Search BIREA:

View:   short pages · print version

Containment testing

Introduction

This section provides guidance on testing methods that can be used for weed and arthropod biological control agents in containment. The results of these tests are a vital component of the information that needs to be given in the application form for general or conditional release of new organisms, in this case, biological control agents.

Weed biocontrol agents

The objective of host specificity assessment in weed biocontrol is to quantify the ability of prospective biological control agents to attack non-target plant species after introduction (Marohasy 1998). Plant species are chosen for host range testing on the basis of their taxonomic relationship to the target species (Wapshere 1989, Harley and Forno 1992). By exposing prospective agents to progressively less related plant species (i.e. to different varieties of the target host, then to different species in the same genus, then to different genera in the same tribe, and so on) it is thought that all plant species that might support populations of the agent will be identified (Tallamy 1999). A more novel approach is to expose the agent to plants with similar signatures of secondary metabolic compounds, regardless of their taxonomic affiliation with the target species (Tallamy 1999).

If the majority of host specificity testing occurs in containment then the risk that laboratory studies will generate conservative results increases. In weed biocontrol, host specificity testing has to balance two potentially conflicting needs; avoiding the introduction of an agent that may cause unacceptable damage to a non-target plant but arguably, just as important is avoiding the rejection of a potentially effective agent unnecessarily.

Parasitoids, predators

Pre-release host specificity testing is carried out to assist in the prediction of the likely post-release field host range and impact of introduced biological control agents. Strictly monophagous parasitoids are rare, and in some taxonomic groups, non-existent. While it is often stated that a high degree of host specificity is desirable in the interests of minimising non-target effects, there is also a view (now less widely advocated) that the availability of alternate hosts can be important for the success of the biological control in the long-term (e.g. Nechols et al. 1992). The decision to be made by the EPA to allow any biological control release is a compromise based on incomplete knowledge, and must be the result of a risk/benefit analysis. The benefit is usually easy to determine in economic terms, but it is far more difficult to place a value on indigenous species, environmental integrity and biodiversity (Barratt et al. 1999).

Test species selection

It is important to choose an appropriate range of species for use in host range testing experiments. Some of the most important issues are discussed below.

A general problem that is gradually becoming much better understood is the existence of previously unrecognised biotypes. Either the weed, the agent or the natural enemy may actually consist of several distinct populations that vary biologically but are lumped under a single taxonomic name. For example, a parasitoid may only attack some of the populations of a host. Conversely, a parasitoid may exist as several distinct biotypes, each of which shows behavioural or physiological adaptation to different hosts.

Because of the far-reaching consequences of biotype specificity and effectiveness, host range testing should be done with the exact geographic population of parasitoid that will be released. This important issue needs to also underline all the considerations listed below.

Weed biocontrol agents

Evidence is available to support the view that that host range expansion by phytophagous insects is limited to plant species which have a similar composition of phytochemicals (Futuyma 2000). Such plants are usually also close relatives (Berenbaum and Zangerl 1992). One could conclude from this that biocontrol practitioners could eliminate from host range tests any plants with very different phytochemicals from the target host plant. However, there are rare examples in the literature of herbivorous insects that have expanded their host range to genetically and chemically quite different plants (Tallamy 1999). Further to this, the lack of information available to biological control practitioners often precludes this sort of reasoning. To date there is no evidence of evolutionary changes in the fundamental host range of weed biocontrol agents after release (van Klinken and Edwards 2002). This conclusion is backed up by the review of both weed and insect agent non-target impacts (Louda et al. 2003). A more recent review of data on host use by 117 biological control agents released against weeds in the continental USA, Hawaii and the Caribbean since 1902, (Pemberton 2000) found only 15 insect agents attacking 41 native non-target plant species. All but one of these plants was closely related to the target weed, and the potential for attack by the agents could have been predicted from adequate host range testing. Therefore according to Pemberton the majority of non-target impact examples would be placed into this category, and true unpredicted attack from weed biological control agents is relatively uncommon.

Selecting the test plant list

Wapshere (1974) set down some criteria for selecting plants for testing to demonstrate host specificity of biological control agents. Previous to this, a small group of related plants and a large series of unrelated crop plants had been selected for testing against the potential weed biocontrol agents (Zw�lfer and Harris 1971). The criteria outlined by Wapshere in 1974 have proven to form a successful foundation for at least three decades of host specificity testing around the world included the following:

As the understanding of host selection behaviour of phytophagous insects improved (Miller and Strickler 1984), Wapshere reconsidered his testing sequence, especially in light of the number of potential weed biocontrol agents being rejected on the basis of apparently greater host range being predicted from host specificity testing in a caged environment, than was observed in field situations (Wapshere 1989). The conclusion was that if any of the caged methods of host specificity testing (generally conducted within quarantine facilities with artificial lighting and climate controls) leads to the by-passing of an important cue within the insects host selection behavioural sequence, that the results may not be predictive. Wapshere considered that since by-passing of cues increased the potential host range of an insect, any plant not attacked or oviposited upon in a cage or after placing the larva directly onto it, could be considered safe from oviposition or feeding. Thus, if it is not possible to test all plants under natural conditions, the most reliable method of estimating host range and reducing the rejection of apparently safe agents was to test selected plants with the agent at the critical host selection phase, and then to test only those affected at the next less restrictive phase, and so on. What should be left after a sequence of such tests are only those few plants that require a field test or the expression of a complete normal sequence of host selection behaviour to indicate the risk to the plant. This was referred to as the "reverse testing sequence" (Wapshere 1989).

For access to current plant species in New Zealand for determining phylogenetic affiliations, the Landcare Research database [http://nzflora.landcareresearch.co.nz] can be used.

Arthropod biological agents

The number of insect species which could be considered for host range testing will always far out-number those for plants, and our knowledge of insect phylogeny is often far less comprehensive than that for plants. Indeed, in New Zealand many insect species remain undescribed, and undiscovered. This makes the adoption of the test species selection method for host specificity testing far much more difficult than it is for weed biological control agents. Furthermore, rearing insects in sufficient numbers for robust tests is often more challenging than for plants. Nevertheless, attempts to develop test insect selection methods have been made.

Selecting the test insect list

Some factors to be considered in selecting test species for host specificity testing have been discussed elsewhere in detail (Goldson and Phillips 1990, Goldson et al. 1992, Barratt et al. 2000). Depending upon the nature of the insect pest that is the proposed target of biological control, sometimes significant amounts of preliminary information can be gleaned on a proposed biological control agent from various sources in the public arena (Hoddle 2004, Sands and Van Driesche 2004). Many potential pitfalls such as taxonomic synonymy and misidentifications can be avoided if awareness of the possibilities is high (Sands and Van Driesche 2004). One process that could be followed in obtaining a suitable host testing list is (Barratt et al. 1999):

Another approach similar to that described above has been published recently where the list compiled using phylogenetic and ecological criteria is filtered by eliminating species which have different spatial, temporal and morphological characteristics which do not coincide with the target species (Kuhlmann et al. 2006).

Host range testing methods

There are several important factors to consider when designing experiments to assess the host range of a prospective biological agent. These are discussed below.

Test designs for weed biocontrol

As mentioned above, the assessment of plants as potential hosts for herbivorous insects began over 70 years ago. A variety of tests have been developed, although those most in favour have changed over time due to changes in the perspective of biocontrol practitioners and developments in the study of insect behaviour. Tests used have focused on oviposition, adult feeding, larval development and survival, oogenesis and multi-generation population persistence, and host preference. There have been a number of thorough reviews that describe the different test method types and discuss and detail the appropriateness of each test type for different types of weed biocontrol agents (Harris and Zw�lfer 1968, Cullen 1989, Wapshere 1989, Heard and Van Klinken 1998, Marohasy 1998, Briese 1999, Sheppard 1999, Heard 2000, van Driesche and Murray 2004). Each of these papers makes a valuable contribution to the appropriateness of different types of test design. For this reason a thorough debate of the pros and cons of various testing designs according to the biology and behaviour of the individual weed biocontrol agent is beyond the scope of this website. Instead, bullet points will be used to summarise main points below and for more details you should consult the above papers.

What we do need to emphasise in applications to the EPA is that the applicant is open and clear in justifying both the way they arrived at the test species selection list as well as the reasoning for approaching the testing methodology in terms of the host selection behaviour and biology of the potential agent. Wapshere (1989) provides a very clear and logical explanation of how both could be approached, and applicants are urged to read that paper.

The possible agents that may be chosen as potential weed biocontrol agents is very large, so there are no strict guidelines on how to undertake testing. Careful consideration of the behavioural characteristics of both the:

The importance of testing the host specificity of both life stages separately is that the dispersive life stage has the ability to choose the plant species that will be contacted by the other life stages. For instance, in many cases it is the adult female insect that is mobile and actively undertakes host searching and host choice for oviposition. Often internally feeding or virtually immobile larvae do not have the ability to move far enough to choose a different, perhaps more palatable species of plant. In such cases, host specificity testing that establishes both the host preference of the adult female, and then the survival and feeding ability of the larva, will capture all the vital information required to make an accurate risk assessment.

There are examples where the reproducing female may be immobile on a plant part, yet produces highly dispersive crawlers or ballooning larvae that themselves move according to weather conditions such as wind and the proximity of other plant species will have a significant influence on what plants they contact. In such a case, host specificity testing that establishes the preference of the neonate larvae to settle on the different plants, and then the survival and reproductive potential of the resultant individuals, will capture all the information required to make an accurate risk assessment.

In other cases the reproductive female may be only loosely selective in choosing a certain habitat in which to oviposit, and the larvae may be highly mobile with the ability to locate different individuals and species of plants through some or all of their damaging life stages. In such a case, host specificity testing will need to both establish the feeding preferences of the mobile life stages of the larvae between different plants, and the oviposition preferences of the adults, as well as the survival and reproductive potential of individuals on individual plant species.

A flow chart generalising decisions involved in the selection of an initial host specificity testing procedure for weed biocontrol agents has been presented by Sheppard (1999). However, Withers (1997) strongly insists that the results of initial screening trials undertaken should not be used to reduce the plant list used for testing in subsequent trial types. This is because host specificity testing on both the dispersive and damaging life stages need to be gathered independently for a quantitative risk analysis to be undertaken.

Test designs for arthropod biological control agents

Similarly to weed biocontrol, there are some extremely valuable publications describing the design of host range tests for parasitoids and predators (van Driesche and Murray 2004). A common generalisation is that estimation of the host range of entomophagous biocontrol agents (parasitoids and predators) is more complex than for phytophagous weed biological control agents. This is primarily because there is an additional trophic layer involved and often an intimate and specific relationship between the target and test organisms and their substrate (usually their food plant). An important consequence of this intimate and specific relationship between the host or prey of entomophagous agents and their substrate is that prior experience of the substrate can affect the organism's responsiveness to cues from this and other substrates.

A second complicating factor for endoparasitoids is that it is not possible, in most cases, to inoculate all test organisms with eggs or neonates to determine "suitability", though exceptions do exist (Morehead and Feener 2000, Fuester et al. 2001). Thus, a program to determine the host range of parasitoids is denied one of the most powerful tools (the so-called physiological host range test on larvae) that is used in determination of host range of phytophagous agents (Hill 1999, van Klinken 2000). This means that, in the host range testing of parasitoids, it is important to employ test procedures that will maximize the probability that the test species will be accepted for oviposition in the first place. This is vital for an accurate risk assessment (Withers and Barton-Browne 2004). It is possible in some cases, however, to distinguish between behavioural host acceptance and physiological compatibility of parasitoids and test species. For example, McNeill et al. (2000) used a pathogen on the ovipositor of parasitoids as an indicator of an oviposition attempt in a test species. Parallel exposure of untreated parasitoids to the test insects gave an estimate of physiological compatibility.

Failing a high tech method such as this, only the careful dissection of all exposed hosts to identify parasitoid ovi-or larvi-position will reliably ascertain whether a failure to develop a parasitoid was the result of a physiological host immune response or a lack of oviposition. This is a laborious and time-consuming exercise but may be necessary, along with carefully chosen controls, to prove that a suitable testing method has been utilised.

Replication

There is no perfect number to choose for how many replicates of a test to undertake. Statistical power analyses exist that can more accurately predict the number of replicates that need to be undertaken to generate data with certain probabilities of type I or type II errors. This calculation is based on the variance from the mean obtained in an a priori trial run or the mortality obtained in tests a posteriori (Hoffmeister 2005). It is possible that variance in biological parameters such as survival and development may differ according to the suitability of the host, and if so, different levels of replication may be required for different test plants.

It is important to avoid pseudo-replication. This can come about in host specificity testing if a shortage of test species results in long-lived individuals being utilised repeatedly in different tests. Pseudo-replication can also occur if the statistical testing is undertaken upon individual insects when in fact the data were gathered from insects caged together as a group (Hoffmeister et al. 2006). Hence, the samples are not independent. As a general rule of thumb true replication is achieved at the 'cage' level within which each individual test was run. Hoffmeister et al. (2006) goes as far as to say that it is essential that insects for the tests on non-target species do not come from one rearing container while control individuals come from another. Also it is essential that non-target hosts are not always tested in the same container or field cage or on the same plant, while target hosts are tested in another cage or on another plant. Equally important is that the location of experimental units (cages) within a chamber or laboratory, need to be randomised to avoid confounding effects of differences in temperature and light conditions etc. Similarly, tests with control and test insects should be carried out in parallel and not at different times. Randomisation of testing order or random assignment to plants or test cages assures that pseudo-replication can be avoided (Hoffmeister 2005). When writing applications to the EPA, it is unwise to expect the reviewers to assume that experiments have been carried out according to 'best practice', rather it is preferable to mention in detail that such steps were taken in the methods section of the host range testing reports.

Controls

Positive controls are essential for oviposition and feeding trials, to validate negative responses by showing that the group from which the test biocontrol agents were drawn, had the capacity for oviposition or feeding. Individuals used in controls should either be:

Negative controls, in which test species are not exposed to parasitoids, or plants not exposed to weed agents, are needed to detect background mortality of test species that is unrelated to the biocontrol agent. For feeding tests with predators, controls in which only water is provided show whether the test prey species provides any nutritional benefit to the predator by assessing survival time with prey versus water alone. For tests measuring survival and development, performance on a host for survival and development of known suitability or an artificial diet could be used as a standard against which reduced survival in a poorer host can be gauged (van Driesche and Murray 2004).

Rigorous positive controls are absolutely critical for the reliable host range testing of parasitoids and predators, particularly when no-choice tests are used. Parasitoids can differ in the way they mature their eggs prior to oviposition. This can be expressed according to the "ovigeny index" (Withers and Barton-Browne 2004). The main implication of the ovigeny index on host range testing methods is for those species that emerge with no or very few mature eggs. There is a risk that young females may not oviposit in any host because of a lack of mature eggs. Futhermore, even young females carrying some mature eggs might not oviposit in lower ranked hosts because of a low egg load. Positive controls needs to be females from the same rearing group exposed to the target pest to confirm egg laying ability of the parasitoid cohort used.

No-choice tests

The no-choice test design was the first approach used in early weed biocontrol projects (van Driesche and Murray 2004). No-choice tests can be used with adult insects to assess feeding, survival to maturity and the willingness to oviposit. With larvae no-choice tests measure the ability of a plant to support the development of the immature life stages. The accuracy of the data obtained from no-choice tests obviously relies upon providing in non-limiting quantities the required part of the plant, whether it be bud, leaf, flower, seed pod etc, in optimum physiological condition. For species with larvae that do not move between plants, use of the no-choice test is ideal because the only choice available to the larva in the field would also be to either feed or die. The strength of no-choice tests is that negative results are very robust and provide convincing evidence that a test species is not likely to be used as a field host, provided of course that the experimental design includes an environment that permits normal behaviour of the biological control agent, as evidenced by a positive response (or excellent larval survival) to the normal host that is provided as a control. Use of no-choice tests early in the testing sequence provides a strong rationale for classifying un-attacked test species as non-hosts.

Unfortunately no-choice development tests sometimes indicate that herbivorous insect larvae can feed successfully on plant species that adult insects did not find or accept for oviposition. Sometimes this is as a result of time-dependent changes in responsiveness, that can be viewed as a high motivation to feed or oviposit when starvation or death is imminent (Barton-Browne and Withers 2002). Also positive responses within no-choice tests can sometimes be argued to have been artificially induced by confinement. The cage environment may bring the agent into close contact with a test species such that important host finding steps may be skipped (since the insect is literally put on the host or very near to it), allowing oviposition or adult feeding to occur on test species that might not have been approached or utilised in the field. This issue can be particularly important with parasitoids (van Driesche and Murray 2004).

So often no-choice tests alone cannot be used for risk assessment without incorporating other assay methods, or potentially safe and effective agents may be rejected unnecessarily Hill (1999). Often when used on phytophagous larvae to test development, no-choice tests are becoming known as the method for ascertaining the "physiological host range" or "fundamental host range" of an insect (van Klinken 2000). Using this terminology reminds the reader or reviewer that the risk assessment is not complete based on no-choice data alone, and data incorporating other testing methods will also be required to reliably predict the "field host range" (van Klinken 2000). Undoubtedly the most important benefit of no-choice tests is the tendency for them to reveal lower-ranked (lower preference or lower palatability) hosts whose status within the physiological host range can be missed if too much reliance is given to the result of choice tests (Withers 1999).

Often survival from immature larva to adult is not an absolute measure obtained from no-choice tests. Instead more careful monitoring is often conducted to obtain survivorship curves on the range of test plant species. When some survival to pupation or adulthood does occur, a single measure of percentage survival to adult of the cohort is often insufficient to enable accurate interpretation of the data. Some insects feeding on sub-optimal hosts conserve body mass at the expense of time taken to reach full size, others conserve development time at the expense of final body size (Barton-Browne 1995). The repercussions for these two strategies can be significant, so where a degree of physiological suitability for development is revealed for an insect feeding on a non-target plant in a no-choice test, both time taken to develop, as well as final body size, recorded in relation to gender, needs to be recorded and analysed appropriately to aid interpretation of the data.

Positive controls are essential to validate negative responses by showing that the group from which the test agents were taken had the capacity for oviposition or feeding. In parasitoids, negative controls are also required, in which test species are not exposed to parasitoids. These are held to detect mortality of test herbivores that is unrelated to the direct action of parasitism (van Driesche and Murray 2004).

Hill (1999) argued that the size and mobility of the agent should influence the choice of test design. Where agents are large, and mobile females can choose to move freely between plants with relative ease, and with reasonable frequency, choice tests may be more appropriate than no-choice tests. In this situation females can select or reject a host, and respond by arresting movement or dispersing once more. No-choice tests may be a more appropriate test for target species that are small relative to their host and disperse passively, or conversely are relatively immobile. In this case the choices available to an individual are to feed or to not to feed, a scenario that can be effectively tested by no-choice tests.

Choice tests

In choice tests, two or more plant or host species are presented to the biocontrol agent simultaneously, and the comparable response is a measure of preference between the two options (. The target species is often, but not always, one of the choices offered. This method is most commonly used to compare oviposition preferences but can also be used for comparing feeding preferences of adults and sometimes of larvae that are sufficiently mobile to move between hosts.

Choice tests rose to a height of popularity in the 70's and 80's because practitioners felt that the no-choice test (especially larval starvation tests) were resulting in too many cases in which a promising agent was rejected when a wider than expected physiological host range was being predicted in the laboratory. Hence there was a time when testing for weed agents relied on the results of choice tests alone. Sometimes choice tests were only conducted on those species that had revealed positive results in no-choice tests, apparently in the mistaken belief that subsequent lack of attack in choice tests would identify which of the positive responses in the no-choice data set were "erroneous" (Fornasari et al. 1991). This line of reasoning has since been shown to be inappropriate as research has now revealed many cases where choice tests failed to reveal a lower but significant host association (Marohasy 1998).

Choice tests have often been used to compare oviposition responses in weed biocontrol agents, parasitoids and predators. In some cases the polyphagy of a species is clearly revealed when all non-target species are attacked at rates similar to the known field host. However, in choice tests, sometimes agents will show strong oviposition preferences towards the target pest or weed, or even fail to attack non-target hosts at all. From such results it is tempting to conclude that the agent is host specific, whereas a more accurate interpretation may be that the preference ranking (or hierarchy) of the agent is that the target host sits most highly ranked out of those test species presented to it.

The greatest concern with choice tests is that they have clearly been shown (Withers et al. 2000, Barton-Browne and Withers 2002) to carry an unsatisfactorily high risk that there will be no attack on less stimulating hosts (less preferred or lower ranked) in choice tests that are run for too short a time or when the insect is introduced to the test in an un-stimulated (such as satiated in relation to food, or not ready to oviposit in relation to reproductive) state. Consequently, such tests can fail to reveal the fundamental host range, and there is a risk of a false negative result. On this basis Barton-Browne and Withers (2002) strongly recommended that choice tests should never be used as the only type of test for assessing risk to non-target hosts, and that they should not be used for a 'screening' phase of host specificity testing programs, if the intention is the use the results to reduce the host list for further testing.

Another way of expressing this, is that for the results of a choice oviposition test to be predictive of field events, (1) the agent must experience the choice in the field � that is, the non-target (lower ranked) host must not be the only possible host encountered, or (2) the non-target (low ranked) host must be so non-preferred that even agents deprived of their preferred hosts for considerable periods will keep searching rather than attack the low ranked species. Since these conditions may not always be met, inferring that a species not attacked in a choice oviposition test is not a host will lead to some unpredicted impacts (van Driesche and Murray 2004). Despite these concerns choice tests remain valuable, and in some cases, absolutely crucial for helping to assess the likely risk of non-target attack in the field, when we are reliant on laboratory testing methods alone.

Sequential testing

A variation on the no-choice and choice test design is called the sequential test, in which the natural enemy is exposed to a series of test species, one at a time. These were called "sequential no-choice tests" by (Withers and Mansfield 2005, but "sequential choice tests" by van Driesche and Murray (2004). The sequence in which species are presented in such tests can vary. For example, the presumably high ranking target species can be presented before or after the presentation of one or a sequence of non-target species, or both. Another possibility is that insects are given access to the target species between each test with a different non-target species. Furthermore the duration of the experimental regime, times during presentations and between them, can vary (Barton-Browne and Withers 2002). All these factors can potentially have a significant effect on the outcome of the test.

A theoretical analysis of the potential outcomes of some sequential no-choice test types (Barton-Browne and Withers 2002) concluded that the outcome of the test varied according to the period of time for which the insects were given no-choice access, particularly access to host B. For instance, if the parasitoids oviposited during the first access to the target host (which was often the aim � to ensure the parasitoid was physiologically and behaviourally ready to oviposit), it may not accept the non-target host when it was first presented with it (note the generality of this outcome will depend upon the species reproductive biology and behaviour, and what latency period occurs between ovipositions). Generally we can conclude that the probability the parasitoid will accept the non-target host during the test will depend on whether the test was run for sufficient length of time for time-dependent processes (Barton-Browne and Withers 2002) to act upon the parasitoid to lower its acceptance threshold to a level whereby the non-target host stimulated attack behaviour (Withers and Mansfield 2005).

The general conclusion around sequential no-choice tests is that they should only be utilised cautiously when the physiology and behaviour of the parasitoid or weed biocontrol agent is very well documented and understood in terms of its temporal pattern of oviposition. Otherwise the validity of the results will be questionable (van Driesche and Murray 2004).

A complicated test design for phytophagous insects was used to establish the rank order (or preference rank) of plant species for oviposition in butterflies (Wiklund 1981). The insects were confined in a large field cage and each day the plant array was altered to ascertain the preference ranking. A similar design has been demonstrated where tethered lepidopterans have been successfully manipulated to ascertain the length of their discrimination phase with regards to different host species (Singer 1982, Jallow and Zalucki 1996) for quantifying the host preferences. However such experiments are relatively technically complicated and not all insects are suited to this type of experimental manipulation or to longer running assays. Marohasy (1998) suggests a similar method of sequential choice test where the plant species that receives the most eggs progressively is removed until there is not oviposition on any plant species. This has been referred to as the Preference Ranking Test (van Driesche and Murray 2004). This type of method is designed to overcome the problem of false negative results owing to unresponsiveness to lower ranked plants species in choice tests. This method has been trialled with some success (Solarz and Newman 1996, Withers et al. 2000) but whether such complicated sequential designs to reveal host preference rankings are worth the effort and can be safely interpreted will depend upon the circumstances and biology of each agent.

Parameters that can be measured

This section is designed to provide ideas of the range of parameters that might be included within an EPA application to summarise the results of host range testing. In reality an in depth discussion of each parameter is not possible here, as the biology of each target and agent organism will need to dictate the appropriate methods used.

Weed biocontrol

  • Oviposition � number of eggs laid
  • Qualitative severity of damage from adult feeding
  • Percentage adult survival when fed only that species
  • Larval development rate fed only that species
  • Larval final body size/ pupal weight fed only that species
  • Sex ratio resulting from larval feeding on only that species
  • Percentage larval survival to pupa or adult fed only that species
  • Oogenesis after larva fed only that species
  • Multi-generation population persistence fed only that species
  • Latency to oviposit or initiate feeding
  • Proportion of population accepting species for oviposition
  • Proportion of population accepting species for feeding
  • Proportion of population located on that plant over time
  • Comparative levels of plant damage The most useful parameters to be measured in any host range test of a weed biocontrol agent is necessarily going to be dictated by the biology and behaviour of the agent. Many of the research papers referred to in this website can assist with designing the most appropriate test and therefore the parameters to be measured. We therefore urge applicants to be familiar with the relevant literature and utilise any specific methods that have been shown to be relevant to biology and behaviour of their proposed biocontrol agent.

    Insect biocontrol (van Driesche and Murray 2004)

    There are almost infinite numbers of parameters that could be measured in host range tests. With parasitoids and predators some common ones are included in the above list. A subset of these are discussed in detail in van Driesche and Murray (2004). The applicant is encouraged to consult this resource as well as being familiar with any behavioural or ecological research on the target or closely related insects. Behavioural research papers can provide invaluable information on the host searching and oviposition or feeding behaviour of the agent. The most appropriate parameters to record can then be identified for each of the host range tests undertaken.

    Statistical analysis

    Most of the traits to be analysed in biocontrol do not follow a Gaussian ('Normal') distribution, and thus standard t-tests, analysis of variance (ANOVA) or regression analyses cannot be used with validity to statistically test the effect of a treatment. All these common 'classical' methods assume that the distribution of residuals around the fitted model (i.e., the error distribution) is normal. Thus data need to be appropriately transformed (Fernandez 1992) to first achieve a Gaussian distribution or preferably different statistical approaches have to be used (Hoffmeister 2005).

    Non-parametric tests like Mann-Whitney U tests or Kruskal-Wallis tests are often the most appropriate test for host specificity testing data, and are more than adequate if the results are clear-cut. However non-parametric tests lack statistical power, which can mean that real, but more minor, differences existing between, for example, test species and target (control) species will tend to be interpreted as not significantly different by a non-parametric test, when in fact this might not be the case. Non-parametric tests will almost certainly have insufficient power to analyse data if sample sizes were forced to be small due to other factors out of the researchers' control. Generalized Linear Models can be used to predict responses both for dependent variables that are not normally distributed and for dependent variables which are nonlinearly related to the predictors (Hoffmeister 2005). Expert statistical advice will probably be required to both appropriately utilise and interpret such models.

    Choice tests are often invaluable for undertaking tests on the preference hierarchies shown by ovipositing or feeding insects. However choice tests are problematic for statistical analyses (Mansfield and Mills 2004, Withers and Mansfield 2005). The same individual insect is confronted with both target and non-target species, and we usually wish to obtain comparative acceptance rates for both target and non-target species as a result of such a test. Thus, a repeated measure design must be used. This renders the data dependent (not independent as required for Gaussian statistics, see above). The acceptance of non-target hosts or prey within such a test may well depend upon the frequency of target and non-target hosts within the cage. Therefore every target that is eaten or accepted and that is not immediately replaced alters the experimental conditions of the experiment, and the acceptance of any given host or prey may depend on the current ratio of availability of alternative hosts or prey. Thus if exploited hosts or prey cannot be replaced immediately, simultaneous choice tests may become almost impossible to interpret (Hoffmeister 2005).

    An alternative to simultaneous choice tests are sequential no-choice tests (Singer 1986). This design does have appropriate statistical analysis methods available to treat and analyse the data (Hoffmeister 2005). However sequential no-choice tests generate a multitude of potential behavioural complications which may be almost impossible to control for or adequately measure (Barton-Browne and Withers 2002, Withers and Barton-Browne 2004, Withers and Mansfield 2005).

    Overseas field surveys or field tests

    The most significant non-containment research that can assist with a biological control application involves research on the host range of the proposed agent in the country of origin, and open field or large cage testing in another country where the proposed agent is either endemic or established. Open-field tests obtain host range data on agents under more natural conditions than those obtained via cage tests. The results are more reliable and more likely to reflect that which will occur should the agent be introduced to New Zealand. It has been argued that such field tests are necessary or highly desirable to ascertain the potential risk to those probably closely related plants whose status as a host cannot be discounted even after the "reverse testing sequence" method has been followed in the laboratory (Wapshere 1989).

    Nonetheless, open-field tests have been criticised on several grounds. It has been argued that agent densities in such situations may be much lower than potential populations in the country of introduction and any changes in host choice induced by conditions of extreme intra-specific competition and/or starvation would not be replicated (McFadyen 1998). This is a valid criticism considering that temporary spill-over feeding as a result of large population densities and/or rapid defoliation of target weeds has been responsible for a significant number of reported non-target attacks in weed biocontrol (Louda et al. 2003). Such conditions would virtually never be recreated either in the country of origin or in a field test.

    So for the EPA to consider data reliable open field tests would have to be considered sufficiently robust including evidence that:

    Briese (1999) has reviewed the potential types of experimental design that can be used in field tests, including when augmentation of background populations are made to enhance the robustness of the data, and identifies a number of issues critical to the success of open field tests. Furthermore he has suggested a new design of open-field test called the "two-phase set design" that will encourage spill-over feeding in a field test, if this is ever likely to occur as a result of the rapid defoliation or death of the target weed in the field (Briese 1999).

    In conclusion, open field tests have not become widely used, in part because they must be done in the country of origin of the agent, in part because of the potential quarantine problems of moving test organisms to that region, and in part because they are often seen as a final step, not a first step (due to cost). Applying open field tests to insect biocontrol agents is even less feasible because of quarantine concerns with the test insect species. It is difficult to conceive of a separate country existing where all the same relevant non-target insects are also pre-existing. Conceivably, simulated open field tests could be constructed using walk-in cages within naturally-lighted, quarantine glasshouses, which would provide large enough spaces for natural insect behaviours to be displayed (van Driesche and Murray 2004).

    References

    Barratt B.I.P., Ferguson C.M., McNeill M.R. and Goldson S.L. (1999). Parasitoid host specificity testing to predict host range. Pp. 70-83 In: Host specificity testing in Australasia: towards improved assays for biological control, T.M. Withers, L. Barton-Browne and J.N. Stanley (Ed.) CRC for Tropical Pest Management, Brisbane, Australia.

    Barratt B.I.P., Goldson S.L., Ferguson C.M., Phillips C.B. and Hannah D.J. (2000). Predicting the risk from biological control agent introductions: A New Zealand approach. Pp. 59-75 In: Nontarget effects of biological control introductions, P.A. Follett and J.J. Duan (Ed.) Kluwer Academic Publishers, Norwell, Massachusetts, USA.

    Barton-Browne L. (1995). Ontogenetic changes in feeding behavior. Pp. 307-342 In: Regulatory Mechanisms in Insect Feeding, R.F. Chapman and G. de Boer (Ed.) Chapman and Hall.

    Barton-Browne L. and Withers T.M. (2002). Time-dependent changes in the host-acceptance threshold of insects: implications for host specificity testing of candidate biological control agents. Biocontrol Science and Technology 12: 677-693.

    Berenbaum M.R. and Zangerl A.R. (1992). Genetics of physiological and behavioral resistance to host furanocoumarins in the parsnip webworm. Evolution 46: 1373-1384.

    Briese D.T. (1999). Open field host-specificity tests: is "natural" good enough for risk assessment? Pp. 44-59 In: Host specificity testing in Australasia: towards improved assays for biological control, T.M. Withers, L. Barton-Browne and J. Stanley (Ed.) Scientific Publishing, Department of Natural Resources, Brisbane.

    Cullen J.M. (1989). Current problems in host-specificity screening. Pp. 27-36 In: Proceedings of the VII International Symposium on Biological Control of Weeds, E.S. Delfosse (Ed.) CSIRO Publications, Melbourne.

    Fernandez G.C.J. (1992). Residual analysis and data transformation: Important tools in statistical analysis. Horticultural Science 27: 297-300.

    Fornasari L., Turner C.E. and Andres L.A. (1991). Eustenopus villosus (Coleoptera: Curculionidae) for biological control of yellow starthistle (Asteraceae: cardueae) in north America. Environmental Entomology 20: 1187-1194.

    Fuester R.W., Kenis M., Swan K.S., Kingsley P.C., L�pez-Vaamonde C. and H�rard F. (2001). Host Range of Aphantorhaphopsis samarensis (Diptera: Tachinidae), a larval parasite of the gypsy moth (Lepidoptera: Lymantriidae). Environmental Entomology 25: 332-340.

    Futuyma D.J. (2000). Potential evolution of host range in herbivorous insects. Pp. 42-53 In: Host-specificity testing of exotic arthropod biological control agents: the biological basis for improvement in safety, R.G. Van Driesche, T. Heard, A.S. McClay and R. Reardon (Ed.) USDA Forest Service Bulletin, Morgantown, West Virginia, USA.

    Goldson S.L. and Phillips C.B. (1990). Biological control in pasture and lucerne and the requirements for futher responsible introduction of entomophagous insects. Bulletin of the Entomological Society of New Zealand 10: 63-74.

    Goldson S.L., McNeill M.R., Phillips C.B. and Proffitt J.R. (1992). Host specificity testing and suitability of the parasitoid Microctonus hyperodae (Hym.: Braconidae, Euphorinae) as a biological control agent of Listronotus bonariensis (Col.: Curculionidae) in New Zealand. Entomophaga 37: 483-498.

    Harley K.L.S. and Forno I.W. (1992). Biological control of weeds. A handbook for practitioners and students. Inkata Press, Melbourne. 74 pp.

    Harris P. and Zw�lfer H. (1968). Screening of phytophagous insects for biological control of weeds. Canadian Entomologist 100: 295-303.

    Heard T.A. (2000). Concepts in insect host-plant selection behavior and their application to host specificity testing. Pp. 1-10 In: Host-specificity testing of exotic arthropod biological control agents: the biological basis for improvement in safety, R.G. Van Driesche, T. Heard, A.S. McClay and R. Reardon (Ed.) USDA Forest Service Bulletin, Morgantown, West Virginia, USA.

    Heard T.A. and Van Klinken R.D. (1998). An analysis of test designs for host range determination of insects for biological control of weeds. Pp. 539-546 In: Proceedings of the Sixth Australasian Applied Entomological Research Conference, M.P. Zalucki, R.A.I. Drew and G.G. White (Ed.) Brisbane, University of Queensland.

    Hill R.L. (1999). Minimising uncertainty - in support of no-choice tests. Pp. 1-10 In: Host specificity testing in Australasia: towards improved assays for biological control, W.T.M., L. Barton Browne and J. N. Stanley (Ed.) CRC for Tropical Pest Management, Brisbane, Australia.

    Hoddle M.S. (2004). Analysis of fauna in the receiving area for the purpose of identifying native species that exotic natural enemies may potentially attack. Pp. 24-39 In: Assessing host ranges for parasitoids and predators used for classical biological control: a guide to best practice, R.G. Van Driesche and R. Reardon (Ed.) USDA Forest Service, Morgantown, West Virginia.

    Hoffmeister T.S. (2005). From design to analysis: effective statistical approaches for host range testing Pp. 672-682 In: Second International Symposium on Biological Control of Arthropods, Davos, Switzerland, 12-16 September, 2005, M.S. Hoddle (Ed.) United States Department of Agriculture, Forest Service, Washington.

    Hoffmeister T.S., Babendreier D. and Wajnberg E. (2006). Statistical tools to improve the quality of experiments and data analysis for assessing non-target effects. Pp. 222-240 In: Environmental impact of invertebrates for biological control of arthropods: methods and risk assessment, F. Bigler, D. Babendreier and U. Kuhlmann (Ed.) CABI Publishing, Wallingford, Oxford.

    Jallow M.F.A. and Zalucki M.P. (1996). Within- and between-population variation in host-plant preference and specificity in Australian Helicoverpa armigera (H�bner) (Lepidoptera: Noctuidae). Australian Journal of Zoology 44: 503-519.

    Kuhlmann U., Schaffner U. and Mason P.G. (2006). Selection of non-target species for host specificity testing. Pp. 15-37 In: Environmental impact of invertebrates for biological control of arthropods: methods and risk assessment F. Bigler, D. Babendreier and U. Kuhlmann (Ed.) CABI Publishing, Wallinford, Oxford.

    Louda S.M., Pemberton R.W., Johnson M.T. and Follett P.A. (2003). Nontarget effects - the Achilles' heel of biological control? Retrospective analyses to reduce risk associated with biocontrol introductions. Annual Review of Entomology 48: 365-396.

    Mansfield S. and Mills N.J. (2004). A comparison of methodologies for the assessment of host preference of the gregarious egg parasitoid Trichogramma platneri. Biological Control 29: 332-340.

    Marohasy J. (1998). The design and interpretation of host-specificity tests for weed biological control with particular reference to insect behaviour. Biocontrol News and Information 19: 13-20.

    McFadyen R.E.C. (1998). Biological control of weeds. Annual Review of Entomology 43: 369-393.

    McNeill M.R., Barratt B.I.P. and Evans A.A. (2000). Behavioural acceptability of Sitona lepidus (Coleoptera: Curculionidae) to the parasitoid Microctonus aethiopoides (Hymenoptera: Braconidae) using the pathogenic bacterium Serratia marcescens Bizio. Biocontrol Science and Technology 10: 205-214.

    Miller J.R. and Strickler K.L. (1984). Finding and accepting host plants. Pp. 127-157 In: Chemical Ecology of Insects, W.J. Bell and R.T. Card� (Ed.) Chapman & Hall, London.

    Morehead S.A. and Feener D.H. (2000). An experimental test of potential host range in the ant parasitoid Apocephalus paraponerae. Ecological Entomology 25: 332�340.

    Nechols J.R., Kauffman W.C. and Schaefer P.W. (1992). Significance of host specificity in classical biological control. Pp. 41-52 In: Selection Criteria and Ecological Consequences of Importing Natural Enemies, W.C. Kaufmann and J.E. Nechols (Ed.) Entomological Society of America, Lanham, Maryland, USA.

    Pemberton R.W. (2000). Predictable risk to native plants in weed biological control. Oecologia 125: 489-494.

    Sands D.P.A. and Van Driesche R.G. (2000). Evaluating the host range of agents for biological control of arthropods: rationale, methodology and interpretation. Pp. 69-83 In: Host-specificity testing of exotic arthropod biological control agents: the biological basis for improvement in safety, R.G. Van Driesche, T. Heard, A.S. McClay and R. Reardon (Ed.) USDA Forest Service Bulletin, Morgantown, West Virginia, USA.

    Sands D.P.A. and Van Driesche R.G. (2004). Using the scientific literature to estimate the host range of a biological control agent. Pp. 15-23 In: Assessing host ranges for parasitoids and predators used for classical biological control: a guide to best practice, R.G. Van Driesche and R. Reardon (Ed.) USDA Forest Service, Morgantown, West Virginia.

    Sheppard A.W. (1999). Which test? A mini review of test usage in host specificity testing. Pp. 60-69 In: Host specificity testing in Australasia: towards improved assays for biological control, T.M. Withers, L. Barton-Browne and J. Stanley (Ed.) Scientific Publishing, Department of Natural Resources, Brisbane.

    Singer M.C. (1982). Quantification of host preference by manipulation of oviposition behaviour in the butterfly Euphydryas editha. Oecologia 52: 224-229.

    Singer M.C. (1986). The definition and measurement of oviposition preference in plant-feeding insects. Pp. 65-94 In: Insect-Plant Interactions, J.R. Miller and T.A. Miller (Ed.) Springer-Verlag, New York.

    Solarz S.L. and Newman R.M. (1996). Oviposition specificity and behavior of the watermilfoil specialist Euhrychiopsis lecontei. Oecologia 106: 337-344.

    Tallamy D.W. (1999). Physiological issues in host range expansion. Pp. 11-26 In: Proceedings: Host specificity testing of exotic arthropod biological control agents: The biological basis for improvement in safety, N.R. Spencer (Ed.) Bozeman, Montana.

    van Driesche R.G. and Murray T.J. (2004). Overview of testing schemes and designs used to estimate host ranges. Pp. 68-89 In: Assessing host ranges for parasitoids and predators used for classical biological control: a guide to best practice, R.G. Van Driesche and R. Reardon (Ed.) USDA Forest Service, Morgantown, West Virginia.

    van Driesche R.G. and Murray T.J. (2004). Parameters used in laboratory host range tests. Pp. 56-67 In: Assessing host ranges for parasitoids and predators used for classical biological control: a guide to best practice, R.G. Van Driesche and R. Reardon (Ed.) USDA Forest Service, Morgantown, West Virginia.

    van Klinken R.D. (2000). Host-specificity testing: why do we do it and how we can do it better. Pp. 54-68 In: Host-specificity testing of exotic arthropod biological control agents: the biological basis for improvement in safety, R.G. Van Driesche, T. Heard, A.S. McClay and R. Reardon (Ed.) USDA Forest Service Bulletin, Morgantown, West Virginia, USA.

    van Klinken R.D. and Edwards O.R. (2002). Is host specificity of weed biological control agents likely to evolve rapidly following establishment? Ecology Letters 5: 590-596.

    Wapshere A.J. (1974). A strategy for evaluating the safety of organisms for biological weed control. Annals of Applied Biology 77: 201-211.

    Wapshere A.J. (1989). A testing sequence for reducing rejection of potential biological control agents for weeds. Annals of Applied Biology 114: 515-526.

    Wiklund C. (1981). Generalist vs. specialist oviposition behaviour in Papilio machaon (Lepidoptera) and functional aspects on the hierachy of oviposition preferences. Oikos 36: 163-170.

    Withers T. and Mansfield S. (2005). Choice or no-choice tests? Effects of experimental design on the expression of host range. Pp. 620-633 In: Second International Symposium on Biological Control of Arthropods, Davos, Switzerland, 12-16 September, 2005, M.S. Hoddle (Ed.) United States Department of Agriculture, Forest Service, Washington.

    Withers T.M. (1997). Changes in plant attack over time in no-choice tests: an indicator of specificity. Pp. 214-217 In: Proceedings of the 50th New Zealand Plant Protection Conference, M. O'Callaghan (Ed.) Lincoln University, NZ., New Zealand Plant Protection Society Inc.

    Withers T.M. (1999). Examining the hierarchy threshold model in a no-choice feeding assay. Entomologia Experimentalis et Applicata 91: 89-95.

    Withers T.M. and Barton-Browne L. (2004). Behavioral and physiological processes affecting the outcome of host range testing. Pp. 40-55 In: Assessing host ranges for parasitoids and predators used for classical biological control: a guide to best practice, R.G. Van Driesche and R. Reardon (Ed.) USDA Forest Service, Morgantown, West Virginia.

    Withers T.M., Barton-Browne L. and Stanley J. (2000). How time-dependent processes can affect the outcome of assays. Pp. 27-41 In: Host-specificity testing of exotic arthropod biological control agents: the biological basis for improvement in safety, R.G. Van Driesche, T. Heard, A.S. McClay and R. Reardon (Ed.) USDA Forest Service Bulletin, Morgantown, West Virginia, USA.

    Withers T.M., McFadyen R.E. and Marohasy J. (2000). Importation protocols and risk assessment for weed biological control agents in Australia: The example of Carmenta nr ithacae. Pp. 195-214 In: Nontarget Effects of Biological Control, P.A. Follett and J.J. Duan (Ed.) Kluwer Academic Publishers, Norwell, MA.

    Zw�lfer H. and Harris P. (1971). Host specificity determination of insects for biological control of weeds. Annual Review of Entomology 16: 159-178.