Introduction and Establishment of Three Parasitoids of the Lily Leaf Beetle, Lilioceris lilii, (Coleoptera: Chrysomelidae) in North America

Following three years of discovery and evaluation, three larval parasitoids were imported from Europe and introduced into North America to control Lilioceris lilii (Coleoptera: Chrysomelidae), an introduced herbivore of native and cultivated lilies. The first species, Tetrastichus setifer Thomson (Hymenoptera: Eulophidae), introduced in Massachusetts in 1999, was found to be established in 2002. We made additional releases of T. setifer, introduced the parasitic wasps, Lemophagus errabundus Szepligeti (Hymenoptea: Ichneumonidae) and Diaparsis jucunda (Holmgren) (Hymenoptera: Ichneumonidae), and evaluated the establishment and distribution of the three parasitoids through 2013. Tetrastichus setifer is now established in Massachusetts, Rhode Island, New Hampshire, Maine, Connecticut, and Ontario, Canada. Lemophagus errabundus is established in Massachusetts and Rhode Island, and D. jucunda is established in Massachusetts, Rhode Island, and Maine. All three parasitoids have spread a considerable distance from release sites. The establishment of these parasitoids is associated with substantial reductions of L. lilii populations in some locations. In time it is likely that the parasitoids will spread throughout the North American range of L.lilii, but it may be useful to redistribute the parasitoids to accelerate this process.


6.
Lemophagus errabundus peak parasitism at recovery sites 42 and distance to nearest release site.

7.
Diaparsis jucunda peak parasitism at recovery site and 43 distance to nearest release site.

INTRODUCTION
The lily leaf beetle, Lilioceris lilii Scopoli, was first found in North America near Montreal, Canada in 1943. The first report of the beetle in the United States was in Cambridge, Massachusetts in 1992    (Livingston 1996, LeSage and. The lily leaf beetle is univoltine, overwinters as an adult, and after initiating feeding in the spring, lays rows of eggs on the undersides of leaves. The larvae carry a fecal shield, which is believed to provide some defense against predators  but may also serve to attract parasitoids (Schaffner and Müller 2001). Larvae complete four instars before pupating in the soil or leaf litter . Lilioceris lilii is present in Europe and Asia as far north as Siberia and to Morocco in the south , and in China . Based on the wide geographic and climatic range of the beetle's native distribution, it is likely to be able to establish itself across all of North America .
Since its introduction into North America, L. lilii has been considered a serious pest of cultivated lilies in Canada (LeSage 1992) and the United States . Adult and larval feeding of L. lilii causes extensive defoliation of lily leaves, buds, and flowers .
Even avid lily growers often give up growing lilies after repeat infestations of L. lilii . Infested lilies continue to send up new sprouts for a few years before they die, but in the meantime the plants and flowers are so damaged that they have lost their aesthetic value ).
An additional concern is that the lily leaf beetle will also feed on native lilies. Northeastern USA is home to four species of native lilies:  , USDA plants database 2014. Lilium superbum is a facultative wetland plant that is common in Rhode Island, but is endangered in New Hampshire and New York. Lilium canadense is threatened in Rhode Island and New York. Lilium philadelphicum is rare or endangered in Rhode Island, New York, Maryland and Ohio, and L. michiganense is considered endangered in New York (USDA plants database). The dispersal of Lilioceris lilii throughout North America is likely to have a negative impact on these and other rare native lilies ).
The genus Lilioceris Reiter is a large genus with the largest concentration of species found in China ) and approximately six species found in Europe  Ichneumonidae) were introduced against the cereal leaf beetle . Tetrastichus asparagi Crawford (Eulophidae) and Lemophagus crioceritor Aubert (Hymenoptera: Ichneumonidae) were released against the common asparagus beetle, and T. crioceridis Graham (Hymenoptera: Eulophidae) and Diaparsis truncatus (Gravenhorst) (Ichneumonidae) were introduced to control the spotted asparagus beetle . Tetrastichus julis and T. asparagi established in the areas of release, dispersed from the release sites, and are considered successful biological control agents of cereal leaf beetle and common asparagus beetle, respectively   . Early trials with the cereal leaf beetle egg parasitoid Anaphes flavipes showed that it could reproduce on L.lilii in Petri dishes but would not attack in larger cages and did not establish in field plots .  referred to a gregarious larval parasitoid in France, and Lemophagus errabundus was also reported from Lilioceris merdigera (L.) in France . These reports led to exploration in France and nearby European countries   . Parasitism is common in lilies growing in gardens, but populations found on native lilies are particularly heavily parasitized (Gold et al. 2001, Haye and. The discovery of these parasitoids led to European studies on parasitoid distribution and biology ) and a series of experiments to reveal host specificity, including parasitism of sympatric field populations of native Lilioceris species (L. tibialis, L. martagon, and L. merdigera), host range testing with these congeneric species in Europe, laboratory tests of chemical ecology, and host range testing in quarantine, as summarized in .
The egg parasitoid, Anaphes sp. and the two dipteran larval parasitoids, Meigenia spp. were rejected as potential biological control agents for L. lilii because they lacked sufficient host specificity. The remaining four were sent to the URI quarantine facility for host-specificity testing. Lemophagus pulcher was rejected as a biological control agent because, as initially indicated by European studies with sympatric populations and chemical ecology, this species may lack adequate host specificity. In quarantine, the parasitoid attacked the North American native Lema trilineata and the asparagus beetle Crioceris asparagi . The three remaining parasitoids, T. setifer, L. errabundus, and D.
jucunda were found to be host specific to the level of the genus Lilioceris, and all were approved by USDA APHIS PPQ for field release .
If in the (distant) future it is determined that these three agents which were released against L. lilii are not providing adequate control, L.
pulcher might warrant additional research. It has the advantage of being multivoltine and its ecological host range may be more limited than indicated by laboratory tests. For instance, if it were to attack asparagus beetles in nature, that would likely have been shown in the European research (Hendrickson et al.1991) that led to biocontrol releases against the asparagus beetles in the USA. Furthermore, Lema trilineata (three-lined potato beetle) is often a serious pest and could itself be a legitimate target of biological control. However, L. pulcher would require extensive host range testing, so we set it aside for now.
Tetrastichus setifer is a gregarious larval parasitoid, found very commonly in commercial lily fields in France and Switzerland. It was described in 1978 and is known from Czech Republic/Slovakia, France, Yugoslavia and Sweden . It is univoltine and overwinters as mature larvae in host cocoons in the soil . Tetrastichus setifer is the most widespread parasitoid of L. lilii in Europe, found in all regions investigated from Bulgaria to the UK and Northern Germany to Italy . This parasitoid is found in a broad range of climatic conditions in Europe from the warmer maritime regions of northern Germany to the colder high altitudes of the Alps.
Based on this European distribution, we expect T. setifer to survive in most of the presently invaded range of the lily leaf beetle.
Lemophagous errabundus  was reported to attack Lilioceris merdigera, in France. It is a solitary, univoltine larval parasitoid that kills L. lilii in the pre-pupal stage and overwinters as a teneral adult in the host cocoon . It is more prevalent in extreme northern Germany, Holland and western France with parasitism rates reaching over 70% among late instars in these areas with ocean-moderated climates . It has recently been found in England (Salisbury 2003). Based upon the European distribution, we consider L. errabundus to be climatically suited for release in coastal southern New England.

Diaparsis jucunda was reported by Horstmann (1971) from
Sweden, Finland, Denmark, Germany, and the Czech Republic. It is a solitary univoltine larval parasitoid which attacks all stages of L. lilii. It was very common in surveys conducted in native lilies in Switzerland , and was the dominant parasitoid of L. lilii in central and southern Europe, which represent the colder range where T. setifer is also found. Total parasitism in the last instar averaged about 60% in lily fields, 74% in gardens, and 90% on wild lilies . Based on USDA plant hardiness zones, New England releases of Diaparsis jucunda are most appropriate for release at inland and northern New England sites.
Surveys by  revealed that throughout Western Europe a complex of parasitoids was found attacking L. lilii with different parasitoids predominating in different regions, but more than one species was important at virtually all sites. We released all three host-specific parasitoids in New England, based on knowledge of their different climatic ranges in Europe  to determine release sites . This document summarizes the parasitoid releases made over the past 14 years and includes results of surveys for establishment, distribution, and apparent impact on L. lilii.

Source of parasitoids
Parasitoids of L. lilii were collected by colleagues from CABI Switzerland beginning in 1998 from France, Switzerland, Germany, Italy, Holland, Belgium, Bulgaria, and Ukraine (Table 1). Field-collected larvae were held in the laboratory at CABI at 25°C and fed lily leaves until pupation in vermiculite and emergence of adult L. lilii. The remaining parasitized pupae were held at 4° C in a growth chamber for a minimum of two months before being shipped to the University of Rhode Island in chilled insulated boxes. After a shipment arrived at the URI Insect Quarantine Laboratory, parasitoids were stored at 4° C, and then moved to and 20 stems in the smaller plots, and we removed 20 fourth instar larvae from these plants (when available) for dissection to determine parasitism by T. setifer. Control plots were sampled in the same manner as release plots (40 stems per plot) to establish baseline data for long-term evaluation of parasitoid releases. We used cages at release sites to allow release of additional parasitoids while sampling for establishment of parasitoids from releases made in the previous year.

Additional Release Sites
In conjunction with more formal release sites where we released one of the three L. lilii parasitoids and monitored the L. lilii population for parasitism, we also collected larvae from residential gardens for dissection and evaluation for parasitism. Once we had documented establishment of all three parasitoids in our release plots we also made additional releases in residential sites in Rhode Island, SE Massachusetts, Maine, and Connecticut (Table 4). In 2010, we shipped T. setifer parasitoids to Dr. N.
Cappuccino who, with a Canadian permit, released them into a research plot near Ottawa, Ontario in Canada. Residential sites for both releases and recovery were identified through a variety of methods: proximity to initial release plots, collaboration with university or state department of agriculture personnel, and contacts through master gardener programs and local newspapers. Recovery sites are listed in Appendix A.

Tetrastichus setifer establishment and spread
The establishment and spread of T. setifer from the first release in 1999 until 2013 are represented here with all release sites and all positive recovery sites (Fig. 4).    (Table   5). In spite of this, we still continue to find many home gardens with damaging L. lilii populations.
Additional T. setifer adults were released in other release plots in New England (Table 1) and in home gardens (Table 4) (Table 6).
Subsequently, it was found in a home garden 2.9 km from the release site in 2006 (Table 6). One home garden recovery site (4 km from the Plainville release site) had steadily increasing parasitism, from a peak of 9 % in 2009, to 50% in 2010, 78% in 2011, and 94% in 2012 (Table 6). In 2013 we visited the site when we should have found peak parasitism, but could not find any larvae to evaluate for parasitism.
The rate of spread of L. errabundus is graphed in Fig. 8 using the date of first recovery for each of the recovery sites in Fig. 5. A linear regression line fit to these data results had a slope of 0.79 km/yr, with all points indicating spread less than 2 km per year.

Diaparsis jucunda establishment and spread
Diaparsis jucunda adults were released in five plots from 2003 to 2007 (Table 3) and in a number of smaller garden sites (Table 4) (Table 7).
The rate of spread of D. jucunda is graphed in Fig. 9 using the date of first recovery for each of the recovery sites in Fig. 6. Linear regression showed a non-significant relationship between years and distance spread.
It is noteworthy that years 4-6 indicate that this parasitoid is capable of moving as much as 4-5 km per year. There are a number of recoveries of D. jucunda in Massachusetts and Maine which indicate that this parasitoid has been found approximately 15-20 km from a release site.

Control plots
No parasitism was found in any of the control plots until T. setifer and D. jucunda were found in the Belmont, MA plot in 2011 (Tables 5 and   7). Control plots in Maine and in Falmouth, Massachusetts were abandoned due to the extensive lily damage caused by the lily leaf beetle.
Also because of high beetle populations, other control plots, such as the Deerfield, NH plot were converted from a T. setifer control plot into a release site for D. jucunda.

DISCUSSION
Since the introduction of L. lilii into North America in 1943, it has experienced a familiar pattern of long-range redistribution with potted plants or bulbs and then a localized spread throughout the new area. In 1992 the beetle showed up in several disjunct sites, including Boston, apparently as a result movement of plants or plant parts . It was only several years later that the population spread throughout the northern New England states, now making a continuous distribution with the Canadian population (Fig 2). The beetle's arrival in Manitoba, Canada was attributed to plant movement   and it now appears that this prediction may come true. If this is the case, then the successful application of a classical biological control program early in the invasion of this introduced species could be an important tool in reducing the severity of the invasion as it reaches new areas. It is also apparent that a complex of parasitoids will be important for managing this pest in different climatic zones.
Tetrastichus setifer was the first parasitoid evaluated, and the first to be released in North America. In European surveys, T. setifer was found in a wide range of climatic conditions , and appeared to have the best chance of establishing throughout North America. In Sweden T. setifer and L. errabundus are the most abundant L. lilii parasitoids ). In the UK T. setifer has been found everywhere that L. lilii is found   In the Wellesley site that received the first releases of T. setifer, densities of L. lilii larvae declined from roughly 7/stem to 0.5/stem in the 7 years following parasitoid release, a decline that was statistically significant compared to a control plot 17 km distant.

Lemophagus errabundus is most prevalent in Europe in areas with
ocean-moderated climates . This knowledge led us to errabundus may have a greater impact on L. lilii than it does in Europe, because it was introduced without its hyperparasitoid, Mesochorus lilioceriphilus, which is very common in the UK, Sweden, and most of Europe .
Diaparsis jucunda is the dominant parasitoid found in native lilies in mountainous sites in Switzerland and in colder areas of central and southern Europe . For this reason we wanted to release It is possible that as parasitoids further increase their distribution and abundance that their rate of spread will increase, but at present it appears that if homeowners are expecting to see positive results in less than a decade, releases may be on the order of 20 km distant. Releases made at 16 sites spread throughout Connecticut in the past two seasons should add considerably to our knowledge of parasitoid spread and also to the optimal numbers of parasitoids to release per site. Based upon our experience to date, it appears that releases of 50-100 T. setifer should result in establishment and ichneumonid releases should be roughly 25 to 50 per plot of 20 lilies with a high infestation of L. lilii.
The success of a classical biological control program depends on the successful establishment and dispersal of the introduced predators or parasitoids. The habitat fragmentation (or patchiness) of lilies, as previously mentioned, will affect the future dispersal of both the beetle and its parasitoids, although predators and parasitoids are often more strongly affected by habitat fragmentation than the abundance and diversity of herbivorous hosts . Specialists such as the lily leaf beetle parasitoids may be better dispersers than generalist predators and parasitoids because they are more susceptible to the patchy distribution of available hosts, whereas generalists may find alternate hosts available . Larger parasitoids (Lemophagus and Diaparsis) could be expected to be better fliers and disperse more easily to new locations with hosts, but smaller ones, such as the eulophid T. setifer, may be more easily dispersed by wind. Small gregarious specialist parasitoids, like Tetrastichus setifer tend to be good dispersers, reported to travel at least 2 km within a season ). The issue of parasitoid dispersal may be more important in a classical biological control program in a landscape setting, such as this one, than in an agricultural setting.
To date, rearing and release of agent has been done through URI at a fairly small scale. As the beetle spreads into more states and provinces there may be interest in a larger program of collecting and redistributing, or rearing of all three parasitoids. The cereal leaf beetle biological control program of the 1970's provides precedence for distribution of close relatives of the three Oulema melanopus (cereal leaf beetle) parasitoids (T. julis, D. temporalis, and L. curtus). This program used large field plots to produce cereal leaf beetle larvae and county agents were brought in for field days to successfully redistribute parasitized larvae throughout the state . This program also benefited from a USDA parasitoid rearing laboratory in Niles, MI which distributed parasitoids throughout the infested states . The cereal leaf beetle was considered a major threat to US agriculture and the biological control program was a well funded priority of the USDA. With limited funding for lily leaf beetle, no major parasitoid redistribution programs have been initiated. The cost and shortage of organic lily bulbs is an important issue: at $0.50 per bulb it costs over $2.00 per square foot to establish a rearing plot vs. pennies to grow oats in the cereal leaf beetle program. We have worked out procedures for laboratory rearing, but it is also quite expensive.
The popular practice of mulching lilies in the garden may not support the development of parasitoid populations. Unlike lily leaf beetle adults which can fly to suitable overwintering habitats, all three parasitoids overwinter as immature larvae or pupae directly under their host plants.
Mulch may not provide adequate protection from cold, desiccation, and predation.  suggested that our initial establishment efforts with T. setifer benefited from the removal of mulch from the plot. This may also be one explanation for the continued prevalence of persistent L. lilii populations in many home gardens in Cumberland, RI, in spite of ten years of high T. setifer parasitism. All three parasitoids we have released are also present in Sweden, but L. lilii still causes significant damage to lilies. It has been suggested that cultivation practices, use of mulch, and the practice of fall digging and spring replanting of bulbs may negatively impact parasitoid success in Sweden ).
Another complication in biological control of L. lilii is the fact that most lily bulbs purchased by gardeners have been treated with the insecticide imidacloprid for protection against aphids. From our rearing experience, we know that this systemic insecticide kills L. lilii feeding on treated plants for at least one season. During that time the lilies will not support L. lilii populations, and therefore will not support development of L. lilii parasitoids. We have used organically-produced lily bulbs in our rearing program for over a decade and gardeners wishing to encourage parasite establishment should consider this as well. Experiments on the impact of mulching on parasitoid populations would also be useful in developing management recommendations to enhance biological control.
We have established in North America the three host-specific parasitoids that are most commonly found attacking lily leaf beetle populations in Europe. Based upon research conducted to date, we believe they have potential to manage this pest in North America and we encourage further work on parasitoid redistribution and subsequent management. As the parasitoids spread throughout North America, there will also be an opportunity to compare the distribution of these three species to the distributions found in Europe using a predictive model such as CLIMEX. Documentation of this program as a successful biological control program will be improved with an understanding of the factors that contributed to the success of the parasitoids in different areas.