Benefits and risks of exotic biological control agents

The use of exotic (=alien) arthropods in classical and augmentative biological control programs has yielded huge economic and ecological benefits. Exotic species of arthropods have contributed to the suppression of key pests in agriculture and forestry or have aided in restoring natural systems affected by adventive species. However, adverse non-target effects of exotic biological control agents have been observed in a number of projects. Non-target effects range from very small effects, e.g. 2% parasitization on a non-target insect on a local level, to massive effects on a large scale. Until now, no consensus on how to judge the magnitude of non-target effects and whether these effects can be tolerated or are unacceptable has emerged. In this paper, we briefly review both the benefits of biological control as well as the associated risks including to human and animal health, plant health and particularly the environment. We also make an attempt at identifying the major challenges for assessing risks and for balancing benefits and risks. There is general agreement that sound risk assessment procedures should precede the release of exotic invertebrate biological control agents and a recent shift??especially for arthropod biological control??from introductions done without meaningful risk assessment studies to projects conducting thorough host range testing can be observed. However, overly stringent regulations that would preclude promising agents from being developed must be avoided.     

外来入侵杂草传统生物防治的生态风险及其防范对策

长期大量实践说明,引进天敌防治外来入侵杂草的传统生物防治方法是治理外来入侵杂草的一条切实可行的有效途径,但对其潜在的生态风险——对本土生物的直接或间接不良影响不容忽视。利用传统评价方法预测候选天敌的生态风险存在缺陷,主要表现在：（1）寄主专一性测定过分依赖室内进行的生理寄主范围测定结果,对生态寄主范围（实际寄主范围）问题重视不够,后者指在新环境中的一系列物理和生物条件下的寄主利用预测;（2）在生理寄主范围测定中,过分依赖完成生长发育的可能性,对行为、遗传性状以及系统发育关系重视不够;（3）在风险评估中,过多强调对经济作物的风险,而对自然生态系统的风险重视不够。对此,建议：（1）鼓励对已释放的天敌进行回顾性跟踪研究,从而为杂草生物防治实践提供生态学理论支撑;（2）在运用生物防治手段对付外来入侵杂草实践中,建议采用“有害推论”的预防性原则,以避免在面临入侵生物重大威胁时草率做出释放天敌的决策;（3）在评估候选天敌风险中重视生态效应的风险评估。     

Evolution of biological control agents following introduction to new environments

The introduction of biological control agents (BCAs) creates the potential for adaptive evolution in translocated organisms. BCAs are confronted with new environments that can promote adaptation to exploit novel resources, even within short ecological time frames. In particular, insect BCAs are amenable to rapid evolution due to their short generation times and relatively large population sizes. These factors hypothetically increase the likelihood that, when exposed to novel habitats, environmental selection could cause BCAs to extend their range to non-target host species. Alternatively, insects may simply extend their range as their generalist or polyphagous habits are fully realized. In this review, we consider recent literature that addresses these topics. Adaptations to environmental conditions have been demonstrated in a number of BCAs. Mechanisms of adaptation include founder effects, hybridization, and endosymbiosis. Yet, there is little evidence of adaptive host range expansions among insect and weed biological control agents to non-target species, albeit existing examples are from limited numbers of studies. Important future directions and current developments in the field incorporate next generation sequencing technology that can promote better resolution of population divergence, possible mechanisms involved in adaptation to novel resources, and insect hybridization. Future studies should also include a careful consideration of the influence of microbes on BCA efficacy and environmental adaptation.     

Does parasitoid state affect host range expression?

The pre-release risk assessment of parasitoids for classical biological control generally involves non-target testing to define the agent’s host range. To ensure that no suitable host species are falsely rejected in these tests, it has been suggested that the physiological and informational state of parasitoids be manipulated to enhance their “motivation to oviposit”. However, the effects of such factors on host acceptance are not consistent across parasitoid species, making it laborious to identify the conditions necessary to maximise host acceptance. Our objective was to determine whether changes in parasitoid state could alter host acceptance behaviour sufficiently to affect host range expression. In addition, we tested the assumption that a state-dependent shift in motivation to oviposit on the target host will translate to a similar change in responsiveness to lower-ranked host species. Three-day-old and 10-day-old females of the candidate classical biological control agent, Diadromus pulchellus, were offered 12 non-target species of varying relatedness to the target pest, Acrolepiopsis assectella, in a series of no-choice and choice oviposition trials. Younger D. pulchellus females had previously demonstrated greater motivation to oviposit in the target pest and were, therefore, predicted to express a broader host range than older females. Parasitoid age had a minor effect on host range expression that was contrary to expectations. Older females more readily attacked one of the non-target species in no-choice tests and inflicted higher mortality in one of the choice tests. Ultimately however, young and old parasitoids still attacked the same four non-target species and their offspring emerged from the same three. There was an interaction between the effects of parasitoid condition and experimental design on responsiveness to low-ranked hosts: increasing non-target density in choice tests significantly altered attack rates by 10-day-old, but not by 3-day-old, parasitoids. The implications of these findings for host specificity testing depend largely on the specific aims of a host range assessment. Parasitoid state influenced the frequency of non-target attack but did not affect which non-target species were attacked.     

Methods Used to Assess Non-target Effects of Invertebrate Biological Control Agents of Arthropod Pests

The number of concerns regarding potential non-target effects of invertebrate biological control agents of arthropods has risen over the last decade and an increasing number of studies have since dealt with this topic. Despite some recent international initiatives aimed at providing guidance for risk assessment of biological control agents, detailed methods on how tests should be designed and conducted to assess for potential non-target effects still need to be provided. It is believed that this review comes at an ideal time, giving an overview of methods currently applied in the study of non-target effects in biological control of arthropod pests. It provides the first step towards the ultimate goal of devising guidelines for the appropriate methods that should be universally applied for the assessment and minimisation of potential non-target effects. The main topics that are reviewed here include host specificity (including field surveys, selection of non-target test species and testing protocols), post-release studies, competition, overwintering and dispersal. Finally, a number of conclusions that have emerged from this comprehensive compilation of studies are drawn, addressing potential non-target effects in arthropod biological control.     

Environmental risk assessment of the egg parasitoid <Emphasis Type="Italic">Anaphes inexpectatus</Emphasis> for classical biological control of the <Emphasis Type="Italic">Eucalyptus</Emphasis> snout beetle, <Emphasis Type="Italic">Gonipterus platensis</Emphasis>

Classical biological control is a valuable tool against invasive pests, but concerns about non-target effects requires risk assessment studies. Potential non-target effects of Anaphes inexpectatus Huber and Prinsloo (Hymenoptera: Mymaridae) were assessed for a classical biological control programme against the Eucalyptus snout beetle, Gonipterus platensis (Marelli) (Coleoptera: Curculionidae). No-choice tests were conducted with 17 non-target species to assess host specificity, including 11 curculionids. In behavioural observations, A. inexpectatus showed no interest in any of the non-target species, but two weevil species were parasitised within five days of exposure, although at significantly lower rates than G. platensis. In choice tests, only one non-target, Hypera postica (Gyllenhal) (Coleoptera: Curculionidae), was parasitised, at a rate of 0.6%, while 50.0% of G. platensis eggs were parasitised. Based on the host specificity test results and the potential host fauna found in the target area, the likelihood of non-target effects resulting from the release of A. inexpectatus is considered to be negligible.     

Host range tests reveal Paectes longiformis is not a suitable biological control agent for the invasive plant Schinus terebinthifolia

The most critical step during a weed biological control program is determination of a candidate agent’s host range. Despite rigorous protocols and extensive testing, there are still concerns over potential non-target effects following field releases. With the objective to improve risk assessment in biological control, no-choice and choice testing followed by a multiple generation study were conducted on the leaf-defoliator, Paectes longiformis Pogue (Lepidoptera: Euteliidae). This moth is being investigated as a biological control agent of Schinus terebinthifolia Raddi (Sapindales: Anacardiaceae), which is one of the worst invasive plant species in Florida, USA. Results from no-choice testing showed higher larval survival on S. terebinthifolia (48 %) and its close relative Schinus molle L. (47 %), whereas lower survival was obtained on six non-target species (<25 %). When given a choice, P. longiformis females preferred to lay eggs on the target weed, but oviposition also occurred on four non-target species. An improved performance on the native Rhus aromatica Aiton was found when insects were reared exclusively on this non-target species for one or two generations. Results from host range testing suggest that this moth is oligophagous, but has a preference for the target weed. Non-target effects found during multiple generation studies indicate that P. longiformis should not be considered as a biological control agent of S. terebinthifolia.     

Predictability of pathogen host range in classical biological control of weeds: an update

Before an exotic pathogen can be released as a classical biological control agent the likely positive and negative outcomes of that introduction must be predicted. Host range testing is used to assess potential damage to non-target plants. To-date 28 species of fungi have been released as classical biological control agents against weeds world-wide. These pathogens have been reported infecting only six non-target plant species outdoors and all of these incidents were predicted. Many more non-target plant species developed disease symptoms in glasshouse tests than in the field. Consequently, data from other sources are needed to ensure potential agents are not prematurely rejected. Predictions of pathogen host range to date have been sufficiently accurate to prevent unpleasant surprises. Exotic pathogens are a safe and useful tool for weed control, especially in natural areas rich in valued non-target species.     

Comparing predictions from mixed model equations with host range determinations from historical disease evaluation data of two previously released weed biological control pathogens

Host range determination is essential to risk assessments of classical biological weed control agents and needed for petitions to release of agents into non-native areas. Recently Henderson’s mixed model equations (MME) were adapted to evaluating host ranges of two biological control pathogens. This was accomplished by incorporating genetic relatedness, in the form of DNA sequences, with disease evaluation data among target and non-target plants to generate best linear unbiased predictors (BLUPs) of disease reaction among these plant species. Although qualities of MME analysis provide inherent validation of the approach, it has been difficult to validate by comparison with historical host range determination or field data. These types of data are virtually non-existent, because original, raw host range data from replicated tests have, for the most part, been lost, and host-range data from tests or monitoring of multiple plant species in the field are extremely limited. In this study, the host-ranges of two biological weed control pathogens previously released in the USA were re-evaluated with the MME and historical host-range data. Results showed that: BLUPs confirmed the narrow host range of the two pathogens; BLUPs supported and substantiated currently available information on host range; BLUPs were accurate predictors of host range, particularly concerning susceptibility of any non-target species in the field; and no new, unforeseen non-target effects could be expected in nature on the basis of BLUPs. Thus, validation was made for MME analysis in predicting known field host-range from greenhouse data.     

Controlling Lygus plant bugs (Heteroptera: Miridae) with European Peristenus relictus (Hymenoptera: Braconidae) in Canada--risky or not?

The European Peristenus relictus Loan (syn. P. stygicus) has been considered for biological control of Lygus plant bugs native to Canada. Laboratory and field studies were conducted in the area of origin to evaluate the host specificity of P. relictus. Laboratory choice and no-choice tests demonstrated that P. relictus attacked all non-target species offered (fundamental host range). However, closely related non-target mirids (tribe Mirini) were generally well accepted by P. relictus, while hosts from the tribe Stenodemini were less frequently attacked and less suitable for parasitoid development. To validate the laboratory results, a thorough examination of the parasitoid complex of common mirids in Europe was conducted to determine which non-target species may serve as alternative hosts for P. relictus in a natural situation (ecological host range). When comparing both approaches, the fundamental host range of P. relictus matched its ecological host range. In addition to three Lygus species, the ecological host range of P. relictus in the area of Schleswig-Holstein, northern Germany, contains at least 16 non-target species, including hosts belonging to the subfamilies Mirinae, Phylinae and Bryocorinae. A broad ecological and fundamental host range suggests that P. relictus is a generalist; however, P. relictus was not the primary contributor to parasitism of most non-target hosts studied. Although P. relictus is assumed to be of minor importance for regulating non-target populations in the area of investigation, the results of the present study indicate that P. relictus has the potential to use non-target host populations for reproduction.     

Avoiding conflicts between insect and weed biological control: selection of non-target species to assess host specificity of cabbage seedpod weevil parasitoids

Abstract:  Classical biological control of insect pests and weeds may lead to potential conflicts, where insect pests are closely related to weed biological control agents. Such a conflict may occur in the classical biological control of the cabbage seedpod weevil, Ceutorhynchus obstrictus (Marsham) in North America, which belongs to the same subfamily, Ceutorhynchinae, as a number of agents introduced or proposed for introduction against non-indigenous invasive weed species. We propose a step-by-step procedure to select non-target species and thereby to develop a non-target species test list for screening candidate entomophagous biological control agents of a herbivore pest insect in a way that would simultaneously evaluate non-target potential on weed biological control agents and other non-target species. Using these recommendations, we developed a non-target test list for host specificity evaluations in the area of origin (Europe) and the area of introduction (North America) for cabbage seedpod weevil parasitoids. Scientifically based predictions on expected host–parasitoid interactions and ecological information about the ecological host range in the area of origin can help avoid conflicts, while still allowing the introduction of safe and effective agents against both insect pests and weeds.     

Beyond Pandora’s Box: quantitatively evaluating non-target effects of parasitoids in classical biological control

A seminal paper by Howarth (Proc Hawaii Entomol Soc 24:239–244, 1983) entitled “Classical biological control: Panacea or Pandora’s Box” ignited a sometimes acrimonious debate over the relative safety of introductions for classical biological control. Extolled for years as environmentally benign, the litany of negative non-target effects profiled by Howarth heightened awareness of this issue. Several factors have muddied this debate including the conflation of frequency of effects with their strength, grouping the effects of disparate biological control agents together, and the lack of quantitative data on either side of the argument. Here, I examine the potential for non-target effects among insect parasitoids, the most common group used for biological control of arthropods. In response to calls for better quantitative studies, I highlight three different techniques, quantitative food webs, life table analysis, and experimental populations, respectively, to quantitatively assess or reassess non-target effects in different systems. I also explore three methodological approaches employed to ascertain the strength of competitive interactions between native and introduced parasitoids, a potential non-target effect that has received little attention in the literature. These types of studies may greatly increase our understanding of the nature of non-target interactions with introduced parasitoids and bring more rigor to a debate often dominated by rhetoric.     

Should the flower bud weevil Anthonomus santacruzi (Coleoptera: Curculionidae) be considered for release against the invasive tree Solanum mauritianum (Solanaceae) in New Zealand?

The invasive tree Solanum mauritianum Scop. has been targeted for biological control in South Africa and New Zealand, by deploying insect agents that could constrain its excessive reproductive output. The flower-feeding weevil Anthonomus santacruzi (Curculionidae) was approved for release in South Africa in 2007 but following the loss of the original culture in quarantine, new stocks were introduced from Argentina in 2008–2009. This study was initiated to confirm that the host range of the new culture was the same as that of the previous one, but also to assess the risks associated with the weevil's release in New Zealand. Different testing procedures, including no-choice tests and multi-choice tests in different arenas, produced inconsistent and ambiguous results. During no-choice tests, oviposition and larval development to adulthood occurred on three non-target species including two native South African and one native New Zealand Solanum species. However, subsequent multi-choice tests and a risk assessment suggested that the risks of anything more than collateral damage to non-target Solanum species are low. Overall, these data do not deviate substantially from the results of the original quarantine tests which facilitated the release of A. santacruzi in South Africa in 2009. Although we argue that none of the New Zealand native and cultivated species are at risk, stronger evidence from open-field trials and chemical ecology studies may be required to convince the regulatory authorities that A. santacruzi is suitable for release in New Zealand.     

Utilization of the Exotic Weed Pluchea odorata (Asteraceae) and Related Plants by the Introduced Biological Control Agent Acinia picturata (Diptera: Tephritidae) in Hawaii

The flower-head feeding fly Acinia picturata (Diptera: Tephritidae) was deliberately introduced from Mexico into Hawaii in 1959 for biological control of the exotic weed Pluchea odorata (Snow) (Asteraceae). Neither field efficacy nor possible non-target effects of the fly have been evaluated in the 40 years since the introduction. We assessed the impact of the fly on both its target host and on seven non-target plant species. The impact on the target weed was limited, with only 5-13% of the developing seeds in P. odorata flowerheads being destroyed by larval feeding. We did not detect any host range expansion of A. picturata onto flowerheads of two exotic or 5 endemic non-target plant species in the family Asteraceae.     

Estimating Fundamental Host Range: a Host-Specificity Study of a Potential Biocontrol Agent for Prosopis Species (Leguminosae)

One approach to predicting non-target attack by potential biological control agents is to first describe their fundamental host ranges and then to predict how it will be expressed under postrelease field conditions. In this paper, we illustrate how the fundamental host range can be estimated experimentally by excluding possible limiting factors such as time-dependent effects. The example we use is a host-specificity study of a leaf-tying moth (Gelechiidae: Evippe sp. #1) which was being assessed for the biological control of mesquite (Leguminosae: Prosopis spp.) in Australia. Females oviposited all eggs on plants, mostly into cracks and fissures. First instar larvae leaf-mined and subsequent instars leaf-tied. Oviposition was not host-specific in cage trials, although ten times more eggs were laid on Prosopis than on non-targets. The fundamental host range for initiation of larval feeding was restricted to Prosopis and Leucaena leucocephala which both belong to the same tribe, and the fundamental host range for complete larval development was restricted to Prosopis . We predict that if released in Australia Evippe sp. #1 will only attack Prosopis spp., although low levels of 'indiscriminate' oviposition might occur on other taxa, and might result in initiation of feeding on L. leucocephala .     

Risk analysis and management decisions for weed biological control agents: Ecological theory and modeling results

Biologically based control methods offer many advantages for the control of invasive plant species; however, these methods are not without risks to native species. Thus, there is a need for more effective and efficient methods of risk analysis for biological control agents. We show how the process of ecological risk assessment established by the United States’ Environmental Protection Agency may be adapted to improve assessment of the risks of proposed biological control agents. We discuss the risks posed by weed biological control agents, and present a simple individual-based model of herbivorous insect movement and oviposition on two species of host plant, a target invasive plant species and a non-target native species, in simulated landscapes. The model shows that risks of non-target impacts may be influenced by the details of the movement behavior of biological control agents in heterogeneous landscapes. The specific details of insect movement that appear to be relevant are readily measured in field trials and the general modeling approach is readily adapted to real landscapes. Current biological control risk assessments typically emphasize effects analysis at the expense of exposure analysis; the modeling approach presented here provides a simple and feasible way to incorporate exposure analyses. We conclude that models such as ours should be given serious consideration as part of a comprehensive strategy of risk assessment for proposed weed biological control agents.     

Risk analysis and management decisions for weed biological control agents: Ecological theory and modeling results

Biologically based control methods offer many advantages for the control of invasive plant species; however, these methods are not without risks to native species. Thus, there is a need for more effective and efficient methods of risk analysis for biological control agents. We show how the process of ecological risk assessment established by the United States’ Environmental Protection Agency may be adapted to improve assessment of the risks of proposed biological control agents. We discuss the risks posed by weed biological control agents, and present a simple individual-based model of herbivorous insect movement and oviposition on two species of host plant, a target invasive plant species and a non-target native species, in simulated landscapes. The model shows that risks of non-target impacts may be influenced by the details of the movement behavior of biological control agents in heterogeneous landscapes. The specific details of insect movement that appear to be relevant are readily measured in field trials and the general modeling approach is readily adapted to real landscapes. Current biological control risk assessments typically emphasize effects analysis at the expense of exposure analysis; the modeling approach presented here provides a simple and feasible way to incorporate exposure analyses. We conclude that models such as ours should be given serious consideration as part of a comprehensive strategy of risk assessment for proposed weed biological control agents.     

Using the literature to evaluate parasitoid host ranges: a case study of Macrocentrus grandii (Hymenoptera: Braconidae) introduced into North America to control Ostrinia nubilalis (Lepidoptera: Crambidae)

We propose a method for using the literature to evaluate host ranges of parasitoids that are candidates for biological control introductions. Data on the parasitoids that attack a given host species can be used as negative evidence concerning the candidate whose host range is being evaluated. By compiling studies for a variety of host species, one can delineate those taxa unlikely to be attacked by the candidate. Using a retrospective case study of a parasitoid introduced into North America, we describe (1) this approach to using the literature to evaluate host range and (2) how well predictions based on such an evaluation match actual host range. Based on the host range of Macrocentrus grandii in Eurasia as reported in the literature, we predicted that the species in the genus Ostrinia are the most likely hosts. Of native North American species, Ostrinia obumbratalis is the only non-target species likely to be attacked by M. grandii. The predicted host range for North America matched the actual host range found in the field. This suggests that a careful literature review could be used as an important source of data on host range of parasitoid species proposed for introduction into a new environment.     

Open-field Tests in Host-specificity Determination of Insects for Biological Control of Weeds

Open-field tests may be used for the host-specificity determination of insects used in the biological control of weeds. Such tests allow insects to exercise free choice of plants without constraints associated with the use of cages. Therefore, this testing method can generate host data on candidate biocontrol agents under more natural conditions than those obtained via cage tests. The literature contains 24 studies of open-field testing, involving 13 target weed species, more than 34 species of insects and one eriophyid mite. Field-test data were used to support the release of 20 of these candidate agents into new countries. Most field tests have been conducted in concert with laboratory host-specificity tests or in response to the results of laboratory tests. This review also provides information on experimental designs, locations, categories of test plants included and the constraints of open-field testing.     

What Magnitude Are Observed Non-Target Impacts from Weed Biocontrol?

A systematic review focused by plant on non-target impacts from agents deliberately introduced for the biological control of weeds found significant non-target impacts to be rare. The magnitude of direct impact of 43 biocontrol agents on 140 non-target plants was retrospectively categorized using a risk management framework for ecological impacts of invasive species (minimal, minor, moderate, major, massive). The vast majority of agents introduced for classical biological control of weeds (>99% of 512 agents released) have had no known significant adverse effects on non-target plants thus far; major effects suppressing non-target plant populations could be expected to be detectable. Most direct non-target impacts on plants (91.6%) were categorized as minimal or minor in magnitude with no known adverse long-term impact on non-target plant populations, but a few cacti and thistles are affected at moderate (n = 3), major (n = 7) to massive (n = 1) scale. The largest direct impacts are from two agents (Cactoblastis cactorum on native cacti and Rhinocyllus conicus on native thistles), but these introductions would not be permitted today as more balanced attitudes exist to plant biodiversity, driven by both society and the scientific community. Our analysis shows (as far as is known), weed biological control agents have a biosafety track record of >99% of cases avoiding significant non-target impacts on plant populations. Some impacts could have been overlooked, but this seems unlikely to change the basic distribution of very limited adverse effects. Fewer non-target impacts can be expected in future because of improved science and incorporation of wider values. Failure to use biological control represents a significant opportunity cost from the certainty of ongoing adverse impacts from invasive weeds. It is recommended that a simple five-step scale be used to better communicate the risk of consequences from both action (classical biological control) and no action (ongoing impacts from invasive weeds).