Cactoblastis cactorum (Berg) (Lepidoptera: Pyralidae) use of Opuntia host species in Argentina

Transcription

1 Biol Invasions (2014) 16: DOI /s ORIGINAL PAPER Cactoblastis cactorum (Berg) (Lepidoptera: Pyralidae) use of Opuntia host species in Argentina Laura Varone Guillermo A. Logarzo Juan A. Briano Stephen D. Hight James E. Carpenter Received: 24 September 2013 / Accepted: 5 March 2014 / Published online: 9 March 2014 Ó Springer International Publishing Switzerland 2014 Abstract A central aspect in biology and ecology is to determine the combination of factors that influence the distribution of species. In the case of herbivorous insects, the distribution of herbivorous species is necessarily associated with their host plants, a pattern often referred to as host use. Novel interactions that arise during a biological invasion can have important effects on the dynamics of that invasion, especially if it is driven by only a subset of the genetic diversity of the invading species. This is the case of the wellknown South American cactus moth, Cactoblastis cactorum, a successfully used biological control agent of nonnative Opuntia cacti in Australia and South Africa, but now threatening unique cactus diversity and agriculture in North America. We studied the patterns of host plant usage by and host plant availability for C. cactorum under field conditions in Argentina, covering the geographical range of the four C. cactorum phylogroups and the recently documented southern distribution. We also assessed female preference and L. Varone (&) G. A. Logarzo J. A. Briano FuEDEI, Fundación para el Estudio de Especies Invasivas, Bolívar 1559 (1686) Hurlingham, Buenos Aires, Argentina lauvarone@fuedei.org S. D. Hight Center for Biological Control, USDA-ARS-CMAVE, FAMU, Tallahassee, FL 32308, USA J. E. Carpenter USDA-ARS-CPMRU, Tifton, GA 31794, USA larval performance under laboratory conditions. Cactoblastis cactorum showed a geographical pattern of host use in its native range that was related to host availability. Laboratory assays of female preference showed some degree of preference to oviposit on O. ficus-indica, O. leucotricha and O. quimilo, but it was not positively correlated with the performance of larvae. These findings contribute to the further comprehension of the host use dynamics of C. cactorum in the insects native range, and could provide useful information for assessing the risk and future spread of this insect in North America. Keywords Cactus moth host Availability Host utilization Native range Introduction A central theoretical aspect in biology and ecology is to identify the factors responsible for determining the distribution of species in their environment (Darwin 1859; von Humboldt 1805). In the case of herbivorous insects, species are necessarily associated with their host plants, and their distribution among host plant species at the landscape level can be described as patterns of host use. In addition, several ecological and evolutionary studies have identified geographical differences in the patterns of host use by populations of phytophagous insects (Wiklund 1974; Fox and

2 2368 L. Varone et al. Morrow 1981; Singer 1983; Thompson 1988b; Thomas et al. 1990; Logarzo et al. 2011). Geographical differences in host use have been generally attributed to one of two driving factors; differences in female preferences among insect populations (Singer 1983; Singer et al. 1991; Forister 2004; Thompson 1988b), or spatial variation in host plant abundance across the geographical environment (Wiklund 1974; Courtney and Forsberg 1988). Preference is considered a behavioral trait and the feeding and/or oviposition on one plant species is more likely to occur than on alternate species (Singer 2000). Female preference is conventionally assessed by offering individuals several hosts simultaneously and expecting that she will follow a hierarchical order of choosing various hosts (Thompson 1988a). Differences in host use patterns identified in the field could simply reflect geographic variation in the availability of acceptable plant species, even though the females have identical host preferences (Wiklund 1974; Thompson 1988b). One method to evaluate the preference among populations across varied geographies is to compare the proportion of resources in the diet of an herbivore as a function of host plant availability (also known as electivity, Singer 2000). If insects encounter hosts in proportion to their abundances, and the probability of accepting each host species does not change with host abundance, then the preference will likely be similar across a landscape with varying relative abundances of host types (Kuussaari et al. 2000). Patterns of host use for herbivorous insects invasive to new areas may be different from patterns in their native homeland because the herbivores will likely encounter novel plant species in the invaded areas. All colonization events are initiated by a female either accepting or not accepting a plant as an oviposition substrate (Renwick and Chew 1994; Stefanescu et al. 2006). Host use patterns are therefore determined by adult behavior through host-selection and female preference (Jermy 1984), but ultimately affected by the survival of the offspring on the oviposited host (Van Nouhuys et al. 2003). As a general trend, the novel interactions between an herbivore and potential hosts that arise during an invasion can have important effects on the dynamics of the invasion (Sih et al. 2010). In addition, if an invasion by a herbivore species includes only a subset of the species genetic diversity present in its native area, i.e., only one ecotype of the species invades the new area, then the complete array of the species co-evolved traits, including preference hierarchies, novel host counterdefense strategies, and/or local climate adaptations, may not arrive with the invading ecotype, resulting in different implications for the invasion dynamics than if another ecotype invaded (Brooks et al. 2012). This is the case of the well-known South American cactus moth, Cactoblastis cactorum (Berg) (Lepidoptera, Pyralidae), successfully used for the biological control of Opuntia spp. in Australia and South Africa. Since 1989, this moth has been unintentionally invading Opuntia spp. in the southeastern United States, threatening the unique cactus diversity and industry in the western United States and Mexico (Hight and Carpenter 2009). In Argentina, the cactus moth is oligophagous on a long list of native and exotic Opuntia spp. (Varone et al. 2012). Larvae of C. cactorum feed gregariously inside cladodes of cacti, consuming the interior of the stems, introducing secondary infections by microbial pathogens, and often leading to plant death (Starmer et al. 1988). Larvae develop through five instars, drop to the ground and spin cocoons usually near the base of the infested plant. Adult moths emerge, mate, and lay coinlike eggs stacked on top of one another to form a small stick ( eggstick ) attached to the stem segments of the host plant (Dodd 1940; Pettey 1948;Mann1969). Marsico et al. (2011) identified four different genetically isolated groups of C. cactorum in Argentina based on structure of COI mitochondrial gene sequence data and concluded that all exotic populations outside South America were derived from a single genetic source in the moth s native range located in Entre Ríos province. The analyzed samples of C. cactorum from Argentina were grouped according to four geographic regions, and a significant portion of the genetic variation was explained by these geographically distinct groups, termed phylogroups. Reasons for the limited gene exchange and genetic structure found among these phylogroups remains unknown and deserve further investigation. No apparent geographic barriers between regions exist to limit gene exchange. Furthermore, the concordance between McFadyen s biotype map based on larval morphology (McFadyen 1985) and the pattern of genetic structure found by Marsico et al. (2011) suggests the potential that these groups represent ecotypes within the species (Brooks et al. 2012). Brooks et al. (2012) suggested that the exportation of a single ecotype of the moth from Argentina would have imposed strong biotic constraints on the insects exotic distribution. For example, if the exported C. cactorum

3 Use of Opuntia host species in Argentina 2369 ecotype s preference and host range included only a subgroup of the entire host range of the species, then this could have implications on the invasion dynamics of this insect. Brooks et al. (2012)demonstrated that hostuse in the native range differed among various genetic groups and provided some evidence that these patterns reflected ecotypic variation. However, the lack of regional data on the relative abundance of host plant species prevented the identification of whether the observed pattern was a function of host availability or host preference. In addition, Jezorek et al. (2010) studied the female preference and larval performance of C. cactorum invading Florida on both Florida native and naturalized common Opuntia species from Mexico and the southwestern United States. While the study demonstrated certain hierarchies in C. cactorum oviposition preference, preference was not correlated with larval performance, as predicted by the optimal oviposition theory (Jaenike 1978). This theory hypothesizes that females lay their eggs on host plants based on the hosts quality and suitability for her offspring. The lack of a relationship between female preference and larval performance for C. cactorum was attributed in part to the lack of coevolutionary history between the cactus moth and the North American Opuntia species. We studied the patterns of field host use by C. cactorum on exotic and native Opuntia spp. commonly occurring in Argentina. The geographical range of all four phylogroups identified by Marsico et al. (2011) and the southern distribution recently documented by Briano et al. (2012) were covered in this study. Field use of C. cactorum was determined based on host availability and host attack. We also assessed the laboratory oviposition preference and the subsequent larval performance for one of the C. cactorum phylogroups. Host use patterns and preferences of C. cactorum in its native range will fill the gap from previous studies and provide information on the potential effect and spread of the cactus moth in its invasive range of North America. Materials and methods Field studies Surveyed area Field surveys to identify host plant availability and utilization by C. cactorum were conducted in Argentina between January 2008 and May Selected Fig. 1 Distribution of Cactoblastis cactorum on Opuntia hosts in Argentina. Shaded areas correspond to geographical regions with different phylogroups of C. cactorum: west (W), northwest (NW), northeast (NE), east (E), a hybrid region with overlapping phylogroups (H) (Marsico et al. 2011), and southern populations (S) that have not been genetically analyzed. Black circles indicate positive patches where eggs or larvae of C. cactorum were found, and white circles correspond to Opuntia patches free of attacks

4 2370 L. Varone et al. sites were within the native distribution of C. cactorum and included both monocultures, typical of the ornamental prickly pear Opuntia ficus-indica (L.) Miller, and patches of native hosts along roadsides and natural areas. The total surveyed area was divided into six regions, four of which corresponded to the different C. cactorum phylogroups identified by Marsico et al. (2011). Phylogroups exhibited a high degree of genetic structure among populations in different geographic regions, and the four isolated groups used in this study were identified based on structure in COI mitochondria. In the fifth region, in the middle of the four distinct phylogroup regions, all four phylogroups were present. The sixth region was called the southern region and consisted of C. cactorum populations which had not been genetically analyzed. The regions were roughly located as follows: (1) western, between 26 S 32 S and 62 W 67 W; (2) eastern, between 28 S 33 S and 57 W 60 W; (3) northwestern, between 23 S 26 S and 62 W 66 W; (4) northeastern, between 23 S 27 S and 58 W 61 W; (5) hybrid, between the four regions; and southern, between 32 S 40 S and 58 W 64 W (Fig. 1). Surveyed plants in each region were selected while driving along highways and roads. Stops were made along the roadway more or less systematically every 30 km after observing an Opuntia spp. patch. Number of patches, size of patches, and C. cactorum presence were recorded per region. DIVA GIS software (Hijmans 2012) was used to place sampled sites onto a map. Host plants studied A total of eight Opuntia species or varieties was found commonly during the surveys: the seven South American native species O. quimilo K. Schum., O. megapotamica Arechav., O.elatavar. elata Link & Otto ex Salm-Dyck, O.elata var. cardiosperma (K.Schum.) R.Kiesling, O. anacantha Speg. O. bonaerensis Speg., and O. sulphurea Gillies ex Salm-Dyck; and the nonnative North American species O. ficus-indica. The following eight Opuntia species were found in fewer than five sites so were arbitrarily considered rare and excluded from the analysis: the three Mexican native species O. robusta H. L. Wendl. ex Pfeiff., O. leucotricha DC., and O. microdasys (Lehm.) Pfeiffer; and the five South American native species O. arechavaletae Speg.,O. penicilligera Speg., O. monacantha (Willdenow) Haworth, O. aurantiaca Gilles ex Lindleyand, and O. salmiana Parmentier ex Pfeiffer. Growth habit of the sampled Opuntia spp. was variable (Kiesling 2005; Kiesling and Meglioli 2003; Anderson 2001). Detailed information on the characteristics of these plants is given in Table 1. Table 1 Growth habit of the sampled Opuntia spp. commonly found during the surveys in Argentina Species/variety Growth/high Spines Cladodes Distribution across regions a Observations Origin North America O. ficus-indica Tree/1-6 m Spineless Oblong All Grown in plantations Origin South America O. megapotamica Shrub Long Rounded W, E, H, S Dense mucilage O. el.var. elata Sub-arborescent Flat Thick, subrhomboids W, E, H, S Spines only present at maturity O. el.var. cardiosperma Shrubby erect/2 m Curved, long, thick Elliptical, narrow all Reddish glochids O. quimilo Tree/4 m Long Large, elliptical to W, NE, NW, H Prominent areoles obovate O. bonaerensis Erect branches/1-2 m Spineless Obovate to elliptical S Non or few glochids O. anacantha Prostrate Reflexed Flat, elongate, lance W, E, NE, NW Clonal mat shaped O. sulphurea Creeping Irregular Round to obovate W, NW, H, S Clonal mat a Distribution in Argentina according to regions of Fig. 1 and Table 2: W western, E eastern, NE northeastern, NW northwestern, H hybrid, and S southern

5 Use of Opuntia host species in Argentina 2371 Host plant availability and use Host plant availability by region was estimated as the number of Opuntia patches that contained at least five individual plants of the same Opuntia spp. If an Opuntia spp. occurred in less than five patches, then this species was not considered in the study and excluded from analysis. A total of 357 patches was checked in all six regions, 23 of which corresponded to one of eight species found less than five times in our survey of Argentina (O. robusta, O. leucotricha, O. microdasys, O. arechavaletae, O. penicilligera, O. monacantha, O. aurantiaca, and O. salmiana). Host use by C. cactorum was measured as the number of patches (as determined above) that supported eggsticks or larvae (attacked plants). Eggsticks and larval presence were combined as a measure of host use because field surveys were not all conducted during times that eggsticks were present in the field. The presence of larvae was confirmed by opening cladodes with signs of internal feeding. In patches containing 5 30 Opuntia plants, all plants were visually inspected for C. cactorum. At larger patches of over 30 plants, only 30 Opuntia plants were randomly selected and inspected for C. cactorum. Bias for patch size and Opuntia species sampled was reduced by conducting the survey over a large geographic area and in a systematic fashion. Patch density was highly variable across the surveyed area, and patch size was not a species specific characteristic. All species were found in low and high density patches. Counting the number of individual plants of Opuntia species in each patch would be a more accurate estimate of host availability, however, distinguishing between plants in large patches, was often impossible given the clonal growth habit of Opuntia. Laboratory experiments Female preference: multiple choice Female oviposition preference was evaluated in 2009 inside mesh field cages ( m) at the Hurlingham, Argentina laboratory with two multiple choice experiments. The first experiment included 10 Opuntia species, six native species most frequently found during the surveys (O. anacantha, O. megapotamica, O. elata var. elata; O. elata var. cardiosperma, O. quimilo and O. sulphurea), one rare native species (O. arechavaletae), and three species native to North America (O. ficus-indica, O. robusta and O. leucotricha). Ten pots, each containing a different rooted Opuntia sp., were randomly placed in a circular arrangement inside the field cage. Ten couples of newly emerged C. cactorum adults were released in the center of the cage. After 5 d (estimated duration of adult survivorship), all plants were carefully inspected for the presence of eggsticks. The number of eggs/ eggstick was counted for each eggstick with a 109 hand lens. The second experiment was conducted in the same manner as the first but with a subset of only six Opuntia species in the cage. The six host plant species were two North American species attacked more often than expected by chance in the first experiment (O. ficus-indica and O. leucotricha), and four South American species, two avoided for oviposition in the first experiment (O. megapotamica and O. elata var. elata); and two identified from the literature as non-hosts for C. cactorum, O. quimilo and O. sulphurea (Zimmermann et al. 2007). Six host plant species were often found growing together at sites during the field surveys. Each of the two experiments was replicated ten times with new potted plants randomly placed in a circular arrangement inside the cage. Replicates were conducted sequentially as only six field cages were available. Potted test plants were initiated 1.5 years before the experiment to stabilize the growth of the plants. Cladodes of each host species were collected in the field, transported to the laboratory, and potted with standard soil in a greenhouse under irrigation. Prior to the experiment, the size of each potted plant was estimated by summing the foliar area of all cladodes on a plant, considering the area of an ellipse [(p)(r 1 )(r 2 ), where r 1 is the major axis length and r 2 is the minor axis length]. Adult C. cactorum used in the caged preference experiments were reared in the laboratory from larvae or eggsticks previously collected on O. ficus-indica in Quilino, Córdoba province (S , W ). This moth population belonged to the western phylogroup. Larvae were reared on excised O. ficus-indica cladodes within a vented plastic container with cat litter (chemically free granulated clay) in the bottom to absorb plant and larval exudates during insect development. Rearing containers were checked three times per week and fresh cactus cladodes were added as needed. When the cladode was completely consumed, larvae exited the

6 2372 L. Varone et al. destroyed cladode and moved into the new cladode. As feeding increased, especially with 3rd 5th instar larvae, containers were checked daily to supply additional food as needed. When larvae were ready to pupate, they would leave the cladode and spin white cocoons on the bottom of the container lid. Cocoons with pupae were carefully removed from the container and placed individually into 50 ml plastic cups, held for adult emergence, and placed into the field cages at the beginning of the experiments. Larval performance: no-choice The performance of larvae that developed from eggsticks laid during the first multiple choice field cage experiments was compared among Opuntia species. Larval survival per replicate (plant) was estimated as the proportion of larvae that reached the pupal stage. Once adults emerged from the pupae, the proportion of females and the number of eggs per female were recorded. Date of oviposition and subsequent larval pupation were recorded to calculate the duration of egg-pupa development time. Larval performance on each host plant species was considered a combination of larval and pupal survival and development time, and also the potential fecundity and sex ratio of the resulting adults, as the components are not always correlated (Thompson 1988a). No-choice larval performance was determined on the potted plants on which their eggsticks were laid in the first multiple choice preference test. For each eggstick, the number of eggs hatched was recorded after the eggstick was removed from the plant. Potted plants containing developing larvae were covered with a voile cage to prevent larvae from escaping and to protect larvae from attack by natural enemies. If the larvae depleted the plant, the last cladode with larvae feeding inside was transferred to a new potted plant of the same species and positioned at the base of the new plant. When larvae were ready to pupate, they left the cladode and spun cocoons on the soil surface of the pot or the cage walls. Cocoons with pupae were carefully removed, placed into 50 ml plastic cups and held for adult emergence. A maximum of five pupae were placed into a single plastic cup. Laboratory colony eggsticks were added to plant species that received few, if any, eggsticks during the female preference test. Eggsticks containing about 30 eggs were added until five plants of any non-preferred species became infested. These supplemental eggsticks were derived from the same population at Quilino, Córdoba. Statistical analysis Differences in host plant species used by C. cactorum were analyzed with a Chi square goodness of fit test following the model of Neu et al. (1974); comparing host plant availability with the proportion of plants used. Availability of an Opuntia sp. was considered as the number of patches found with that particular Opuntia sp. The proportion of patches found with attack by C. cactorum was a measure of plant use by this insect. Assumptions of the statistical model were: (1) each attack on a host plant species found in the field, either eggstick or larvae, was considered an independent observation; (2) availability of each host plant for attack was the same for all cactus moth specimens. Two null hypotheses were tested: (1) C. cactorum uses host plant species in proportion to their availability, considering all host plant species simultaneously; and (2) C. cactorum uses plant species in proportion to their availability, considering all of the host plant species separately. A Chi square goodness of fit was used to test hypothesis 1. When a difference in plant use was detected, simultaneous confidence intervals (CI) were calculated utilizing the following Bonferroni t-statistic to test the second hypothesis: qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi CI ¼ ^p j t a=2k:n 1 ^p j 1 ^p j =n where ^p j is the proportion of plants used by C. cactorum and the a/2k is used to ensure that k (number of patches analyzed) simultaneous CIs have an overall a = This way, three categories of host use were established: (1) preference, if host use was higher than host availability (i.e., plant availability was lower than the lower limit of the CI); (2) proportional use, if host use was proportional to host availability (host plant availability was within the CI); and (3) avoidance, if host use was lower than host availability (i.e., plant availability was higher than the upper limit of the CI). Female C. cactorum oviposition preference from the multiple choice experiments was also analyzed with the Neu method; host plant availability (equal to 1 for all the options) was compared with the relative proportion of eggsticks laid on each test plant species. Number of eggs laid was compared among the Opuntia species with the non-parametric Friedman

7 Use of Opuntia host species in Argentina 2373 test because the data lacked independence. To determine if plant size could predict number of eggs laid, least squares linear regression was performed. In the larval performance experiment, proportions of larvae that survived were compared among Opuntia spp. using a single factor ANOVA, after the data were arcsine square root transformed to satisfy the assumptions of the ANOVA. Data on duration of egg to pupa stages and on female fecundity were square root transformed to satisfy the assumptions of the ANOVA, before being compared among Opuntia spp. using one-way nested ANOVA, with the factor individual plant being nested within Opuntia species. Only species of Opuntia that produced adults in two or more replicates were used for estimating measurements of development time and estimates of fecundity. Proportion of females of the offspring was non-normal and did not respond to transformations; they were compared among species using Kruskal Wallis. Means are reported ±SE and were separated by the Tukey Kramer statistic at P = Taxonomic identifications Cactoblastis cactorum was identified as larvae following McFadyen (1985) and confirmed with molecular analysis performed by T. Marsico at Mississippi State University, USA. Opuntia species were identified by F. Font (School of Pharmacy and Biochemistry, Herbario Museo de Farmacobotánica Juan Domínguez, Buenos Aires, Argentina). Insect and plant voucher specimens were deposited at the FuE- DEI collection. Results Field studies Host plant availability and use A total of 334 patches was found that contained at least one of the eight common Opuntia spp. (Table 2). Eggsticks and/or larvae of C. cactorum were found at 119 of those patches (35.6 %) on seven of the eight Opuntia species. Opuntia ficus-indica was the only species found in all six regions and was the most available occurring species over all six regions. Opuntia elata var. cardiosperma was the next most available species over all the regions, even though this species was found only in the eastern and northeastern regions. These two species also had the highest percentage of total attacked patches over all regions. Opuntia bonaerensis was found only in the southern region and had the lowest availability and use by C. cactorum of all attacked species. Opuntia sulphurea occurred in three regions, but was never attacked by C. cactorum, only by the congeneric species Cactoblastis doddi Heinrich (14.3 % of patches were used by C. doddi, data not shown). Of all 6 regions, the western region had the most available Opuntia spp. patches, followed by the northeastern and southern regions (Table 2). The use of these patches by C. cactorum was the highest in the eastern region and lowest in the southern and northwestern regions (Table 2, Fig. 1). Pooling each host plant species over all the regions and estimating frequency of attack, revealed that O. elata var. cardiosperma showed the highest percentage of attacked patches (56.3 %), followed by O. ficusindica, O. megapotamica, O. elata var. elata, and O. anacantha (36 45 %), and O. bonaerensis and O. sulphurea with the lowest percentage of attacked patches (6.7 and 0 %, respectively). Availability of specific host plant species varied between the six regions (Table 2). The western, eastern, and southern regions each had four host species; all contained the species O. ficus-indica and O. megapotamica, but different combinations of third and fourth species. Except for the eastern region, O. ficus-indica and O. megapotamica were used more by C. cactorum than the other host species in their regions. The two varieties of O. elata were used most often by C. cactorum in the eastern region. The remaining northeastern, northwestern, and hybrid regions each had three Opuntia host species, but only O. ficus-indica was present in all three. The northwestern region had the lowest number of Opuntia host species attacked by C. cactorum; only O. ficus-indica and O. quimilo. Cactoblastis cactorum showed no bias towards the use of any particular Opuntia host species for oviposition and larval development; most species were used in proportion to their availability, with the exception of O. bonaerensis and O. sulphurea that were avoided (X 2 = 25.1, P = , df = 7; proportion of O. bonaerensis plants available was 0.04, confidence interval of proportion of insect use of O. bonaerensis

8 2374 L. Varone et al. Table 2 Availability of Opuntia host plant species and their use by Cactoblastis cactorum in six regions of Argentina Species/variety Region Western Eastern Northeastern Northwestern Hybrid Southern Total Number of patches available (percentage of relative frequency) O. ficus-indica b 45 (48.9) 7 (14.0) 10 (17.0) 24 (61.5) 15 (40.5) 13 (22.8) 114 (34.1) O. el. var. cardiosperma 0 a 18 (36.0) 30 (50.8) 0 a 0 a 0 a 48 (14.4) O. megapotamica 13 (14.2) 11 (22.0) a 20 (35.1) 44 (13.2) O. quimilo 22 (23.9) 0 0 a 8 (20.5) 9 (24.3) 0 39 (11.6) O. el. var. elata 0 a 14 (28.0) (31.7) 0 a 27 (8.1) O. sulphurea 12 (13.0) (17.9) 0 a 9 (15.8) 28 (8.4) O. anacantha 0 a 0 a 19 (32.2) 0 a (5.7) O. bonaerensis (26.3) 15 (4.5) Total 92 (27.5) 50 (15.0) 59 (17.6) 39 (11.7) 37 (11.1) 57 (17.1) 334 (100) Number of patches used (percentage of relative frequency) O. ficus-indica b 22 (64.7) 5 (16.6) 2 (8.0) 6 (85.7) 8 (57.1) 4 (44.4) 47 (39.5) O. el. var. cardiosperma 11 (36.7) 16 (64.0) 0 27 (22.7) O. megapotamica 8 (23.6) 6 (20.0) 4 (44.4) 18 (15.1) O. quimilo 4 (11.7) 1 (14.3) 2 (14.3) 7 (5.9) O. el. var. elata 8 (26.7) 4 (28.6) 12 (10.1) O. sulphurea O. anacantha 7 (28.0) 0 7 (5.9) O. bonaerensis 1 (11.2) 1 (0.8) Total 34 (28.6) 30 (25.2) 25 (21.0) 7 (5.9) 14 (11.7) 9 (7.6) 119 (100) Percentage of patches used/available a Species present in the region but excluded from the statistical analysis (Neu Method, see text) due to their low occurrence b Origin North America was CI 0.03/-0.02; proportion of O. sulphurea plants available was 0.08; proportion of insect use of O. sulphurea was 0.0). When availability of host plant and use of host plant by C. cactorum were analyzed within each region, no significant interaction was found between availability and use. In most regions, C. cactorum attacked Opuntia species proportional to the species availability (east: X 2 = 0.23, P = 0.97, df = 3; northeast: X 2 = 7.36, P = 0.06, df = 3; northwest: X 2 = 0.43, P = 0.56, df = 1; hybrid: X 2 = 1.77, P = 0.44, df = 2 and south: X 2 = 4.42, P = 0.22, df = 3). The western region was different in that a significant positive interaction was found between host plant availability and host plant use (X 2 = 7.98, P = , df = 2). However, there still was no bias for plants attacked more than expected by chance in the western region, with O. ficus-indica and O. megapotamica attacked at random, and O. quimilo and O. sulphurea avoided (proportion of plants available was 0.28 and 0.13; confidence interval of proportion of insect use was CI = 0.12/-0.06 and 0.0 for O. quimilo and O. sulphurea, respectively). Laboratory experiments Female preference: multiple choice A total of 3,290 eggs was laid in 97 eggsticks on the ten different potted Opuntia species during the trial 1 experiment, including eggs from all ten replicates (Table 3). The number of eggsticks per Opuntia species ranged from 0 to 24, and the three species O. ficus-indica, O. quimilo and O. leucotricha were chosen as oviposition substrates more than expected by chance (X 2 = 44.59; gl = 9; P \ ). The highest number of eggs was laid on O. quimilo and on

9 Use of Opuntia host species in Argentina 2375 Table 3 Multiple choice oviposition preference experiments in field cages with 10 (trial 1) and 6 (trial 2) potted Opuntia species exposed to 10 Cactoblastis cactorum females Species/variety No. eggsticks laid (mean ± SE) Oviposition preference 1 No. of eggs laid (mean ± SE) Total plants attacked (percentage) Trial 1 Trial 2 Trail 1 Trial 2 Trial 1 Trial 2 Trial 1 Trial 2 Origin North America O. ficus-indica 13 (1.3 ± 0.2) 23 (2.3 ± 0.7) ± 24.7 ab 69.4 ± 28.9 bc O. leucotricha 16 (1.6 ± 0.2) 11 (1.1 ± 0.1) ± 16.9 b 35.3 ± 7.0 abc O. robusta 2 7 (0.7 ± 0.2) ± 21.2 ab 30 Origin South America O. megapotamica 6 (0.6 ± 0.1) 24 (2.4 ± 0.2) ± 11.5 ab 74.9 ± 24.2 bc O. el.var. elata 3 (0.3 ± 0.1) 3 (0.3 ± 0.1) ± 4.1 a 7.5 ± 5.0 a O. el.var. cardiosperma 2 8 (0.8 ± 0.1) ± 24.4 ab 40 O. quimilo 24 (2.4 ± 0.5) 23 (2.9 ± 0.3) ± 26.7 b 80.7 ± 23.5 c O. arechavaletae 2 9 (0.9 ± 0.2) r 24.6 ± 16.2 ab 30 O. anacantha O. sulphurea 11 (1.1 ± 0.1) 6 (0.6 ± 0.1) r ± 16.7 ab 23.8 ± 12.7 ab Total ,290 2,554 Significant differences (P \ 0.05) are indicated by different letters within each column 1 According to the Neu method (see text)?species selected for oviposition more than expected by chance -species selected less than expected by chance r species selected at random 2 Species not tested in trial 2 O. leucotricha, both species had 60 % of potted plants attacked. This high egg number was significantly different only from the small number of eggs laid on O. elata var. elata (Friedman T 2 = 1.32; df = 9; P = 0.03), which also had a fewer (20 %) number of plants attacked. Overall, however, C. cactorum females did not show a marked oviposition preference for a single Opuntia species measured in either number of eggs or number of eggsticks laid. In addition, plant size was not related to oviposition (r 2 = 0.05; P = 0.2). When six host plant species were evaluated for oviposition preference in ten replicated field cages, female C. cactorum laid a total of 2,554 eggs in 90 eggsticks, with 3 24 eggsticks per Opuntia species (Table 3). Opuntia ficus-indica, O. leucotricha and O. quimilo were again selected more often for oviposition than was expected by chance, with the inclusion of O. megapotamica (X 2 = 28.89; gl = 5; P \ ). The highest number of eggs laid was again laid on O. quimilo, with 70 % of the plants attacked. The high number of eggs on O. quimilo was significantly greater than the number of eggs laid on O. elata var. elata and on O. sulphurea, which had 20 % and 40 % of their plants attacked, respectively. Again, C. cactorum females did not show a marked oviposition preference for a single Opuntia species. Larval performance: no-choice Larval survival was significantly different among the Opuntia species (F = 4.64; df = 8, 59; P = 0.002) (Table 4). Survival on the North American species was higher than survival on the native species, with O. robusta significantly higher than all native species. Although the native species were not significantly different from one another, larval survival of the two native O. elata varieties was extremely low. Opuntia sulphurea was excluded from the analysis because all 382 larvae from 11 eggsticks laid on five plants (replicates) failed to develop and died when attempting to feed on this host plant species. Development

10 2376 L. Varone et al. Table 4 Larval performance of Cactoblastis cactorum on South and North American Opuntia spp. evaluated with laboratory nochoice trials Species/variety (n) Larval survival (proportion of eggstick) (no. eggsticks) Egg pupa development time (days) (no. eggs) No. eggs/female (no. females) Proportion of females (no. females) Origin North America O. ficus-indica (9) 0.45 ± 0.14 ab (9) ± 1.20 a (153) ± 7.71 b (66) 0.57 ± 0.07 O. leucotricha (9) 0.34 ± 0.14 ab (9) ± 2.39 b (108) ± b (43) 0.58 ± 0.13 O. robusta 0.86 ± 0.05 b ± 0.82 a (292) ± 3.48 b (94) 0.51 ± 0.03 Origin South America O. megapotamica 0.13 ± 0.12 a ± 1.02 c (60) ± 5.24 ab (19) 0.54 ± 0.12 (2) O. el.var. elata (5) 0.09 ± 0.07 a (5) O. el.var. cardiosperma (7) 0.07 ± 0.06 a (7) ± 1.91 c (22) ± ab 0.45 ± 0.05 (2) O. quimilo (9) 0.23 ± 0.10 a (9) ± 1.25 bc (126) ± 6.43 ab (31) 0.40 ± 0.10 O. arechavaletae 0.15 ± 0.07 a ± 1.22 b (58) ± 0.07 (4) O. anacantha (5) 0.22 ± 0.07 a (8) ± 4.95 b (54) 53.9 ± a (15) 0.30 ± 0.13 (8) O. sulphurea 2 (5) Significant differences (P \ 0.05) are indicated by different letters within each column. Mean ± SE are reported 1 Not included in statistical analysis because adults not produced in 2 or more replicates 2 Not included in statistical analysis because no larvae survived time from egg to pupae was also significantly different among the Opuntia spp. (F = 26.33; df = 7, 865; P \ ), being significantly shortest on O. robusta and O. ficus-indica, and longest on O. elata var. cardiosperma and O. megapotamica (Table 4). In addition, the nested analysis indicated differences in the duration of egg to pupa stage among individual plants of the same Opuntia species (P \ ), identifying high variation among replicates (plants) of the same species (treatments). Number of eggs per female was significantly different among the seven analyzed Opuntia hosts (F = 25.0; df = 6, 241; P \ 0.001) (Table 4). Females with the highest number of eggs were reared on the three non-native species O. ficus-indica, O. robusta and O. leucotricha, and the lowest on O. anacantha. Nested analysis again indicated differences between the replicates within each treatment (P \ ), meaning high variability within the treatments. The overall mean proportion of females of the offspring was not significantly different among the host plant species (Kruskall-Wallis H = 12.65; df = 8; P = 0.08), ranging between 0.14 and 0.58 (Table 4). Discussion In Argentina, C. cactorum showed a geographical pattern of host use unrelated to the phylogenetic groups corresponding to the insects geographic range. For example, O. ficus-indica was attacked across all regions sampled, but, in contrast, O. anacantha and O. bonaerensis were attacked only in the northeastern and southern regions, respectively. However, when

11 Use of Opuntia host species in Argentina 2377 host availability was considered, we found that host use was related to host availability in each region. Brooks et al. (2012) previously suggested that the genetic structure of C. cactorum populations in Argentina was related to host use patterns, because they observed a positive association for larvae from the eastern and northeastern regions with O. elata var. cardiosperma, and for larvae from the west and northwest regions with O. ficus-indica. However, our extended survey highlighted that the high frequency of attacks on those species was a consequence of high host plant availability of those species in those regions (Table 2). As a general trend, the North American species O. ficus-indica was the most common host of C. cactorum throughout Argentina (of the 114 available patches, 47 patches were attacked), followed by O. elata var. cardiosperma (of the 48 available patches, 27 patches were attacked). Considering the regions separately, the highest number of O. ficusindica patches attacked was concentrated in four regions (western, northwestern, hybrid, and southern), and O. elata var. cardiosperma in two (eastern and northeastern). In all regions, the most frequently attacked host plant species were the most abundant, and use was proportional to host plant availability. The laboratory multiple choice experiments conducted to investigate female host selection behavior identified that all hosts exposed to C. cactorum, with the exception of O. anacantha, were selected as oviposition substrates. Even though laboratory assays did not show a significant oviposition preference for a single Opuntia species, more eggs were laid on certain species (O. quimilo and O. leucotricha) than others (O. elata var. elata). The lack of congruency between field host use (proportional to availability) and the degree of oviposition preference for some species found in the laboratory might be explained if females did have preferences in the field. However, as the lifespan of the adult cactus moth is short (5 8 days) (Zimmermann et al. 2007), females may not find the preferred host and oviposit on the first host encountered as her life is ending, as suggested by Jaenike (1978) and Kareiva (1982). In either of the two situations (presence or lack of field female preference), it is clear that females oviposit on a wide variety of hosts in all regions, and the establishment and dispersion of the cactus moth will depend on larval performance more than the choice by females. Expansion of C. cactorum host range in introduced areas will occur because of the females low selectivity for Opuntia spp. hosts. Host range expansion will primarily be determined by the capacity of the larvae to develop on the host species selected by their mother. Furthermore, previous studies on larval performance indicated that the encounter and development with newly associated hosts, like O. ficus-indica, might provide defense-free space (Gandhi and Herms 2010) toc. cactorum and allow for better herbivore performance (Varone et al. 2012). According to the preference-performance hypothesis, females should lay more eggs on those hosts in which survival of their offspring will be the highest (Jaenike 1978). However, our results suggest that C. cactorum failed to identify such hosts, because larval survival on, for example, O. quimilo was low compared to other hosts, yet the laboratory preference for that species as an oviposition substrate was one of the highest. In addition, field observations indicated that O. quimilo was one of the two least selected hosts by C. cactorum. However, this apparent poor selection by the female may be a reflection of the poor performance and early death of the larvae on this host plant species, precluding the ability to score oviposition attempts on field plants. Information from life table studies on C. cactorum throughout Argentina indicated that O. quimilo was successfully used as a host in the western region, implying that this cactus moth population is potentially locally adapted to this host species (GL unpublished data). This type of local adaptation of C. cactorum to native host species was also detected with reciprocal crossing experiments with two C. cactorum populations; superior performance was found for the population originally collected and reared on O. megapotamica compared to the population from O. ficus-indica switched onto O. megapotamica (Varone et al. 2012). Alternatively, previous studies on C. cactorum demonstrated that females were attracted to oviposit on Opuntia plants previously attacked by C. cactorum (Robertson 1987; Myers et al. 1981), generating an egg clumping phenomenon. We also observed plants heavily attacked with eggsticks beyond the average number of eggsticks/plant, such as an attacked O. quimilo plant receiving 12 eggsticks, well above the average attack rate of 2.5 eggsticks/plant (unpublished data). The lack of congruency between larval performance and oviposition preference has been recorded

12 2378 L. Varone et al. for various insect species, and has been explained by diverse mechanisms (Thompson 1988a; Mayhew 1998; Forister 2004). For example, the time hypothesis proposed that an insufficient time had occurred for adaptation to be observed, and the enemy-free space hypothesis postulated that performance may be influenced by plant characteristics responding to defense against natural enemies (Thompson 1988a). In the case of C. cactorum, Jezorek et al. (2010) proposed the time hypothesis to explain the lack of congruency between preference and performance identified when evaluating C. cactorum on North American opuntiod cacti: without a co-evolutionary history between C. cactorum and the increasing number of new association host species, female C. cactorum that encounter new hosts do not always know what is best for their offspring. However, our results showed that even in the native range, C. cactorum females still did not select hosts based on suitability for their offspring. Regarding the enemyfree space hypothesis, life table studies conducted on O. quimilo (native and defended species) and O. ficusindica (non-native and not defended species) showed similar effects from natural enemies on population mortality factors for both host species (GL unpublished data). As a general trend, survivorship of C. cactorum has been found to differ significantly among host taxa (Pettey 1948; Robertson 1987; Johnson and Stiling 1996; Mafokoane et al. 2007), but where oviposition preference has been identified, host quality for larvae does not correspond with oviposition preference. Alternative hypotheses to explain this lack of congruency between oviposition preference and larval performance need to include whether female choice is adaptive in ways other than finding the better host for the performance of her offspring, as suggested by Scheirs et al. (2000) and Mayhew (2001). Plant size is another factor that could affect oviposition preference (Thompson and Pellmyr 1991). Myers et al. (1981) found some preference of C. cactorum females to oviposit on larger individuals, however, other studies have shown varying responses to plant size (Johnson and Stiling 1998; Tate et al. 2009; Robertson 1987; Jezorek et al. 2010). We found no effect of plant size on oviposition preference in laboratory trials. Despite our selection for study plants of the same age, plant structure of the different species could influence differences in size. The existence of different native phylogroups of C. cactorum (Marsico et al. 2011) may have been influenced by differences in host plant preference of the C. cactorum phylogroups, leading to local adaptation and genetic divergence. Because our surveys were conducted throughout the year, some patches were evaluated at times when females were not ovipositing and field preference could not be evaluated in the strict sense (eggstick number). However, the presence of feeding larvae were a substitute for eggsticks at these times and when larvae were found it signified that an eggstick was present, giving some idea of female field preference. We observed that patterns in the insects native range of host use were reflected by the variation in host availability, not female preference. Therefore, the factors determining the population structure and the gene flow restriction remain unclear. Even though the evaluation for laboratory female preference was conducted only with the western genetic group, we determined that C. cactorum had the same host use pattern (proportional to availability) for all phylogroups, including the phylogroup exported to Australia (and elsewhere in the world) as a biological control agent (eastern region). This finding suggests different consequences for the C. cactorum invasion of North America than if the exported moth haplotype had diverged in its host use pattern from other haplotypes in the native range. A lack of host use restriction represents a worst case scenario related to invasion consequences, because the exported haplotype exhibits the same preferences as all other haplotypes, and, in this system, the full breadth of host range because all haplotypes attack hosts based on their availability. Because the exotic populations display the entire host range in invaded areas, the spectrum of host plants suitable as hosts will be much wider than if the exported ecotype were a subgroup of the species, with a limited host preference. This study contributes to the understanding of the dynamics of host use by C. cactorum in the native range, and provides useful information for assessing the risk and future spread of this insect in North America. The rates, routes, and dynamics of the invasion of North America by this insect will be influenced by the quality and availability of hosts that the insect finds during its dispersion from the southeastern US to western US and Mexico.

13 Use of Opuntia host species in Argentina 2379 Acknowledgments We thank Fabián Font (University of Buenos Aires) for identification of Opuntia spp., Mariana Manteca Acosta for laboratory assistance, and Travis Marsico, Gary Ervin and Chris Brooks (Mississippi State University) for providing information on cactus moth genetic analysis. We also thank two anonymous reviewers for comments and suggestions that improved the manuscript. References Anderson EF (2001) The cactus family. Timber Press, Portland Briano J, Varone L, Logarzo G, Villamil C (2012) Extended geographical distribution and host range of the cactus moth, Cactoblastis cactorum (Lepidoptera: Pyralidae) in Argentina. Fla Entomol 95(1): Brooks C, Ervin G, Varone L, Logarzo G (2012) Native ecotypic variation and the role of host identity in the spread of an invasive herbivore, Cactoblastis cactorum. Ecology 93(2): Courtney SP, Forsberg J (1988) Host use by two pierid butterflies varies with host density. Funct Ecol 2(1):67 75 Darwin C (1859) On the origin of species by natural selection. John Murray, London Dodd AP (1940) The biological campaign against prickly-pear. Commonwealth Prickly Pear Board, Brisbane Forister ML (2004) Oviposition preference and larval performance within a diverging lineage of lycaenid butterflies. Ecol Entomol 29(3): Fox LR, Morrow PA (1981) Specialization: species property or local phenomenon? Science 211: Gandhi K, Herms DA (2010) Direct and indirect effects of alien insect herbivores on ecological processes and interactions in forests of eastern North America. Biol Inv 12: Hight SD, Carpenter JE (2009) Flight phenology of male Cactoblastis cactorum (Lepidoptera: Pyralidae) at different latitudes in the southeastern United States. Fla Entomol 92(2): Hijmans RJ (2012) DIVA-GIS (version ). Free mapping program, Jaenike J (1978) On optimal oviposition behavior in phytophagous insects. Theor Popul Biol 14(3): Jermy T (1984) Evolution of insect/host plant relationships. Am Nat 124: Jezorek HA, Stiling PD, Carpenter JE (2010) Targets of an invasive species: oviposition preference and larval performance of Cactoblastis cactorum (Lepidoptera: Pyralidae) on 14 North American opuntioid cacti. Environ Entomol 39: Johnson DM, Stiling PD (1996) Host specificity of Cactoblastis cactorum (Lepidoptera: Pyralidae), an exotic Opuntiafeeding moth, in Florida. Environ Entomol 25: Johnson DM, Stiling PD (1998) Distribution and dispersal of Cactoblastis cactorum (Lepidoptera: Pyralidae), an exotic Opuntia-feeding moth, in Florida. Fla Entomol 81:12 22 Kareiva P (1982) Experimental and mathematical analysis of herbivore movement: quantifying the influence of plant spacing and quality on foraging discrimination. Ecol Monogr 52(3): Kiesling R (2005) Cactaceae, Cactáceas. In: Troncoso NS, Bacigalupo NM (eds) Flora Ilustrada de Entre Ríos. Dicotiledoneas Arquiclamídeas. B: Geraniales a Umbelliflorales, vol IV, 4b. Colección Científica INTA, Buenos Aires, Argentina, pp Kiesling R, Meglioli S (2003) Cactaceae, Cactáceas. In: Kiesling R (ed) Flora de San Juan, República Argentina, vol II., Dicotoledóneas Dialipétalas (Segunda parte: Oxalidáceas a Umbelíferas)Estudio Sigma, Buenos Aires, pp Kuussaari M, Singer M, Hanski I (2000) Local specialization and landscape-level influence on host use in an herbivorous insect. Ecology 81(8): Logarzo G, Casalinuovo M, Piccinali R, Braun K, Hasson E (2011) Geographic host use variability and host range evolutionary dynamics in the phytophagous insect Apagomerella versicolor (Cerambycidae). Oecologia 165(2): Mafokoane LD, Zimmermann HG, Hill MP (2007) Development of Cactoblastis cactorum (Berg) (Lepidoptera: Pyralidae) on six North American Opuntia species. Afr Entomol 15(2): Mann J (1969) Cactus-feeding insects and mites. United States National Museum Bulletin, vol 256, Smithsonian Institution Press, Washington DC, USA Marsico TD, Wallace LE, Ervin GN, Brooks CP, McClure JE, Welch ME (2011) Geographic patterns of genetic diversity from the native range of Cactoblastis cactorum (Berg) support the documented history of invasion and multiple introductions for invasive populations. Biol Invasions 13(4): Mayhew PJ (1998) Testing the preference-performance hypothesis in phytophagous insects: lessons from chrysanthemum leafminer (Diptera: Agromyzidae). Environ Entomol 27:45 52 Mayhew PJ (2001) Herbivore host choice and optimal bad motherhood. Trends Ecol Evol 16(4): McFadyen RE (1985) Larval characteristics of Cactoblastis spp. (Lepidoptera: Pyralidae) and the selection of species for biological control of prickly pears (Opuntia spp.). Bull Entomol Res 75: Myers JH, Monro J, Murray N (1981) Egg clumping, host plant selection and population regulation in Cactoblastis cactorum (Lepidoptera). Oecologia 51(1):7 13 Neu CW, Byers CB, Peek JM (1974) A technique for analysis of utilization-availability data. J Wildl Manag 38: Pettey FW (1948) The biological control of prickly pear in South Africa. Sci Bull Dep Agric Union S Afr 271:1 163 Renwick JAA, Chew FS (1994) Oviposition behavoir in Lepidoptera. Annu Rev Entomol 39: Robertson HG (1987) Oviposition site selection in Cactoblastis cactorum (Lepidoptera): constraints and compromises. Oecologia 73: Scheirs J, Bruyn LD, Verhagen R (2000) Optimization of adult performance determines host choice in a grass miner. Proc R Soc Lond B Biol Sci 267(1457): Sih A, Bolnick DI, Luttbeg B, Orrock JL, Peacor SD, Pintor LM, Preisser E, Rehage JS, Vonesh JR (2010) Predator prey naïveté, antipredator behavior, and the ecology of predator invasions. Oikos 119(4):