dry in a ventilated repeated measure. The transformation did not com- greenhouse. On 6 August, all remaining P. perfoliatum pletely normalize the data, but skewness and kurtosis seeds, stems, and leaves were collected, allowed to moments (1.44 and 2.04, respectively) were within dry, and subsequently weighed. acceptable limits, and the mean was within the inter- Mean plant survival times from planting to a max- quartile range. A one-way repeated measures ANOVA imum of91 d (when the experiment ended) were with sex and plant part combined and with nominated analyzed using a two-way ANOVA by block and treat- posterior contrasts was used for mean separation. ment followed by TukeyÕs test for mean separation. Number ofeggs deposited on P. perfoliatum during the Seed number and plant biomass data were not nor- period that females were ovipositing (days 9Ð63) was mally distributed and did not have homogeneous vari- analyzed using PROC GENMOD with a negative bi- ances even when transformed; therefore, these data nomial probability distribution, type III contrasts, log were analyzed using the nonparametric Kruskal- linkages, and cages as the repeated measure (SAS Wallis test (PROC GLM on ranks; SAS Institute 1999), Institute 1999). The algorithm converged, and data with mean separation based on least signiÞcant dif- were properly dispersed (␹2 ϭ 656.4632 and ␹2/df ϭ ference (LSD). Untransformed means and SEs are 0.8683). This procedure produced a test similar to a shown in the Þgures. one-way repeated measures ANOVA by plant part but did not require the errors to be independent, have Results equal variances, or be normally distributed (Orelien 2001). Nominated posterior contrasts were employed Rhinoncomimus latipes Feeding and Oviposition for mean separation. The relationships between feed- Behavior. The total amount offoliageconsumed per ing location and weevil age and oviposition location weevil per 3 d averaged 0.39 Ϯ 0.03 cm2 (SEM) for and weevil age were subjected to regression analysis females and 0.23 Ϯ 0.03 cm2 for males. Females con- (PROC REG; SAS Institute 1999). Mean emergence sumed signiÞcantly more P. perfoliatum overall (N ϭ times ofmale and femaleadult R. latipes were com- 20, df ϭ 1, F ϭ 74.71, P Ͻ 0.0001) and signiÞcantly more pared using a two-sample t-test. capitula (Fig. 2) than males. August 2004 COLPETZER ET AL.: R. latipes ON P. perfoliatum 993

Fig. 2. Rhinoncomimus latipes male and female consump- tion (mean Ϯ SEM) of different parts of P. perfoliatum. Fig. 4. Rhinoncomimus latipes oviposition (mean Ϯ Means sharing the same letters are not signiÞcantly different SEM) on different parts of P. perfoliatum. Means sharing the (ANOVA, type III contrasts). same letters are not signiÞcantly different (PROC GEN- MOD, type III contrasts).

Females consumed more capitula than ocreae or declined and capitula feeding increased slightly leaves (Fig. 2), and capitula consumption increased (Fig. 3B). signiÞcantly with age, whereas leaffeedingdeclined Females laid more eggs on P. perfoliatum capitula (Fig. 3A). Males consumed more ocreae than leaves or than on any other plant part (N ϭ 10, df ϭ 3, ␹2 ϭ 9.82, capitula (Fig. 2). Male preference for feeding on P ϭ 0.0202; Fig. 4), and this preference became more ocreae increased with age, whereas leaffeeding pronounced as the weevils aged (Fig. 5). After the Þrst 9 d, mean egg production ßuctuated between 6 and 12 eggs per female per3dfortheremainder ofthe experiment (Fig. 6). Male R. latipes emerged from pupation Ϸ2 d before females, with time to emergence averaging 25.1 Ϯ 0.1 d (N ϭ 28) for males and 26.8 Ϯ 0.1d(N ϭ 30) for females (t ϭ 46.65, P Ͻ 0.0001). Simulated Damage to P. perfoliatum. All plants that were clipped to simulate R. latipes damage beginning when they were small died by 6 August, whereas only one ofthe large clipped plants died beforethe exper- iment was terminated. Mortality ofmedium-sized plants ranged from 25% at the lowest level of damage to 100% at the highest level. Small clipped plants sur- vived for an average of 15Ð21 d, whereas medium-sized

Fig. 3. Female (A) and male (B) R. latipes consumption of P. perfoliatum over time. For females on capitula, Y ϭ 0.01160 ϩ 0.00497X (R2 ϭ 0.5378; F ϭ 22.11; P ϭ 0.0002); on ocreae, Y ϭ 0.16495 Ϫ 0.00141X (R2 ϭ 0.1074; F ϭ 2.29; P ϭ Fig. 5. Rhinoncomimus latipes oviposition on P. perfolia- 0.1470); and on leaves, Y ϭ 0.15953 Ϫ 0.00176X (R2 ϭ 0.2277; tum over time. On capitula, Y ϭ 0.78810 ϩ 0.09459X (R2 ϭ F ϭ 5.60; P ϭ 0.0287). For males on capitula, Y ϭ 0.00002 ϩ 0.6258; F ϭ 31.77; P Ͻ 0.0001); on leaves, Y ϭ 2.26286 Ϫ 0.00072X (R2 ϭ 0.2528; F ϭ 6.43; P ϭ 0.0202); on ocreae, Y ϭ 0.00710X (R2 ϭ 0.0060; F ϭ 0.11; P ϭ 0.7396); on ocreae, Y ϭ 0.04260 ϩ 0.00258X (R2 ϭ 0.5385; F ϭ 22.17; P ϭ 0.0002); and 1.16619 Ϫ 0.00879X (R2 ϭ 0.0305; F ϭ 0.60; P ϭ 0.4488); and on leaves, Y ϭ 0.15044 Ϫ 0.00195X (R2 ϭ 0.2908; F ϭ 7.79; P ϭ on stems, Y ϭ 0.34762 ϩ 0.00216X (R2 ϭ 0.0387; F ϭ 0.77; P ϭ 0.0116). ▫, capitula; E, ocreae; ‚, leaves. 0.3952). , capitula; ‚, leaves; E, ocreae; , stems. 994 ENVIRONMENTAL ENTOMOLOGY Vol. 33, no. 4

Fig. 6. Rhinoncomimus latipes age-speciÞc fecundity (mean Ϯ SEM). plants survived for 51Ð91 d, depending on level of damage (Fig. 7). Fig. 8. Plant biomass (A) and seed production (B) on 6 The mean amount ofbiomass produced by large August 2002 for undamaged (control) and P. perfoliatum plants subjected to four levels of simulated R. latipes damage plants clipped to simulate damage by one to four (1Ð4 beetles/plant) beginning when medium-sized or large. beetles, and medium-sized plants clipped to simulate Means sharing the same letters are not signiÞcantly different damage by one beetle, did not differ from the control (Kruskal-Wallis test, LSD). (Fig. 8A), but the medium-sized plants damaged at the two-, three-, and four-beetle levels produced signiÞ- cheap to produce, whereas ova are large and relatively cantly less biomass than the control plants. The num- costly to produce, females are expected to consume ber ofseeds produced by large plants at any damage more resources than males while attempting to max- level did not differ signiÞcantly from the number pro- imize their Þtness (Alcock 2001). duced by the control plants (Fig. 8B), but plants dam- Males and females fed selectively on different plant aged at any level beginning when they were medium- parts, with females feeding more on capitula and males sized produced signiÞcantly fewer seeds than the feeding more on ocreae and leaves. Female selective control. Small plants are not included in Fig. 8 because feeding on capitula, which increased with age, may be they all died before 6 August and produced no biomass an attempt to acquire sufÞcient protein for oviposi- or seed. tion, especially once initial stores have been depleted. Production ofova requires a substantial amount of Discussion protein (Nation 2002), and pollen is a good protein source (Roulston et al. 2000). The overall greater consumption of P. perfoliatum Male preference for ocreae and leaves may be a by female than by male R. latipes is predicted by mechanism to increase mating opportunities. R. latipes anisogamy, the differential investment in gametes by pupate in the soil, and adults emerge and ascend P. the sexes. Because sperm are small and relatively perfoliatum stems to feed. Because males emerge from pupation Ϸ2 d before females, they may maximize their reproductive success by feeding low on P. per- foliatum and close to the stems (i.e., on ocreae) while waiting for females to emerge. In our experiment, males and females were held individually on plants (but mixed together for mating every 3 d), and it is possible that their feeding site selections may differ when the sexes are present on plants together. Oviposition primarily on P. perfoliatum capitula may result in part from the femalesÕ preference for feeding on capitula, but probably also serves to increase larval survival. Apical portions ofthe stems are tenderer than basal portions, and numerous larvae were observed Fig. 7. Mean survival time of P. perfoliatum plants sub- during rearing that seemed to have died because they jected to four levels of simulated R. latipes damage (1Ð4 were unable to penetrate the semiwoody lower por- beetles/plant) beginning when small, medium-sized, or tion ofa P. perfoliatum stem. large. Means sharing the same letters are not signiÞcantly The feeding and oviposition preferences of female different (ANOVA, TukeyÕs test). R. latipes for plant capitula suggest that host speciÞcity August 2004 COLPETZER ET AL.: R. latipes ON P. perfoliatum 995 tests should be conducted with plants that are ßow- Blossey, B. 1995. A comparison ofvarious approaches for ering, at least for assessing maximum potential weevil evaluating potential biological control agents using in- success. Also, ifthe weevil is introduced into North sects on Lythrum salicaria. Biol. Control. 5: 113Ð122. America, postrelease monitoring for the insects could Colpetzer, K., and J. Hough-Goldstein. 2004. A rapid ger- focus on plant terminals, where adult female feeding, mination protocol for mile-a-minute weed, Polygonum eggs, and larval infestation will be concentrated. perfoliatum L. Seed Sci. Technol. (in press). The results ofthe simulated damage study suggest Colpetzer, K., J. Hough-Goldstein, J. Ding, and W. Fu. 2004. that R. latipes could cause substantial seedling mor- Host speciÞcity ofthe Asian weevil, Rhinoncomimus lati- pes Korotyaev (Coleoptera: Curculionidae), a potential tality in P. perfoliatum ifplants are attacked when biological control agent ofmile-a-minute weed, Polygo- small, and reduced seed production ifsomewhat larger num perfoliatum L. (Polygonales: Polygonaceae). Biol. plants are infested. The relative phenology of the plant Control. 30: 511Ð522. and the insect in North America is difÞcult to predict Connett, J. F., L. M. Wilson, J. P. McCaffrey, and B. L. before release, and thus, the extent to which such Harmon. 2001. Phenological synchrony of Eustenopus damage will occur is not known. Although the accu- villosus (Coleoptera: Curculionidae) with Centaurea sol- racy ofplant clipping in simulating weevil damage has stitialis in Idaho. Environ. Entomol. 30: 439Ð442. not been validated, plant mortality and reduced seed DeClerck-Floate, R, and R. S. Bourchier. 2000. Ecological production were also observed on potted plants dur- principles ofbiological control: frompopulation theory to ing weevil rearing. Attempts to rear R. latipes on pot- weed biocontrol practice, pp. 517Ð520. In N. R. Spencer ted plants without conÞning them to stem apices were (ed.), Proceedings X international symposium on biolog- unsuccessful, because as few as three pairs of adults ical control ofweeds. Montana State University, Boze- and their progeny killed the plants before any larvae man, MT. left the stems to pupate. When R. latipes were con- Ding, J., W. Fu, R. Reardon, Y. Wu, and G. Zhang. 2004. Þned to stem apices for oviposition (and moved to new Exploratory survey in China for potential insect biocon- Polygonum perfoliatum terminals and new plants as needed), an average of trol agents ofmile-a-minute weed, L., in eastern USA. Biol. Control. 30: 487Ð495. four pairs of weevils and their progeny did not kill Fornasari, L., and R. Sobhian. 1993. Life history of Eusteno- plants, but the plants rarely developed seeds, whereas pus villosus (Coleoptera: Curculionidae), a promising bi- seeds were removed weekly from plants of similar size ological control agent for yellow starthistle. Environ. En- and age in the greenhouse. tomol. 22: 684Ð692. Polygonum perfoliatum is an annual species with a Fornasari, L., C. E. Turner, and L. A. Andres. 1991. Eusteno- persistent seed bank (Van Clefand Stiles 2001), and pus villosus (Coleoptera: Curculionidae) for biological therefore reduction in P. perfoliatum populations control ofyellow starthistle (Asteraceae: Cardueae) in solely through reduced seed production will be slow. North America. Environ. Entomol. 20: 1187Ð1194. Predicting the extent to which combined seedling Kluge, R. L. 2000. The futureofbiological control ofweeds mortality and reduced seed production will impact the with insects: no more ÔparanoiaÕ, no more ÔhoneymoonÕ, weedÕs population will require additional study ofthe pp. 459Ð467. In N. R. Spencer (ed.), Proceedings X in- life history and ecology of P. perfoliatum in its intro- ternational symposium on biological control ofweeds. duced range. The results ofthis study suggest that R. Montana State University, Bozeman, MT. latipes could have a substantial impact ifintroduced Korotyaev, B. A. 1997. New and little known species ofwee- into North America for control of P. perfoliatum. vils from East Asia (Coleoptera: Apionidae, Curculion- idae). Zoosystematica Rossica. 5: 285Ð288. Lamont, E. E., and J. M. Fitzgerald. 2001. Noteworthy plants reported from the Torrey RangeÑ2000. J. Torrey Acknowledgments Bot. Soc. 128: 409Ð414. Louda, S. M., R. W. Pemberton, M. T. Johnson, and P. A. We thank R. C. Reardon and D. W. Tallamy for valuable Follett. 2003. Nontarget effectsÑthe AchillesÕ heel of consultation and guidance throughout the study and P. Tobin biological control? Retrospective analyses to reduce risk for statistical advice. We also thank R. Fuester (director of associated with biocontrol introductions. Annu. Rev. En- the USDA-ARS BeneÞcial Insects Introduction Research tomol. 48: 365Ð396. Unit, Newark, DE) for Þnding space for us in their quarantine facility and D. Jianqing (Biological Control Institute, Chinese McCormick, L., and N. Hartwig. 1995. Control ofthe nox- Academy ofAgricultural Sciences, Beijing, China) forcol- ious weed mile-a-minute (Polygonum perfoliatum)inre- lecting and shipping R. latipes to Delaware. This study was forestation. Northern J. Appl. Forestry. 12: 127Ð132. supported in part by a Forest Health Technology Enterprise McEvoy, P. B., and E. M. Coombs. 2000. Why things bite Team, USDA Forest Service, Cooperative Agreement with back: unintended consequences ofbiological weed con- JHG, and a University ofDelaware undergraduate research trol, pp. 167Ð194. In P. A. Follett and J. J. Duan (eds.), grant (Science and Engineering Scholars Program) to K.R.H. Nontarget effects of biological control. Kluwer, Boston, MA. McEvoy, P., C. Cox, and E. Coombs. 1991. Successful bio- logical control ofragwort, Senecio jacobaea, by introduced References Cited insects in Oregon. Ecol. Appl. 1: 430Ð442. Alcock, J. 2001. The evolution ofreproductive behavior, pp. Moul, E. 1948. A dangerous weedy Polygonum in Pennsyl- 317Ð356. In behavior: an evolutionary approach, vania. Rhodora. 50: 64Ð66. 7th ed. Sinauer, Sunderland, MA. Mountain, W. L. 1989. Mile-a-minute (Polygonum perfolia- Baldwin, I. T. 1990. Herbivory simulations in ecological re- tum L.): updateÑdistribution, biology, and control sug- search. Trends Ecol. Evol. 5: 91Ð93. gestions. Regulatory Hortic. 15: 21Ð24. 996 ENVIRONMENTAL ENTOMOLOGY Vol. 33, no. 4

Nation, J. L. 2002. Nutrients for oogenesis, pp. 429Ð430. In Roulston, R. H., J. H. Cane, and S. L. Buchmann. 2000. What Insect physiology and biochemistry. CRC, Boca Raton, governs protein content ofpollen: pollinator preferences, FL. pollen-pistil interactions, or phylogeny? Ecol. Monogr. Oliver, J. D. 1996. Mile-a-minute weed (Polygonum perfo- 70: 617Ð643. liatum L.), an invasive vine in natural and disturbed sites. SAS Institute. 1999. SAS/STAT userÕs manual, version 8.0. Castanea. 61: 244Ð251. SAS Institute, Cary, NC. Orelien, J. G. 2001. Model Þtting in PROC GENMOD, Tiffin, P, and B. D. Inouye. 2000. Measuring tolerance to Paper 264-26. In SUGI 26 proceedings. SAS Institute, herbivory: accuracy and precision ofestimates made us- Cary, NC. Available: www2.sas.com/proceedings/sugi26/ ing natural versus imposed damage. Evolution. 54: 1024Ð proceed.pdf. 1029. Papaj, D. R. 2000. Ovarian dynamics and host use. Annu. Van Clef, M., and E. W. Stiles. 2001. Seed longevity in three Rev. Entomol. 45: 423Ð448. pairs ofnative and non-native congeners: assessing inva- Price, D. L., J. Hough-Goldstein, and M. T. Smith. 2003. sive potential. Northeastern Naturalist. 8: 301Ð310. Biology, rearing and preliminary evaluation ofhost range Wu, Z., and P. H. Raven (eds.). 2004. Flora ofChina, vol. 5 oftwo potential biological control agents ofmile-a- (Ulmaceae through Basellaceae). Science Press, Beijing, minute weed, Polygonum perfoliatum L. Environ. Ento- China. mol. 32: 229Ð236. Riefner, R. 1982. Studies ofthe Maryland ßora VIII: range extension of Polygonum perfoliatum L., with notes on introduction and dispersal in North America. Phytologia. 50: 152Ð159. Received 9 July 2003; accepted 13 May 2004.