Use of Risk Analysis for Screening Weed Biocontrol Agents: Altica carduorum Guer. (Coleoptera: Chrysomelidae) from China as a Biocontrol Agent of Cirsium arvense (L.) Scop. in North America

Candidate weed biocontrol agents must be screened to exclude those that could threaten desirable plants . Traditionally , this has been done by rejecting species that develop on economically important plants in laboratory no - choice tests . However , because congeneric plants often support development in these tests , even when they are not utilized in nature , the tests do not meet legislated requirements for rare plant species or the increasing public concern for native plants . Plant suitability for larval development is a poor predictor of host range because insects use a sequence of steps in which the early steps , such as host finding and acceptance for oviposition , tend to be stronger than the later ones , such as suitability for development . This study is a trial of a new approach to screening insects as weed biocontrol agents that uses risk analysis to quantify the suitability of a plant as a host on the basis of inset performance at various stages in its life cycle . The insects used for the study was a NW Chinese biotype of the leaf beetle , Altica carduorum, which in terms of climate adaptation and damage it inflicts on the weedy thistle Cirsium arvense, is a promising biocontrol agent for Canada . However , its ability to develop on all North American Cirsium spp . in laboratory no - choice tests currently excludes its release . We show by risk and factor analyses , with five sequential host - selection parameters , that the suitability of these thistles to A. carduorum is so low that the beetle would not jeopardize the continued existence of rare native thistles , so its release should not be a problem .     

Evaluation of host range and larval feeding impact of Chrysolina aurichalcea asclepiadis (Villa): considerations for biological control of Vincetoxicum in North America

A biological control program has been initiated against European swallow-worts Vincetoxicum nigrum (L.) Moench. and V. rossicum (Kleopow) Barbar., which are invasive in North America. A population of the leaf beetle Chrysolina aurichalcea asclepiadis (Villa) originating from the western Alps has been under evaluation as a part of this program. The preliminary host range of C. a. asclepiadis was determined among 37 potential host plants. In addition, a prerelease impact study was conducted to determine the effect of larval feeding on the performance of V. nigrum. Under no-choice conditions beetle larvae completed development on nine plant species within the genera Artemisia and Tanacetum (Asteraceae) and Asclepias and Vincetoxicum (Apocynaceae). The host range of adults is broader than larvae (13 plant species within five genera received sustained feeding). Three of the six nontarget species supporting larval development are native to North America, however in separate oviposition tests, female beetles failed to produce eggs when confined to these hosts. In multiple-choice tests, neither larvae nor adults preferred Vincetoxicum spp. to nontarget species. Larval damage by C. a. asclepiadis at densities at and above five larvae per plant substantially reduced growth, biomass, and delayed reproduction of V. nigrum. However, this population of C. a. asclepiadis is polyphagous and unsuitable for biological control of Vincetoxicum because of potential risk of attack to Asclepias tuberosa L. and native North American Asteraceae, particularly Artemisia.     

Individual variation in two host plants of the ladybird beetle,Epilachna pustulosa (Coleoptera: Coccinellidae)

Individual variation in two species of host plants (thistle,Cirsium kamtschaticum, and blue cohosh,Caulophyllum robustum) of the herbivorous ladybird beetleEpilachna pustulosa was examined under laboratory conditions for their acceptability to adult beetles as a food resource, for adult preference and for larval performance. When clones of these plants were subjected to non-choice tests using posthibernating female beetles, there was found to be significant intraspecific variation among clones in terms of their acceptability, but interspecific variation was not detected. Significant intraspecific as well as interspecific variation were frequently detected in the two host plants when clones of these plants were subjected to choice tests using posthibernating female beetles; the magnitude of interspecific plant variation for beetle preference is not necessarily larger than that of intraspecific plant variation. Individual variation across plant species with respect to beetle larval performance was also significant. A positive correlation between adult preference and larval performance is suggested across the two taxonomically remote host plant species, thistle and blue cohosh, although this needs further investigation.     

Preference-performance linkage in a herbivorous lady beetle: consequences of variability of natural enemies

We investigated the relationship between oviposition preference and offspring performance in a herbivorous lady beetle Epilachna pustulosa on two co-occurring plant species, thistle Cirsium kamtschaticum and blue cohosh Caulophyllum robustum, in 1994 and 1995. The relative importance of bottom-up effects by host plants and top-down effects by natural enemies on offspring performance were determined using field and laboratory experiments. In both years, egg density on blue cohosh was significantly higher than on thistle. A laboratory experiment demonstrated that larval survival from hatching to adult emergence was significantly higher, and developmental period shorter when larvae were reared on blue cohosh compared to thistle. The positive preference-performance linkage varied between years in the field. Top-down effects had a different impact on larval survival on the two host plant species. Arthropod predators, a lady beetle Harmonia axyridis and an earwig Forficula mikado, considerably depressed immature survival on thistle, while they were negligible on blue cohosh. Although the lack of effective predation increased larval survival on blue cohosh, it led to defoliation due to increased larval feeding late in the season. Because of severe intraspecific competition, old larvae had significantly lower survival on blue cohosh than on thistle. In 1994, as larval survival decreased due to defoliation on blue cohosh, the overall survival rate was significantly higher on thistle than on blue cohosh. This survival pattern was opposite to that found in the laboratory experiment. In contrast, in 1995, the increase in predatory lady beetles on thistle caused greater larval mortality. Thus, the overall survival was significantly lower on thistle than on blue cohosh, although severe intraspecific competition occurred on blue cohosh as it had in 1994. Consequently, the offspring performance on the two host plants is largely determined by the relative importance of arthropod predation determining larval survival on thistle and host plant defoliation reducing late larval survival on blue cohosh. These results indicate the important role of spatial and temporal variability of natural enemies on the preference-performance linkage of herbivorous insects. Received: 19 August 1998 / Accepted: 11 January 1999     

Effect of Herbivore Density, Timing of Attack and Plant Community on Performance of Creeping Thistle Cirsium arvense (L.) Scop. (Asteraceae)

Creeping thistle, Cirsium arvense (L.) Scop. (Asteraceae), is one of the most serious weeds in ecological compensation areas (within-field or border refugia) in Europe. Since conventional weed control measures are restricted in compensation areas, augmenting indigenous agents for biological control of the weed may be a feasible alternative. In this paper, we studied the effect of density of the shield beetle, Cassida rubiginosa Muller (Coleoptera, Chrysomelidae), and two vegetation types typical for ecological compensation areas, on the performance of creeping thistle plants in an open field experiment. Early-season larval feeding had no measurable effect on creeping thistle growth, while late-season feeding significantly reduced shoot growth. These findings were attributed to higher feeding rates of the herbivores at higher ambient temperatures late in the season. Defoliation had a strong effect on the above-ground performance of C. arvense plants, but not on the below-ground performance. In contrast, the plant community affected all below-ground performance parameters measured, but only some of the above-ground performance parameters of creeping thistle. A combination of high levels of plant competition and herbivory by C. rubiginosa larvae led to 50% mortality in C. arvense plants during the growing season. We conclude that augmentation of indigenous herbivores of C. arvense in combination with breaking up the root system by tillage and the establishment of a highly competitive plant community of beneficial herbs may be a feasible way to control this problematic weed in ecological compensation areas.     

Relationship between adult and larval host plant selection and larval performance in the generalist moth, Trichoplusia ni

Adult oviposition preferences are expected to correlate with host plant suitability for the development of their offspring. For most lepidopteran species, this is particularly important as the hatching neonate larvae of many species are relatively immobile. Thus, the site of oviposition chosen by a female adult can greatly influence the probability of survival for her offspring. In the present study, we investigated the oviposition preference of adult Trichoplusia ni moths for six plant species to determine whether they could accurately rank the suitability of the plants for larval development. We also compared oviposition preferences to neonate larval acceptance and preference to determine whether the adult host range matched that of larval diet breath. Our results indicate that in two-choice and no-choice tests adult T. ni were able to rank the plants accurately, with the exception of anise hyssop. However, when given a choice of all six plants together, they laid more eggs on a plant that was not suitable for larval survival. Larvae accepted and fed on all plants in no-choice tests, and accurately ranked them according to larval performance. We conclude that neonate larvae are better able than adults to rank plants according to larval performance, and that larval diet breadth is wider than the range of plants accepted by adults. We also provide a discussion of the reduced accuracy of adult oviposition preference with increased plant choices.     

Adaptive change and conservatism in host specificity in two local populations of the thistle-feeding ladybird beetle Epilachna niponica

We investigated geographic differences in the host specificity of Epilachna niponica Lewis (Coleoptera: Coccinellidae). The Yuwaku population feeds mainly on Cirsium matsumurae Nakai (Asteraceae) and secondarily on Cirsrium kagamontanum Nakai. The Asiu population, located 150 km away from the Yuwaku, feeds exclusively on C. ashiuense Yokoyama et T. Shimizu. Under laboratory conditions, we examined the differences between the two populations in adult feeding acceptance, adult feeding preference, and larval performance, using several closely related thistle species and varieties, including their native hosts. In the Asiu population, adult beetles clearly avoided the host of the Yuwaku population, C. kagamontanum, and no larvae were able to complete their development, whereas in the Yuwaku population, adults accepted and even preferred it to some other thistle species, and about 10% of first instar larvae became adults. This indicates that the Yuwaku population evolved its feeding preference and physiological adaptation to C. kagamontanum through a utilization of this low‐ranked host under natural conditions. Apart from C. kagamontanum, the two populations showed a similar host susceptibility pattern, indicating that this ladybird beetle has a conserved hierarchy in feeding preference and growth performance. We also observed adult leaf choice behavior when given different thistle species, and found that difference in biting rate after palpation determined the leaf areas consumed, implying that factors on the leaf surface played an important role in the choice.     

Host range of the leaf-tier Tentamen atrivirgulatum Metz (Lepidoptera: Gelechiidae); an unsuitable candidate for biological control of Brazilian peppertree

ABSTRACT

The life history and host range of the South American leaf-tier Tentamen atrivirgulatum Metz, 2019 (Lepidoptera: Gelechiidae) was evaluated to determine its suitability for classical biological control of invasive Brazilian peppertree, Schinus terebinthifolia, in the USA. Larvae were collected during a survey in the native range of Brazil feeding on Brazilian peppertree. A quarantine colony was established and no-choice tests were conducted to determine the host range of this species. The results of these tests indicated that the larvae fed and completed development on seven of the eleven non-target species tested. Percent survival on these non-targets ranged from 20% to 100%. Larval survival was 50% when fed the target weed Brazilian peppertree. When larvae were fed leaves of these non-targets, pupal weights were significantly lower only for those fed Pistacia chinensis. Development time to the pupal and adult stages did not differ with larval diet. These no-choice results indicate T. atrivirgulatum larvae have a broad host range and thus this species will not be developed as a biological control agent for Brazilian peppertree in the USA.     

Host utilization of field-caged native and introduced thistle species by Rhinocyllus conicus

Rhinocyllus conicus Fr?elich was introduced from Europe into North America as a biological control agent of the exotic weed Carduus nutans L. Concern exists over the feeding of this weevil on at least 25 species of native Cirsium thistles. Beginning in 2008, cage studies isolating adults of R. conicus on buds and flower heads of all eight thistle species (native and introduced) recorded from Tennessee were conducted to test if R. conicus could use these species for reproduction and what impacts larval feeding of R. conicus may have on seed production. Larvae of R. conicus completed development in heads of the native species C. carolinianum (Walter) Fernald and Schubert. and C. horridulum Michaux, and significant reductions in seed numbers of both species occurred during 2008. Rhinocyllus conicus oviposited on both C. carolinianum and C. horridulum at significantly greater levels than the introduced species C. arvense (L.) Scopoli and C. vulgare (Savi) Tenore. Infested heads of C. carolinianum contained numbers of R. conicus per centimeter of plant head width similar to Ca. nutans in 2008, and both native species contained numbers of R. conicus per centimeter of plant head width similar to C. arvense and C. vulgare in 2009. Body length was similar between R. conicus reared on native thistles and its target host Ca. nutans. This report is the first documentation of R. conicus feeding and reproducing on C. carolinianum and C. horridulum. Although R. conicus has been observed only on introduced thistles in naturally occurring populations in this region, the utilization of C. carolinianum and C. horridulum as host species in controlled conditions warrants continued monitoring of field populations and further investigation into factors that may influence nontarget feeding in the future.     

Pollen availability,seed production and seed predator clutch size in a tephritid—thistle system

Summary We develop a simple model explaining clutch size behaviour ofOrellia ruficauda on its principle host in North America,Cirsium arvense. Offspring of flies feed solely on thistle seeds and seed production is pollen-limited. Thus, female flies risk reduced offspring fitness when committing large clutches to hosts (female flower heads) occurring in localities where male plants are locally absent. We therefore predict that attacked hosts will contain fewer eggs in such localities, a prediction that is consistent with data obtained in the field: large clutches are never laid in flower heads in low-pollination localities. However, larvae reared from such low-quality hosts are significantly smaller on average and will therefore carry smaller egg loads as adults. Small clutches in poor-quality hosts may thus be an expression of lower per-adult fecundity. Nevertheless, sufficient numbers of large, fecund flies are produced in low-pollination localities to make this last explanation less convincing.     

Biology and host range of Tecmessa elegans (Lepidoptera: Notodontidae), a leaf-feeding moth evaluated as a potential biological control agent for Schinus terebinthifolius (Sapindales: Anacardiaceae) in the United States

During surveys for natural enemies that could be used as classical biological control agents of Schinus terebinthifolius Raddi (Brazilian pepper), the caterpillar, Tecmessa elegans Schaus (Lepidoptera: Notodontidae), was recorded feeding on the leaves of the shrub in South America. The biology and larval and adult host range of this species were examined to determine the insect's suitability for biological control of this invasive weed in North America and Hawaii. Biological observations indicate that the larvae have five instars. When disturbed, the late instar larvae emit formic acid from a prothoracic gland that may protect larvae from generalist predators. Larval host range tests conducted both in South and North America indicated that this species feeds and completes development primarily on members of the Anacardiaceae within the tribe Rhoeae. Oviposition tests indicated that when given a choice in large cages the adults will select the target weed over Pistacia spp. However, considering the many valued plant species in its host range, especially several North American natives, this species will not be considered further for biological control of S. terebinthifolius in North America.     

Morphological variation in different populations of <Emphasis Type="Italic">Aceria anthocoptes</Emphasis> (Acari: Eriophyoidea) associated with the Canada thistle, <Emphasis Type="Italic">Cirsium arvense</Emphasis>, in Serbia

The russet mite, Aceria anthocoptes (Nal.), is the only eriophyid that has been recorded on Canada thistle, Cirsium arvense (L.) Scop. It has been noted in several European countries and recently in the USA. With its apparent host specificity and because of the damage it causes to its host plant, A. anthocoptes is being studied as a potential candidate for classical biological control. The aim of the present study was to examine quantitative morphological traits in four populations of A. anthocoptes living on two infraspecific host plant taxa (C. arvense var. arvense and C. arvense var. vestitum) in two geographically separate areas of Serbia in order to test the hypothesis of absence of the possible host plant impact on mite morphology. MANOVA analysis revealed significant differences between populations from different localities in Serbia. Populations of A. anthocoptes inhabiting two thistle varieties in the vicinity of Belgrade differed significantly from mites inhabiting the same two host varieties in the vicinity of the town of Ivanjica. Canonical discriminant analysis showed that the trait which best discriminates the populations of A. anthocoptes is the number of dorsal annuli. It was not possible to ascribe morphological differences to the impact of the host plant.     

Life history and host range of Sauris nr. purpurotincta,an unsuitable biological control agent for Chinese tallowtree

Foreign surveys in China discovered a defoliating insect species feeding on the leaves of Chinese tallowtree (Triadica sebifera), an invasive weed of the southeastern U.S.A. The life history of this species, Sauris nr. purpurotincta (Lepidoptera: Geometridae), was examined and larval no-choice and adult multiple-choice host range tests were conducted in quarantine to evaluate their suitability for biological control of Chinese tallowtree. The results indicated that the larvae have five instars and require approximately 22 days to complete development to the adult stage. Host range tests indicated that the larvae could not feed and complete development on most species tested. However, 40% of the larvae survived when fed leaves of Hippomane mancinella, a state-listed endangered species in Florida, and all larvae survived when fed Morella cerifera, a common native species of the southeastern U.S.A. Multiple-choice oviposition tests indicated eggs were laid on leaves of both a south Florida native plant Gymnanthes lucida and Chinese tallowtree. Considering this broad host range, this species will not be considered further for biological control of Chinese tallowtree in the U.S.A.     

Asymmetry of larval diet breadth and oviposition preference in Leptinotarsa undecimlineata

The relationship between preference and performance was studied under laboratory conditions for larvae and adult females of Leptinotarsa undecimlineata (Stål) (Coleoptera: Chrysomelidae). Leptinotarsa undecimlineata feeds exclusively on Solanum lanceolatum L. (Solanaceae) in the centre of Veracruz, Mexico. A common sympatric species, Solanum myriacanthum Dunal, is not used as host. We conducted a series of laboratory experiments to detect reasons for the exclusive use of one host plant. Leptinotarsa undecimlineata larvae were reared under laboratory conditions on both plant species, Solanum lanceolatum and S. myriacanthum . Length of larval stages, pupal weight, and percent adult eclosion were compared, and larval preference for both Solanum species was tested. Upon eclosion, adult females were exposed to each of the plants, and the effect of larval diet on adult preference was determined. Oviposition on both Solanum species was observed in no-choice tests. Larvae showed no significant preferences for either plant species. Pupal weight, however, was significantly higher on S. lanceolatum . Female beetles significantly preferred S. lanceolatum leaves, and no effect of preimaginal conditioning was found. Females reproduced in lower numbers after feeding on S. myriacanthum , oviposition initiated later, and females laid fewer eggs than females fed with S. lanceolatum . No female accepted S. myriacanthum for oviposition. Results obtained suggest that the potential host breadth of L. undecimlineata is broader in larvae than in female beetles.     

Host finding and recognition by Altica carduorum, a defoliator of Cirsium arvense

Altica carduorum Guer. (Coleoptera: Chrysomelidae) populations aggregate on vigorously growing stands of its host, the thistle Cirsium arvense (L.) Scop. (Asteraceae). The beetle needs contact to recognize a Cirsium species and biting to distinguish its host C. arvense. Feeding or mechanical damage on C. arvense, but not other Cirsium species, aggregates the beetles. Also feces from feeding on C. arvense, but not other Cirsium species, aggregates the opposite sex. Adults also aggregate on larval feces. We suggest that initial host location in the spring depends on random encounter by beetles overwintering in a stand of C. arvense. The beetles commonly make short flights after feeding and either land on another C. arvense plant, which establishes a secondary aggregation centre, or return to the original one. In this manner beetles released on a stand of C. arvense in China spread a radius of 600 m in a year, but those released 35 m away from their host failed to find it. Altica carduorum will develop on any Cirsium species to which it is confined, but is monophagous in the field because host finding is dependent on aggregation to wound and feces substances that are specific to C. arvense.     

Adult female and neonate larval plant preferences of the generalist herbivore, Epiphyas postvittana

The polyphagous leafroller moth, Epiphyas postvittana, is a pest of many fruit crops in New Zealand. Since the larva is highly mobile, host selection in this insect may involve both the adult female and the larva. In order to test the relative importance of the adult female and the neonate larva in the selection of host plants, the ovipositional preferences of females, and the preferences or acceptances of neonate larvae towards 26 plant species, consisting of 15 plants considered hosts and 11 not considered hosts, were investigated. In the ovipositional tests, the mean preferences of females for hosts and non-hosts were very similar. In contrast, larvae showed a significantly greater mean preference or acceptance towards hosts than to non-hosts, in both choice and no-choice bioassays, respectively. There were highly significant correlations between the preferences and acceptances of larvae for plants in the choice and no-choice tests. In the no-choice tests, there was a highly significant correlation between the acceptances of neonate larvae towards plants after one and three days (i.e., acceptances changed little over time). Moreover, in these no-choice tests, there was a significant negative correlation between larval acceptance at 1 day and larval mortality after 3 days; that is, the less acceptable a given plant at 1 day, the more likely larvae would fail to establish, feed, and survive on it by three days. Female and larval preferences towards the various plants were also negatively correlated. Together, these data suggest that the selection of a plant for the neonate larva to feed on is largely governed by the preferences of the larva, rather than by the preferences of the female. However, selection of a plant for oviposition by the female, may be important in host selection for reasons unrelated to larval preferences, for example, by encouraging dispersal, perhaps to other plant species, of the neonate larvae and thereby decreasing intersibling competition.     

Leaf traits of congeneric host plants explain differences in performance of a specialist herbivore

1. Within the host range of herbivorous insects, performance hierarchies are often correlated with relatedness to a primary host plant, as plant traits are phylogenetically conserved. Therefore, it was hypothesised that differences in herbivore performance on closely related plant species are due to resistance traits that vary in magnitude, rather than in the nature of the traits. 2. This hypothesis was tested by manipulating putative resistance traits of three congeneric thistle species (Cirsium arvense, Cirsium palustre, and Cirsium vulgare) and assessing the performance of the oligophagous, leaf‐feeding beetle, Cassida rubiginosa. Measurements were done of survival, weight gain, and development time of the beetle on its primary host, C. arvense, and two alternative hosts under low and high nutrient availability, and on shaved and unshaved leaves. 3. Survival of C. rubiginosa was strongly dependent on plant species with final mean survival rates of 47%, 16%, and 8% on C. arvense, C. palustre, and C. vulgare, respectively. Survival was primarily explained by leaf trichome densities, and to a lesser extent by specific leaf area. Leaf flavonoid concentrations did not explain differences in beetle survival, and there were no differences in beetle weight gain or development time of individuals that survived to adulthood. 4. No beetles survived on unshaved (hairy) C. vulgare plants, but manipulating leaf trichome densities of the thistle species by shaving the leaves moderated the plant‐specific resistance, and equalised the survival rates. Survival of C. rubiginosa on alternative congeneric hosts was explained by a common physical resistance trait that varied in magnitude.     

Indirect interaction between two native thistles mediated by an invasive exotic floral herbivore

Spatial and temporal variation in insect floral herbivory is common and often important. Yet, the determinants of such variation remain incompletely understood. Using 12 years of flowering data and 4 years of biweekly insect counts, we evaluated four hypotheses to explain variation in damage by the Eurasian flower head weevil, Rhinocyllus conicus, to the native North American wavyleaf thistle, Cirsium undulatum. The four factors hypothesized to influence weevil impact were variations in climate, weevil abundance, phenological synchrony, and number of flower heads available, either on wavyleaf thistle or on the other cooccurring, acquired native host plant (Platte thistle, Cirsium canescens), or on both. Climate did not contribute significantly to an explanation of variation in R. conicus damage to wavyleaf thistle. However, climate did influence weevil synchrony with wavyleaf flower head initiation, and phenological synchrony was important in determining R. conicus oviposition levels on wavyleaf thistle. The earlier R. conicus was active, the less it oviposited on wavyleaf thistle, even when weevils were abundant. Neither weevil abundance nor availability of wavyleaf flower heads predicted R. conicus egg load. Instead, the strongest predictor of R. conicus egg load on wavyleaf thistle was the availability of flower heads on Platte thistle, the more common, earlier flowering native thistle in the sand prairie. Egg load on wavyleaf thistle decreased as the number of Platte thistle flower heads at a site increased. Thus, wavyleaf thistle experienced associational defense in the presence of flowering by its now declining native congener, Platte thistle. These results demonstrate that prediction of damage to a native plant by an exotic insect may require knowledge of both likely phenological synchrony and total resource availability to the herbivore, including resources provided by other nontarget native species.     

Biology and host specificity of Rhinusa pilosa,a recommended biological control agent of Linaria vulgaris

Linaria vulgaris Mill. (Plantaginaceae), common or yellow toadflax, is a Eurasian short-lived perennial forb invasive throughout temperate North America. Rhinusa pilosa (Gyllenhal) (Coleoptera, Curculionidae) is a univoltine shoot-galling weevil found exclusively on L. vulgaris in Europe. Under no-choice test conditions, 13 non-native Linaria species exposed to R. pilosa were accepted for oviposition and most were found to be suitable, to varying degrees, for gall and larval development. Adult feeding and survival was minimal on native North American species in the plant tribe Antirrhineae which includes the target plant. In no-choice tests with 63 native North American species and 24 other non-target species outside Linaria, oviposition was limited to four native North American species. Only three larvae developed to the adult stage on Sairocarpus virga (A. Gray) D.A. Sutton, with no negative impact on plant growth. Risks to native flora from the release of R. pilosa are therefore expected to be minimal. The Technical Advisory Group for the Biological Control of Weeds (TAG—BCW) has recommended release of R. pilosa in September 2013.     

Galerucella birmanica (Coleoptera: Chrysomelidae), a promising potential biological control agent of water chestnut,Trapa natans

Water chestnut, Trapa natans, has become a major invasive plant in shallow water bodies in the northeastern United States. The failure of chemical and mechanical means to provide long-term and economically sustainable suppression of the species resulted in interest in the development of biological control. Field surveys in Asia and Europe identified a number of potential biological control agents in the native range of T. natans. The most promising species appeared to be the leaf beetle Galerucella birmanica, which is considered a pest of farmed T. natans in China. However, initial attempts to develop biological control faltered when field observations in China suggested that G. birmanica may not be host specific. Of particular concern was attack on water shield, Brasenia schreberi, a species native to China and North America. We conducted a number of laboratory and field investigations in China to assess preference and performance of G. birmanica on T. natans and B. schreberi. Initial no-choice experiments using 19 different plant species in 13 different families demonstrated that G. birmanica oviposited and was able to complete development only on Trapa spp. and B. schreberi. In larval no-choice tests G. birmanica was able to complete development on B. schreberi, however, larvae showed a 20% increase in mortality and longer development time compared to larvae developing on T. natans. In laboratory and field choice tests adults strongly preferred T. natans and in the field only occasionally laid eggs on B. schreberi. In addition, adults emerging from larvae reared on B. schreberi were less fit with reduced feeding and a declining oviposition rate. Their strong preference for T. natans was maintained in the field, even when T. natans was completely defoliated and adults were forced to migrate. We found only occasional “spill-over” of beetles onto B. schreberi and our data indicate that G. birmanica is a more promising biological control agent of T. natans than previously thought, although additional host specificity tests with many more North American plant species need to be completed.