Multiple-choice and no-choice tests were conducted at the Department of Agriculture, Division of Plant Industry Quarantine facility in Gainesville, FL to determine the specificity of the Brazilian leaf-beetle Gratiana graminea Klug, a candidate for biological control of Solanum viarum, tropical soda apple. One hundred fifteen plant species in 32 families were included in the feeding-oviposition multiple-choice tests including the target weed and the 5 major cultivated Solanaceae Capsicum annuum L., Lycopersicon sculentum Mill., Nicotiana tabacum L., Solanum melongena L., and Solanum tuberosum L. Eight to 12 plant species, including always the main target weed, growing in 1-gallon pots were simultaneously exposed to 20 G. graminea adults (10 males and 10 females that most of the time had recently emerged from pupae) in an aluminum cage (60 × 60 × 60 cm). At the beginning of each test the insects were placed at the bottom center of each cage to allow them to orient by themselves to the tested plants. Plant species in each test were replicated 3–4 times (one replication of tested plants in each separate cage). Plants tested were exposed to G. graminea adults from 3–6 weeks. Observation of oviposition and feeding were made during almost all the weekdays. No-choice host specificity tests were conducted with G. graminea adults on potted plants in cages made of clear-plastic cylinders and with G. graminea larvae placed on cluster of leaves of each individual plant tested. Ten G. graminea adults were exposed to 29 plant species individually tested during 3 to 5 weeks, and 10 neonate larvae were exposed to 31 plant species. Plant species in each test were replicated 3–4 times. Results indicated that G. graminea fed and developed only on the target weed. The tests indicated that a host range expansion of G. graminea to any of the major cultivated Solanaceae species is highly unlikely. A petition for field release in Florida was submitted to the Technical Advisory Group for Biological Control Agents of Weeds (TAG) in Sep 2008.
Tropical soda apple, Solanum viarum Dunal (Solanaceae), is a perennial weed, originally from northeast Argentina, southern Brazil, Paraguay, and Uruguay, that has been spreading throughout Florida at an alarming rate during the last two decades. The pasture-land infested in 1992 was estimated to be approximately 60,000 hectares (Mullahey et al. 1993), and increased to more than 300,000 hectares in 1995–96 (Mullahey et al. 1997). Currently, the infested area is estimated at more than 400,000 hectares (Medal et al. 2008). Tropical soda apple, first reported in the United States in Glades County, Florida in 1988 (Coile 1993; Mullahey & Colvin 1993), is also present in Alabama, Georgia, Mississippi, North Carolina, South Carolina, Texas, and Puerto Rico (Bryson & Byrd, Jr. 1996; Dowler 1996; Mullahey et al. 1997; Medal et al. 2003). The potential range of tropical soda apple in the United States may be extended even further based on studies of the effects of temperatures and photoperiods conducted by Patterson (1996) in controlled environmental chambers. This invasive exotic weed was placed on the Florida and Federal Noxious Weed Lists in 1995.
In addition to its invasion of pasture lands and reduction of cattle carrying capacity (Mullahey et al. 1993; Bredow et al. 2007), tropical soda apple is known to harbor at least 6 viruses that affect cultivated solanaceous crops such as tomato, tobacco, and pepper (McGovern et al. 1994a, 1994b, 1996). Tropical soda apple is also an alternative host for key pests such as the Colorado potato beetle, Leptinotarsa decemlineata (Say) (Coleoptera: Chrysomelidae), major defoliating insect pest of potato in North America; the tomato hornworm, Manduca quinquemaculata (Haworth) and the tobacco hornworm, Manduca sexta (L.), (Lepidoptera: Sphingidae), major pests of tomato and tobacco plants; the silverleaf whitefly, Bemisia argentifolii Bellows and Perring (Homoptera: Aleyrodidae) one of the most troublesome insect pest worldwide of many field and vegetable crops; the tobacco budworm, Heliothis virescens (Fabr.) (Lepdoptera: Noctuidae) one of the most destructive pests of tobacco; the green peach aphid, Myzus persicae (Sulzer) an important pest of peach trees and vector of plant viruses to solanaceous plants and other food crops (Homoptera: Aphididae); the southern green stinkbug, Nezara viridula (L.) (Hemiptera: Pentatomidae) an important pest of soybean and vegetable crops; and the suckfly, Tupiocoris notatus (Distant) (Hemiptera: Miridae) a pest of several crops including tobacco (Habeck et al. 1996; Medal et al. 1999a; Sudbrink et al. 1999). Although it is very difficult to estimate the real (direct and indirect) economic losses due to this invasive weed, the production loss to Florida ranchers by tropical soda apple was estimated from $6.5–16 million annually (Thomas 2007).
Although tropical soda apple is able to spread vegetatively from the root system, the primary method of dispersal is by seed dissemination (Bryson et al. 1995; Medal et al. 1999b), which occurs mainly by livestock and wildlife that feed on the fruits and scarify the seeds (Akanda et al. 1996; Brown et al. 1996). A single plant of tropical soda apple can produces up to 150 fruits per year, with each fruit containing on average 400 seeds. The estimated seed production is 60,000 seeds/ plant/season with a viability of more than 75% (Mullahey & Colvin 1993; Pereira et al. 1997).
Currently recommended management practices for this invasive plant in southeastern United States include herbicide applications and mechanical techniques (mowing/tilling) (Mislevy et al. 1996; Mullahey et al. 1996; Sturgis & Colvin 1996; Akanda et al. 1997). These control tactics provide temporary weed suppression at an economic cost estimated at $62 and $47 per hectare in chemical and mechanical control methods, respectively (Thomas 2007). However, application of these control methods is often difficult to employ in remote and/or inaccessible areas.
A biological control project for tropical soda apple was started in Dec 1996 by the University of Florida in collaboration with the Universidade Estadual Paulista, Jabotical campus, Brazil, Universidade Federal do Parana in Curitiba, Brazil, Universidade Regional de Blumenau, Santa Catarina state, Brazil, Universidade Centro-Oeste in Irati, Parana state, Brazil, and the USDAS-ARS, Biological Control Laboratory in Hurlingham, Buenos Aires Province, Argentina. The release of the Brazilian leaf-beetle Graminea graminea Klug (Coleoptera: Chrysomelidae) in Florida will complement the defoliation effects that Gratiana boliviana Spaeth has been making on tropical soda apple plants during the warm season in Florida since it was released in the summer 2003 (Medal et al. 2008; Overholt et al. 2008). Tropical soda apple defoliation by G. boliviana and G. graminea, in southern Brazil is causing a major suppressive effect on tropical soda apple density (Gandolfo et al. 2007; Medal et al. unpublished data). These two leaf-feeder beetles have a synergistic effect on tropical soda apple defoliation and occupy different niches in somewhat overlapped geographical regions in southern parts of Brazil.
In this paper we report the results of the host-specificity tests conducted at the Florida Department of Agriculture-Division of Plant Industry quarantine facility in Gainesville with the leaf-beetle G. graminea as a potential biological control agent of the non-native weed tropical soda apple.
MATERIALS AND METHODS
Host-Feeding Specificity Tests
Plant-host specificity tests with Gratiana graminea adults and first instars were conducted from Sep 2000 to Aug 2004 at the Florida Department of Agriculture and Consumer Services-Division of Plant Industry quarantine facility in Gainesville, Florida. Additional feeding/oviposition tests with G. graminea adults were conducted at the Gainesville quarantine facility from May to Sep 2008. Gratiana graminea (all developmental stages) were collected on tropical soda apple plants in Rio Grande do Sul, Brazil, introduced into Florida-quarantine, placed on caged plants of tropical soda apple growing in 1-gallon pots and eggs were removed twice a week to provide the insects required for testing.
Multiple-Choice Feeding and Oviposition Tests
One hundred fifteen plant species in 32 families were included in the feeding and oviposition preference tests in quarantine (Table 1). The plants tested included 56 species in the family of the target weed (Solanaceae) of which 29 were from the genus Solanum and 27 from 15 other genera that include plants of agricultural or ecological importance. Ten species representing 5 families (Boraginaceae, Convolvulaceae, Ehretiaceae, Nolanaceae, Polemoniaceae) that are very closely related phyllogenetically to the Solanaceae and in the same order Polemoniales (Heywood 1993) were included. Forty-nine plant species representing 26 families, most of them with an economically and/or environmentally value in North America, were also tested. The major target weed (tropical soda apple), and 10 plant species in the Solanaceae were tested at least 3 times (Table 1). They included Solanum donianum Walpers that is in the list of Florida threatened plants (Coile 1998); 4 secondary target-weeds (Solanum tampicense Dunal, Solanum torvum Sw., Solanum capsicoides All., Solanum elaeagnifolium Cav.); and the 5 major cultivated Solanaceae (Capsicum annuum L., Lycopersicon esculentum Mill., Nicotiana tabacum L., Solanum melongena L., Solanum tuberosum L.). Eight to 12 plant species, including always the main target weed, growing in 1-gallon pots were simultaneously exposed to 20 G. graminea adults (10 males and 10 females which were newly emerged from pupae most of the time) in an aluminum cage (60 × 60 × 60 cm). At the beginning of each test the insects were placed at the bottom center of each cage to allow them to orient by themselves to the tested plants. Plant species in each test were replicated 3–4 times (1 replication of tested plants in each separate cage). Plants were exposed to G. graminea adults from 3–6 weeks. Observations of oviposition and feeding were made during most of the weekdays. Plants consumed were replaced as needed. Plants were checked for oviposition sites and eggs were removed and counted weekly. On the last day of each experiment, plants were checked for feeding and eggs laid on them. Leaf area consumed was measured with a Portable Area Meter Model LI-3000 (Lambda Instrument Corporation) and the leaf-feeding area is reported on a scale from 0–5 (0 = no feeding, 1 = probing or <5% of leaf area consumed, 2 = light feeding or 5–20% of the area, 3 = moderate feeding or 21–40%, 4 = heavy feeding or 41– 60%, and 5 = intense feeding or >60% of the leaf area consumed).
No-Choice Larval Feeding Tests
No-choice host specificity tests were conducted with G. graminea neonate larvae in an environmental chamber at a temperature of 22 ± 2°C, relative humidity of 55–65%, and a photoperiod of 12:12 (L:D). Recently hatched non-fed larvae were exposed to 31 plant species including 30 species in the family of the target weed (Solanaceae) and 1 species in the family Convolvulaceae. The species tested included 7 genera of plants very closely related phyllogenetically in the same family as the target weed, and with an economical and/or environmental value in North America (Table 2). Larvae were exposed to clusters of leaves of each individual plant tested by placing the clusters individually in a 30-mL plastic-cup containing water and fitted with a paper lid to avoid insect contact with the water. The leaf petiole was inserted through a hole (3–4 mm diameter) made in the middle of the paper-lid. The cup and plant cluster was placed inside a clear-plastic container covered with a plastic-lid having 6–7 small holes to allow air circulation. Moistened tissue paper was placed at the bottom of the plastic container and under the plastic-lid to provide moisture. The plants (treatments) were arranged in a completely randomized design. Three to 4 replications were used with 10 one-d- old larvae per replication. Each group of 10 G. graminea larvae was provided with only 1 plant species which they fed on or died. Daily observations of feeding were made and leaves were replaced as needed. Larval mortality counts were made 3 and 7 d after the experiment started.
No-Choice Adult Feeding Tests
No-choice host specificity tests were conducted with G. graminea adults at the Gainesville quarantine facility with potted plants (20–60 cm height) in cages. Gratiana graminea adults were exposed to 29 plant species including S. donianum in the list of Florida threatened plants, all major cultivated Solanaceae, and 7 exotic (Table 3). Five to 6 plant species were individually tested each time due to limitation in availability of cages. Ten G. graminea adults (5 males, 5 females) per replication (3–4 replications) were exposed to plants during 21 to 35 d. Cages were made of clear plastic cylinders (15 cm diameter, from pupae or were still young (1–2 week old) and showing the intense green color that differentiate them from pale-yellow older adults. Eggs laid on plants, if any, were removed weekly and plants replaced as needed. At the end of the testing periods, feeding and adult mortality were recorded.
TABLE 1.
HOST RANGE ADULT FEEDING AND OVIPOSITION TESTS WITH GRATIANA GRAMINEA IN FLORIDA.

(CONTINUED)

(CONTINUED)

(CONTINUED)

TABLE 2.
HOST RANGE OF GRATIANA GRAMINEA FIRST INSTAR IN NO-CHOICE FEEDING TESTS IN FLORIDA QUARANTINE.

RESULTS AND DISCUSSION
Multiple-Choice Feeding-Oviposition Tests
In the quarantine multiple-choice tests (Table 1), Gratiana graminea adults fed heavily to intensively (41–100% of the leaf area offered) on the major target weed tropical soda apple. It fed lightly to moderately (5–40% of the leaf area offered) on turkeyberry, Solanum torvum Sw. (native to west Africa and on the list of Federal noxious weeds and on Florida's invasive species list of the Florida Exotic Pest Plant Council; webpage: http://www.fleppc.org/97list.htm). Minor or exploratory feeding (<5% of the leaf area offered) was observed on the non-native red soda apple, Solanum capsicoides All. (prickly weed of South American origin also present and spreading in Florida), on eggplant, Solanum melongena L. (crop of economic importance), on the non-native Solanum mamosum L. (native from Central America, not growing in USA), on the non-native Solanum tampicense Dunal (weed of Mexico, Central America, and Caribbean origin and now established and expanding in south Florida; also on the Florida's invasive species list of the Florida Exotic Pest Plant Council). No feeding was observed on any of an additional 109 plant species in 32 families that were tested. The adults laid from 58–104 eggs (average: 86) on tropical soda apple, and from 0–2 (average: 0.1 eggs) on eggplant (Table 1). No eggs were deposited on any of the other 113 plant species tested, including the threatened S. donianum. Although some minor feeding on eggplant has occurred in quarantine, this insect has never been recorded attacking eggplant in South America. Expanded host ranges of weed biocontrol candidates under confined quarantine laboratory conditions have been reported by South African researchers (Neser et al. 1989; Hill & Hulley 1995; Olckers et al. 1995; Hill & Hulley 1996; Olckers 1996). They indicated that almost all the agents that have been tested for biocontrol of Solanum weeds have shown feeding on closely related plant species, but they are never attacked under natural conditions. For example, Gratiana spadicea (Klug) (Coleoptera: Chrysomelidae) as a biocontrol agent against Solanum sisymbrifolium Lam. in South-Africa (Hill & Hulley 1995), fed and was successfully reared on eggplant in laboratory tests. This insect was field released in South Africa in 1994 based mainly on the lack of records as a pest of eggplant in South America. This insect has become established on S. sisymbriifolium with no reports of attacks of eggplant fields in South Africa.
TABLE 3.
HOST RANGE OF GRATIANA GRAMINEA ADULTS IN NO-CHOICE FEEDING TESTS IN FLORIDA QUARANTINE.

No-Choice Larval Feeding Tests
Larvae of G. graminea exposed to individual plants (31 species) in quarantine growth chambers (Table 2) completed development on the target weed tropical soda apple (74% reached the pupae stage, n = 70, 2 trials). Some feeding (5–20%) was observed on Solanum capsicoides (red soda apple), and also some probing or exploratory feeding (<5% of the leaf area offered) was observed on Solanum melongena (eggplant) and on Solanum torvum (turkeyberry), but larvae stopped feeding and died within a week after the experiment began. The rapid death of the G. graminea larvae with no feeding and no development on the 30 non-target plant species tested clearly indicated no risk of attack on these plants by this beetle.
No-Choice Adult Feeding Tests
Starvation tests (no-choice) with G. graminea adults exposed to individual potted plants (29 species) in cages at the quarantine facility (Table 3) indicated that the insect fed and laid eggs (47–61 eggs per female; average: 57 per female) only on tropical soda apple. Feeding on tropical soda apple was intense (>60% of the leaf area offered) compared to a probing or exploratory feeding (<5%) observed in S. melongena (eggplant cultivars Black Beauty and Market), on S. tampicense, and on S. elaeagnifolium (secondary target weeds). Although there was some feeding (5–20% of the leaf area offered) on S. torvum (secondary target weed), and on S. mammosum (non-native), the females did not lay eggs on these plants. No eggs were laid on any of the 28 non-target plant species tested including the 9 eggplant cultivars (Black Beauty, Classic, Ichiban, Italian-Nadia, Market, Neon, Orient Charm, Orient Express, and Thai).
The high specificity shown by this beetle in the host range feeding tests and development only on the target weed, indicated no adverse impacts would be expected on the 6 solanaceous species that were not tested and are listed as threatened or endangered in Hawaii and Puerto Rico. Indirect beneficial effects on wildlife populations associated with release and establishment of G. graminea may be expected due to recolonization by native plants that have been displaced by the rapidly growing and highly competitive tropical soda apple plants.
The host specificity tests in quarantine indicated that G. graminea is safe to release. Occasional temporary feeding might occur on the nonnative weeds S. torvum and S. tampicense (in the Federal Noxious Weed list), and S. capsicoides, a prickly weed introduced from South America (Kissman & Groth 1995). Noticeable damage to eggplant is unlikely to occur based on our host tests. The lack of a record as a crop pest in the native range of the beetle support our findings on the specificity and safety of G. graminea as a biocontrol agent of tropical soda apple.
Based on the specificity of G. graminea feeding and developing only on the target weed, we consider this beetle safe for field release against tropical soda apple. Therefore, a petition to release the Brazilian leaf-beetle G. graminea for the control of tropical soda apple in the southeastern United States was submitted to the USDA-APHIS-PPQ Technical Advisory Group (TAG) members on Sep 2008.
ACKNOWLEDGMENTS
We thank Howard Frank (University of Florida, Entomology and Nematology Department), and Julieta Brambila (United States Department of Agriculture, Animal and Plant Health Inspection Service) for reviewing the manuscript. We thank Zundir Buzzi (Universidade Federal do Paraná, Curitiba, Brazil) for identification of Gratiana graminea. This research was funded by USDA-APHIS, and by the Florida Department of Agriculture and Consumer Services, Division of Plant Industry.