DEFRA Ref: CR 0492 CABI Ref: TR10086
Total Page:16
File Type:pdf, Size:1020Kb
Understanding and addressing the impact of invasive non-native species in the UK Overseas Territories in the South Atlantic: A review of the potential for biocontrol DEFRA ref: CR 0492 CABI ref: TR10086 Preliminary results Ascension Island
Norbert Maczey, Rob Tanner and Richard Shaw
March 2012 Contents
Norbert Maczey Senior Ecologist/Entomologist
CABI Bakeham Lane Egham, Surrey TW20 9TY UK
T: +44 (0) 1491 829029 E: [email protected] Project introduction Recent reviews of invasive non-native species (INNS) on the South Atlantic UKOTs, revealed that there is a considerable on-going threat by these species to the endemic biodiversity on these islands. Management plans and capacity building efforts to combat invasive plants have already been developed, or are in the process of being developed, by various conservation bodies and other research institutions. However, to-date, there are few plans to include classical biological control in these new schemes despite this management practise having a long history of successful use on islands in the past. We understand that many species including rare, endangered and/or iconic species are negatively impacted upon by introduced invasive plants or invertebrate species and that the reduction of their impact through biological control could bring tangible benefits to biodiversity in these fragile habitats. At present there is little known about the potential of this tool in South Atlantic UKOTs and therefore this feasibility study focuses on the future potential usage of this method. A major component of this study is a desk-based review assessing the impact of non-native plant and invertebrate species in the South Atlantic UKOTs. We rank the need and suitability for biological control of individual invasive species, from high to low, using a recently-developed weed biocontrol prioritisation tool, developed initially for Australia, but already proving to be very useful on the islands in the South Pacific. Priority species are selected on the basis of population dynamics, reported impact on biodiversity, impact on livelihoods, costs for on-going control measures other than biological control and others. In addition, two detailed case studies will evaluate the potential of implementing a biological control programme against priority species for the Falklands and South Georgia more closely. These detailed assessments will combine all current information about the target pest species including information on their native range, areas of introduction and recorded natural enemies in their areas of origin. Ranked in order of priority, and associated costs of available control options, we will include a catalogue of recommended strategies for each priority target species aimed to halt the biodiversity loss through invasive plant and invertebrate species more effectively.
Biological control of weeds A major focus of this study is the assessment of invasive weeds on all South Atlantic UK OTs since they are often the most numerous, widespread and damaging introduced non-native species able significantly to alter whole ecosystems. In comparison to mechanical or chemical control, which often fails to persistently control the target species, and which needs to be repeatedly applied throughout seasons and year on year, biological control is often the only available practical and economically feasible option, which can be applied over a large geographical area. It impacts very little on the environment in contrast to the associated disturbance caused by manual or other mechanical clearing, or the application of herbicides. When successful, biological control is permanent and self-sustaining (Wittenberg & Cock 2001). On average, more than 75% of target weeds have been significantly or permanently controlled using biocontrol in the past century. Success rates range from 50% to 87% and depend heavily on the amount of effort and funding applied to required research (Myers and Bazely 2003, MacFadyen 1998, Hoffmann 1995, Fowler et al. 2000). Integrated biological control can be regarded as a safe and cost efficient method provided adequate risk assessment procedures and current protocols such as the International Plant Protection Convention's Code of Conduct for the Import and Release of Exotic Biological Control Agents (International Plant Protection Convention 1996, Food and Agriculture Organization of the United Nations 1997) are followed. Possible non-target effects are quite predictable by the safety- testing that is now routinely carried out (Pemberton 2000). However, we also recognise that the introduction of a biocontrol agent without considering any conflicts of interest is unwise and ultimately unproductive so we will also assess the general public’s opinions on invasive species and biological control as an important part of the detailed feasibility assessments planned for the Falklands Islands and South Georgia. Management plans and capacity building efforts to combat the threat of invasive species have already been developed or are in the process of being developed by various conservation bodies such as RSPB, Kew and other research institutions (i.e. Glass et al. 2006, Whitehead 2008). However, biological control currently plays only a very small role in any new control schemes and we aim to close this gap through the work in this study. We have started this project with a desk-based review and results of the recently completed ‘South Atlantic Invasive Species Project’ have already provided detailed baseline information for this. In addition, CABI has collated information on the impacts of non-native plants and invertebrates using literature reviews and current relevant databases and added this to the results of recent surveys and reviews. Based on this, non-native species were ranked with the aim to form a list of priority candidate species for potential future biological control programmes. For this process a recently- developed weed biocontrol prioritisation tool, developed by Paynter et al. (2009) for Australia, was used. This decision tool is currently being applied to the islands of the South Pacific and is proving to be very useful. It ranks many factors associated with weed biocontrol to produce a score to aid decision makers in allocation of limited resources focussing on likely success versus scale of the problem. These include:
• Is opposition to biological control likely? • Does the weed species have socioeconomic value? • Is the weed sufficiently widespread/intractable/important to justify investment in biological control? • Has the weed been/is it a subject of adequately resourced biological control program elsewhere? • What is the accessibility and ease of working in its native range? • Is literature available regarding natural enemies; is it well known/accessible? • How closely related to indigenous/valued plants is the target weed? • Which habitats are invaded? • What type of is the life cycle needs to be taken into account? • In which way reproduces the target species? • Is it a weed in its native range? • Is it difficult to target multiple forms of weed, or probability of replacement of the weed by forms or congeners of the target following successful biological control thereby negating benefits? • Is it growing in competitive environment (agricultural vs. environmental)? • Is it a native/valued exotic congener?
Arthropods Introduced insects can cause widespread and serious problems on islands from reduced biodiversity and the threat of extinction through reduced agricultural productivity to impacts on human health and quality of life. This was the case with the introduced ensign scale, Orthezia insignis which not only threatened many horticultural plants on St Helena but was severely damaging and in many cases killing the national tree, Commidendrum robustum (Asteraceae). Thankfully a specialist coccinellid beetle had been used successfully against this pest elsewhere in the world and CABI was able to facilitate its introduction in the 1990s. This action is believed to have saved the tree from extinction (see Booth et al. 1995; Fowler 1996). The feasibility for biological control of introduced invasive terrestrial arthropods was assessed in the same way as non-native weeds, and only small adaptation to the selected prioritisation tool were required. Although the in-depth data for many non-native arthropods in the OTs is patchy, a good list exists in most cases. CABI holds an extensive database on previous insect biocontrol activities around the globe and this was used to determine whether any key arthropod pests have been subject to attention in the past. Target species were prioritised based on likely safety and success and the reported scale of the problem.
Adaptation of prioritization tool In this study we use Paynter et al. (2009) as a tool suitable to estimate the feasibility for biological control of weeds and invertebrates on the South Atlantic OTs. Paynter et al. (2009) use a scoring system widely established in Australia through their Weeds of National Significance system (WoNS). A comparable system for the SAOTs does not yet exist. The scoring system developed by Whitebread (2008) for the Falkland Islands assesses the risk of invasiveness and not the actual impact and can therefore not be used for an accurate adjustment of the Paynter et al. method. Developing impact assessment systems comparable to the WONS in Australia are complex and are based on the input of a large number of data over long time periods by many experts eventually collating substantial background information. As a workable replacement we developed a slightly simplified approach, which still allows the accurate ranking of priority species feasible for biological control. We suggest the following adjustments within the three main modules of this method:
Module 1: Weed importance and desirability of biocontrol Based on the information available from the SAIS project, the JNCC database and additional literature covering reported impacts of relevant species in other countries we suggest the scoring system described in table 1 to estimate weed importance:
Table 1: Questions to determine whether a species is sufficiently widespread/intractable/important to justify investment in biocontrol. Score if question is Question answered with yes Is the species only recently introduced/recorded on the island(s)? 5 (0-no, or no data available; 5-yes) Does the species currently have a widespread recorded distribution on the island(s)? 5 (0-no, or no data available; 5-yes) Is the species currently expanding its range on the island(s)? 5 (0-no, or no data available; 5-yes) Is the species known to be invasive in other geographical regions? 5 (0-no, or no data available; 5-yes) Is the species known to have negative impacts on biodiversity in other geographical regions? 5 (0-no, or no data available; 5-yes) Is the species known to cause economic damage in other geographical regions? 5 (0-no, or no data available; 5-yes) Is the species known to cause threats to human or animal health in other geographical regions? 5 (0-no, or no data available; 5-yes) Are habitats on the island(s) which can potentially be invaded of conservation importance? 5 (0-no, or no data available; 5-yes) Does the SAIS assessment suggest a likely negative impact on biodiversity, agriculture or health 10 (0-no; 5-yes) Have ecological impacts been recorded on the island(s) for this species? 10 (0-no; 10-yes) Have economic impacts (agriculture and others) been recorded on the island(s) for this species? 10 (0-no; 10-yes) Have impacts on human or animal health recorded on the island(s) for this species? 10 (0-no; 10-yes)
Only species scoring at least 20 in the questionnaire above are considered as suitable targets for biocontrol and thus further assessed. One option would have been to include current control methods in the above approach, something Paynter et al. have suggested themselves as a possibility for their method applied to Australian weeds. However, the necessary inclusion of data for each species turned out to be above the scope of this project. It may also be of influence to account for public perception in this tool as it could be that a nuisance species might have more support for biocontrol than one actually causing economic or ecological harm.
Module 2: Effort required to obtain and host-range test biocontrol agents We kept this this module unchanged. Paynter et al. give a minimum score for each question in this module to accommodate the fact that because of the unique flora of Australia even in cases of successful existing programs in other geographical areas a certain amount of additional testing is required. We can most likely assume this also for the South Atlantic OTs. Due to the unique nature of the indigenous flora on these islands efforts for host plant testing very similar to Australia can be assumed.
Module 3: Predicting the potential impact of biocontrol Again we kept this this module unchanged. There is the possibility to apply a different weighting of questions 9-14. Since the basis of the weighting within Paynter et al. is based on large combined data sets from Australia, South Africa and North America one can assume that no other datasets able to provide similar accuracy exist, which could deliver a better weighting better suited to the flora of the SAOTs.
Adaptation for invasive alien invertebrate species We used a similar approach as for the weed species. The only difference being a deletion of questions 9-11 within module 3. These questions are not suitable for invertebrates and no datasets exists to replace these factors with others and to develop an according weighting system. We have however, added one new question: Asking whether biological control programs have been conducted (successful or unsuccessful) for closely related species, which would increase the possibility for any successful control efforts of the relevant species. If no biological control of even related species has been recorded the species will be regarded as currently unsuitable for biological control. Preliminary results for Ascension
Weeds Thirty-three non-native plant species had an initial impact score of 20 or above and were processed further within the selected prioritisation tool. Preliminary results of this assessment are given in table 2 below. As a cut-off point below we regard BC not to be feasible anymore we have chosen an overall score of 50. However, a high overall score of an individual species does not automatically result in an immediate recommendation to give BC of this species a high priority. This needs to be looked into in more detail through a case by case assessment taking other control options and costs versus benefits into account. Although scoring very high Lantana camara and Opuntia stricta are not likely to be feasible candidates for future BC as they have by subject to this control method in the past. We decided, however, to process these two species further as at this time as we do not want to exclude them fully from any further considerations before having reviewed their current status in more detail. We are planning to discuss the current impact of these species in more detail conservation officers during a visit to Ascension in March 2012. In case these two species pose still a severe threat to the indigenous fauna and flora they would immediately become highly suitable and recommendable control targets as is backed up by their very high overall scores and the availability of additional effective and tested BC organisms. Aside of these two species we currently regard particularly Prosopis juliflora, already a subject of partially successful biological control, and Cryptostegia grandiflora as highly suitable candidates for future BC projects. Cryptostegia grandiflora seems to be a rather recent introduction and eradication may still be a feasible option taking priority over BC. However, in case this is not the case and rubbervine has already started to spread more widely (we would suspect this to become likely in a relatively short amount of time on the assumption that habitat conditions on Ascension are relatively similar to parts of Madagascar, the area of origin for this species) BC can provide a readily available ’off the shelve’ solution, which has been very successful in other geographical regions. For Prosopis juliflora one very promising candidate agent sticks out at present and more may well appear with further developments in South Africa. The existing candidate agent is a small gelechiid moth, Evippe sp., which has caused spectacular damage to Prosopis spp. dry tropical parts of Australia (van Klinken et al. 2009; Fowler 2011) Biological control using the thrip Liothrips urichi could be very successful for Clidemia hirta. However, this will depend on how much of the population of C. hirta grow in open areas with direct sunlight as the control agent is only effective under such conditions (see table 2). For some species no readily available BC agents have been tested and proven to be effective, but research efforts are on-going as these species cause severe problems in other geographical regions. This group includes Leucaena leucocephala, Tecoma stans, Casuarina equisetifolia, Argemone mexicana, Sonchus asper, S. oleraceus and Cynodon dactylon. For these species suitable agents may become available in the future. Alternatively, independently funded scoping studies and host range testing can be initiated. Such larger scale projects need of course be set in relation to the current or imminent threats these species pose to indigenous species and habitats. Although scoring relatively low, it may well be worth to look in more detail into the potential of BC for Alpinia zerumbet, Buddleja madagascariensis and Rubus pinnatus as other species within these genera are currently subject of BC programmes. The lack of closely related species on Ascension could make them suitable targets based on the results of these on-going projects. In particular complementary initial scoping studies looking for suitable BC agents can be highly recommended. Psidium guajava, Juniperus bermudiana, Heliotropium curassavicum, Begonia hirtella, Nicotiana glauca, Commelina diffusa, Chenopodium murale, Adiantum raddianum, A. capillus-veneris, Paspalum scrobiculatum and Spermacoce verticillata belong to a group where no BC projects have been initiated so far or only insufficient efforts have been conducted in the past. Initiating new control projects would be costly compared to ‘off the shelve’ solutions and need to be justified in case by case studies into the scale of current and expected impacts. We currently regard the impact these species have on the indigenous habitats as too low to justify the implementation of full scale BC projects. However, this preliminary judgement may turn out to be unjustified for some of these species and remains open for reassessment at any time.
Table 2: Preliminary results feasibility for BC for non-native weeds on Ascension BI O TOTA C L O SCOR NT E IM EF R (speci PA FO BC OL es CT RT fea SPECIES FEASIBILITY OF BIOCONTROL FE IMPA SC SC sib AS CT x O O le IBI Bioco RE RE LI ntrol TY x SC 1/Effo O rt) RE Lantana camara 45 4 100 1125.0 Y Worldwide well over 200 releases have been made (39 different natural enemies have been released in 29 countries). However, in the majority of cases the control agents either failed to become established or became established without achieving control. Despite this limited success, classical biological control is still considered to be the only viable, long-term control option, since it offers a safe, economic and environmentally benign method of suppressing the weed. Most of the releases have been carried out in the Pacific, South Africa and Australia (for historical details see Taylor 1989; Cilliers & Neser 1991; Davis et al. 1992; Denton et al. 1991; Swarbrick et al. 1995). The most widely established species include Ophiomyia lantanae, Uroplata girardi and Octoma scabripennis. Day et al. (2003) have produced a detailed review of 48 of these control agents. Teleonemia scrupulosa has been successfully used to control L. camara on St Helena (Ashmole & Ashmole 2000). Lantana camara appears to be under successful biological control on Ascension Island from attack by the deliberately released tingid bug Teleonemia scrupulosa, the hispid beetle Uroplata girardi, and the accidentally introduced ‘Jacaranda bug’. The infamous ‘Jacaranda bug’, Orthezia insignis, nearly caused the demise of the endemic gumwood trees on St Helena before its successful biological control in 1993-4. However, it does not appear to attack any other plant species on Ascension Island to any great extent, so its present role seems beneficial (Fowler 2011). Cryptostegia grandiflora 20 4 179 895.0 Y The leaf-feeding caterpillar, Euclasta whalleyi from Madagascar, was BI O TOTA C L O SCOR NT E IM EF R (speci PA FO BC OL es CT RT fea SPECIES FEASIBILITY OF BIOCONTROL FE IMPA SC SC sib AS CT x O O le IBI Bioco RE RE LI ntrol TY x SC 1/Effo O rt) RE released in Queensland, Australia in 1988-1991 (McFadyen & Harvey 1990; McFadyen & Marohasy 1990), despite the fact that it is not specific to the target genus Cryptostegia but only within the sub-family Periplocoidae. However, its impact on the weed has been insignificant (Tomley, 1995), probably due to parasitism, although more recent reports indicate that permanent populations may be establishing (Mo et al., 2000). Assessments of the Madagascan rust fungus, Maravalia cryptostegiae, showed that this pathogen had the highest biocontrol potential of the natural enemies surveyed: being highly damaging; specific at the genus level, as well as climatically adapted to Queensland conditions (Evans 1993; Evans & Fleureau 1993; Evans & Tomley 1994). This potential has now been proven and a strain of the rust from south-west Madagascar, introduced in 1995, has had an enormous impact on the weed throughout its invasive range following a mass production and release programme (Evans 2002; Tomley & Evans 2004). Rust-induced defoliation has resulted in a significant reduction in weed biomass leading to almost complete loss of fecundity, as well as to widespread plant death (Tomley & Evans 2004). The population currently recorded on Ascension seems to be still small, but considering the high risk this species poses to indigenous habitats and the good availability of tested and effective control agents this species should certainly be considered for BC at least in the near future. Opuntia stricta var. dillenii 35 4 100 875.0 Y The release of the cactus moth, Cactoblastis cactorum, in 1926 to biologically control O. stricta that had invaded about 24 million ha in Queensland and New South Wales, Australia resulted in spectacular control of the weed during the following nine years. Dodd (1940) wrote “… the most optimistic scientific opinion could not have foreseen the extent and completeness of the destruction. The spectacle of mile after mile of heavy (prickly) pear growth collapsing en masse and disappearing in the short space of a few years did not appear to fall within the bounds of possibility”. Today remnant infestations of O. stricta are limited to regions where C. cactorum is less effective (Hosking et al. 1993). Cactoblastis cactorum was introduced to Ascension in 1974 (against O. vulgaris [misident.?] and the introduction was at least partially successfull (Ashmole & Ashmole 2000). In case Opuntia still impacts heavily on native species and habitats and a survey does show insufficient control by C. cactorum the introduction of additional BC agents can be considered. At least 17 insects and mites have been identified feeding on O. stricta within its native range (Mann 1969), which is assumed to be a conservative estimate. The diversity of associated natural enemies is highest on mainland USA and is considerably less on Caribbean islands. Several were introduced to Australia for the biological control of the two O. stricta subspecies but only three species have eventually become established; the coreid bug, Chelinidea tabulata, the cerambycid beetle, Moneilema variolare and the cochineal, Dactylopius opuntiae (Julien and Griffiths, 1998) [extract from CABI ISC 2011]. BI O TOTA C L O SCOR NT E IM EF R (speci PA FO BC OL es CT RT fea SPECIES FEASIBILITY OF BIOCONTROL FE IMPA SC SC sib AS CT x O O le IBI Bioco RE RE LI ntrol TY x SC 1/Effo O rt) RE Prosopis juliflora 45 13 80 276.9 Y Several biological control programmes using species of seed-feeding bruchid beetles have been developed and implemented. The advantage with bruchids is their observed host specificity, with many species found to feed only on Prosopis, and some only on a single species. Other insect species known to have a deleterious effect on native and exotic Prosopis in the Americas, mainly twig girdlers and psyllids, have also been suggested as possible biological control agents. The twig girdler Oncideres limpida attacks P. pallida in Brazil (Lima, 1994), whereas Oncideres rhodostricta is seen as a serious pest of P. glandulosa in the USA (Polk & Ueckert, 1973). Psyllids are known to severely affect the growth of Prosopis (Hodkinson, 1991) and have been suggested for use in controlling invasions. The same two bruchid species were also introduced to Ascension Island in an attempt to control P. juliflora which is present on 80% of the island, often in dense thickets. Two other species, one a psyllid and the other a mirid, were identified as attacking P. juliflora on Ascension Island and were thought to have been introduced accidentally from the Caribbean. The mirid Rhinocloa sp. causes widespread damage and is thought to lead to substantial mortality of trees (Fowler, 1998) [extract from CABI ISC 2011]. White (2009) recorded stunting from these insects on 98% of Mexican thorn trees she sampled, with 30% showing heavy dieback. High levels of damage from these insects was not apparent in visits to the island in 1993 and 1995, but were very obvious in 1997 (Fowler 1998). [Fowler 2011]. Fowler (2011) recommends to target limited resources on integrated management of Mexican thorn – existing site-led control at key sites, and the release of further biocontrol agents if readily available with a proven track record. One candidate agent sticks out at present and more may well appear with further developments in South Africa. The existing candidate agent is a tiny gelechiid moth, Evippe sp., which has caused spectacular damage to Prosopis spp. dry tropical parts of Australia [van Klinken et al. 2009; Fowler 2011]. Leucaena leucocephala 25 15 91 151.7 Y BC was first contemplated in Hawaii (Smith, 1985) and South Africa (Neser, 1994), but is frustrated by the economic importance of L. leucocephala, although proposals to release the seed-eating bruchid Acanthoscelides macrophthalmus in South Africa were pursued (Neser, 1994) and have now been released (Henderson, 2001), and have also been accidentally introduced into Australia [extract from CABI ISC 2011]. The accidental spread of the psyllid insect defoliator Heteropsylla cubana in the mid 1980s can cause cyclical defoliation, but does not kill trees and the psyllid appears to have been brought under control by a number of generalist local (and in some cases introduced) psyllid predators and parasites [extract from GISD 2011]. On Ascension there is currently some damage reported by the beetle Acanthoscelides suramerica. This beetle probably arrived accidentally on Ascension Island from the Caribbean in 1997 (Fowler 2011). BI O TOTA C L O SCOR NT E IM EF R (speci PA FO BC OL es CT RT fea SPECIES FEASIBILITY OF BIOCONTROL FE IMPA SC SC sib AS CT x O O le IBI Bioco RE RE LI ntrol TY x SC 1/Effo O rt) RE Tecoma stans 30 15 70 140.0 Y Host specificity tests on two rust fungus species, namely, the microcyclic Prospodium transformans and the macrocyclic P. appendiculatum from Mexico are in progress in South Africa. P. appendiculatum is already present in Brazil and Argentina but is not contributing much to the suppression of populations. Further surveys for additional host-specific natural enemies are planned. A raceme-feeding membracid and the pyralid pod-feeding moth Clydenopteron sp. are to be introduced into quarantine in South Africa for possible biological control [extract from CABI ISC 2011]. Clidemia hirta 20 15 101 134.7 Y Biological control using the thrip Liothrips urichi was initiated in Fiji in the early 1930s and two decades later in Hawaii (Mune & Parham, 1967; Wester & Wood, 1977). L. urichi seriously affects the growth of C. hirta in open, sunny areas, whereas in shaded areas (forest or frequent cloud cover) it is not effective. The thrips failed to establish following their introduction to the Solomon Islands (Julien, 1987). Over the past four decades extensive searches of biological control agents have been made to control C. hirta in Hawaiian forests (Nakahara et al., 1992). A pyralid moth, Blepharomastix ebulealis [Ategumia ebulealis], released in 1970 has been heavily parasitized and has been ineffective in controlling C. hirta. Several of 14 species of insects, recently evaluated in Trinidad, can be considered for introduction into Hawaii and the release of four pathogens is envisaged. A leaf spot fungus, Colletotrichum gloeosporioides f.sp. clidemiae, introduced from Panama to Hawaii for range studies shows promise as a biocontrol agent. The introduction of effective biological control agents into Hawaii must be considered with care. The potential sudden death of large monotypic stands of C. hirta, found on steep mountain sides, could result in either severe soil erosion or the establishment of other invasive species (Smith, 1992) [extract from CABI ISC 2011]. BI O TOTA C L O SCOR NT E IM EF R (speci PA FO BC OL es CT RT fea SPECIES FEASIBILITY OF BIOCONTROL FE IMPA SC SC sib AS CT x O O le IBI Bioco RE RE LI ntrol TY x SC 1/Effo O rt) RE Casuarina equisetifolia 25 12 62 129.2 Y Smith (1998) reports that on Hawaii, C. equisetifolia has not yet been evaluated for biological control because of conflicts of interest with its perceived uses. Swearingen (1997) reports a similar lack of a biological control agent for use in Florida, USA [extract from CABI ISC 2011]. USDA Agricultural Research Service (ARS) scientists have been searching in Australia's outback and coastlines for insects that could be key bio- controls for C. equisetifolia (Flores 2008). From 300 species, including wasps, weevils, stem-borers, sap-suckers and seed-eaters about 12 candidates have been identified. Not only do these agents attack C. equisetifolia but many also attack the related invasives C. glauca and C. cunninghamiana. The most promising bio-control agents include the seed- feeding wasp (Bootanelleus orientalis), which is host-specific to Australian pine, and the defoliator moth (Zauclophora pelodes). These insects are still undergoing testing to determine their suitability for use as bio-control agents in the United States. Please follow this link to read more about this research in the September 2008 issue of Agricultural Research magazine. Many pathogens threaten plantations of Casuarina in many parts of the world. In India a stem borer kills shoots and seedling damping-off by Rhizoctonia spp. occurs in nurseries (Binggeli 1997). Nursery seedlings in India are attacked by various insect species. In China the lymantriid moth (Lymantria xylina) is described as one of the worst pests of C. equisetifolia (Elfers 1988). In Florida, USA, there has been a high rate of root rot caused by the fungus Clitocybe tabescens. In Puerto Rico stem canker and dieback attributable to the fungus Diplodia natalensis have been recorded on C. equisetifolia. In Puerto Rico natural regeneration is rare because ants eat nearly all the seeds (Binggeli 1997). Ants have also reportedly been a major source of control in India [extract from GISD 2011]. There is a high likelyhood of competition with Ascension spurge. On the other hand the species has some values as ornamental tree (Ashmole & Ashmole 2000). Alpinia zerumbet 35 29 101 121.9 Y Alpinia zerumbet is invading suitable habitats of native endemic ferns (Lamdon et al. 2009). A mechanical control is difficult as small fragments of rhizome are quickly leading to recolonization of cleared areas. Biological control is feasible in case no other members of this family are commercially used on the island as currently a number of insects or in the process of being tested to control Wild Ginger (Hedychium) on Hawaii, in South Africa and NZ. Some of these species are feeding on several genera within the Zingiberaceae and their potential for the control of Alpinia could easily be tested. Argemone mexicana 30 15 54 108.0 Y This native of Mexico is naturalized in most warm countries of the world in sub-humid as well as semiarid regions. A bacterial wilt, Xanthomonas papavericola [X. campestris pv. papavericola], periodically checks the growth of A. mexicana, in some areas (Holm et al., 1977) but there is a dearth of information on natural enemies of this weed. A biological control programme of A. mexicana and of the closely related A. ochroleuca has been initiated in Australia. This project sought natural enemies in Mexico and identified several predatory insects including an extremely damaging species of root-breeding and leaf-feeding weevil (CSIRO, 1999; Julien, BI O TOTA C L O SCOR NT E IM EF R (speci PA FO BC OL es CT RT fea SPECIES FEASIBILITY OF BIOCONTROL FE IMPA SC SC sib AS CT x O O le IBI Bioco RE RE LI ntrol TY x SC 1/Effo O rt) RE 2002)[extract from CABI ISC 2011].
Psidium guajava 40 31 77 99.4 N Biological control is problematic for this species because there are direct conflicts of interest with fruit growers (Smith, 1998), and none has been implemented [extract from CABI ISC 2011]. Futile attempts to control Psidium guajava on Ascension were made around 1900 (Ashmole and Ashmole 2000) Juniperus bermudiana 40 28 65 92.9 N threatened species in area of origin! (GISD 2011) Heliotropium curassavicum 40 31 70 90.3 N no information on BC is currently available Begonia hirtella 30 31 80 77.4 N no information on BC available
Nicotiana glauca 40 35 62 70.9 N Successful control has been achieved where the plants were sprayed with herbicide and the exposed to the beetle Malabris aculeata (Cronk & Fuller 2001, in PIER 2007). The Management Plan developed for Ascension recommends removing this species from all beaches and attempting to push invasion fronts back and control seed spread, but for St Helena Ashmole & Ashmole (2003) observe that the complete removal of wild tobacco is probably impracticable [extract from GISD 2011]. Commelina diffusa 30 31 70 67.7 N There have not been any attempts to use biological control against Commelina spp. and the possibilities have not been explored. However, Waterhouse (1993) notes that although Commelina spp. are believed to be of Old World origin, it is curious that there are no records of agromyzid leaf miners, except from the Americas, and therefore tropical and subtropical areas of the Americas may be promising sources of candidate biological control agents [extract from CABI ISC 2011]. Sonchus asper 25 21 52 61.9 N Biological agents are currently being researched as a possibility for managing S. asper and other Sonchus species in Australia (CSIRO, 2007). Sonchus species have previously been the target of biological control in Canada (Peschekn, 1984). Preliminary surveys for biological control agents have been made in southern France and northern Europe (Scott & Jourdan, 2005; Peschken, 1984). Currently the rust fungus Miyagia pseudosphaeria and an unidentified eriophyid mite species are two widespread organisms in Australia with potential for control. The CSIRO project will focus on determining the impact of these potential control agents on related native Australian species (Sonchus hydrophilus and Actites megalocarp and on weedy S. oleraceus and S. asper (CSIRO, 2007) [extract from GISD 2011]. BI O TOTA C L O SCOR NT E IM EF R (speci PA FO BC OL es CT RT fea SPECIES FEASIBILITY OF BIOCONTROL FE IMPA SC SC sib AS CT x O O le IBI Bioco RE RE LI ntrol TY x SC 1/Effo O rt) RE Rubus pinnatus 30 40 77 57.8 Y Rubus pinnatus is rapidly grows to a much larger size than the common bramble of Europe (Rubus fruiticosus). Attempts to extirpate it have failed. The roots grow to a great size and every bit left in the ground grows (Jackson 1905) [extract from GISD 2011]. Nothing is known about BC of this species but other Rubus species such as R. niveus and R. ellipticus are currently subject to research into BC and it seems feasible that a control of R. pinnatus in the absence of any commercially valuable closely related species is possible. Chenopodium murale 25 24 52 54.2 Y Despite the long list of natural enemies which attack or are harboured by this weed (see Natural Enemies), only two have been tested for biological control. These were Indian ring dove (Streptopelia decaoto) and the common house sparrow (Passer domesticus). Both were found to feed on seeds of this weed species (Sharma, 1977; Tomar and Singh, 1980) [extract from CABI ISC 2011]. Cynodon dactylon 30 40 69 51.8 Y Drechslera cynodontis, Ustilago cynodontis, Puccinia cynodontis, and Fusarium poae as fungal pathogens and a specimen of Thripidae family were identified on C. dactylon and are being studied for potential use in biological control (Uygur, 2000) [extract from CABI ISC 2011]. Adiantum raddianum and A. capillus-veneris 30 38 65 51.3 Y no information on BC available Paspalum scrobiculatum 30 40 68 51.0 Y no information on BC available Spermacoce verticillata 30 40 68 51.0 Y no information on BC available; there is also a theoretical threat that any BC agent could endanger the endemic Oldenlandia adscensionis, which although now believed to be extinct is closely related to S. verticillata and herbivore may be able to attack also this species. Buddleja madagascariensis 20 31 77 49.7 Y Nothing is known about BC of this species but other species within this genus are currently subject to research into BC and it seems feasible that a control of Buddleja is possible in the future. In 2006, Cleopus japonicus was introduced and released as a potential biocontrol agent for B. davidii in New Zealand (Zhang et al., 1993; Kriticos, 2006; Watson, 2007). Further releases were made in 2007 and 2008 following careful monitoring of weevil behaviour and establishment (Watson 2008). As of 2009, it was still considered too early to judge the field effectiveness of C. japonicus. A second species under consideration for biological control of B. davidii in New Zealand is the stem weevil, Mecysolobus erro. The adults feed on the tender terminal shoots causing tips to wither and die. Host-range testing of this species is still underway (Kay, 2002) [extract from CABI ISC 2011]. BI O TOTA C L O SCOR NT E IM EF R (speci PA FO BC OL es CT RT fea SPECIES FEASIBILITY OF BIOCONTROL FE IMPA SC SC sib AS CT x O O le IBI Bioco RE RE LI ntrol TY x SC 1/Effo O rt) RE Sonchus oleraceus 20 21 52 49.5 Y Biological agents are currently being researched as a possibility for managing S. oleraceus and other Sonchus species in Australia (CSIRO, 2007). Sonchus species have previously been the target of biological control in Canada (Peschekn, 1984). Preliminary surveys for biological control agents have been made in southern France and northern Europe (Scott & Jourdan, 2005; Peschken, 1984). Currently the rust fungus Miyagia pseudosphaeria and an unidentified eriophyid mite species are two widespread organisms in Australia with potential for control. The CSIRO project will focus on determining the impact of these potential control agents on related native Australian species (Sonchus hydrophilus and Actites megalocarp and on weedy S. oleraceus and S. asper (CSIRO, 2007) [extract from GISD2011]. Ageratum conyzoides 30 38 51 40.3 N no information on BC available Melinis minutiflora 20 40 75 37.5 N no information on BC available Digitaria ciliaris 25 38 54 35.5 N There are no reports of serious interest in, or attempts at biological control (CABI ISC 2011). Digitaria sanguinalis 25 37 52 35.1 N There are no reports of serious interest in, or attempts at biological control. Setaria verticillata 25 40 56 35.0 N no information on BC available Conyza bonariensis 25 38 49 32.2 N no information on BC available Euphorbia hirta 30 50 37 22.2 N Jeritta & David (1986) reported that Haplothrips euphorbiae was specific to E. hirta, however no information on BC is currently available (CABI ISC 2011). Solanum nigrum 30 50 36 21.6 N No natural enemies have been recorded for S. nigrum, although it has been shown to be a host for many nematode and virus pests, and the thrip pest Frankliniella occidentalis (Lacasa et al., 1996) [extract from CABI ISC 2011].
Arthropods Only nine non-native arthropod species had an initial impact score of 20 or above and were processed further within the selected prioritisation tool. Preliminary results of this assessment are given in table 3 below. As with the weed species we regard BC not to be feasible below a cut-off point of an overall score of 50. However, a high overall score of an individual species does not automatically result in an immediate recommendation for BC of this species as a high priority. Further evaluation is needed on a case by case assessment taking other control options and costs versus benefits into account. Currently we regard only one arthropod species, the Hemiptera Icerya purchasi, a worthwhile target for classical BC. However, this species has been the subject of at least partially successful biological control in the past and renewed efforts should only be taken up in case current control levels by Rodiola cardinalis (particularly on the endemic spurge (Euphorbia origanoides) are not sufficient. In addition, augmentative control of Periplaneta americana is feasible, particularly as there are no native or endemic cockroach species known from Ascension. For other species where BC is feasible as such but no effective biological control agents have been identified in the past, we currently regard the impact these species have on the indigenous habitats as too low to justify the implementation of full scale BC projects. However, this preliminary judgement may turn out to be unjustified for some of these species and remains open for reassessment at any time. Species belonging to this group are Musca domestica and Cryptophlebia leucotreta. Despite a total score of above 50 we do not regard BC for the termite Cryptotermes brevis currently as feasible. Many efforts over long time periods have been made to find suitable BC option for termites, but so far no effective agents have been identified despite the huge amount of efforts being put into this topic.
Table 3: Preliminary results feasibility for BC for non-native arthropods on Ascension TOTA L SCOR BIOC E IM EF ONTR (speci PA FO OL BC es CT RT FEASI fea SPECIES FEASIBILITY OF BIOCONTROL IMPA SC SC BILIT sib CT x O O Y le Bioco RE RE SCOR ntrol E x 1/Effo rt) Icerya purchasi; Hemiptera (Margarodidae) 35 4 123 1076.3 Y Rodiola cardinalis was released on Ascension Island in 1976 to control cottony cushion scale on ornamental plants (Cronk 1980, Pickup 1999). The beetle also provided protection for the endemic spurge Euphorbia origanoides, which was suffering decline due to scale attack (Cronk 1980, Fowler 2004). By 1995 scale density on the island was low, and the scale was completely absent on sampled E. origanoides populations (Cronk 1980). Icerya purchasi was also absent on E. origanoides when the plant was surveyed in 1997 (Ashmole & Ashmole 1997, van Driesche 2010). Rodolia cardinalis was also introduced in 1896 and 1898 to control this pest on citrus on St. Helena (Greathead 1971), but nothing is known about the success on citrus and ornamental plants on Ascension. It may therefore be worth to follow up the success of this control programme. More agents (as released on Sao Tome, before 1960) are readily available for the control of I. purchasi (Cryptochaetum iceryae; Cryptochetidae, Diptera) (Greathead 1971). Multiple introduction of Rodolia plus Cryptochetum may be promising (CABI ISC 2011). TOTA L SCOR BIOC E IM EF ONTR (speci PA FO OL BC es CT RT FEASI fea SPECIES FEASIBILITY OF BIOCONTROL IMPA SC SC BILIT sib CT x O O Y le Bioco RE RE SCOR ntrol E x 1/Effo rt) Periplaneta americana; Blattodea (Blattidae) 30 12 63 157.5 Y Little information is available regarding classical BC of cockroaches, but there are numerous cases of augmentative BC for this group. Dolichurus stantoni (Hym. Ampulicidae), Rhinopsis caniculatus (Hym. Ampulicidae), Sceliphron haematogastrum (Hym. Sphecidae) have been introduced against cockroaches in general into Hawaii (Lai & Funasaki 1983, Biocat 2010). Ampulex compressa (Hym. Sphecidae) was introduced into Guam & Hawaii specifically to control this species (Lai & Funasaki 1983, Rao et al. 1971, Biocat 2010). Development of classical biological control programs would require a large scale research efforts and efficacy of this methods for this taxonomic group remains speculative. Therefore such efforts can only be justified by accordingly large impacts on native species and habitats, which have not been recorded from Ascension yet. Musca domestica; Diptera (Muscidae) 20 14 81 115.7 Y There is a long history of previous CBC of Musca domestica and Muscidae in general. Consequently, a number of suitable host specific control agents are available (Biocat 2010). However, even relatively minor costs for 'of the shelf' solutions need be set against substantial impacts on native species and habitats or as a nuisance pest, which have not been recorded from Ascension yet. Cryptophlebia leucotreta; Lepidoptera (Tortricidae) 25 22 69 78.4 Y BC of this species has been done in the past but is currently not deemed to be cost effective (Newton & Odendaal 1990). Development of a biological control program for this species applied on Ascension would require rather large scale research efforts. Therefore, such efforts can only be justified by accordingly large recorded impacts on native species and habitats, which have not been recorded from Ascension yet. Cryptotermes brevis; Isoptera (Calotermitidae) 20 25 66 52.8 N Despite long running attempts to find suitable BC agents, so far no sufficient control has been achieved within Isoptera (Chouvenca et al. 2011). Scolopendra morsitans; Chilopoda (Scolopendridae) 25 29 37 31.9 N There is no history of previous BC of Chilopoda in general. Development of biological control programs would require a large scale research efforts and efficacy of this methods for this taxonomic group remains speculative. Therefore such efforts can only be justified by accordingly large recorded impacts on native species and habitats, which have not been recorded from Ascension yet. Paratrechina longicornis; Hymenoptera (Formicidae) 25 31 32 25.8 N Biological control agents exist for ants, such as phorid flies (Vazquez et al., 2006). However, the influence of such agents in regulating any ant population is yet to be demonstrated and generally CBC for ants has not been developed to level, where it can be successfully applied yet (CABI ISC 2011, Rabitsch 2011). Pheidole megacephala; Hymenoptera (Formicidae) 30 31 25 24.2 N Biological control agents exist for ants, such as phorid flies (Vazquez et al., 2006). However the influence of such agents in regulating any ant population is yet to be demonstrated and generally CBC for ants has not been developed to level, where it can be successfully applied yet (CABI ISC 2011, Rabitsch 2011). TOTA L SCOR BIOC E IM EF ONTR (speci PA FO OL BC es CT RT FEASI fea SPECIES FEASIBILITY OF BIOCONTROL IMPA SC SC BILIT sib CT x O O Y le Bioco RE RE SCOR ntrol E x 1/Effo rt) Erechthias minuscula; Lepidoptera (Tineidae) 20 48 11 4.6 N There is no history of previous CBC of this species, however for many Lepidoptera in general. Development of a biological control program would therefore be likely to be successful, but would still require a substantial amount of research efforts. Such efforts can only be justified by accordingly large recorded impacts on native species and habitats, which have not been recorded from Ascension yet.
Annexes
Annex 1: Evaluation tool (adjusted from Paynter et al. 2009) SPECIES IMPORTANCE & DESIRABILITY OUTCOME Prosopis juliflora OF BIOCONTROL 1. Is the species native or at least possibly native? a. Yes and in its natural range Biological control is undesirable Biological control may be ecologically feasible, go b. Yes, but occurring outside its natural range to question 2.
c. No, but possibly colonised island(s) naturally Biological control is undesirable d. No Go to question 2. Y 2. Is opposition to biocontrol likely? Does the species have socioeconomic value a. Yes and value of species demonstrably >cost of control/detrimental impacts Biological control is undesirable i. species valuable as control agent of introduced other non-native species Biological control is undesirable Cost-benefit data is likely to be required for b. Yes, but cost-benefit analysis data does not biocontrol to proceed, delaying biocontrol, adding exist to cost & uncertainty of success c. Yes and value of species demonstrably the island? a. No, only records of single specimen(s) or doubtful records Biological control is unfeasible b. No, no recent records Biological control is unfeasible c. Establishment unknown Biological control is unfeasible d. Yes Go to question 4. Y 4. Has the species already been subject of biological control on the island(s) a. Yes, successful control achieved Further biological control is of low priority b. Yes, but control insufficient or success not sufficiently monitored Go to question 5. Y c. No Go to question 5. 5. Has eradication recently recommended as suitable control method a. Eradication has been recommended and only small populations locally restricted exist Biological control is undesirable b. Eradication has not been recommended but the occurance and life history of the species does not warrant biological control measures Biological control is undesirable c. Eradication has been suggested as control method but species is already widespread Go to question 6. d. Eradication has not been recommended as as suitable control method Go to question 6. Y Cont. annex 1: Evaluation tool (adjusted from Paynter et al. 2009) 6. Is the species sufficiently widespread/intractable/important to Question justify investment in biocontrol? Is the species only recently introduced/recorded on the island(s)? (0-no, or no data avaliable; 5-yes) 0 Does the species currently have a widespread recorded distribution on the island(s)? (0-no, or no data avaliable; 5-yes) 5 Is the species currently expanding its range on the island(s)? (0-no, or no data avaliable; 5-yes) 5 Is the species known to be invasive in other geographical regions? (0-no, or no data avaliable; 5-yes) 5 Is the species known to have negative impacts on biodiversity in other geographical regions? (0-no, or no data avaliable; 5-yes) 5 Is the species known to cause economic damage in other geographical regions? (0-no, or no data avaliable; 5-yes) 5 Is the species known to cause threats to human or animal health in other geographical regions? (0-no, or no data avaliable; 5-yes) 0 Are habitats on the island(s) which can potentially be invaded of conservation importance? (0-no, or no data avaliable; 5-yes) 5 Does the SAIS assessment suggest a likely negative impact on biodiversity, agriculture or health (0-no; 5-yes) 5 Have ecological impacts been recorded on the island(s) for this species? (0-no; 10-yes) 5 Have economic impacts (agriculture and others) been recorded on the island(s) for this species? (0-no; 10-yes) 5 Have impacts on human or animal health recorded on the island(s) for this species? (0-no; 10-yes) 0 species IMPACT SCORE 45 Ashmole & Ashmole 2000; References records, impact and Belton 2008; Lambdon et al. invasiveness in other geographical 2009; Varnham 2009; CABI regions ISC 2011, GISD 2011 Cont. annex 1: Evaluation tool (adjusted from Paynter et al. 2009) 7. Is impact score equal or above 20 if yes, go to next question a. No Biological control is undesirable b. Yes Go to next set of questions Y EFFORT REQUIRED TO OBTAIN & HOST-RANGE TEST BIOCONTROL AGENTS 8. Has the species been/is it a subject of adequately resourced biocontrol program elsewhere? If specific agents are already known & host-range testing has already been conducted overseas, then programme is a. Yes, successful program likely to be cheaper 1 1 Law of diminishing returns - if current known suite of agents is ineffective, finding b. Yes, unsuccessful program new ones will be harder 15 c. Current target/too early/insufficient data to assess success elsewhere or variable success Potential for cost savings, but uncertainty elswhere factored into score 8 d No, never 20 9. Accessibility and ease of working in native range a. Difficult 5 b. Moderate 3 c. Easy 2 d. Not applicable (if repeat programme) 1 1 10. Literature regarding natural enemies well known/accessible a. Yes 1 1 Formal identification of candidate agents (required for import/release permits) may b. No be time consuming, delaying a program 5 11. Phylogeny: How closely related to indigenous/valued species is the target species? Cheaper no-choice tests may be sufficient, a. None in same family larget pool of candidate agents 1 b. Same Family 10 10 More extensive host-range testing may be required, more species may require testing before a sufficiently specific species is c. Same Genus identified 20 EFFORT SCORE 13 Cont. annex 1: Evaluation tool (adjusted from Paynter et al. 2009) PREDICTING THE POTENTIAL IMPACT OF BIOCONTROL 12. Has the species been a subject of adequately resourced biocontrol programme overseas? a. Yes, successful target overseas 1 or more Maximum score: do not go to next set of occasions questions 100 b. Yes, only one example or with varying degrees of success or partial success i. Reason for variable success known (e.g. agent only attacks certain forms of species, or is restricted to certain habitats/climates) and considered unlikely to be a problem Do not go to next set of questions 80 80 ii. Reason for variable success unknown Do not go to next set of questions 60 iii. Reason for variable success known and considered likely to be a problem Do not go to next set of questions 40 c. Unsuccessful target overseas only once 30 d. Unsuccessful target overseas more than once 20 e. Not a target elsewhere or too early to assess success of overseas program Go to next set of questions 1 13. Habitat Aquatic/wetland Higher probability of success 35 Terrestrial Lower probability of success 14 14. Life cycle Predominantly temperate annual 3 Predominantly tropical/sub tropical annual 5 Biennial/perennial 5 15. Reproduction Vegetative (+/- seed/spore) 25 Seed/spore only 10 16. Weed in native range a. Yes Lower probability of success 3 b. No Higher probability of success 10 17. Difficulty targetting multiple forms of weed, or probability of replacement of the weed by forms or congeners of the target following successful biological control thereby negating benefits a. Likely Lower probability of success 0 b. Unlikely Higher probability of success 5 c. Unknown 2 18. Occurring in competitive environment (agricultural vs environmental) a. Predominantly agricultural Lower probability of success 1 b. Predominatly environmental Higher probability of success 5 c. Unknown/both equally 3 19. Native/valued exotic congener a. Yes 0 b. No 15 References for efforts, potential impact and on Belton 2008; CABI ISC BC in other geographical areas 2011, GISD 2011 BIOCONTROL FEASIBILITY SCORE 80 TOTAL SCORE (species IMPACT x Biocontrol x 1/Effort) 276.92 Comments Several biological control programmes using species of seed-feeding bruchid beetles have been developed and implemented. The advantage with bruchids is their observed host specificity, with many species found to feed only on Prosopis, and some only on a single species. Other insect species known to have a deleterious effect on native and exotic Prosopis in the Americas, mainly twig girdlers and psyllids, have also been suggested as possible biological control agents. The twig girdler Oncideres limpida attacks P. pallida in Brazil (Lima, 1994), whereas Oncideres rhodostricta is seen as a serious pest of P. glandulosa in the USA (Polk and Ueckert, 1973). Psyllids are known to severely affect the growth of Prosopis (Hodkinson, 1991) and have been suggested for use in controlling invasions. The same two bruchid species were also introduced to Ascension Island in an attempt to control P. juliflora which is present on 80% of the island, often in dense thickets. Two other species, one a psyllid and the other a mirid, were identified as attacking P. juliflora on Ascension Island and were thought to have been introduced accidentally from the Caribbean. The mirid Rhinocloa sp. causes widespread damage and is thought to lead to substantial mortality of trees (Fowler, 1998) [extract from CABI ISC 2011]. contact CABI europe CABI Head Office Nosworthy Way, Wallingford, Oxfordshire, OX10 8DE, UK T: +44 (0)1491 832111 CABI Bakeham Lane, Egham, Surrey, TW20 9TY, UK T: +44 (0)1491 829080 CABI Rue des Grillons 1, CH-2800 Delémont, SWITZERLAND T: +41 (0)32 4214870 asia CABI C/o Internal Post Box 56, Chinese Academy of Agricultural Sciences, 12 Zhongguancun Nandajie, Beijing 100081, CHINA T: +86 (0)10 82105692 CABI 2nd Floor, CG Block, NASC Complex, DP Shastri Marg, Opp. Todapur Village, PUSA, New Delhi – 110012, INDIA T: +91 (0)11 25841906 CABI PO Box 210, 43400 UPM Serdang, Selangor, MALAYSIA T: +60 (0)3 89432921 CABI Opposite 1-A, Data Gunj Baksh Road, Satellite Town, Rawalpindi-PAKISTAN T: +92 (0)51 9290132 africa CABI ICRAF Complex, United Nations Avenue, Gigiri, PO Box 633-00621, Nairobi, KENYA T: +254 (0)20 7224450/62 americas CABI UNESP- Fazenda Experimental Lageado, Rua: José Barbosa de Barros, 1780 Botucatu – SP, CEP: 18610-307, BRAZIL T: (14) 3882 - 6300 / 3811 - 7127 CABI Gordon Street, Curepe, TRINIDAD AND TOBAGO T: +1 868 6457628 CABI 875 Massachusetts Avenue, 7th Floor, Cambridge, MA 02139, USA T: +1 617 3954051