Cold treatment for management of Drosophila Suzukii 13_ECCT_2013_Dec
A BRIEF OVERVIEW OF COLD TREATMENT RESEARCH METHODOLOGY
2013-11-07
(Prepared by Guy J. Hallman, USDA/ARS, USA)
Background
[1] Phytosanitary cold treatments have been a mainstay of phytosanitation for a century, and research on them has been continually carried out during this entire time period (Gould 1994). Thus, research has been conducted by generations of scientists under differing circumstances using equipment varying in ability to consistently sustain and record temperatures. Earlier work especially did not present sufficient research methodology to repeat the studies nor were some properly published. The result is that today there is a confusing array of results that are difficult to synthesize and use to establish treatments.
[2] Furthermore, although some aspects of cold treatment efficacy have received considerable research attention (such as most tolerant stage of pests), we seem no closer to answering fundamental questions on cold treatment efficacy and factors that might affect efficacy than we were years ago.
[3] This Expert Consultation on Cold Treatments (ECCT) begins the process to make cold treatment research and application more directly applicable to the issues that the treatment faces in facilitating trade in agricultural commodities. An overview of cold treatment research methodology summarizes techniques that have been used to investigate cold treatments, assumptions made (both stated and not stated), and ramifications for cold treatments. We hope that this will be a continuing process that will lead to more effective and applicable phytosanitary cold treatments.
Abstract
[4] The following is a draft position paper on the current status of phytosanitary cold treatment research methodology. It is apparent that cold treatment research methodology is not consistent in application and interpretation. Although we do not aim to inflexibly prescribe cold treatment methodology, there are basic points that can be made regarding its effective conduction, and this ECCT has the objective to identify those points.
[5] It is expected that readers, attendees at the ECCT, and other researchers will add additional information to this position paper and that it will continue to develop, providing guidance to research.
A brief overview of cold treatment research methodology
[6] Cold treatments are not unique in having issues with research methodology. Heather and Hallman (2008) devote a chapter (#6) of their book, Pest Management and Phytosanitary Trade Barriers, to development of postharvest phytosanitary treatments with particular emphasis on research methodology. Their chapter on cold treatments (#7) also discusses research methodology.
[7] Research supporting a cold treatment is often conducted with small quantities of fruit in small treatment facilities that may reach treatment temperatures quicker than large commercial facilities where the treatment is applied. For example, Gould and Hennessey (1997) found that rapid cooling (in 45 min) of carambolas infested with Anastrepha suspensa to 1.1°C reduced the time required to kill all insects by 1/3rd compared with cooling over a 24 h period, as would happen with large commercial lots. Therefore, treatments based on research done using rapid cooling may fail when applied commercially.
[8] It is logical to assume that temperature of a cold treatment is directly related to the amount of time required to achieve complete control of a pest. However, Powell (2003), reviewing Back and Pemberton (1916), suggests that temperatures between 0-2.2°C might be equally lethal to Ceratitis capitata, and data from De Lima et al. (2007) tend to extend that hypothesis up to 3°C. If true that temperatures between 0-3°C may be considered equal in their lethality to C. capitata or possibly other tephritids, this would greatly facilitate research and application of cold treatments. Research methodology to substantiate this concept for low temperatures would need to be carefully considered.
[9] Acclimation of pests to cold temperatures could affect their susceptibility to cold treatments (Iwata et al. 1992). Therefore, research methodology should describe the temperature regimes that were used for rearing and manipulating the pests for cold treatment research before the actual treatment itself was conducted and consider how closely those temperatures might correspond to temperatures occurring in the field.
[10] Age and precise developmental stage of pests treated should be given in research methodology. Simply noting the general developmental stage may be insufficient. For example, Myburgh and Bass (1969) found that middle-aged pupae of Thaumatotibia leucotreta are more cold tolerant than younger or older pupae, meaning that research confirming efficacy of the treatment should be conducted with middle-aged pupae.
[11] Differences in sample size between research units could affect apparent efficacy. Myburgh (1965) concluded that T. leucotreta reared on diet were more cold tolerant than those reared on orange and, thus, conducted phytosanitary research with insects reared on diet. However sample size was too small for the insects reared on oranges (mean = 493) compared with those reared on diet (mean = 15,320) to be able to determine mortality at high levels of control (Table 1). I.e., 0.1% of 493 insects is not even ½ insect while 0.1% of 15,000 insects translates to 15 insects.
Table 1. Percentage survival of Thaumatotibia leucotreta larvae reared on diet and orange and subjected to 1.1°C.
Days at 1.1°C / Percentage survival when reared onDiet / Orange fruit
8 / 7.7 / 56.3
9 / 3.5 / 33.4
10 / 2.2 / 17.5
11 / 2.0 / 7.9
12 / 0.9 / 2.5
13 / 1.8 / 1.6
14 / 1.1 / 1.4
15 / 0.11 / 0
16 / 0.11 / 0
17 / 0.07 / 0
[12] Furthermore the data indicate much higher levels of mortality in diet-reared insects treated for 8-11 d. Mortality of non-treated controls is not provided; if mortality in the control is >5% then mortality in the treated cohorts should be adjusted by the following formula:
[13] Ya = 100% - [(X – Y)/X](100%)
[14] where Ya is the adjusted percentage surviving in the treated cohort, X is the percentage surviving in the control, and Y is the percentage surviving in the treated cohort.
[15] Infestation methodology could affect survival. Grout et al. (2011) concluded that orange was not an ideal host of Bactrocera invadens based on high mortality of the immature stages reared in orange (~80%) compared with immatures reared on diet (~20%). However, the fruit infestation technique consisted of injecting eggs suspended in distilled water into a 6 mm hole bored into the stem end of the fruit and supplemented with diet. The hole was then plugged with cotton and hot wax. This unnatural manipulation of the eggs could have killed many, and it would be prudent to test artificial manipulations of research subjects for possible detrimental effects that could affect the conduction of the tests.
[16] Can comparisons between species be used to justify phytosanitary treatments, even if the comparisons are done in vitro? Hallman et al. (2011) found B. invadens to be more cold tolerant than C. capitata when 3rd instars were treated at 0.9°C in diet. This was substantiated when they were treated while infesting oranges (Hallman et al. 2013). It seems logical to conclude that this type of comparison done in fruit would validate a treatment for any species found to be no more cold tolerant than species for which treatment schedules already exist.
[17] Do differences in cold tolerance exist among different populations of the same species? Hallman et al. (2013) discuss that possibility for most cold-tolerant stage of C. capitata. Literature from Australia (De Lima et al. 2007, Jessup et al. 1993, Hill et al. 1988) found the 2nd instar to be more cold-tolerant than the 3rd while studies done in Argentina (Willink et al. 2007), Egypt (Hashem et al. 2004), Hawaii (Back and Pemberton 1916), and South Africa (Ware et al. 2005 as interpreted by Hallman et al. 2013) found no difference or the 3rd instar was more cold-tolerant than the 2nd. Therefore, it is a reasonable hypothesis that the population of C. capitata in Australia differs from those tested elsewhere for this particular trait. A comparison among populations done at a neutral site, such as the International Atomic Energy Agency laboratories in Seibersdorf, Austria, could shed light on that question.
[18] Does host affect cold-tolerance? De Lima et al. (2007) estimated that 9.7 and 13.6 d, respectively, were required to achieve 95% mortality of 2nd instar (most tolerant stage) C. capitata in ‘Lisbon’ lemon and ‘Valencia’ orange. This topic deserves further research. Perhaps it is simply a question of host suitability with insects being less cold tolerant on less suitable hosts.
[19] Different techniques may be used to measure efficacy, from prevention of insect movement after the treated sample is allowed to warm to ambient temperatures for a time period to prevention of adult emergence from treated larvae. Because phytosanitary inspectors usually consider any moving insects found at any time post-treatment as survivors, prevention of movement should be the measure of efficacy for cold treatments.
[20] It has been suggested that surrogate species (closely related to the target organism) could be used to develop cold treatments. Given the example that A. ludens is considered considerably more cold-tolerant than other species in the genus Anastrepha (APHIS 2013), using surrogate species to determine cold treatments for other species does not seem prudent. However, surrogate species could be used to develop broad generic treatments that would incorporate data from a number of species in the group for which the generic treatment is being developed.
Statistical analyses
[21] Phytosanitary treatment research is often done using standard statistical approaches of regression analysis and confidence intervals of 95% for means testing. The former is for predicting efficacious doses and the latter (means testing) is for determining most tolerant stage or other comparisons of factors that might affect efficacy.
[22] The problem with prediction of efficacious doses is that the level of efficacy demanded of phytosanitary treatments is 99.99%, and prediction of that extreme of a level of efficacy with moderate sample sizes does not carry much confidence. That is precisely why large-scale confirmatory testing is required. An argument could be made that small-scale dose-response testing with the objective of using the data to predict efficacious doses is of no help, thus, is unnecessary and that researchers should concentrate on large-scale testing using an iterative approach. I.e., choose an initial dose likely to fail, begin testing, and raise that dose by an amount that depends on how badly and quickly it failed as well as considerations on the difficulty of doing the research. This approach may not seem as “scientific” as small-scale, dose-response analysis, but the latter may simply carry a façade of scientific accuracy.
[23] Using means testing to determine significant differences between insect growth stages, populations, infestation techniques, or other factors that could affect efficacy is statistically valid. However, the level of confidence may need to be made more liberal because the consequences of type I and type II errors are very different. A type I error (false positive) erroneously concludes that there is a difference when there is not. A type II error (false negative) says there is not a difference when there is. There is essentially no consequence for the commercial application of a phytosanitary treatment when a type I error is committed. I.e., if the mean cold treatment time needed to kill all eggs and 3rd instars of a pest were 11.1 and 12.7 d, respectively, but analysis of variance found no significant difference at 95% confidence, there would be no consequence for treatment efficacy by simply picking the one with highest mean (3rd instar), although the research might be more difficult to perform. (Difficulty should never be a reason for taking risky shortcuts in research.) On the other hand, choosing eggs as the most tolerant stage because no significant difference was found and it is easier to work with eggs risks developing a treatment that will not kill all 3rd instars. It would be more prudent in cases like this to use a much lower level of confidence in means testing.
[24] Also, testing significant differences among factors should be done at high levels of control, 95 and <100%. This is because slopes might not be parallel and the part of the relationship of interest is that part that achieves high levels of control. When comparing factors, control should be a little less than 100% for at least all but one of the factors or there is no test. I.e., one cannot detect differences among factors at levels of control that are 100% for those being compared.
Generic cold treatments
[25] Generic phytosanitary treatments may be applied across groups of pests and/or hosts even though research may not have been conducted on all of the species covered by the generic treatment. Some generic cold treatments already exist. E.g., APHIS (2013) treatment schedule T107-c is for all species of Anastrepha except ludens on 9 fruits. There is no technical reason why the generic concept could not be expanded for cold treatments. E.g., a generic cold treatment for all tropical Tephritidae (Anastrepha, Bactrocera, Ceratitis, Dacus) would be based on the most tolerant species, which is considered to be A. ludens, which would make the dose at 1.1°C 20 d. Compare that with 15 d for all other species of Anastrepha and 14 d for C. capitata. In this case tolerance of the fruits being treated and time would be the factors upon which treatment application would be based.
[26] Differences among hosts have been noted for cold treatments (De Lima et al. 2007). Therefore, for a cold treatment to be generic for host the dose for the host requiring the most days would be used. In the case of B. tryoni on Australian fruit 16 d might be the generic dose required at 3.0°C given that it is the highest dose of those scheduled by one account (APHIS 2013).
Table 2. APHIS (2013) treatment schedules at 3.0°C for Australian fruit at risk of infestation with Bactrocera tryoni.
Fruit / Exposure period (d)Cherry / 15
Citrus limon / 14
C. nobilis / 16
C. reticulata / 16
C. sinensis / 16
Effect of cold treatment on commodity quality
[27] The effect of a cold treatment on commodity quality is not a necessary part of research to develop a phytosanitary dose. However, for practical considerations of commercial application the commodity in question must tolerate the treatment. Three critical concerns for commodity quality are: