Larval morphology and host use confirms ecotypic variation in Cactoblastis cactorum (Berg)
Despite their recognized importance in the literature, the contribution of native-range species interactions to invasion success has been inadequately studied. Previous authors have suggested that biases in the sampling of propagules from the native range might influence invasion success, but most contemporary invasion hypotheses focus on the development of novel interactions or a release from native consumers and competitors. When ecotypic variation exists in native host-consumer associations, the specific pattern of sampling across ecotypes could determine invasion success, especially when the genetic diversity among exotic propagules is low. The South American cactus moth, Cactoblastis cactorum (Berg), is an oligophagous consumer whose larvae feed on prickly pear cacti (subfamily Opuntioideae). The moth was collected from a small geographic area along the Argentina-Uruguay border in 1925 and was introduced to multiple continents as a biological control species, which has subsequently invaded North America. Here we show that groups defined by genetic structure in this species’ native range are concordant with distinct patterns of host association and larval morphology. Furthermore, in Florida populations, morphological traits have diverged from those found in the native range, and patterns of host association suggest that strong biases in host preference also occur in invasive populations. The documented history of C. cactorum introductions confirms that multiple attempts were made to export the moth, but that only a single ecotype was exported successfully. Additional work will be necessary to determine whether the observed host biases in North America reflect a rapid adaptation to naïve hosts or a conservation of traits related to specific aspects of the host-consumer association.
KeywordsEcotype Cactoblastis cactorum Argentina Florida Opuntia
Invasion success is determined by the complex interplay among ecological and evolutionary mechanisms that govern the relative importance of propagule pressure, gene by environment interactions, and the turnover in communities of interacting species from native to exotic landscapes (Catford et al. 2009; Kanarek and Webb 2010). Likewise, the almost universal impossibility of monitoring invaders as they become established and of acquiring data on exportation events—where they occurred and how many individuals were involved—has obfuscated the search for a general understanding of how genetic diversity and adaptation influence establishment success. The dearth of broad support for any specific mechanism governing successful invasion is reflected in the development of at least 29 hypotheses to explain invasion success (see Catford et al. 2009 for an excellent review).
The process of disentangling the relative ecological importance of propagule pressure, abiotic and biotic effects on invasion success is further complicated by a lack of knowledge about the exact number and geographic source(s) of introduction(s) and the consequent genetic diversity of founding populations for most invaders (Holland 2001; Novak and Mack 2005; Wares et al. 2005; Roman and Darling 2007). Bias in the geographic sampling of propagules can influence the potential for an invader to become established in an exotic landscape because of the differential representation of traits necessary for establishment and spread (Colautti et al. 2004). In host-consumer systems, for example, geographic mosaics of coevolutionary selective pressures between hosts and their consumers in the native range (Thompson 1982, 1994, 2005) can lead to the emergence of geographically isolated host and/or consumer ecotypes. Different ecotypes, defined as genetically distinct groups of populations that possess distinct morphological and functional traits, may possess a greater potential for invasion than others (e.g., Brunner and Frey 2010) if functional traits associated with these ecotypes influence successful establishment and spread in novel geographic ranges. Despite broad interest in the geographic mosaic theory of coevolution (Thompson 1982, 1994, 2005), and the recognition that ecotypic variation can be an important component of success in biological control (e.g., Ellison et al. 2008; Goldson et al. 1997), explicit study of native ecotypic structure and its influence on invasion success has been largely absent from the invasion literature (but see Neuffer and Hurka 1999; Linde et al. 2001).
Biological control species provide an excellent platform for studying how ecotypic variation might influence invasion success. The primary reason is that biological control introductions represent careful, well-documented invasions that lack confounding processes that frequently plague similar analyses of less well-documented invasions. For example, biological control programs can produce enormous propagule pressure which can increase the probability of establishment. The number of introductions and the geographic origin for many biological control organisms has also been recorded; this represents knowledge that can be used to better understand the role of genetic diversity and adaptation in invasions. Biocontrol organisms also generally are introduced without the suite of interacting species that exist within their native communities, and it has become standard practice that biological control introductions occur only when there is documented evidence that potential unintended interactions can be avoided (Blossey 1995). Despite these precautions and owing to unexpected complexities of host-consumer systems, biological control organisms occasionally attack non-target species. Both the success of the biological control agent and the likelihood of its ‘escape’ may be a consequence of the ecotypes represented in the original stock collected for biocontrol. Understanding how ecotypic variation in a biological control organism relates to the pattern of non-target host use could reveal its importance in determining invasion success.
The South American cactus moth, Cactoblastis cactorum (Berg), is an escaped biological control agent and an excellent candidate for the study of native ecotypic variation and propagule bias on consumer invasion. C. cactorum is a cactophagous moth whose larvae feed inside the stem segments of prickly pear cacti (genus Opuntia Miller). This moth is well-known as one of the most successful biocontrol organisms ever used. Its global spread began with the collection of larvae by Alan P. Dodd (1940) from native hosts that, based on current species availability, would probably be identified as O. elata Salm-Dyck var. elata and/or O. elata var. cardiosperma (K. Schum.) R. Kiesling, near Concordia, Argentina in January of 1925 (Brooks et al. 2012). These larvae were reared and mated in Argentina and approximately 2,750 eggs transported to Australia. Subsequent introductions were made from Australia into South Africa (18,000 eggs in December 1932; Pettey 1948) and from South Africa into the West Indies (100 larvae and 300 eggs in March 1956; Simmonds and Bennett 1966). While the spread of C. cactorum across the islands of the Caribbean and ultimately into the United States is not as well documented, recent work based on mitochondrial sequence data (Marsico et al. 2011) confirms McFadyen’s (1985) statement that all non-native populations of the moth are derived from Dodd’s original collection sites.
Across Argentina, C. cactorum exhibits a high degree of genetic structure among populations in different geographic regions. Marsico et al. (2011) identified at least four isolated groups across Argentina based on structure in COI mitochondrial haplotypes. We have previously demonstrated that host use in the native range differs among those genetic groups (Brooks et al. 2012). Thus, native-range genetic structure could have important consequences for the success of exported populations if it reflects ecotypic differences between populations in different regions (e.g., McFadyen 1985; Brunner and Frey 2010). This is particularly important for specialist species that interact with a limited number of species (e.g., specialist herbivores or parasites), because the signature of local coadaptation in the native landscape may influence the suite of potentially suitable hosts in non-native landscapes, even where environmental conditions are expected to be suitable.
Here we combine analysis of larval morphology and host association to evaluate the hypothesis that the genetic groups identified by Marsico et al. (2011) represent native ecotypes. McFadyen (1985) previously defined six C. cactorum “biotypes” using twelve morphological characters based on the pattern of black pigment on sixth instar larvae. Those biotypes were based on the relative frequency with which pigment characters occurred and anecdotal information on host associations, but no statistical analyses of larval morphology or host association were conducted to delimit biotypes. Our approach is to refine McFadyen’s (1985) work and use a formal statistical approach to assess the degree to which patterns of host use and larval morphology correspond to previously described geographic patterns in genetic structure (Marsico et al. 2011). Our understanding of the patterns of ecotypic variation in the native range and how interactions of those ecotypes have shifted after invasion will provide important insight into how the conservation of ecological traits might influence invasive spread.
Materials and methods
We examined the relationship between the genetic groups defined in Marsico et al. (2011) in order to explore the potential that these groups represent ecotypic variation. Larval morphology was scored using the twelve traits defined by McFadyen (1985). The reader is referred to McFadyen’s (1985) work for detailed descriptions and drawings of each larval trait. Larvae from each collection site were examined at 15× magnification and each character was recorded as either present or absent. The presence of faint pigmentation was noted so that we could maintain flexibility in defining trait presence as a binary state and test whether such a distinction was useful in identification. As noted by McFadyen (1985), larval markings “change somewhat in each instar.” Based on this and our own analysis of assignment probabilities, we excluded individuals with a head capsule width less than 1.67 mm because scoring of such specimens was unreliable. This resulted in the examination of a median of two individuals per Argentine collection and three per Florida collection. Host plant identification was independently confirmed by L. Majure (U. Florida) for collections made in Florida and by Fabián Font (U. Buenos Aires, Herbario Museo de Farmacobotánica Juan Dominguez, Buenos Aires, Argentina) for Argentine samples. Host plant taxonomy follows Majure et al. (2012) for North American species and Kiesling (2005) for Argentine species.
The probability of assigning a particular collection (defined as a group of larvae collected from a unique plant-site-date) to a region is simply the product of the probabilities of group assignment for each individual in the collection. Binomial tests were used to assess the association between the previously defined genetic groups, larval morphology, and host association by assessing whether the probability of correct assignment for each collection to the haplotype groups was better than would be expected if assignment occurred randomly (H0: p = 0.2). Analyses were conducted for larval morphology alone and then for the combination of larval morphology and host association.
The degree to which host associations (defined as the relative frequency of infestations on a host taxon) were a function of the relative frequency of host species in the region was examined using binomial tests. Data on the relative abundance of uninfested and infested plants across Florida (see Sauby et al. 2012 for sampling details) were used to determine whether the number of infestations across the plants examined differed from the relative frequency of that taxon across sites. Binomial tests were conducted for each of the five most abundant host plant species to test the null hypothesis that the proportion of plants infested is equal to the relative abundance of that species across the state. Data for the relative abundance of uninfested plants were not available in Argentina, rather we had collected data only on the presence of species at sampling sites. Thus, we conducted binomial tests to examine whether the relative frequency of sites in which a host taxon was infested in Argentina differed from the relative frequency of sites in which a host taxon occurred.
The proportion of individuals possessing the twelve morphological traits described by McFadyen (1985) (based on the bootstrapped data), the probability that each collection would be correctly assigned to group and the p value associated with a binomial test where the null hypothesis is that correct assignment occurs at random (i.e., Pr (correct) = 0.2
Two host species native to Florida, O. humifusa (Raf.) Raf. and O. stricta (Haw.) Haw., had the highest relative abundance (0.368 and 0.292, respectively) and accounted for 66 % of all plants surveyed across the state. Only O. ficus-indica (L.) P. Mill., a species native to Mexico but widely cultivated, occurred in both Florida (relative abundance = 0.009) and Argentina. Given observed differences in relative abundance in Florida, O. ficus-indica (2.7 % of plants infested; p = 0.024) and O. stricta (73.1 % of plants infested; p < 0.001) were infested more frequently than expected. Two other host species native to Florida: O. humifusa (5.9 % of plants infested; p < 0.001), and O. pusilla (Haw.) Haw. (3.2 % of plants infested, rel. abundance = 0.122; p < 0.001) were significantly less infested than predicted by their relative abundance.
There was a similar pattern of biased host association that occurred across sites in the Argentine portion of the moth’s native range. Sites in the native range which contained O. elata var. cardiosperma were more likely than expected to be infested (Expected Pr [infested] = 0.40; Pr [infested] = 0.706; p = 0.013). The numbers of sites infested for O. megapotamica Arechav. (present at 0.056 of sites (hereafter, site frequency), infested at 62.5 % of sites present; p < 0.001), O. e. var. cardiosperma (site freq. = 0.118, infested at 70.6 % of sites present; p < 0.001), and O. e. var. elata (site freq. = 0.056, infested at 75.0 % of sites present, p < 0.001) were higher than expected, based on the relative frequency of sites in which each of these host species was found. Infestation of these three native host species was exclusive to the western (W), northeastern (NE), and eastern (E) groups, respectively. The host that had the highest site frequency, the Mexican species O. ficus-indica (present at 50.7 % of all sites), was not infested more than expected by its relative frequency; the species was infested at 42.5 % of sites at which it occurred (p = 0.162). Three host species (O. e. var. elata, O. e. var. cardiosperma, O. ficus-indica, and O. megapotamica) accounted for 93.1 % of all site-level infestation across Argentina.
The proportion of collections containing larvae collected from a particular host species in each of the five regions along with the probability of correct assignment of collection to group using morphological and host association data (all p-values were < 0.0001)
O. elata var. elata
O. elata var. cardiosperma
O. ammophila Small
Analyses of larval morphology and patterns of host associations confirm the presence of distinct ecotypes in the native range of C. cactorum and suggest that knowledge of ecotypic variation is potentially important to understanding the dynamics of invasive, North American populations. These ecotypes match the geographic extent of the six biotypes described by McFadyen (1985), but differ morphologically and in patterns of host use from those previously proposed groupings. Direct comparisons between McFadyen (1985) and our own collections were difficult because of the lack of precise locations for McFadyen’s original collections. Even with the difficulty in making quantitative comparisons with McFadyen’s results, we did observe some common patterns in our data. Two characters (labeled as ‘j’ and ‘c’ by McFadyen (1985)) were relatively consistent across groups, and were consistent with patterns observed across larvae examined by McFadyen (1985). The mesothoracic spot ‘j’ identified by McFadyen (1985) in C. doddi (Heinrich) was absent from all specimens examined, regardless of ecotype. Likewise, the lateral spot ‘c’ was present in more than 70 % of the larvae examined across ecotypes as noted by McFadyen (1985). Few other general patterns from McFadyen’s analysis corresponded to differences among the previously defined genetic groups in the present study.
While many invasive species have a history of multiple introductions (e.g., Kang et al. 2007; Novak and Sforza 2008) that can provide the necessary genetic variation for rapid adaptation in exotic landscapes, C. cactorum is unique in having a single successful ‘export’ of individuals, representing fewer than 30 matings, from a single locale near Concordia, Argentina (Dodd 1927, 1940). This bias in the export of propagules from a small portion of the native range can have important consequences for future spread (Colautti et al. 2004). Marsico et al. (2011) confirmed that all North American populations of C. cactorum were derived from this collection along the Argentina-Uruguay border near Concordia. Curiously, however, data here suggest that the larval morphology of Florida populations is more distinct from populations in the eastern region (its origin) than was observed in the structure of COI haplotypes. The greater morphological divergence between the Florida and Eastern groups (Fig. 2) suggests a potential lack of stability in the characters chosen, or a more rapid divergence in morphological traits than has occurred at the COI mitochondrial locus. The apparent recent colonization of the western region of Argentina following the expansion of the cultivation of O. ficus-indica (Ervin 2012) is reflected in both the lack of clear genetic differentiation of larvae from that region, described by Marsico et al. (2011), as well as the highly variable morphology and host use differences observed in this study.
This morphological divergence may be useful as a tool to identify ecotypes across the native range, but it is the functional ecological variation (i.e., host use) associated with the observed morphological patterns and genetic isolation which may provide the greatest potential insight into mechanisms behind the successful invasion and continued spread of C. cactorum in North America. The lack of any bias towards the infestation of the abundant non-native host O. ficus-indica in Argentina, combined with regionally-biased infestation of native host taxa, O. e. var. elata by the eastern (E) group, O. e. var. cardiosperma by the northeastern (NE) group, and O. megapotamica by the western (W) group, suggests a potential for host traits to influence invasion success and patterns of host use in the exotic range.
Strong host preferences that occur in Florida may reflect the bias of exporting only the eastern ecotype. The observed over-representation of O. ficus-indica and O. stricta among infested plants across Florida may reveal important host traits that can drive patterns of infestation. These biases are likely driven by an increased relative fitness on some host species. These patterns might also arise if oviposition preferences reflected the quality of hosts for larvae, but field experiments conducted by Jezorek et al. (2010) suggest that oviposition preferences are decoupled from host quality. Sauby et al. (2012) found that sites in Florida containing O. stricta were more likely to be infested than sites in which this species was absent. Within infested sites however, Sauby et al. (2012) found that there was no apparent pattern in the rate of infestation for one taxon over another. Likewise, observed preferences among ovipositing females are inconsistent, especially for host taxa that occur in Florida (Robertson 1987; Johnson and Stiling 1996; Mafokoane et al. 2007; Tate et al. 2009; Jezorek et al. 2010). Together, these results suggest that any oviposition preference that might exist within sites is weak.
Survivorship and fecundity 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 see Woodard et al. 2012; Varone et al. 2012), but where oviposition preferences have been identified, host quality does not consistently correspond with oviposition preferences (Jezorek et al. 2010). Resource quality and availability play a key role in defining opportunities for consumer invasions (Andow 1991; Ostfeld and Keesing 2000a, b). Numerous authors have shown that various resource characteristics can influence the establishment and abundance of myriad exotic consumers (e.g., Holt et al. 2003; Andow 1991; Barbosa et al. 2009). Any differential fitness among larvae feeding on different host species is likely to include both secondary defenses (e.g., Woodard et al. 2012) and the genetic and environmental determinants of available resources within host tissues. Determination of the relative roles of plant defense response and the nutritional quality of host tissues on cactus moth invasion dynamics will require a combination of field and laboratory experiments to determine the actual suite of host and consumer traits involved—both in the native and the exotic ranges of the moth.
If native ecotypic variation reflects an underlying set of selective forces in this host-consumer association, then elucidation of the traits driving this association will be critical to understanding the relative importance of rapid evolution versus conservation of key traits in the success of the C. cactorum invasion of North America. There is some anecdotal evidence that local adaptation and the subsequent development of ecotypic variation is key to the moth’s success as both a biological control agent in Australia, and the observed pattern of infestation in Florida. Johnson and Tryon (1914) document a collection of C. cactorum from a cultivated cactus garden in La Plata, Buenos Aires. Four species of cactus were attacked in these gardens: O. quimilo Schum., O. ficus-indica (listed as O. decumana Mill.), O. robusta Wend, and O. e. var. cardiosperma (listed as O. chakensis Speg.). Of these, only O. ficus-indica and O. e. var. cardiosperma are known to occur in the eastern region (the source of larvae for Dodd’s eventual, successful introduction) (Brooks et al. 2012). This unsuccessful introduction was originally attributed to climatic differences between sites (Johnson and Tryon 1914), but the subsequent successful introduction from Concordia (approximately 400 km to the north) appears to refute this hypothesis. McFadyen (1985) suggested that this unsuccessful attempt (and another attempt to introduce C. doddi) failed because larvae suffer low fitness on some host species relative to others. Dodd’s (1940) eventual success may have resulted from a chance bias in the sampling of propagules from an ecotype whose traits made it a suitable candidate for controlling the Opuntia species invasions in Queensland and New South Wales. This sampling and the subsequent mass rearing associated with the introduction of C. cactorum may have also had consequences for the moth’s success in North America. Observed preferences for C. cactorum in Florida were for two species (O. stricta and O. ficus-indica) to which the moth has been previously exposed during its global spread for biological control (Dodd 1940; Pettey 1948). Uncovering the specific traits that might underlie this differential success and elucidating the importance of ecotypic variation on invasion success will require extensive experimentation and sampling in both the native and exotic ranges of C. cactorum.
This work was supported in part by grants from the U.S. Geological Survey Biological Resources Discipline (08HQAG0139) to CPB and GNE, (04HQAG0135) to GNE and U.S. Department of Agriculture (2007-55320-17847) to GNE. Additional funds were provided by the Mississippi State University Office of Research and Economic Development and the College of Arts and Sciences to CPB and GNE. We thank Anastasia Woodard for assistance with sample collection and Florida State Parks for access to sites for sampling. The Nokuse Plantation (located in Bruce, FL, USA) provided lodging during some of the sampling trips.
- Dodd AP (1927) The biological control of prickly pear in Australia. Council for Scientific and Industrial Research, Commonwealth of Aust., Bull. No. 34Google Scholar
- Dodd AP (1940) The biological campaign against prickly-pear. Commonwealth Prickly Pear Board, BrisbaneGoogle Scholar
- Futuyma DJ, Keese MC, Funk DJ (1995) Genetic constraints on macroevolution: the evolution of host affiliation in the leaf beetle genus Ophraella. Evolution 49(797):809Google Scholar
- Johnson DM, Stiling PD (1996) Host specificity of Cactoblastis cactorum (Lepidoptera: Pyralidae), an exotic Opuntia-feeding moth, in Florida. Environ Entomol 25:743–748Google Scholar
- Johnston TH and Tryon H (1914) Report of the prickly-pear travelling commission. Commonwealth Prickly-Pear BoardGoogle Scholar
- Kiesling R (2005) Cactaceae. In: N.M. Bacigalupo (ed.): Flora Ilustrada de Entre Ríos, parte IV Dicotiledoneas Arquiclamideas B: Geraniales a Umbeliflorales. Colección Científica del Instituto Nacional de Tecnología Agropecuaria 4:401–444Google Scholar
- Marsico TD, Wallace LE, Ervin GE, 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 13:857–868CrossRefGoogle Scholar
- Novak SJ and Mack RN (2005) Genetic bottlenecks in alien plant species: influence of mating systems and introduction dynamics. pp. 95–122 In: Exotic Species—Bane to Conservation and Boon to Understanding: Ecology, Evolution and Biogeography. Sax DF, Gaines SD, and Stachowicz JJ (eds.). Sinauer, MAGoogle Scholar
- Novak SJ and Sforza R (2008) Genetic analysis of native and introduced populations of Taeniatherum caput-medusae (Poaceae): Implications for biological control. Proceedings 12th International symposium biological control weeds. 22nd-27th April 2007, La Grande Motte, France. pp 422–28Google Scholar
- Pettey FW (1948) The biological control of prickly pears in South Africa. Department of agriculture science bulletin 271. Government Printer, Pretoria, p 163Google Scholar
- Thompson JP (1982) Interaction and coevolution. J Wiley 179 ppGoogle Scholar
- Thompson JN (1994) The coevolutionary process. Univ of Chicago Press 387 ppGoogle Scholar
- Thompson JN (2005) The geographic mosaic of coevolution. Univ of Chicago PressGoogle Scholar
- Wares JP, Hughes AR, Grosberg RK (2005) Mechanisms that drive evolutionary change: insights from species introductions and invasions. pp. 229–257 in: Species Invasions: Insights into Ecology, Evolution, and Biogeography. Sax DF, Stachowicz JJ, Gaines SD, (eds). Sinauer, MAGoogle Scholar