CG Reply at 151. This fact is explained, however, by the historical application of diazinon at much higher rates and by the relative newness of the current four-pound rate. In any event, diazinon applications at four pounds or less have in fact been associated with several waterfowl kills, including Canada geese, mallards, and the 85 wigeon killed during one of Ciba-Geigy's field studies (applied at an average rate of 2.6 lbs. ai/A).26

Based on the evidence summarized above, there is no question in my mind that the Agency met its burden of coming forward with an affirmative case regarding the risks of diazinon. I reject the suggestion (e.g., CG Ex. 34 (p. 5)) that cancellation should never be based on lab data, exposure information and bird kills standing alone.27 In fact, the Notice of Intent to Cancel in this case was issued in the absence of field-study data. Ciba-Geigy is correct, however, that the best evidence of risk consists of validly conducted field studies. Such studies may be used to rebut inferences based solely on lab data.28 Thus, the issue reduces to whether Ciba-Geigy's five field studies are sufficient to rebut the Agency's prima facie

26 Initial Decision at 45; see also EPA W-5 (pp. 15-17). Ciba-Geigy's contention (CG Reply at 153) that this large wigeon kill was attributable to the use of a boomless cluster sprayer is refuted by the witness who conducted the field study in question, Dr. Kendall, who concluded that wigeon remain at lethal risk if exposed to residues from the proposed 2+2 rate. CG Ex. 28 (pp. 8–9). I similarly find Ciba-Geigy's challenge to the accuracy of other kills at low rates (CG Reply at 156-60) to be speculative and unpersuasive.

27 See EPA W-19 (pp. 3-5) (risk assessment premised on lab toxicity data and exposure information "is based on guidance recommended by the American Institute of Biological Services (AIBS) and has been peer-reviewed by the FIFRA Scientific Advisory Panel"; such data "are routinely utilized at EPA in assessing and managing risks"); see also EDF v. EPA, 548 F.2d at 1005 (lab data are a "sufficient basis" for cancellation or suspension); EDF v. EPA, 489 F.2d 1247, 1254 (D.C. Cir. 1973) (same).

28 Initial Decision at 54-55. Lab data do suffer from certain inherent defects. For example, they fail to determine whether birds in the field might avoid treated turf. Initial Decision at 13-14. LCso values are derived from tests on young chicks and do not show whether adult birds are less susceptible to diazinon poisoning. Id. at 14-15. On the other hand, this criticism does not apply to LDso values, which show median lethal dosages for adult birds and more appropriately reflect the risk to waterfowl because they feed continuously on turf for extensive periods of time. EPA W-13 (p. 12). Moreover, lab data might underestimate the actual risk in certain respects. Birds in the field are subjected to low temperatures and other routine environmental stress, or might be more sensitive than the surrogate species used in the tests due to different feeding patterns or physiology. EPA W-2 (p. 25); EPA W-19 (pp. 3-5). Field residues might be higher than those predicted due to mechanical and human error. Id.

case, and, if so, to what extent will birds remain at risk under the proposed label.29

F. Ciba-Geigy's Field Studies

Shortly after the Notice of Intent to Cancel was issued, CibaGeigy commissioned five level-one field studies to determine whether its proposed label amendments would eliminate the risk to birds posed by diazinon use. A level-one field study is a screening study designed to determine only whether adverse effects occur, not the extent of those risks. RX 274 (Preface). Four studies were conducted on golf courses and focused on waterfowl, using the Canada goose or the wigeon as representative species. The fifth study was conducted on home and commercial lawn sites and examined the effects of diazinon on songbirds. All five are described in more detail in the Initial Decision at 53-58.

Based on his review of the field studies, the Presiding Officer determined that "the bird mortality indicated by the laboratory and residue data is not likely to occur with any predictable frequency or in any predictable numbers, although [the field studies] do not enlighten us as to the specific reasons why this is so." Id. at 61. He further found that "an undetermined number of deaths" would occur under the proposed label, but that there "is not sufficient evidence to quantify the adverse effects with any greater degree of precision." Id. at 62.

In my view, the field studies are insufficient to show that use under the proposed label would eliminate avian risk altogether, or even reduce that risk to low levels. In particular, two critical defects undercut the significance of these studies.

First, each study was premised on the assumption that any significant mortality resulting from diazinon poisoning would occur relatively quickly and that the bird would remain on the site to die.30 Thus, the carcass searches were confined to the perimeters of the testing sites, and no radiotelemetry was used to determine if birds were dying offsite. A very recent field study conducted by EPA, however, demonstrates that this assumption is incorrect. Dr. Bill Williams, Team Leader of the Wildlife Toxicology Research Team at

29 The Presiding Officer found, and I agree for the reasons set forth in the Initial Decision, that Ciba-Geigy's simulated field studies and pen studies "have little weight as evidence of what is likely to occur under actual field conditions." Initial Decision at 53; see id. at 13-16, 51-52.

30 See, e.g., CG Ex. 33 (pp. 3–4); CG Ex. 43 (p. 20); CG Ex. 57 (p. 5).

EPA's laboratory in Corvallis, Oregon, testified about the results of this study, which have not yet been drafted into a report (EPA W20 (p. 9)); Ciba-Geigy's avian experts were thus apparently unaware of this study. It demonstrates, however, that birds exposed to organophosphates like diazinon left the exposure site to die elsewhere. Id. Dr. Williams concluded that "[i]t is probable, in fact, that birds that ingest less than immediately lethal levels of an [organophosphate] such as diazinon, will be able to fly away to other sites, making them more vulnerable to predators *** and environmental stresses in the near term and more susceptible to disease, immune dysfunction and reproductive impairment in the long term." Id.; accord, EPA W-2 (pp. 17-18); EPA W-8 (pp. 3-4).31

The Presiding Officer concluded that the field studies nonetheless suggest that on-site mortality would be reduced under the new label. Initial Decision at 60. Even if this is true, the issue of offsite mortality continues to loom large. Two Ciba-Geigy experts, Drs. Kendall and Best, testified that diazinon intoxication and sublethal effects warrant further study.32 Because Ciba-Geigy's field studies were based on the assumption that any significant mortality would be immediate, the extent of offsite mortality remains uninvestigated, and the significance of Ciba-Geigy's field studies is seriously eroded.

More importantly, I am troubled by the small number of studies upon which Ciba-Geigy bases its conclusions. EPA's Guidance Document for Conducting Terrestrial Field Studies recommends that fourteen studies be conducted to ensure that statistically significant data exist to show that no adverse effect has occurred. RX 274 (pp. 15– 22). Although the Guidance Document notes that as few as eight sites may be adequate in certain circumstances (id.), here we have only four studies from which to draw conclusions regarding waterfowl, and only one study that focused on passerines.33 The Presiding Offi

31 Ciba-Geigy's field studies themselves revealed several instances where Canada geese, the species used as a surrogate for all waterfowl, suffered diazinon intoxication from residues applied at the 2+2 rate. Initial Decision at 56-61. When these birds flew off, they were presumed to have recovered, (e.g., CG Ex. 28 (p. 8)), but Dr. Williams' testimony shows that they might well have suffered continued adverse effects and died offsite.

32 Tr. 950, 953, 1067, CG Ex. 125 (p. 46). Dr. Kendall does not believe that diazinon's sublethal effects would lead to widespread mortality. Tr. 950. It is not entirely clear how he reached this conclusion given the limited carcass-search area and absence of radiotelemetry in his field studies. In any event, risks of less than widespread mortality must still be considered and require cancellation if not justified by diazinon's benefits. See Section II.B, supra.

33 Although some of the other studies also included observations of songbirds, together these studies still fall far short of the number needed to generate statistically

cer concluded that this flaw precludes drawing any conclusions about effects on bird populations (Initial Decision at 60), but its significance goes deeper than that. Level-one field studies are merely screening studies to determine whether adverse effects occur; they are generally not used to quantify adverse effects or determine impacts on populations. EPA W-19 (pp. 10-11); RX 274 (Preface & p. 10).34 The small number of sites here calls into question the very heart of Ciba-Geigy's risk case that there will be no significant adverse effects on waterfowl under the proposed label-because the field studies fail to provide statistically significant evidence that no adverse effects will occur (as level-one studies are designed to do). EPA W-19 (pp. 14-15); EPA W-20 (pp. 10-12).

I realize that it might have been impossible to complete eight or fourteen field studies prior to the hearing to examine whether the proposed label will cause significant mortality. Because CibaGeigy bears the burden of proof, however, its risk case necessarily suffers from the lack of field experience under its many recent label amendments and proposed label revisions.35

G. Ultimate Risk Findings

In arriving at an ultimate finding regarding avian risk, I have carefully considered the testimony of the impressive array of avian experts assembled by Ciba-Geigy.36 To the extent these experts state that the risk to birds would be "acceptable" or "reasonable," their conclusions cannot be accorded much weight because they failed to weigh risk against the benefits of continued use as required by significant data. Although Ciba-Geigy contends that the openness of the golf courses makes it unlikely that mortality would go unobserved (CG Reply at 83), the concern here is not the accuracy of the observations but whether they resulted in statistically significant data. EPA W-19 (pp. 14-15).

34 The more extensive "level-two" field study is needed to quantify the magnitude of the effects when an adverse impact is observed. RX 274 (Preface & p. 10).

35 The problem raised by the small number of sites is exacerbated by other flaws in the field studies. For example, the hazing that occurred at the Sudden Valley and Sea Links studies raises serious questions about the extent to which they replicated actual field conditions. EPA W-12 (pp. 10, 15, 27); EPA W-20 (p. 11). The extent to which Canada geese, the surrogate species for waterfowl, were actually feeding on treated turf is also unclear, particularly in the Connecticut study. EPA W-12 (pp. 5, 11, 18, 21–22, 24–25); EPA W–10 (p. 15). The Atlanta study was described as so defective that it "has very little to contribute to the assessment of the risks of diazinon." EPA W-12 (p. 24). Ciba-Geigy expert Dr. Best also criticized the studies as being "narrow in their temporal focus" due to time constraints in preparing for the hearing, and recommended additional studies during significant breeding seasons and other times of the year. CG Ex. 125 (pp. 45–46).

36 See, e.g., CG Exs. 28, 32–33, 43–44, 69–70, 112, and 125.

FIFRA. Their collective testimony does convince me, however, that diazinon use under the proposed label would probably have a minimal adverse effect on overall bird populations (in contrast to individual birds).37 Moreover, the proposed geographic and seasonal restrictions would undoubtedly reduce the risk associated with diazinon use under existing labels.

Due to the two critical defects in the field studies identified above, however, I cannot find that the avian risk would be as low as Ciba-Giegy contends.38 At a minimum, EPA has established "the risk of the unknown." 39 Although the risk under the proposed label would be less than under the existing label, the risk to individual birds might well remain widespread and continuous.

Moreover, even if one were to accept Ciba-Geigy's field studies as perfectly sound, certain identifiable risks would remain. The kill of 85 wigeon during the Sudden Valley field study demonstrates conclusively that wigeon exposed to residues under the proposed label would probably be killed. Although the proposed seasonal restrictions on the West Coast would reduce the risk to migratory wigeon in that area, the small non-migratory resident wigeon population would remain at some risk.40 More significantly, wigeon outside the West Coast would also remain at risk. Contrary to Ciba-Geigy's assertion that these wigeon do not feed on turf (CG Reply at 108-15), the record demonstrates that the wigeon is an "opportunistic omnivore" and could easily adapt to feeding on turf outside the West Coast.41

37 Id.

38 Many of Ciba-Geigy's avian experts direct their overall conclusions on risk primarily (if not exclusively) to diazinon's effects on bird populations. See, e.g., Tr. 3773– 74 (Klem); CG Ex. 44 (pp. 4-5) (Jaber); CG Ex. 112 (p. 55) (Dickson); CG Ex. 125 (p. 47) (Best), CG Ex. 130 (pp. 3–5) (Klem); CG Ex. 99 (pp. 6–7) (Prince). Although some also state that the risks under the proposed label would be "minimal" or "insignificant," the repeated references to bird populations suggest that population effect was the benchmark used for many of these conclusions. For other witnesses, the primary concern apparently was the prevention of large kills or widespread mortality. Dr. Ringer, who reviewed Ciba-Geigy's field studies, concluded that its proposed label was "adequate" and posed no "undue threat." CG Ex. 33, Attachment E (pp. 2, 8). These rather vague assertions are put in context by his more specific and limited conclusion that the proposed label would "protect against catastrophic acute toxicity die-offs.” Id. (p. 8). Thus, there is no uniform opinion among Ciba-Geigy's experts that the proposed label would reduce the risk to individual birds to negligible, trivial, or low levels as Ciba-Geigy contends.

39 In re Stevens Industries, Inc. (DDT), 37 Fed. Reg. 13369, 13373 (July 7, 1972). 40 Tr. 1513. Ciba-Geigy witness Dr. DiGuilo recommended that diazinon exposure to breeding wigeon in Washington be assessed. CG Ex. 70 (p. 13).

41 Initial Decision at 50 n.137. Ciba-Geigy's attempt (CG Reply at 114) to limit this finding to the wigeon breeding season is unavailing. See Tr. 3835-37.