(i) Nectarines

89. According to the United States, in the consultations Japan had suggested that a scientific basis for requiring testing for each variety arose from USDA data for the development of an MB quarantine treatment for codling moth in nectarines for export to Japan. The research alluded to by Japan was presented in two articles by American scientists, one in 1987 and the second in 1990. In the first study (1987), dose-mortality tests were conducted on six early- to mid-season varieties of nectarines: May Grand, Fire Brite, Red Diamond, Spring Red, Summer Grand and Fantasia. The test results had shown that the level of methyl bromide estimated to produce 50 per cent mortality in codling moth on Summer Grand was significantly lower than for the remaining five varieties. ⁷⁵ In 1990, three additional varieties of nectarines were tested (May Diamond, Mayfire and May Glo). The dose-response of these suggested that a lower amount of methyl bromide was necessary to achieve 100 per cent mortality. ⁷⁶ Nevertheless, the higher dose established in 1987 was used for the confirmatory tests. The United States argued that the principal reason a higher dose had been established in 1987, was because the Summer Grand variety that had "stuck out" in dose-mortality tests in 1987, had also required the highest minimum dose to achieve 100 per cent mortality.
90. In the 1990 study on nectarines, the authors had looked at different insect and fruit interactions to evaluate the necessity for varietal testing. Dose-response data had been obtained for eggs fumigated on different fruit substrates (i.e., different species: nectarines, plums, peaches, apples) as well as eggs on waxed paper inserted into the fumigation chamber with a load of fruit. The authors compared LD₅₀ and LD₉₅ for codling moth eggs. The scientists found that while there might have been slight apparent numerical differences in codling moth susceptibility to methyl bromide (dose-response) on various substrates, this did not affect the efficacy of the quarantine treatment, nor had these differences been anything but natural variations in the response of the test insects. The highest minimum doses tested that caused 100 per cent mortality of codling moth on nectarines, peaches, plums and apples, and waxed paper during dose-mortality tests were well below the 48 g/m³ methyl bromide that was used in the ultimate quarantine treatment for nectarines established through confirmatory tests. Based on these results, the authors of the study had concluded that testing for efficacy of methyl bromide against codling moth eggs on every nectarine cultivar proposed for export to Japan was unnecessary to show the efficacy of the quarantine treatment. ⁷⁷
91. The United States claimed that, in consultations, Japan had pointed to the relative susceptibility to methyl bromide of codling moth eggs on Summer Grand and noted differences in the physical characteristics of the fruit as an indicator of significant differences among varieties of the same agricultural product which warranted varietal testing. Yet the United States was of the view that the authors' comment regarding flaws in Summer Grand did not state or imply that these flaws were characteristic of the variety. Because varieties ripen at different times, scientists had only a small window of opportunity to test the quarantine treatment simultaneously on several varieties. Thus idiosyncratic aspects of the physical condition of the specific fruit due to harvest practices, weather conditions, the amount of time one variety may have been stored while waiting for other varieties to be harvested for testing, and other external factors were reflected in the dose-mortality tests. The condition of any particular fruit could affect dose-response results.
92. Moreover, although Summer Grand required less methyl bromide than the other varieties to achieve 50 per cent mortality of the test insects, it required more methyl bromide to achieve 100 per cent mortality of the test insects than the other five varieties tested. This apparently contradictory information indicated that the natural variation of the insect, and factors such as experimental error, fruit condition, and factors that could affect the amount of fumigant to which insects are exposed, such as sorption by the packing material and load in the chamber, all played a role in differing dose-response results. ⁷⁸
93. It was also significant to note that over the summer of 1997, Vail and Yokoyama had re-tested two varieties of nectarines - May Grand and Summer Grand. In the dose-mortality tests, the level of methyl bromide required to achieve a lethal dose for 50 per cent and 95 per cent of codling moth eggs did not differ significantly. ⁷⁹ This was in contrast to the initial 1987 nectarine study results. The United States argued that these 1997 results confirmed that the differences noted in 1987 were due to natural variation of the insects and other experimental variables. The authors of the 1997 study had stated the "extremely high numbers of insects killed in large-scale confirmatory tests and the virtual non-host status of nectarines attest to the high level of security afforded to this treatment to prevent introduction of codling moth to Japan via nectarine cultivars". In re-testing this variety in 1997, however, results of dose-mortality testing had indicated that the highest minimum dose for Summer Grand was now a dose closer to the six varieties tested in 1990. This indicated that a dose closer to that seen with dose-mortality tests for the 1990 batch might be the more appropriate level of treatment for nectarines.
94. More significantly, the confirmatory test applied in 1987 had established the efficacy of the treatment on all nectarines tested. A dose that was 20 per cent greater than the highest minimum dose required for 100 per cent mortality during the dose-mortality testing had been proposed as a quarantine dose for these six varieties. ⁸⁰ This treatment had proven successful to achieve Japan's level of protection and had been accepted by Japan for all six varieties. As it was the large-scale confirmatory tests at a lower dose had not been done in 1990, because the 1987 results indicated that a higher dose was necessary. Thus confirmatory tests at 48 g/m³ were completed (which was the dose established in 1987 that was derived from the highest minimum dose believed to achieve Japan's level of protection in the dose-mortality tests with 10-20 per cent added) and showed that this dose was effective for all nectarine varieties. Hence, the United States reaffirmed that confirmatory tests were the only relevant indicator of efficacy of a quarantine treatment. The United States maintained that Japan's claim that variations in dose-mortality tests obtained from two groups of varieties tested three years apart were the result of inherent varietal differences was scientifically invalid. Such a position ignored natural variation in responses of insects, and lack of a common control among tests (i.e., no one variety was tested in both years). Dose-mortality testing would inevitably show natural variability among dose-response results. This did not justify re-establishment of confirmed quarantine treatments for codling moth on a given product.
95. The United States noted that similar findings on different host fruit species and other insect species had been published in Japan; Misumi had investigated quarantine control through MB fumigation of the Japanese mealybug and the citrus mealybug naturally infesting Satsuma mandarins. ⁸¹ All stages of the two species of pests were reared on fresh pumpkins as an alternative to Satsuma mandarins because of difficulty in rearing mealybugs on the Satsuma mandarin. The physical and chemical differences between Satsuma mandarins and pumpkin were considerable. Yet despite the use of different host plants, the authors concluded this was of no significance to the efficacy of the fumigation process. Hence, Japanese scientists were willing to accept efficacy of quarantine treatment for a pest as it related to two radically different products, pumpkins and mandarins. This experiment supported the position of the United States that requiring exhaustive data on individual varieties of a product was not necessary. The study demonstrated that methyl bromide, when applied at specified concentrations over specified periods of time, killed codling moth in the same way, regardless of variety, and even regardless of substrate.
96. Japan recalled that at the time of the additional lifting of the ban for the 1990 varieties, even though a dose of 20 g/m³ had been proven effective to ensure 100 per cent mortality, the United States had chosen to propose 48 g/m³ as the treatment level. ⁸² The United States did have the option to propose 20 g/m³ or 24 g/m³, adding a "buffer", as the treatment for the additional three varieties. Japan accepted the application of the existing treatment ⁸³ as the on-site confirmatory test had demonstrated its efficacy. This did, however, raise a practical question as to what level of treatment the United States would have proposed had the six varieties (tested in 1987) not been approved initially. It was reasonable to assume that they would not have proposed 48 g/m³, but would have chosen a level around 24 g/m³ instead. The United States could not argue that "confirmatory tests ¼ showed that this dose (of 48 g/m³) was effective for all nectarine varieties" (paragraph 4.94). At the most, the United States could argue that this dose was effective for the six varieties tested in 1987, and might be effective for the same varieties in the future. ⁸⁴ A similar problem arose with respect to New Zealand cherries of the Bing variety (from paragraph 4.103).

Table 7 Doses which Achieved 100 per cent Mortality of One-day-old Eggs
Year	Varieties	Dose (g/m3)
1987	Summer Grand	40
	May Grand	35
	Firebrite	35
	Spring Red	35
	Fantasia	30
	Red Diamond	30
1990	May Diamond	15.0
	Mayfire	17.5
	May Glo	20.0

97. Japan acknowledged that the issues of natural variation within the same insect group or year-to-year variation of the fruits were scientifically genuine. There would always be variables other than varietal differences. The existence of various exogenous variables did not by itself prove that observed differences were not attributable to varieties. The task of a scientific demonstration began, not ended, with the discovery of variables. Therefore, Japan had no reason to doubt the 1997 data on two cultivars of nectarines (May Grand and Summer Grand) which indicated, contrary to the 1987 test data, absence of a statistically significant difference between LD values. Explanatory variables other than varietal differences could include natural variation in insects or crops. However, Japan pointed out, once it was assumed that populations were different year-to-year, or case-by-case, any increase in the sample number was pointless, and large-scale confirmatory tests could not predict efficacy of a treatment either, because the confidence in the tests would end at the termination of that particular crop year (for the particular crop population), or with the last individual of that particular group of codling moth (for the particular insect population). This also belied the US claim that "the confirmatory test applied in 1987 established the efficacy of the treatment of all nectarines tested" (paragraph 4.94) The United States should have argued that the test did not confirm efficacy of the treatment for any variety in any year except 1987. Alternatively, the United States meant to say that there were unknown exogenous variables which significantly affected the results of these studies. However, from a practical point of view, this argument was equally empty; efficacy of a treatment could never be established.
98. Japan noted, in respect of the US reference to the 1987 nectarine study (paragraph 4.89), that the authors had stated that "no differences in egg susceptibility to MB were found among five infested cultivars". Hence, the conclusion had applied only with respect to the five varieties out of total of six varieties tested. In fact, a difference in efficacy had been found between the last variety (Summer Grand) and the others tested. The authors acknowledged that "[a] comparison of the LD₅₀'s showed that eggs on Summer Grand were significantly (non-overlap of 95 per cent CL) more susceptible to MB fumigation than eggs on other cultivars".⁸⁵
99. In respect of the US reference to the 1990 nectarine study (paragraph 4.90), Japan noted that the authors had stated, "[w]e propose that a c x t product of 68.0�3.0 gh/m³ methyl bromide ... would be a useful measurement to help maintain treatment security for control of codling moth on all nectarine cultivars". ⁸⁶ The proposed measurement of the CxT product was the basis for the cited conclusion. However, the present treatment of nectarines did not control the CxT value during the fumigation process. Consequently, its efficacy across all cultivars could not be assumed from the authors� statement.
100. In respect of the US reference to the Vail et al., 1997 re-test of Summer Grand and May Grand nectarines (paragraph 4.93), Japan pointed out that the authors had not demonstrated that a "high level of security" was equivalent of Japan's level of protection. Furthermore, the authors had not denied contribution of varietal differences; they had stated "the source of variation cannot be attributed solely to cultivar differences". (emphasis added).
101. The United States refuted Japan�s claim that the order in which dose-mortality tests were conducted could result in the establishment of a treatment for one variety that would fail on additional varieties. As it had not been shown that variations in dose-mortality test data reflected underlying varietal differences affecting treatment efficacy, there was no evidence to support this assertion. The 1997 retest of the 1987 Summer Grand nectarine data indicated that the 1987 test results were anomalous. Had the results been in reverse order, there would have been no reason to believe that a treatment based on the lower dosage levels would not have been effective. And had the anomalous 1987 data been too low rather than too high, this would not have resulted in an ineffective quarantine treatment since the confirmatory test on the same variety would have failed. The United States stressed that an ineffective treatment would not be established in the first place, let alone be permitted to remain in place for future varieties. In addition, the United States noted that a confirmatory test could never indicate whether a dose was too high, only if was too low.
102. Japan noted, in this respect, that the results which the United States now described as "anomalous" were the United States' own test results which had formed the basis for lifting the ban on the importation of the products at issue.

(ii) Cherries

103. Japan recalled the conclusion of the authors of a 1987/1988 New Zealand study on cherries:

"For the range of mortalities that we considered, the relationship between the c x t sum and the injected dose is close to linear, with a non zero constant term that varies between cultivars and between season. Thus, the injected dose that is required to achieve a given mortality will vary between cultivars. Factors that may affect the sorption pattern of any one cultivar require further investigation." ⁸⁷

104. In the case of New Zealand cherries, Japan noted that the same treatment level had been initially established for the Dawson and Bing varieties. When Rainier, Sam and Lambert varieties were additionally approved, test results showed a significantly higher level of efficacy for the Bing variety in comparison to Dawson, Rainier and Sam in terms of non-overlap of 95 per cent confidence intervals of LD₅₀ and LD₉₉. As it was, since the treatment level had been set at a level which would ensure effective treatment of Dawson, the less susceptible variety, its efficacy was confirmed for the additional three varieties as well. Similar to the case of nectarines, the data implied a possibility, however, that, had Bing alone been approved initially, the treatment could have been established at a level which would have been found ineffective for Dawson, Rainier or Sam (Table 5, above).
105. Hence, Japan argued there were cases where, while the actual treatment levels were not modified, variations did exist to such a degree that different treatment levels could have been proposed depending on the sequence in which the varieties were tested. By no means did this imply that varietal difference did not exist because the same treatment of the pest had been found effective for all approved varieties of the respective products.
106. The United States pointed out that the authors of the New Zealand study on cherries had concluded that "although the results indicated a higher injected dose when 'Rainier' had been fumigated compared to 'Bing', the large change between seasons in the results for 'Sam' indicated that seasonal differences (perhaps associated with differences in maturity) may have been more important than cultivar". ⁸⁸ Many factors could have contributed to a variation in Bing in 1987. The authors had noted that "sources of variation that may affect results include the insect material, cherry sorption, and measurement and circulation of methyl bromide within the chambers". ⁸⁹ For example, Sam cherries had different sorption patterns between 1987 and 1989. Again, the Japanese overlooked the existence of natural variation of the test insects, whether it occurred in the United States or another exporting nation.
107. Most significantly, the United States claimed that Japan had also ignored the stated conclusion of the authors that the developed quarantine treatment for cherries was applicable to all cherry varieties. "Where complete kill is the objective, the commercial rate used by cherry exporters is 64 g/m³, 12� C, 2 hours, and 40% load. This treatment controls codling moth eggs potentially infesting any of the cultivars considered". ⁹⁰ [emphasis added]. For all season-cultivar combinations, in both years, the highest lethal dose levels of methyl bromide at the LD₉₅ had still been at least 10 g/m³ below the "complete kill" commercial dose used to do the confirmatory test. As the authors further noted in the article abstract: "The commercial treatment thus affords a high level of security". Japan had incorrectly interpreted the significance of these studies and erroneously concluded that the efficacy of treatment could vary within a product.
108. The United States argued that there was always a degree of variability in any dose-mortality test from variety to variety and even within the same fruit variety. Such variations were the inevitable result of differences in natural conditions and testing environments, from crop to crop and year to year. Japan had already accepted a certain dose-response variability in allowing imports of the varieties that it had already approved. This normal variability in testing results could not constitute a legitimate basis for denying approval for other varieties of the same products. Japan maintained that the existence of various exogenous variables did not in itself prove that the observed differences were not attributable to varieties.

To continue with CxT Values

⁷⁵ Op. cit., pp. 840-842. (US Exhibit 14)

⁷⁶ Yokoyama, Miller and Hartsell, "Evaluation of a Methyl Bromide Quarantine Treatment to Codling Moth (Lepidoptera: Tortricidae) on Nectarine Cultivars Proposed for Export to Japan," 83 J. Econ. Entomol. 466-471, 1990. (US Exhibit 12)

⁷⁷ Ibid., pp. 466, 468, 470. (US Exhibit 12)

⁷⁸ The United States noted that probit model of analysis looked only at the linear relationship between dose and mortality (as stated in US Exhibit 28). There were other, more sophisticated statistical models which included consideration of natural variation of the test insect population. Such a model had been used to re-analyze all varieties in the 1987 and 1990 studies by Yokoyama et al., Taking natural variation into consideration, it had been shown that the LD₅₀s among varieties were not significantly different (in Robertson and Yokoyama, "Effect of Nectarine Cultivar on Response of Codling Moth (Lepidoptera: Tortricidae) to Methyl Bromide Fumigation," Unpublished, 1998. (US Exhibit 15)).

⁷⁹ Vail and Yokoyama, "Nectarines: The Issue of Varietal Testing", unpublished. (US Exhibit 16)

⁸⁰ MB fumigation at 48 g/m³ for 2 hours at 21� C or above and 50 per cent or less load.

⁸¹ Misumi, Kawakami, Mizobuchi and Tao. No. 30 Res. Bull. PL. Prot. Japan 30: 57-68, 1994. (US Exhibit 13, in Japanese)

⁸² Japan noted that the scientists had determined the treatment level of 48 g/m³ by multiplying the complete mortality dose of 40 g/m³ by 1.2 times (buffer).

⁸³ Paragraph 4.94.

⁸⁴ Japan noted that Robertson (Op. cit., US Exhibit 15) re-analyzed the data on American nectarines (1987 and 1990) by a statistical method developed by the author and concluded that no statistical difference was observed. However, the analysis failed to reject even a 190-times difference of confidence level. Such a wide degree of ambiguity was not acceptable to Japan.

⁸⁵ Supra note , p.841. (US Exhibit 14)

⁸⁶ Supra note , p.470. (US Exhibit 12)

⁸⁷ Maindonald, Waddell and Birtles, "Response to Methyl Bromide Fumigation of Codling Moth (Lipidoptera: Tortricidae) Eggs on Cherries", New Zealand. J. Econ. Entomol. 85(4): 1220-1230, 1992. (Japan, Exhibit 21 and US Exhibit 4)

⁸⁸ Ibid., p.1227. (Japan, Exhibit 21 and US Exhibit 4)

⁸⁹ Ibid., p.1224. (Japan, Exhibit 21 and US Exhibit 4)

⁹⁰ Ibid., p.1229. (Japan, Exhibit 21 and US Exhibit 4)