Tortricid Moths Reared from the Invasive Weed Mexican Palo Verde, Parkinsonia aculeata, with Comments on their Host Specificity, Biology, Geographic Distribution, and Systematics

As part of efforts to identify native herbivores of Mexican palo verde, Parkinsonia aculeata L. (Leguminosae: Caesalpinioideae), as potential biological control agents against this invasive weed in Australia, ten species of Tortricidae (Lepidoptera) were reared from Guatemala, Mexico, Nicaragua, and Venezuela: Amorbia concavana (Zeller), Platynota rostrana (Walker), Platynota helianthes (Meyrick), Platynota stultana Walsingham (all Tortricinae: Sparganothini), Rudenia leguminana (Busck), Cochylis sp. (both Tortricinae: Cochylini), Ofatulena duodecemstriata (Walsingham), O. luminosa Heinrich, Ofatulena sp. (all Olethreutinae: Grapholitini), and Crocidosema lantana Busck (Olethreutinae: Eucosmini). Significant geographic range extensions are provided for O. duodecemstriata and R. leguminana. These are the first documented records of P. aculeata as a host plant for all but O. luminosa. The four species of Sparganothini are polyphagous; in contrast, the two Cochylini and three Grapholitini likely are specialists on Leguminosae. Ofatulena luminosa is possibly host specific on P. aculeata. Host trials with Rudenia leguminana also provide some evidence of specificity, in contrast to historical rearing records. To examine the possibility that R. leguminana is a complex of species, two data sets of molecular markers were examined: (1) a combined data set of two mitochondrial markers (a 781-basepair region of cytochrome c oxidase I (COI) and a 685-basepair region of cytochrome c oxidase II) and one nuclear marker (a 531-basepair region of the 28S domain 2); and (2) the 650-basepair “barcode” region of COI. Analyses of both data sets strongly suggest that individuals examined in this study belong to more than one species.


Introduction
Parkinsonia aculeata L. (Leguminosae: Caesalpinioideae), commonly known as Jerusalem thorn, Mexican palo verde, and/or jellybean tree, is a large shrub or small tree native to North and South America that shows marked genetic divergence among populations (Hawkins et al. 2007). It is used as an ornamental in tropical and subtropical climates, and it has been used to re-vegetate desertified regions throughout the pantropics (van Klinken et al. 2008). It has escaped cultivation and become established in many places around the globe. It was introduced into Australia as an ornamental and shade tree around 1900 (Woods 1992) and by 1906 was considered weedy in some parts of Queensland (Bailey 1906). Now it is regarded as one of the most troublesome invasive weeds in northern Australia and is recognized as a Weed of National Significance. Parkinsonia aculeata currently infests over 800,000 hectares, mainly along watercourses, in Western Australia, Queensland, and the Northern Territory. It has the potential to invade most of the semi-arid to subhumid tropical areas in Australia (van Klinken et al. 2008).

Platynota stultana Walsingham (Figures 7, 8)
Platynota stultana is one of the smaller species in the genus. It ranges throughout the southwestern United States and northern Mexico; it is adventive in California (Powell 1983), Hawaii (Miller and Hodges 1995), Florida (Kimball 1965), Virginia, and Washington, D.C. (USNM).
Biology. This species is known in the American economic literature as the omnivorous "leaf-roller". Larval hosts in native situations include western ragweed (Ambrosia psilostachya DC.; Asteraceae), gooseberry (Ribes sp.; Grossulariaceae), and youngberries (Rubus sp.; Rosaceae). Agricultural and ornamental hosts include a wide range of native and cultivated plants (Atkins et al. 1957;MacKay 1962;Powell 1983;Miller and Hodges 1995). Rolled leaves of Parkinsonia aculeata harboring larvae of Platynota stultana were collected at several sites in a large swamp known as Laguna

Rudenia leguminana (Busck) complex (Figure 9)
Rudenia leguminana is widely distributed in North America, ranging from southern Connecticut to Florida, and from the midwestern U.S. to California. It is abundant in some xeric areas of the southwestern U.S. Rudenia leguminana was reared from P. aculeata in Guatemala (Jutiapa), Mexico (San Luis Potosí, Oaxaca, and Veracruz), Nicaragua (Granada, Managua, and Matagalpa), and Venezuela (Miranda and Anzoátegui), and from Pithecellobium dulce (Leguminosae) in Mexico (San Luis Potosí). On P. aculeata eggs are laid on leaves (rachis and pinnules) and tips of other vegetative parts. First instars feed inside the rachis. They then move to the axil and make a tunnel within it. Larvae leave the tunnel at night to feed on the pinnules and rachis of leaves. Frass can be seen protruding from tunnels especially during the mid-life of larvae when the tunnel is being actively excavated. Pupation usually occurs in tunnels, and when adults emerge the exuvium is partly exuded from the tunnel. Adults of Rudenia leguminana emerged from stems (n = 417 individuals), flowers (n = 90), and pods (n = 29) of the host.
Host Specificity Trials. The large number of individuals of R. leguminana reared from P. aculeata in the field indicated that this herbivore was worthy of further consideration as a biocontrol agent, so additional studies were conducted. "No-choice" host specificity trials were performed in 2006. Three females and 4 males were placed in bags covering tips of 6 test plant species and the control, P. aculeata (Table 1). Successful development on 5 non-target hosts resulted: Parkinsonia praecox (Ruiz & Pav.) J. A. Hawkins, Mimosa asperata L., Mimosa pigra, Delonix regia, and Acacia farnesiana. However, oviposition, development, and adult emergence of R. leguminana on non-target hosts may have been an artifact of the confined conditions of the trials. For example, R. leguminana were never reared from Mimosa pigra in the field despite considerable research on the herbivores of this plant over many years in the same geographic region.
To better establish the field host range, an open-field trial was conducted in a plot at La Aguada, Veracruz, Mexico. Eleven plant species were grown in a plot: P. aculeata, Mimosa asperata, Mimosa pigra, Mimosa pudica L., Acacia farnesiana, Caesalpinia pulcherrima, Calliandra grandiflora (L'Her.) Benth., Desmanthus virgatus (L.) Willd., Inga jinicuil G. Don, Leucaena leucocephala, and Tamarindus indica (all Leguminosae). Two releases of lab-reared adults were made; the first consisted of 42 females and 42 males on 24 July 2006 and the second of 52 females and 52 males on 9 August 2006. When damage to tips of the plants was observed, the terminal area was bagged to capture emerging adults. In these trials, R. leguminana successfully developed only on P. aculeata (20 from the first trial and 34 from the second). Although R. leguminana successfully developed on Mimosa asperata, Mimosa pigra, and Acacia farnesiana in the no-choice trials, no feeding or adult emergence was detected on these plant species in the openfield trials. This suggests that if not confined with the non-target hosts, R. leguminana may not oviposit on them, or at least development is far less likely.

Molecular Analyses.
Because few native tortricids range from the northeastern U.S. to Venezuela, it is possible that more than one species is concealed within the material examined. Razowski (1985) indicated that specimens from Sonora, Sinaloa, and Baja California have genitalia very similar to "R. leguminana and may be conspecific with that species," even though they exhibit some external and genital variation. Although some variation (in facies and genitalia) is present in the material reared from P. aculeata, the variation does not co-vary and is not concordant with geography.
Hence, four genetic markers were analyzed to see if molecular data could differentiate groups within this broad geographic distribution. In one data set (Data Set I) two mitochondrial markers were examined, a 781basepair region of cytochrome c oxidase I (COI) and a 685-basepair region of cytochrome c oxidase II (COII), and a 531basepair region of the nuclear marker 28S domain 2 (D2). In Data Set I 5 individuals were sampled from Mexico, 3 from Nicaragua, 1 from Arizona, and 1 from Nebraska. In addition, another data set (Data Set II) was analyzed which consisted of the barcode region of COI. In Data Set II 29 individuals of Rudenia leguminana were sampled ranging from Virginia to Venezuela, and 3 other tortricids as outgroups (Acleris semipurpurana, Aethes biscana, and Eugnosta busckana). Evolutionary pairwise distances in PAUP* 4.0b10 (Swofford 2002) were calculated using a maximum likelihood model with parameter estimates from a maximum likelihood analysis carried out in Garli 0.951 (Zwickl 2006) with the GTR+I+G model, selected as best by Modeltest 3.7 (Posada and Crandall 1998) based on the AIC criterion. A tree based on Data Set II was inferred using MrBayes 3.1.2 (Huelsenbeck and Ronquist 2001) with a data set partitioned by codon position into 3 subsets, with uninformative priors except for the branch length prior, 5 million generations (burnin=2 million), 4 chains (1 cold, 3 heated), and 2 concurrent runs. For establishing branch length prior, the empirical Bayes method was employed, calculating mean branch length for the maximum likelihood tree from 4 separate Garli analyses and using the resulting value for creating an exponential distribution for this prior. Whether the MCMC chains converged was assessed by examining the potential scale reduction factors and loglikelihood plots over time.
Evolutionary distances between individuals in both Data Set I and Data Set II strongly suggest that these individuals belong to more than one species. In Data Set I, evolutionary distances for individuals from the same geographic area are at the level expected for conspecifics (below 1% for both mitochondrial markers) (Hebert et al. 2003), whereas distances for individuals from more distant locations show a divergence an order of magnitude greater (Tables 2 and 3). There is very little variation in D2. Based on the analysis of the Data Set II (COI barcode) it appears that there are at least three different species (Figure 15), with distances within species ranging 0.0-2.3% and between species 4.9-13% (Table 4). Divergence levels of greater than 3% frequently are interpreted to represent separate species (Hebert et al. 2003;Hebert et al. 2004). It is odd that one of the individuals from Venezuela (VE 1) is about 5% distant from the three other Venezuelan specimens (however, it is possible that this is the result of contamination). Also, the origin of the four specimens from Mexico (MX 1-4) that cluster with the Venezuelan outlier is uncertain because these specimens are larvae that were intercepted at U.S. ports-of-entry from "Mexico" on Pithecellobium dulce (i.e. the exact point of origin is unknown, although it is likely to be northern Mexico).
Although geographic sampling is limited, these results strongly suggest that more than one species is present in the material examined: species A, including individuals from the eastern U.S. (e.g. Kansas, Louisiana, Mississippi, Texas, and Virginia); species B,  (Figure 15). Owing to weak support for the rest of the tree, we refrain from drawing conclusions regarding the assignment to species of the remaining individuals.  LPL 11154-157, 11193, 12208, 12224, 12397, 12409, 12417, 12422, 12457, 12458, 12459. Tamos, La Cortadura, 22° 10.98' N, 98° 1.36' W, LPL 10849, 10853, 10905, 12410, 12420, 12427, 12461   Cochylis sp. (Figure 10) Two females of an undetermined species of Cochylis were reared from Veracruz, Mexico. The taxonomic disarray of the group and the large number of undescribed taxa combine to inhibit accurate species-level identification.

Summary
Biology. Although most Cochylini worldwide feed on Asteraceae (Razowski 1970), there are numerous deviations from this pattern, and Leguminosae is not an unusual host plant family for members of this tribe.

Ofatulena Heinrich, 1926
As defined by Heinrich (1926), Ofatulena, included two species: O. duodecemstriata (Walsingham) and O. luminosa Heinrich. The two share a whitish gray, finely striated forewing with a distinct ocellar patch in the tornal region; as in most Grapholitini, males of Ofatulena lack a costal fold. The male genitalia (illustrated by Heinrich 1926) have a distally swollen cucullus with a row of long, stout, flattened, marginal spiniform setae along the outer portion of the valva. "Enarmonia" jamaicana Walsingham was transferred by Razowski to Ofatulena without comment in the checklist of Neotropical Lepidoptera (Powell et al. 1995). However, because it lacks the typical forewing pattern and shape, and the male genitalia deviate considerably from those of O. duodecemstriata (Walsingham) and O. luminosa Heinrich, we remove it from the genus and transfer it provisionally to Cydia (new combination), with which the facies and genitalia are more similar. Ofatulena duodecemstriata usually can be distinguished superficially from O. luminosa by its slightly greater forewing length and absence of peach or orange scaling in the forewing ocellar patch.
The male genitalia of O. duodecemstriata have a linear patch of larger spiniform setae along the lower edge of the valva subapically and an extra, stout spiniform seta near the middle of the apical region. In O. luminosa the setae are more numerous, but much finer and thinner, and usually are easily dislodged in slide-mounted preparations. Specimens reared from Parkinsonia praecox in Oaxaca, Mexico apparently represent a third and undescribed species of Ofatulena. Based on collection records (USNM) and published literature (Heinrich 1926;MacKay 1959;Woods 1992), Ofatulena has been reared only from Leguminosae.

Ofatulena duodecemstriata (Walsingham) (Figure 11)
Ofatulena duodecemstriata ranges across the western United States from California to Texas, north to Utah and south into Mexico. In addition, a series of this species (n = 6) was discovered in the USNM with the following data: Venezuela, Lara, Puente Torres, 24 km E Carora, 10 March 1978, thorn forest, blacklight, J. B. Heppner. The latter represent a considerable increase in the previously documented geographic range of the species. Although the Venezuelan specimens are considerably smaller than those from North America, the genitalia are identical to North American specimens.

Biology. The primary larval host in North
America is mesquite ("mesquite beans"), Prosopis sp. (Leguminosae) (Heinrich 1926;MacKay 1959). Although there is a specimen in the USNM reared from Verbascum thapsus L. (Scrophulariaceae), this is almost certainly an error. One specimen from P. aculeata was reared in Mexico and two in Venezuela.
Biology. Specimens of O. luminosa were reared from P. aculeata in Oaxaca, San Luis Potosí, and Veracruz, Mexico and from P. texana in San Luis Potosí. The abundance of individuals in the field and the damage it causes indicated that this herbivore could have potential as a biocontrol agent, so additional observations and studies were conducted.
Larvae bore in leaf and stem tips, mature green stems, and green seeds. The stem may appear swollen as a result of the presence of the larvae. Most damage is concentrated in the pith in the center of the stems. When mature, the larva prepares an exit hole distinguished by a frass and silk structure. From 218 bagged stems which showed evidence of damage, 96 adults emerged. Two to 3 damaged stems were observed per plant on 2 transects showing that this insect is consistently and abundantly available (Table 5). Overall, adults of Ofatulena luminosa emerged from stems (n = 164 individuals), pods (n = 55), and flowers (n = 2) of the host. Adults live for a mean of 6 days.

Surveys of Natural Host Plant Use.
Twentythree legume species growing in the same habitat as P. aculeata were recognized (5 species of Caesalpinioideae, 11 species of Mimosoideae, and 7 species of Faboideae) ( Table 6). Stems with evidence of damage were bagged for adult emergence. In October 2008 when the site was threatened with flooding, all stems were collected and taken to the lab where they were held in plastic bottles until insects emerged, or were dissected if the stem showed signs of drying out. All insects that emerged were pinned, labeled, and identified. Ofatulena luminosa emerged only from Parkinsonia aculeata and P. texana (A. Gray) S. Watson var. macra (I.M. Johnst.) Isely.

Crocidosema lantana Busck (Figure 14)
Crocidosema lantana is indigenous to Mexico; it is recorded from Florida to Costa Rica. It was described from Hawaii (Busck 1910) based on specimens imported from Mexico for biological control of weedy lantana (Lanana camara L.: Verbenaceae) (Perkins and Sweezy 1924;Zimmerman 1978). It subsequently was introduced into Australia (Tyron 1914;Common 1957) and Miocronesia (Zimmerman 1978) for biological control.

Concluding Remarks
The ten species of Tortricidae reared from P. aculeata in this study show a range of feeding modes from leaf rollers to borers in stems and reproductive parts. Their systematic positions vary from clearly defined species to possible species complexes, and include members of both of the largest subfamilies -Tortricinae and Olethreutinae. Their host specificity also varies widely from extreme generalists to possible specialists on the genus Parkinsonia. Although 7 of the 10 species are clearly not suitable as biocontrol agents, 2 species show some potential. Rudenia leguminana is probably a complex of species, one or more of which may be a specialist. Further work is needed to understand the systematics of this group. Ofatulena leguminosa is possibly a specialist on the genus Parkinsonia. This level of specificity may be suitable for the importation of this species into Australia as no other species of Parkinsonia occur there. Further work is ongoing to determine its host specificity and its impact on the target plant.