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。     

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.     

Translating host-specificity test results into the real world: The need to harmonize the yin and yang of current testing procedures

In its modern era, the discipline of biological control can be perceived as being subject to a series of tensions due to differences in philosophy, different needs, and different practices; these include the view that biological control is environmentally friendly and desirable vs. the view that any organism alien to a particular habitat should be considered an undesirable invasive, the need to protect nontarget species vs. the need to introduce the most effective biological control agents and the need to know what an agent might attack under normal field conditions vs. the restrictive nature of current testing procedures. These tensions are particularly manifest in the area of host-specificity testing where researchers interact most directly with regulators and policy-makers and recent “incidents” of nontarget use by biological control agents in three regions have brought them to the fore. The empirical tension between needing more certainty about agent host-range in the real world and being obliged to test this primarily under highly restricted conditions is of most concern to researchers. This paper suggests ways to reduce this through improvements in (1) what plants are tested, (2) how the selected plants are tested, and (3) how the test data are interpreted and communicated to regulators. It is argued that the currently used centrifugal phylogenetic method for selecting test plant lists should be modernized to accommodate the many recent improvements in knowledge of plant phylogenetic relationships and insect host-choice evolution and behavior. The reliance on quarantine-based testing is briefly examined and low- and high-technology alternatives considered. Both these are rejected, at least in the short-term, as being impractical, and the use is proposed of comparative laboratory-based vs. open-field host tests against a few key nontarget species to proactively calibrate subsequent quarantine data and aid interpretation of results obtained under artificial conditions. While these improvements should help better translate host-test results to real-world outcomes, it is essential that there be increased transparency in the communication of these results to regulators, with discussions of different components of risk, such as localized vs. widespread and short-term collateral damage vs. long-term evolutionary impact. Harmonizing the tensions that currently impinge on biological control can only be achieved through improving the quality of information provided, which will help regulators make decisions that are based more on knowledge and less on the often ill-perceived fears of the public.     

Translating host-specificity test results into the real world: The need to harmonize the yin and yang of current testing procedures

In its modern era, the discipline of biological control can be perceived as being subject to a series of tensions due to differences in philosophy, different needs, and different practices; these include the view that biological control is environmentally friendly and desirable vs. the view that any organism alien to a particular habitat should be considered an undesirable invasive, the need to protect nontarget species vs. the need to introduce the most effective biological control agents and the need to know what an agent might attack under normal field conditions vs. the restrictive nature of current testing procedures. These tensions are particularly manifest in the area of host-specificity testing where researchers interact most directly with regulators and policy-makers and recent “incidents” of nontarget use by biological control agents in three regions have brought them to the fore. The empirical tension between needing more certainty about agent host-range in the real world and being obliged to test this primarily under highly restricted conditions is of most concern to researchers. This paper suggests ways to reduce this through improvements in (1) what plants are tested, (2) how the selected plants are tested, and (3) how the test data are interpreted and communicated to regulators. It is argued that the currently used centrifugal phylogenetic method for selecting test plant lists should be modernized to accommodate the many recent improvements in knowledge of plant phylogenetic relationships and insect host-choice evolution and behavior. The reliance on quarantine-based testing is briefly examined and low- and high-technology alternatives considered. Both these are rejected, at least in the short-term, as being impractical, and the use is proposed of comparative laboratory-based vs. open-field host tests against a few key nontarget species to proactively calibrate subsequent quarantine data and aid interpretation of results obtained under artificial conditions. While these improvements should help better translate host-test results to real-world outcomes, it is essential that there be increased transparency in the communication of these results to regulators, with discussions of different components of risk, such as localized vs. widespread and short-term collateral damage vs. long-term evolutionary impact. Harmonizing the tensions that currently impinge on biological control can only be achieved through improving the quality of information provided, which will help regulators make decisions that are based more on knowledge and less on the often ill-perceived fears of the public.     

Biology and preliminary host range assessment of two potential kudzu biological control agents

Two insect species from China, Gonioctena tredecimmaculata (Jacoby) (Coleoptera: Chrysomelidae) and Ornatalcides (Mesalcidodes) trifidus (Pascoe) (Coleoptera: Curculionidae), were studied in quarantine in the United States as potential biological control agents for kudzu, Pueraria montana variety lobata (Willd.) Maesen and S. Almeida. Adults of G. tredecimmaculata were ovoviviparous and reproduced throughout the summer, producing offspring that had an obligate adult diapause. In no-choice tests, adult and larval G. tredecimmaculata rejected most of the plant species tested, but consumed foliage and completed their life cycle on soybean (Glycine max L. Merr.) and on a native woodland plant, hog-peanut (Amphicarpaea bracteata L. Fernald), which are in the same subtribe as kudzu (Glycininae). Insects showed similar responses to field- and greenhouse-grown soybean and kudzu foliage, despite measurable differences in leaf traits: field-grown foliage of both plants had greater leaf toughness, higher total carbon content, higher trichome density, and lower water content than greenhouse foliage. O. trifidus adults also rejected most of the plants tested but fed on and severely damaged potted soybean and hog-peanut plants in addition to kudzu. Further tests in China are needed to determine whether these species will accept nontarget host plants under open-field conditions.     

Prospective Biological Control Agents of Varroa destructor n. sp., an Important Pest of the European Honeybee, Apis mellifera

This paper reviews prospective biological control agents of the varroa mite, Varroa destructor n. sp. (Acari, Mesostigmata). This ectoparasite has caused severe damage to populations of the European honeybee, Apis mellifera , world-wide in recent years. To date, no promising natural enemies of varroa species have been identified on A. mellifera or its original host, Apis cerana . Therefore, biological control will probably require natural enemies from other hosts. The following groups of organisms were reviewed as potential biological control agents: predatory mites, parasitoids and entomopathogens (nematodes, protozoa, viruses, Bacillus thuringiensis , rickettsiae, and fungi). The candidate groups were ranked according to their lethality to Acari, likely ability to operate under the physical conditions of honeybee colonies, ease of targeting, and ease of mass-production. Preferential consideration was given to the natural enemies of Acari that occupy taxonomic groups close to varroa. Entomopathogenic fungi, which kill a wide range of acarine species, were identified as prime candidates for screening against varroa. Bacillus thuringiensi s also requires study, particularly strains producing novel toxins active against non-insect hosts. Entomopathogenic protozoa and nematodes show less potential for varroa control, but nonetheless warrant preliminary investigation. We consider predators, parasitoids, viruses and rickettsiae to have little potential to control varroa. Because the physical conditions within honeybee colonies are similar everywhere, it is very likely that a biological control agent of varroa could be used successfully throughout the world.     

Suitability of two root-mining weevils for the biological control of scentless chamomile, Tripleurospermum perforatum, with special regard to potential non-target effects

The biology and host range of the two root-mining weevils Diplapion confluens Kirby and Coryssomerus capucinus (Beck), two potential agents for the biological control of scentless chamomile Tripleurospermum perforatum (Mérat) Laínz, were studied in the field in southern Germany and eastern Austria, and in a common garden and under laboratory conditions in Delémont, Switzerland from 1993 to 1999. Both weevils were univoltine, and females started to lay eggs in early spring. Diplapion confluens had three and C. capucinus five instars. Larvae of both species were found in the field from mid-April until the end of July; later instars preferentially fed in the vascular cylinder of the shoot base, root crown or root. Although larvae of both species occupy the same temporal and spatial niche within their host plants, they occurred at all investigated field sites together, and showed a similar distribution within sites. No negative or positive interspecific association was detected. Host-specificity tests including no-choice, single-choice, and multiple-choice tests under confined conditions, as well as tests under field conditions with natural and augmented insect densities revealed that both herbivores were specific to plant species in the tribe Anthemideae. However, their development to mature larva or adult on several cultivated plants, as well as on one plant species native to North America, rendered them unsuitable for field release in North America. It was concluded that to investigate non-target effects reliably, host-specificity tests with biological control agents should be carried out under a variety of conditions, particularly with augmented insect densities, as are expected to occur naturally after release.     

Combined use of the larvo‐pupal parasitoids Diachasmimorpha longicaudata and Aganaspis daci for biological control of the medfly

In biological control programmes, it is very common to employ multiple species to manage a single insect pest. However, the beneficial effects of natural enemies are not always additive because of several factors, including interspecific competition between these biocontrol agents. For this reason, in the present study we assessed several biological parameters (percentage parasitism, fertility, induced mortality and population reduction) of the parasitoids Diachasmimorpha longicaudata and Aganaspis daci when used together against the medfly Ceratitis capitata under laboratory and greenhouse conditions. Our results showed that, under laboratory conditions, fertility and percentage parasitism corresponded to a different functional response for each species (D. longicaudata: type II; A. daci: type III), whilst under greenhouse conditions, and unlike what occurs with single releases, both parasitoids showed a type III functional response; this is the only response which may lead to direct density dependence when host densities are low. Our results also revealed that when both species acted together, they produced a very high total percentage parasitism compared to that reported for single releases under both laboratory (64–76%) and greenhouse (21–51%) conditions. The parasitism was also higher for A. daci except when medfly larvae were provided in an artificial diet. Furthermore, host mortality induced by the two parasitoids acting together was very high, especially at low‐host densities; medfly population was almost completely reduced under greenhouse conditions. In summary, the data reported here supports the combined use of these species in biological control programmes against the medfly and highlights the importance of several factors, such as climatic conditions and host density, when planning their field releases.     

Indirect nontarget effects of host-specific biological control agents: Implications for biological control

Classical biological control of weeds currently operates under the assumption that biological control agents are safe (i.e., low risk) if they do not directly attack nontarget species. However, recent studies indicate that even highly host-specific biological control agents can impact nontarget species through indirect effects. This finding has profound implications for biological control. To better understand the causes of these interactions and their implications, we evaluate recent case studies of indirect nontarget effects of biological control agents in the context of theoretical work in community ecology. We find that although particular indirect nontarget effects are extremely difficult to predict, all indirect nontarget effects of host specific biological control agents derive from the nature and strength of the interaction between the biological control agent and the pest. Additionally, recent theoretical work suggests that the degree of impact of a biological control agent on nontarget species is proportional to the agent’s abundance, which will be highest for moderately successful control agents. Therefore, the key to safeguarding against indirect nontarget effects of host-specific biological control agents is to ensure the biological control agents are not only host specific, but also efficacious. Biological control agents that greatly reduce their target species while remaining host-specific will reduce their own populations through density-dependent feedbacks that minimize risks to nontarget species.     

Indirect nontarget effects of host-specific biological control agents: Implications for biological control

Classical biological control of weeds currently operates under the assumption that biological control agents are safe (i.e., low risk) if they do not directly attack nontarget species. However, recent studies indicate that even highly host-specific biological control agents can impact nontarget species through indirect effects. This finding has profound implications for biological control. To better understand the causes of these interactions and their implications, we evaluate recent case studies of indirect nontarget effects of biological control agents in the context of theoretical work in community ecology. We find that although particular indirect nontarget effects are extremely difficult to predict, all indirect nontarget effects of host specific biological control agents derive from the nature and strength of the interaction between the biological control agent and the pest. Additionally, recent theoretical work suggests that the degree of impact of a biological control agent on nontarget species is proportional to the agent’s abundance, which will be highest for moderately successful control agents. Therefore, the key to safeguarding against indirect nontarget effects of host-specific biological control agents is to ensure the biological control agents are not only host specific, but also efficacious. Biological control agents that greatly reduce their target species while remaining host-specific will reduce their own populations through density-dependent feedbacks that minimize risks to nontarget species.     

Making host specificity testing more efficient: Exploring the use of abridged test plant lists

Testing the specificity of candidate agents is a key component of risk analysis in weed biological control. This step is often time-consuming due to the numerous plant species that need to be tested under quarantine conditions in the invaded country of the weed species. Here, we examined whether an abridged phylogenetically based test list could be used in the weed's native range to quickly screen the host specificity of candidate agents. Ten plant species were used to test the host specificity of a promising candidate for the biological control of Sonchus oleraceus in Australia, the gall midge, Cystiphora sonchi. No-choice and choice tests were carried out in the native Mediterranean range of the midge. The results showed the midge has potential to threaten native Australian species, as those species showed high infestation levels in no-choice tests and produced significantly higher numbers of galls in choice tests. As a result of this approach, C. sonchi was rapidly discarded from the list of agents to be imported into Australian quarantines for further tests. This study demonstrates that testing a few key phylogenetically related species in the native range may save cost and effort in a weed biological control programme.     

生物防治利用生物多样性保护生物多样性

本文论述了生物防治与保护生物多样性的关系,提出生物多样性是生防作用物的必要来源,生物防治是保护生物多样性的重要措施。文中分析了自然界和农田生态系统中天敌的多样性和寄主专一性,并从外来种的治理、濒危物种和栖境的保护等几个方面探讨了生物防治对于保护生物多样性的作用。作者还强调应加强国际间天敌资源的交换,建立严格的天敌引种释放法规,以便开展更多的安全有效的生物防治项目。     

In search of artificial domatia for predatory mites

Banker plants can enhance biological pest control by providing both floral resources and appropriate oviposition sites, e.g. through acarodomatia, to predator species. The use of materials mimicking domatia i.e. artificial domatia may be an economically favourable alternative to the use of banker plants bearing domatia. The aim of the present study was to identify materials that are able to host eggs of the Neoseiulus californicus predatory mite but not those of the Tetranychus urticae pest mite. In a laboratory experiment, the oviposition of predatory and phytophagous mites were compared in Petri dishes containing leaves. The different modalities compared were (i) natural domatia of Viburnum tinus or (ii) one of twelve potential artificial domatia materials. The overall oviposition response of predatory mites to all artificial domatia was similar to that of the natural domatia. The oviposition of the Tetranychus urticae pest mite did not increase in response to the artificial domatia. Five artificial domatia hosted as many eggs of the predatory mite as observed in the natural domatia. The effect of the physical properties of artificial domatia was also tested and N. californicus was found to favour the artificial domatia that had high heat retention capacities for oviposition. Three of these artificial domatia were tested on rose plants in a greenhouse experiment; none of which enhanced the biological control on the plants under these conditions. The present study highlights the difficulty in identifying and using suitable artificial domatia as substitutes to banker plants in biological pest control efforts.     

Identification of a universally primed-PCR-derived sequence-characterized amplified region marker for an antagonistic strain of Clonostachys rosea and development of a strain-specific PCR detection assay

We developed a PCR detection method that selectively recognizes a single biological control agent and demonstrated that universally primed PCR (UP-PCR) can identify strain-specific markers. Antagonistic strains of Clonostachys rosea (syn. Gliocladium roseum) were screened by UP-PCR, and a strain-specific marker was identified for strain GR5. No significant sequence homology was found between this marker and any other sequences in the databases. Southern blot analysis of the PCR product revealed that the marker represented a single-copy sequence specific for strain GR5. The marker was converted into a sequence-characterized amplified region (SCAR), and a specific PCR primer pair was designed. Eighty-two strains, isolated primarily from Danish soils, and 31 soil samples, originating from different localities, were tested, and this specificity was confirmed. Two strains responded to the SCAR primers under suboptimal PCR conditions, and the amplified sequences from these strains were similar, but not identical, to the GR5 marker. Soil assays in which total DNA was extracted from GR5-infested and noninoculated field soils showed that the SCAR primers could detect GR5 in a pool of mixed DNA and that no other soil microorganisms present contained sequences amplified by the primers. The assay developed will be useful for monitoring biological control agents released into natural field soil.     

Prioritisation of potential agents for the biological control of the invasive alien weed,Pereskia aculeata (Cactaceae), in South Africa

Pereskia aculeata Miller (Cactaceae) is an invasive alien species in South Africa that is native in Central and South America. In South Africa, P. aculeata outcompetes native plant species leading to a reduction in biodiversity at infested sites. Herbicidal and mechanical control of the plant is ineffective and unsustainable, so biological control is considered the only potential solution. Climatic matching and genotype matching indicated that the most appropriate regions in which to collect biological control agents were Santa Catarina and Rio de Janeiro provinces in Southern Brazil. Surveys throughout the native distribution resulted in 15 natural enemy species that were associated with the plant. Field host range data, as well as previous host plant records, were used to prioritise which of the species were most likely to be suitably host specific for release in South Africa. The mode of damage was used to determine which species were most likely to be damaging and effective if released. The most promising species prioritised for further study, including host specificity and impact studies, were the stem-wilter Catorhintha schaffneri Brailovsky & Garcia (Coreidae); the stem boring species Acanthodoxus machacalis Martins & Monné (Cerambycidae), Cryptorhynchus sp. (Curculionidae) and Maracayia chlorisalis (Walker) (Crambidae) and the fruit galler Asphondylia sp. (Cecidomyiidae). By prioritising the potential biological control agents that are most likely to be host-specific and damaging, the risk of conducting host specificity testing on unsuitable or ineffective biological control agents is reduced.     

Identification of a Universally Primed-PCR-Derived Sequence-Characterized Amplified Region Marker for an Antagonistic Strain of Clonostachys rosea and Development of a Strain-Specific PCR Detection Assay

We developed a PCR detection method that selectively recognizes a single biological control agent and demonstrated that universally primed PCR (UP-PCR) can identify strain-specific markers. Antagonistic strains of Clonostachys rosea (syn. Gliocladium roseum) were screened by UP-PCR, and a strain-specific marker was identified for strain GR5. No significant sequence homology was found between this marker and any other sequences in the databases. Southern blot analysis of the PCR product revealed that the marker represented a single-copy sequence specific for strain GR5. The marker was converted into a sequence-characterized amplified region (SCAR), and a specific PCR primer pair was designed. Eighty-two strains, isolated primarily from Danish soils, and 31 soil samples, originating from different localities, were tested, and this specificity was confirmed. Two strains responded to the SCAR primers under suboptimal PCR conditions, and the amplified sequences from these strains were similar, but not identical, to the GR5 marker. Soil assays in which total DNA was extracted from GR5-infested and noninoculated field soils showed that the SCAR primers could detect GR5 in a pool of mixed DNA and that no other soil microorganisms present contained sequences amplified by the primers. The assay developed will be useful for monitoring biological control agents released into natural field soil.     

Ecological role of control agent,and not just host‐specificity,determine risks of biological control

Biological control agents used to manage alien vegetation are generally viewed as providing an ecosystem service, owing to reduced ecological and economic costs of invasion following their release. In particular, gall‐formers are popular as biological control agents because they are host‐specific and therefore considered low risk. However, galls can also be considered to be ecological engineers, because they provide nutritional resources for native invertebrates. We tested whether native invertebrates had formed associations with the gall‐forming fungus Uromycladium tepperianum, introduced into South Africa to control the Australian invasive alien tree Acacia saligna, by collecting U. tepperianum galls and monitoring emergence. We found that a number of invertebrates had formed associations with the biological control agent, among which was the important citrus pest, Thaumatotibia leucotreta (false codling moth). We used pheromone‐baited traps to ascertain if this supplementary source of T. leucotreta increased their abundance in orchards close to patches of gall host, but did not find this to be the case. We did find, however, that control measures used by farmers explained T. leucotreta abundances in traps, which may have obscured detection of any effects of a nearby host for the pest. Nevertheless, this study illustrates the first case of a host‐specific classical biological control agent providing resources for an economically significant crop pest. We conclude that although biological control agents are strictly vetted to ensure host‐specificity, introduced biological control agents that become abundant and can act as ecological engineers pose risks when native biota form associations with them, resulting in a number of possible cascading ecosystem effects. In addition, there could be economic consequences when these associated species include agricultural pests. We conclude that not just host specificity, but potential ecological effects of biological control agents, should be considered in their selection.     

Environmental risk assessment of exotic natural enemies used in inundative biological control

In the past 100 years many exotic naturalenemies have been imported, mass reared andreleased as biological control agents. Negativeenvironmental effects of these releases haverarely been reported. The current popularity ofinundative biological control may, however,result in problems, as an increasing number ofactivities will be executed by persons nottrained in identification, evaluation andrelease of biological control agents.Therefore, a methodology for risk assessmenthas been developed within the EU-financedproject `Evaluating Environmental Risks ofBiological Control Introductions into Europe[ERBIC]' as a basis for regulation of importand release of exotic natural enemies used ininundative forms of biological control (i.e.not in `classical biological control' thoughsome of the same principles and approachesapply). This paper proposes a general frameworkof a risk assessment methodology for biologicalcontrol agents, integrating information on thepotential of an agent to establish, itsabilities to disperse, its host range, and itsdirect and indirect effects on non-targets. Ofthese parameters, estimating indirect effectson non-targets will be most difficult, asmyriads of indirect effects may occur whengeneralist natural enemies are introduced. Theparameter `host range' forms a central elementin the whole risk evaluation process, becauselack of host specificity might lead tounacceptable risk if the agent establishes anddisperses widely, whereas, in contrast, amonophagous biological control agent is notexpected to create serious risk even when itestablishes and disperses well. Drawing onpublished information and expert opinion, theproposed risk assessment methodology is appliedto a number of biological control agentscurrently in use. These illustrative casehistories indicate that the risk assessmentmethodology can discriminate between agents,with some species attaining low `risk indices'and others scoring moderate or high. Riskindices should, however, not be seen asabsolute values, but as indicators to which ajudgement can be connected by biologicalcontrol experts for granting permission torelease or not.     

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.     

Scientific advances in the analysis of direct risks of weed biological control agents to nontarget plants

Research on host specificity testing protocols over the last 10 years has been considerable. Traditional experimental designs have been refined and interpretation of the results is benefiting from an improved understanding of agent behavior. The strengths, weaknesses, and best practice for the different test types are now quite clearly understood. Understanding the concept of fundamental host range (the genetically determined limits to preference and performance) and using this to maximize reliability in predicting field host specificity following release (behavioral expression of the fundamental host range under particular conditions) are still inconsistently understood or adopted despite having been identified as the critical steps in analyzing the threats posed by biological control agents to the agriculture and biodiversity of novel environments. This needs to be consistently understood and applied so the process of testing can follow a recognized process of risk analysis from hazard identification (identifying life stages of the agent that pose a threat and defining their fundamental host range) to uncertainty analysis based on the magnitude (predicted field host specificity following release) and likelihood of threats (predicted actual damage and impact) to nontargets. Modern molecular techniques are answering questions associated with subspecific variation in biological control agents with respect to host use and the chance of host shifts of agents following release. Guidelines for assessment of nontarget impacts need to recognize and adopt such recent developments and emphasize a general increased understanding of the evolution of host choice and the phylogenetic constraints to shifts in host use. This review covers all these recent advances for the first time in one document, highlighting how inconsistent interpretation by biological control practitioners can be avoided.