When release from natural enemies is the basis for invasiveness of an exotic species, biological control could be a valuable and potentially safe method of management of the invasive species. Classical biological control (CBC) is the method of introducing an herbivore that either specializes on the target invasive species or has a diet sufficiently narrow in breadth for it to have a significant impact on the target species without posing a significant risk to nontarget species. Perhaps the best-known successes of CBC are the control of cottony-cushion scale (Icerya purchasi) in California and the control of prickly pear (Opuntia spp.) in Queensland, Australia. It is therefore not surprising that these two cases serve as illustrative examples of CBC in many textbooks (DeBach 1974, Hajek 2004). Although these and other successes have been the subject of many papers and books, the idiosyncrasies that may have critically contributed to their success have seldom been highlighted.

Our aim is neither to denigrate nor to praise the practice of biological control. We believe it is a vital tool in the management of some invasive species, but one that can be significantly enhanced through a better understanding of the scientific reasons behind past successes and failures. Our objectives in this article are threefold. First, we wish to bring to light information that is seldom acknowledged (or simply not known) on the use of the Cactoblastis cactorum (Cactoblastis hereafter) in the control of Opuntia species in Australia. Having been based at Alan Fletcher Research Station (the former headquarters of the Commonwealth Prickly Pear Board), which was responsible for the successful management of Opuntia in Australia, we were able to delve into its archives to access this information. Second, we want to temper the expectations of people hoping to mimic Cactoblastis-like successes when funding or requesting biological control as a management option. We hope to encourage biocontrol practitioners to elucidate the reasons for successful CBC. Finally, we argue for caution in predicting the effects of Cactoblastis as a North American invasive species on the basis of its success as a biocontrol agent elsewhere, given the idiosyncrasies associated with its historical use as a biocontrol agent.

A brief history of Opuntia in Australia

European settlement of Australia was accompanied by the introduction of hundreds of plant and animal species to support the settlers, including 29 different prickly pear species, 9 of which were to become invasive and pose significant economic and ecological threats (Walton 2005). The earliest records of the introduction of prickly pears go back to 1788, when the first British fleet arriving in Australia brought both the drooping prickly pear (Opuntia vulgaris) and the cochineal (Dactylopius spp.) to produce the red dye needed for the coats of soldiers (Mann 1970). The first records of an Opuntia pest species go back to 1839, when Opuntia stricta (= Opuntia inermis) was moved from Parramatta (near Sydney) to various parts of rural New South Wales and Queensland to be used as a hedge plant around homesteads. In 1870, the plant was found as far north as Rockhampton in Queensland, and by 1884, the problem of its spread was reported in the media. By 1900, some 4 million hectares (ha) of Queensland were infested by prickly pears. The use of the prickly pear as drought fodder further enhanced its rate of spread (Dodd 1940); by 1925, some 24 million ha of Queensland and New South Wales were infested (figure 1), with densities of up to 16,000 plants per ha and an estimated biomass of 250,000 kilograms (kg) per ha. The rate of spread over the previous 10 years was estimated to be about 320,000 ha per year (Walton 2005).

Figure 1.

Spatial extent of the Opuntia infestation in Australia in 1924. District boundaries are indicated. Redrawn from Mann (1970).

Various control methods were attempted, with results that ranged from ineffective (e.g., slashing, mulching, and burning) to effective but dangerous (e.g., chemical control involving the use of some 3 million kg of blends of arsenic pent-oxide and sulfuric acid between 1912 and 1932; Melville 1935, Payne 1936). Resort to dangerous control measures reflected the desperation associated with having most of the arable land under invasion.

Spatial extent, method of release, and time taken to affect Opuntia populations.

After mass rearing in four dedicated facilities, a large workforce (> 200 workers) delivered a coordinated program releasing almost two billion egg sticks (each egg stick containing 50 to 100 eggs) in a four-year period across the entire range of the Opuntia invasion (table 1; Power et al. 1928, 1929,1930). These numbers do not include the release of 10,196,150 egg sticks across 19 locations in Queensland and New South Wales between 1927 and 1928, and another 179,395,650 egg sticks between 1931 and March 1933 (Mann 1970). The release information for individual districts for the 1930s dates was not detailed in the archives, but we presume that this was as spatially extensive as that detailed in table 1. In addition, some restricted releases occurred in subsequent years, with 1,350,000 egg sticks released in 1934–1935 and 480,000 in 1935–1936 (Melville 1935, Payne 1936), in response to Opuntia regrowth after the initial impacts of Cactoblastis.

Table 1.

Number of egg sticks of Cactoblastis cactorum released by the Commonwealth Prickly Pear Board from 1927 through 1930 across districts infested with Opuntia species.

The release efforts detailed above were only those undertaken by the officers of the Commonwealth Prickly Pear Board and by state government officials. Landholders who made an application for this agent were also supplied with some 1,292,882,500 egg sticks of C. cactorum, and approximately 692,700,000 egg sticks were distributed from motor vehicles to control roadside infestations and infestations on Crown (public) lands (Power et al. 1930, Mann 1970). One can only imagine the number of unrecorded redistributions and releases that were made in addition to those accounted for by the Commonwealth Prickly Pear Board.

The method of release is a part of what made this effort unique. Egg sticks were collected from colonies and pasted on a square piece of paper or on short wax quills, which were then pinned on pear plants (Power et al. 1930, Mann 1970) to ensure that the agents found the plants. This is unlike many modern efforts, which typically involve the release of adults; the endogenous dispersal drive of adults of many insect species (den Boer 1990) means that there is an increased chance that they may fly away from a potential host patch and may not encounter another suitable one. The method of releasing a large number of eggs directly onto plants also meant that when adults emerged, mate finding (a vital part of establishing local populations) was significantly enhanced, and the likelihood of Allee effects was low.

Ecologically, successful biocontrol agents are hypothesized to be r strategists, capable of rapid reproduction and strong dispersal mechanisms (Waage 1991). In the case of Cactoblastis, these traits were engineered through arguably the most efficient and intentional human-aided dispersal and spread of a species thus far. It is unlikely that such an intensive rearing and spatially extensive release effort has since been undertaken for any other weed biocontrol agent. This is perhaps most starkly reflected in the time taken for biological control to produce drastic reductions in pear populations. Even by 1928, two years after the first release efforts, some large infestations of prickly pear (ranging from 600 to 2000 acres, or about 240 to 810 ha) had been completely destroyed (Power et al. 1928). By 1936, the annual report of the Prickly Pear Land Commission declared, “The prickly pear menace has been overcome and the devastation it wrought a short ten years ago is now becoming merely a memory,” and the Cactoblastis Memorial Hall at Boonarga was built on a site that had once been covered with a dense infestation of O. stricta (Payne 1936).

Cactoblastis versus contemporary classical biological control.

Although the success of Opuntia management can be attributed to biological control, it appears to be more a case of atypical biological control than of CBC. A relative assessment of contemporary rearing and release practices in CBC may shed light on the validity of this claim. Even though some release densities may be as high as those in the Cactoblastis example, we very much doubt that the sustained rearing and release efforts that almost certainly led to the successful management of Opuntia are mimicked in contemporary cases.

To illustrate this point, we present the case of Lantana camara L., a modern-day equivalent of Opuntia in Australia that infests a comparable area. Lantana camara was introduced as an ornamental plant in the 1840s; it has since escaped and covers more than 4 million ha along eastern Australia. Biological control efforts began in Australia in 1914, and 30 insect and fungal agents have been introduced against L. camara since then (Julien and Griffiths 1998), of which 16 have established at last count (Day et al. 2003). Two of the most recent agents released (representing some of the largest release numbers for this weed's agents) include a leaf-sucking mirid, Falconia intermedia, and a stem-sucking membracid, Aconophora compressa. The rearing and release efforts of these species present a stark contrast to the Cactoblastis example (tables 1, 2). This might in part explain the moderate success in controlling L. camara with biological control.

Table 2.

Rearing and release efforts of two recent biocontrol agents for Lantana camara in Queensland and New South Wales, Australia.

This discrepancy may simply reflect how easy Cactoblastis was to rear or, conversely, how difficult the Lantana agents were to rear. However, even in releases of chrysomelid beetles that are easy to rear and have been extensively used as weed biocontrol agents, the release numbers are orders of magnitude below those of Cactoblastis. For example, an effort targeting Lythrum salicaria in Illinois involved the intensive rearing and extensive statewide distribution of Galarucella species, including the release of some three million beetles between 1995 and 2005 (Susan Post, Illinois Natural History Survey, Champaign, IL, personal communication, 7 August 2007). This is similar to releases of the tropical chrysomelid Zygogramma bicolorata targeting Parthenium hysterophorus (Kunjithapatham Dhileepan, Alan Fletcher Research Station, Biosecurity Queensland, Australia, personal communication, 6 August 2007). However, the spatial extent of these releases was typically a small subset of the overall range of infestation of the weed.

Perhaps the rearing and release efforts are substantially better in the case of similar extensive programs against aquatic weeds in South Africa (Zimmermann et al. 2004a), but such information is seldom published. Even in these cases, one suspects that the spatial extent of the Cactoblastis release effort may have been difficult to match.

Cactoblastis as a biocontrol agent in South Africa and the Caribbean

In South Africa, the initial target for Cactoblastis was Opuntia ficusindica, which occupied some 900,000 ha in the Cape Province alone in 1942 (Zimmermann and Moran 1991, Zimmermann et al. 2004b). An intense Cactoblastis rearing program released some 580 million egg sticks between 1933 and 1941 (Petty 1948, Zimmermann et al. 2004b). However, the “damage caused [to Opuntia] by Cactoblastis in South Africa was not as great or as extensive as that in Australia,” although it did significantly retard “the spread of the weed by reducing its fruiting capacity and killing the seedlings” (Zimmermann et al. 2004b). This diminished impact may in part be due to a substantially greater predation rate (from baboons, rodents, and insects) in South Africa (Petty 1948, Zimmermann et al. 2000, 2004b) than in Australia, where both the moth and the prickly pear had few natural enemies. Better control of O. ficusindica was provided by Dactylopius opuntiae. Despite the Cactoblastis release effort in Eastern Cape Province, large nearby populations of O. stricta (the target for Cactoblastis in Australia) in Kruger National Park were not colonized by Cactoblastis for more than 70 years. In 1988, limited numbers of Cactoblastis (60 egg sticks) were released into a 19,000-ha cactus infestation (Hoffmann et al. 1998a). This established Cactoblastis populations, but the impacts on O. stricta populations were limited, and other methods of control were required to bring weed populations under control (Hoffmann et al. 1998a, 1998b). These results from South Africa suggest that C. cactorum may exhibit subspecific variation in host range.

Opuntia triacantha (Spanish lady cactus) had been a major cause for concern as a weed in pastureland on Nevis and other Leeward Islands (West Indies). Nevis is a small island (93 square kilometers [km² ]), with a productive coastal plain approximately 1 km wide. The success of biological control in reducing populations of O. stricta in Australia made it an appealing prospect for controlling O. triacantha in Nevis (Simmonds and Bennett 1966). Cactoblastis and two Dactylopius species were released in Nevis from material obtained from South Africa, but only Cactoblastis established and had a significant impact: “It was difficult to find clumps more than one foot in height and most clumps [had] been reduced to the point where only six or eight single rooted cladodes remained” (Simmonds and Bennett 1966). Unfortunately, there are few published data on the size of the O. triacantha infestations before and after biological control. In the absence of such data, it is difficult to critically evaluate the true impact of Cactoblastis on the basis of comments indicating that “the biological control programme was successful, and the degree of control compares favourably to the classic success obtained by the introduction of Cactoblastis into Australia”(Simmonds and Bennett 1966). The reported success on Nevis prompted releases on other Leeward Islands, with good effect, but again, the role of Cactoblastis in controlling O. triacantha on these islands is difficult to evaluate because information on release numbers, the spatial extent of the releases, and the area of Opuntia before and after initiation of biological control is not published (Simmonds and Bennett 1966).

Cactoblastis as an invasive species in North America

The detection of Cactoblastis in North America, its potential to spread, and the impact on natural and cultivated Opuntia species are a significant cause for concern (Stiling 2002). This has prompted investigations of the direct effects (based on the physiological and ecological host range of Cactoblastis) and indirect effects (on Opuntia-dependent species) of Cactoblastis invasion (e.g., Johnson and Stiling 1996, Vigueras and Portillo 2001, Stiling 2002, Stiling et al. 2004). Certainly Opuntia species that are already rare and threatened in North America (e.g., O. triacantha, Opuntia corallicola [semaphore cactus]) are likely to be threatened by the invasion of Cactoblastis, because of direct herbivory, pathogenic decay facilitated by wounds created during larval feeding (Starmer et al. 1988), and other indirect effects (Stiling et al. 2004).

Inferring the risks of Cactoblastis invasion on the basis of its success as a biocontrol agent

To estimate the threats posed by Cactoblastis in North America on the basis of a direct comparison with its effects in the biological control of O. triacantha in Nevis, O. ficusindica in South Africa, and O. stricta in Australia would not be entirely appropriate. Because of differences in the area of infestation, the relative density of the release of Cactoblastis, the habitat context (small island, large island, or continent), and the Opuntia species targeted, inferences about any of these scenarios based on the others are limited and require qualification.

One common feature of situations in which Cactoblastis has moderate to high impacts on Opuntia species appears to be high release densities. Currently, pasture is about 3 percent of the area of Nevis ( www.cia.gov/library/publications/the-world-factbook/geos/sc.html). Assuming that when biological control was undertaken in the 1950s, the area of pasture was twice this size (6 km²), and that O. triacantha covered all the land under pasture, the release of 5200 Cactoblastis (egg sticks and small larvae) (Simmonds and Bennett 1966) equates to a potential density of 867 egg sticks per km² of Opuntia infestation. Releases in South Africa had a potential density of 64,444 egg sticks per km² of Opuntia infestation. The density of the Cactoblastis releases in Australia (total from 1927–1930) was 7320 ± 2589 egg sticks per km² of O. stricta infestation (mean ± standard error, assuming that the entire district was covered with Opuntia; range 85 to 48,653 egg sticks per km²; n = 21 districts; table 1). This estimate does not include subsequent releases totaling 2,177,004,300 egg sticks, whose spatial distribution is not evident from the archives (potential density = 9071 egg sticks per km², calculated over the entire Opuntia distribution in Australia). Information on the spatial extent of releases vis-à-vis the distribution of the Opuntia species in South Africa and the Caribbean is not available for comparison with the Australian efforts. Release density alone, however, may not be an adequate predictor of risk. Although Cactoblastis was not sufficient to control O. stricta in Kruger National Park, populations that established from a release of 60 egg sticks did have a measurable impact on cactus populations (Hoffmann et al. 1998a, 998b).

Determining the relative contribution of natural dispersal rates and human movement of cacti (e.g., through the nursery trade) is key to understanding the risks of Cactoblastis as an invader. The hypothesized spread rate of Cactoblastis in North America (160 km per year; Hight et al. 2002, Solis et al. 2004) is far greater than natural dispersal rates recorded in North America and other parts of the world, which suggests that human-aided dispersal plays an important role in the spread of this species (Johnson and Stiling 1998, Zimmermann et al. 2000). In South Africa, it was not until biocontrol practitioners moved Cactoblastis egg sticks into Kruger National Park that some level of control of O. stricta was effected (Hoffmann et al. 1998a, 1998b). Although Cactoblastis's spread was rapid on the Caribbean island of Antigua, its distribution was patchy eight years after release (Simmonds and Bennett 1966). Meaningful inference of Cactoblastis dispersal rates in Australia is confounded by the spatially extensive release strategy (table 1). Perhaps ongoing molecular phylogeography studies (e.g., Simonsen et al. 2006) will reveal more about Cactoblastis's dispersal mechanisms and the factors influencing and limiting its spread.

In all of the regions where significant populations of Opuntia species were successfully managed with biocontrol, Cactoblastis was not the sole agent of control. This fact is significant but often inadequately acknowledged. Cochineal species (Dactylopius spp.) were as important as Cactoblastis in suppressing Opuntia and preventing populations from becoming weedy in these situations (Zimmermann et al. 2004b, Walton 2005). The paucity of information on the impacts of cochineal species limits our ability to evaluate and understand the relative roles of different herbivore species in bringing weedy Opuntia species under control.

The invasion of Cactoblastis into the southern United States and Central America is a source of concern, given the diversity, endemism, and economic significance of various cactus species in this region (Zimmermann et al. 2000, Vigueras and Portillo 2001, Stiling 2002). Because Cactoblastis can feed on a wide range of hosts, it may pose a greater risk of invasion where prickly pear diversity is high—reservoir populations could become established in good or poor seasons, and those could prove difficult to control. Although this threat must be taken seriously, focusing only on the historic impacts of the biological control programs in Australia, South Africa, and the Caribbean, without considering the idiosyncratic features that contributed to Cactoblastis's success as a biocontrol agent, could lead to exaggeration of the nature of the hazard. As stressed by Zimmermann and colleagues (2000), the true nature of the risk will “be governed by the local climate, parasites, predators and diseases, host-plant characteristics and many biotic and abiotic influences on the cactus moth itself, including the vagaries of natural- or human-aided dispersal,” and by how these factors influence the strength of the interactions between the moth and Opuntia populations.

Conclusions

Our objective has not been to deny the value of biological control as a tool in the management of invasive species, but rather to highlight the often overlooked facts and nuances associated with Cactoblastis, a poster child of classical biological weed control. We inferred that the spatial extent and density of the egg releases played a significant role in the establishment of the moth and in the dramatic rate of control of Opuntia in Australia; however, the continued and persistent control of Opuntia to this day is in apparent concordance with the more typical predator–prey cycles expected in CBC. Whether Cactoblastis would have been as successful in Australia without inundative releases is unfortunately not an empirically testable hypothesis, but we are exploring this question theoretically by building on preexisting Cactoblastis–Opuntia interaction models (Caughley and Lawton 1981).

Some of the features of the Australian Opuntia biological control program that contributed to its success—the length of the program, interstate cooperation, and financial input—are only rarely found in other biological control programs. Without these features, future Cactoblastis-like successes may be difficult to achieve. If the value of weed biological control is to be understood from iconic case studies, we should perhaps look to CBC successes such as rubber vine in Queensland and water hyacinth in Papua New Guinea and Africa (Walton 2005). Even in these examples, we recommend exploring the factors that contribute to success in each project before making generalizations, as each project is likely to have its own idiosyncrasies.

We hope that the historical details help clarify the ghost of Cactoblastis past that often serves as a benchmark to promote, to judge, and more recently, to haunt the discipline of weed biological control.