Irradiation as a Quarantine Treatment: Research Protocols (Guideline II)
Irradiation as a Quarantine Treatment for Commodities including Fresh Fruit, Cut Flowers and Durables against Insects, other Arthropod Pests and Nematodes. Guideline II Research and Development of a Single Component Quarantine Treatment or Multiple Components Applied Concurrently.
The research and development of single treatments or multiple component treatments applied concurrently is divided into two discrete phases: (1) preliminary dose efficacy tests and (2) large-scale confirmatory tests. Use of precise experimental irradiation practices is essential together with high standards of dosimetry. Preliminary tests are conducted at graded doses or levels of treatment severity, and the results are analysed to identify the most tolerant stage. These are also used to estimate the dose necessary to attain quarantine security, either directly or in conjunction with iterative tests on larger numbers. The procedure followed will depend on the experimental accuracy achievable with the organism and the commodity. Large-scale confirmatory tests are then conducted to obtain statistical assurance that the treatment efficacy will satisfy quarantine security requirements. Sometimes, commercial scale trials will be required before final acceptance. Importing countries have differing requirements, so close liaison at the planning stage is needed. Special problems will arise where the pest cannot be reared in a laboratory eg it may require developing fruit and sufficient numbers are not available in the field.
The following requirements are essential in the development of a quarantine treatment:
The level of treatment efficacy (quarantine security) and the associated statistical assurance to be achieved must be agreed with the importing country eg 99.9968%, 99.99% or 99.5% at confidence levels (CL) of 95%. The highest risk life stage of the target pest likely to be present at the time of treatment must also be agreed. The behaviour and biology of the pest organism needs to be considered when determining the highest risk (usually most tolerant) life stage. For example, especially with electron beam irradiation, where two life stages of a pest species found in a commodity are equally tolerant of a candidate treatment, the stage normally located deeper in the commodity should be selected as the target stage because treatment effects tend to attenuate towards the commodity centre. If one of these stages is normally predominant it should be selected in preference to one rarely present. Statistical regression analyses are normally used to characterise stages in terms of susceptibility. Statistical packages such as "Genstat" or "SAS" include appropriate routines. The experimental design must utilise a sample size appropriate to the statistical test used to determine the most tolerant stage. Pest organisms may be required to be treated either outside the commodity or in the commodity to determine both the most tolerant life stage and for prediction of the treatment dose which will be required, depending on the country to which the commodity is to be exported. The numbers of target pests used in controls also must be adequate for the analysis used.
For each stage in each cultivar to be tested, 5 evenly spaced doses between about 2 and 98-100% mortality are needed in addition to an untreated control sample, for adequate reliability of analysis. The dose or time range will probably need to be identified in a preliminary "range-finding" test. A minimum of 50 (some recommend 200) in each sample, replicated 3 times, should be treated at each of the 5 doses. A full comparison of stages can be done only if response lines are parallel, and for this a separate test of parallelism is necessary, otherwise, comparisons are only valid at equivalent points, e.g. LD50. If survival is the criterion and the exact number of organisms actually treated is unknown an appropriate form of analysis needs to be used such as the Wadley's method given by Finney (1971) or its equivalent. Where regression lines are not parallel, additional doses or larger sample numbers at mortalities approaching 100% will assist in determining the true response at low levels of survival. It is not unusual to find that lines cross so that a stage which is less tolerant than another at the LD50 is the more tolerant at the LD99.
Alternatively, an Analysis of Variance of responses of stages can be used in conjunction with a single dose. This involves identification of a dose at which there are significant numbers of survivors e.g. between 10 and 90% in all or all but the most tolerant stage. Replicated tests are then done on 50-200 of each stage for comparison on the basis of mortality or survival. This model is very useful for natural infestation tests because of characteristically high variance caused by uneven pest numbers in or on each host unit (piece of fruit or flower stem) and the difference in survival due to unit to unit variation. It is analogous to pointwise comparisons of LD or LT values frequently necessary because regression lines for stages or cultivars are often not parallel
Selection of the experimental method and the determinant population parameter eg LD50
should be made in consultation with the importing country.
The treatment parameters (dose, temperature, time) to provide the required level of quarantine security need to be determined on the stage of highest quarantine risk. Regression analysis eg log-dose-probit, complementary-log-log or logit, to predict the treatment parameters should be based on a number of tests in the range 50-100% mortality. Because of inherent imprecision at the high mortalities involved eg 99.5 to 99.9968%, it may be advisable to use the upper 95% fiducial limit (CL) which is typically wide at the low mortalities involved. Alternatively, the lowest dose at which there were no survivors can be used as the treatment for the first replicate of a large scale confirmatory trial as some importing countries require these tests to be made up of replications of >7 500 or >10 000. If the replicate is successful, remaining replicates can be attempted but if it were not, the dose would be increased for subsequent replicates. This represents an iterative approach and helps to ensure that the treatment is no more severe than necessary to meet the quarantine security efficacy requirement. It is especially important to avoid excessive treatment on commodities susceptible to irradiation injury. Therefore the dose to be used commercially must be the least dose which will achieve the required quarantine security level.
The numbers to be used in a large scale confirmatory trial depend on the efficacy required of the treatment; usually expressed as percentage mortality with a CL, typically 95%, which is an essential. Examples are 99.5% at 95% CL, no survivors from 598; 99.99% at 95% CL, no survivors from 29 956; 99.9968% at 95% CL, no survivors from 93 616. These numbers are derived from established sampling theory. Where the importing country requires it, the total may need to be made up of minimum sized replicates such as 7 500 or 10 000 depending on the total and the number of replicates required. If there are survivors from larger numbers tested the efficacy requirement may still be met with statistical validity but at least one major importing country may not accept a treatment from which there are any survivors regardless of the number tested. Requirements such as this must be established by liaison prior to commencement of research.
The total number of individuals required to be treated without any survivors is determined by the level of security desired by the importing country. For example, for many insects, including fruit flies, USA has historically required a level of security equal to a total treatment mortality of probit 9 (LD99.9968) with 95% confidence. This requires the treatment of at least 93 613 target organisms with no survivors. No survivors from a treated population of 29 956 target pests may be adequate for some temperate climate pests for USA. For many countries, this is the phytosanitary quarantine requirement for all serious pests. For pests of lesser risk some countries will accept 99.5% efficacy (no survivors from 598) and this may be acceptable also where there is difficulty in obtaining larger numbers of pests to test.
If the commodity is artificially infested with a known number of target organisms the number of individuals actually treated must be determined through the recovery of living and dead individuals. In cases where unknown numbers of target organisms are treated, as for both naturally and cage-infested commodities, the number of individuals treated is estimated using the number of individuals recovered from a corresponding control sample of infested but untreated commodity as a parallel sample. Care must be taken in developing mortality assessment criteria. For example, normal adult fruit flies sometimes emerge from deformed puparia. Therefore, scoring all deformed puparia as dead may provide inaccurate mortality data. When the number of individuals treated is known, the natural mortality is determined for each test replication and the treatment mortality must be adjusted for natural mortality (e.g., Abbott's formula).
Sensory and recording equipment used in efficacy tests must meet standards for sensitivity and methods of operation that are agreed upon before tests are initiated. Irradiation dosimetry must be done to a Standard such as that of the ASTM for irradiation.
Because irradiation treatment is non-uniform among individual commodity units during a treatment, the acceptable treatment level will be the minimum level, not the mean level, required for the treatment to provide quarantine security and this will be set at the maximum level of the dose range in the experiment that proved the treatment.
It may be a further requirement to test a proposed treatment under actual or simulated commercial conditions. The design of this type of trial can expected to differ for country, pest and commodity, so it will need to be negotiated with authorities of the importing country. Some of these trials will involve only dosimetry, others may require infested commodity to be included and assessed for mortality.