DocketNumber： 05-75255, 05-76807
Judges： Pregerson, Ikuta, Moskowitz
Filed Date： 9/19/2008
Status： Precedential
Modified Date： 11/5/2024
Opinion by Judge PREGERSON; Partial Concurrence and Partial Dissent by Judge IKUTA.
Petitioners are two environmental groups challenging the Environmental Protection Agency’s (“EPA”) establishment of tolerances for seven pesticides used mostly on fruit and vegetable crops. We grant the petition in part, deny it in part, and remand to the EPA.
BACKGROUND
I.
The EPA regulates pesticides under two statutes: the Federal Insecticide, Fungicide, and Rodenticide Act (“FIFRA”), 7 U.S.C. § 136-136y, and the Federal Food, Drug, and Cosmetic Act (“FDCA”), 21 U.S.C. § 346a.
Under FIFRA, pesticides sold in the United States must be registered by the EPA. 7 U.S.C. § 136a. The EPA may not register a pesticide unless the pesticide will perform its intended function without causing “any unreasonable adverse effects on the environment.” 7 U.S.C. § 136a(e)(5)(C).
The FDCA authorizes the EPA to set “tolerances” for pesticide residues in food.
Tolerances are established by rulemak-ing. 21 U.S.C. § 346a(d). Pesticide manufacturers initiate the rulemaking process by petitioning the EPA to set a tolerance. 21 U.S.C. § 346a(d)(l). The EPA publishes notices of these petitions in the Federal Register. 21 U.S.C. § 346a(d)(3). After reviewing a petition and any comments received on it, the EPA may issue: (1) a final rule establishing a tolerance, (2) a proposed rule, or (3) an order denying the petition. 21 U.S.C. § 346a(d)(4).
Once the EPA takes final action on a petition, any affected party has sixty days to file objections with the EPA and to seek an evidentiary hearing on those objections. 21 U.S.C. § 346a(g)(2). The EPA’s final order in response to those objections is subject to judicial review. 21 U.S.C. § 346a(h)(l).
II.
In 1996, Congress amended the FDCA by enacting the Food Quality Protection Act (“FQPA”), Pub.L. No. 104-170, 110 Stat. 1489. One of the key provisions of the FQPA requires the EPA to give special consideration to risks posed to infants and children when establishing pesticide tolerances. Specifically, the FQPA directs the EPA to use “an additional
Unfortunately, the FQPA does not define “reliable data.” The dispute before us turns on the definition of this term.
Between December 2001 and April 2002, the EPA published seven regulations establishing tolerances for the pesticides ace-tamiprid, fenhexamid, halosulfuron-methyl, isoxadifen-ethyl, mepiquat, pymetrozine, and zetacypermethrin used on many foods, including fruits, vegetables, nuts, cereal grains, milk, and eggs.
The EPA did not apply the presumptive lOx child safety factor to any of these seven pesticides. The EPA reduced the lOx child safety factor to 3x for four of the pesticides (acetamiprid, fenhexamid, isoxa-difen-ethyl, and pymetrozine). For the remaining three pesticides (halosulfuron, me-piquat, and zeta-cypermethrin), the EPA did not apply a child safety factor at all.
In 2002, Natural Resources Defense Council, Inc. (“NRDC”) filed objections to each of the tolerances based on the EPA’s decision to reduce or remove the child safety factor.
On August 10, 2005, the EPA issued its final order rejecting NRDC’s objections. See Order Denying Objections to Issu-ances of Tolerances, 70 Fed.Reg. 46,706 (Aug. 10, 2005) (hereinafter “Final Order”). The Final Order upheld the 3x child safety factor for acetamiprid, fenhex-amid, isoxadifenethyl, and pymetrozine, and a lx child safety factor (i.e., no safety factor) for halosulfuron, mepiquat, and ze-taeypermethrin. Id. at 46,711-13, 46,736.
NRDC and the Northwest Coalition for Alternatives to Pesticides (“NCAP”) each petitioned for review of the Final Order.
JURISDICTION
We have jurisdiction under the FDCA, which provides that “any person who will be adversely affected” by an EPA order denying tolerance objections may file a petition for review in the Court of Appeals within 60 days of the EPA’s order. 21 U.S.C. § 346a(h)(l).
NRDC filed administrative objections to the tolerances in 2002. The EPA published its Final Order denying those objections on August 10, 2005. Within 60 days, NRDC filed a petition for review in the Second Circuit, and NCAP filed a petition for review in this court. The Judicial Panel on Multidistrict Litigation transferred NRDC’s petition for review to this court pursuant to 28 U.S.C. § 2112, and NRDC and NCAP filed joint briefs on appeal.
STANDARD OF REVIEW
The FDCA does not establish a standard for reviewing pesticide tolerances that are established without a public evi-dentiary hearing. See 21 U.S.C. § 346a(h)(2). In the absence of such a standard, our review is governed by section 706 of the Administrative Procedure Act (“APA”). See 5 U.S.C. § 706; City of Sausalito v. O’Neill, 386 F.3d 1186, 1205-06 (9th Cir.2004).
Under the APA, we must set aside an agency’s decision if it is “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706(2)(A). An agency decision would normally be arbitrary and capricious if “the agency has relied on factors which Congress has not intended it to consider, entirely failed to consider an important aspect of the problem, offered an explanation for its decision that runs counter to the evidence before the agency, or is so implausible that it could not be ascribed to a difference in view or the product of agency expertise.” Mfrs’ Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983).
Although we will “uphold a decision of less than ideal clarity if the agency’s path may be reasonably discerned,” we “may not supply a reasoned basis for the agency’s action that the agency itself has not given.” Id. (citations omitted).
ANALYSIS
Petitioners challenge seven tolerances (acetamiprid, fenhexamid, halosulfuron-methyl, isoxadifen-ethyl, mepiquat, pyme-trozine, and zeta-cypermethrin) based on the EPA’s failure to gather drinking water data as part of its exposure analysis. Petitioners also challenge three tolerances (acetamiprid, mepiquat, and pymetrozine) based on the absence of certain toxicity studies, and based on the EPA’s failure to explain why it had reliable data in the absence of those toxicity studies.
I.
The FQPA directs the EPA to use an additional tenfold margin of safety to take into account completeness of data with respect to pesticide “exposure to infants and children.” 21 U.S.C. § 346a(b)(2)(C) (emphasis added).
As required by the FQPA, the EPA monitors whether infants and children will be exposed to unsafe levels of pesticides in drinking water. However, the EPA lacks sufficient drinking water exposure data for several pesticides, including the ones at issue in this case. Therefore, the EPA used computer modeling to determine drinking water exposure for the seven pesticides. Because the modeling results revealed little-to-no risk of pesticide exposure in drinking water, the EPA reduced or removed the child safety factor for each of the pesticide tolerances.
Petitioners argue that because the EPA used modeling instead of actual sampling data to determine drinking water exposure to these pesticides, the EPA does not have “reliable data” justifying its use of a lower than lOx margin of safety for the pesticide tolerances. See 21 U.S.C. § 346a(b)(2)(C). In short, Petitioners argue that modeling results can never constitute reliable data within the meaning of the FQPA.
We disagree. Petitioners have presented no evidence that modeling does not yield reliable data. There is nothing inherently unreliable about the use of models in scientific assessments. See, e.g., Small Refiner Lead Phase-Down Task Force v. EPA 705 F.2d 506, 535 (D.C.Cir.1983) (“[Administrative agencies have undoubted power to use predictive models.”). Moreover, because of the difficulty in sampling the entire nation’s water supply, modeling is necessary to determine whether drinking water has been contaminated by pesticides. Topography, geology, and hydrology differ greatly across the nation and constantly change. In many cases, computer modeling can more accurately incorporate these elements and provide
Therefore, we are unwilling to adopt the narrow construction of “reliable data” that Petitioners advance. Although the FQPA does not define “reliable data,” we are confident that modeling results can satisfy this statutory requirement.
Furthermore, in this case, the EPA provided a specific, detailed explanation of why its drinking water exposure models do yield reliable data. The Final Order states:
Lack of comprehensive drinking water (DW) monitoring data. NRDC contends that, because EPA used a model for calculating drinking water exposure, EPA does not have, as a definitional matter, “reliable data” for choosing a factor different than the 10X default value. Similar comments were made during the development of EPA’s Children’s Safety Policy. This issue was addressed at length in the response to the imida-eloprid objections. That response is incorporated herein and is summarized below.
Although the availability of drinking water monitoring data has increased dramatically in the last several years, EPA still finds it necessary to rely for most pesticides upon various exposure models to estimate exposure levels in drinking water. These models are based on generic data regarding fate and transport of pesticides in the environment, and they operate by combining this generic data with pesticide-specific data on chemical properties to estimate exposure. EPA has primarily used its drinking water models to “screen” those pesticides that may pose unacceptable risks due to exposures in drinking water from pesticides not likely to result in such exposures. To accomplish this goal, the models are based on data from studies at sites that are highly vulnerable to runoff of pesticides to surface water or leaching of pesticides to ground water. If a pesticide fails this conservative (health-protective) screen, EPA would investigate whether the model is significantly overstating the residue levels that actually occur.
EPA has developed models for estimating exposure in both surface water and ground water. EPA uses a two-tiered approach to modeling pesticide exposure in surface water. In the initial tier, EPA uses the FQPA Index Reservoir Screening Tool (FIRST) model. FIRST replaces the Generic Estimated Environmental Concentrations (GEN-EEC) model that was used as the first tier screen by EPA from 1995-1999. If the first tier model suggests that pesticide levels in water may be unacceptably high, a more refined model is used as a second tier assessment. The second tier model is actually a combination of the models, Pesticide Root Zone Model (PRZM) and the Exposure Analysis Model System (EXAMS). For estimating pesticide residues in groundwater, EPA uses the Screening Concentration In Ground Water (SCI-GROW) model. Currently, EPA has no second tier groundwater model.
Whether EPA assesses pesticide exposure in drinking water through monitoring data or modeling, EPA uses the higher of the two values from surface and ground water in assessing overall exposure to the pesticide. In most cases, pesticide residues in surface wa*1050 ter are significantly higher than in ground water.
In the Imidacloprid Order, EPA analyzed each of its water models extensively. Based on the results of design characteristics of the models, outside peer review of the models, validation of the models, and comparison between the models’ predictions and extensive water monitoring data, EPA concluded that the models are based on reliable data and will produce estimates that are unlikely to underestimate exposure to pesticides in drinking water.
Accordingly, EPA reaffirms its earlier conclusion that its drinking water models provide a reliable basis for finding that exposure to pesticide residues in water are not underestimated.
Final Order at 46,726-29 (internal citations omitted) (emphasis added). Thus, the EPA addressed at length the reliability of its models. The EPA explained that the models yield conservative data because the models incorporate the higher of the two values from surface and ground water in assessing the overall risk of exposure to the pesticides. Petitioners have not established that the EPA’s explanation regarding the reliability of its models is faulty or suspect.
Accordingly, we conclude that the computer modeling used by the EPA to calculate the safety of drinking water was neither contrary to law nor arbitrary and capricious.
II.
In addition to NRDC’s challenge to the completeness of exposure data, Petitioners also raise two arguments regarding whether reliable toxicity data supported a reduction in the child safety factor.
A.
NRDC first challenges the EPA’s failure to wait for the results of certain developmental neurotoxicity studies (“DNT studies”) before establishing tolerances for acetamiprid, mepiquat, and pymetrozine. The EPA specifically required the registrants of these pesticides to conduct DNT studies. However, the EPA did not wait for the results of the DNT studies before making its tolerance determinations.
Petitioners contend that DNT studies are essential for assessing pesticide effects. Petitioners explain that DNT studies are the best available studies for examining neurological effects in children because DNT studies are more sensitive than other studies, address areas that are not covered by other studies, and sometimes reveal developmental harm that is significant enough to require more stringent regulation. Petitioners argue that the EPA acted arbitrarily and capriciously by requesting DNT studies but failing to wait for the results of those studies before removing the child safety factor.
Petitioners’ position, however, is inconsistent with the Supreme Court’s decision in National Ass’n of Home Builders v. Defenders of Wildlife, — U.S. -, 127
Thus, Home Builders appears to foreclose Petitioners’ argument. Even though the EPA initially thought that the DNT studies were needed, the EPA later determined that the DNT studies were not needed. Under Home Builders, the EPA’s later determination does not automatically render the EPA’s Final Order arbitrary and capricious. Notably, Petitioners have not identified any procedures that were violated when the EPA decided not to wait for the requested DNT studies before setting the tolerances. Moreover, the Final Order provided a reasoned explanation of why DNT studies are not actually required in every case where there is a reduction of the lOx child safety factor. Final Order at 46,711-12, 46,723-24.
Therefore, we reject Petitioner’s argument that the EPA acted arbitrarily and capriciously by establishing the pesticide tolerances for acetamiprid, mepiquat, and pymetrozine before receiving the DNT studies. While Petitioners’ concerns about the EPA’s apparent change of heart are understandable, on such questions of scientific judgment, we are required to defer to the EPA.
B.
NRDC also disputes that the EPA’s decision to deviate from the lOx presumptive child safety factor imposed by Congress was supported by reliable data establishing, with reasonable certainty, that the lower margin of safety for acetamiprid, mepiquat, and pymetrozine would not harm infants and children. 21 U.S.C. § 346a(b)(2)(A), (C). Even though the EPA was not necessarily required to wait for the results of the DNT studies, the decision to reduce the lOx child safety factor for these three pesticides must still be based on reliable data. Moreover, the EPA must tell the public why its data are reliable.
The problem in this case is that the EPA’s Final Order is vague, making it impossible for us to determine whether the EPA’s deviations from the lOx child safety factor for acetamiprid, mepiquat, and py-metrozine were in fact supported by reliable data, even in the absence of the DNT studies. For example, with respect to py-metrozine, the Final Order states:
NRDC challenged a December 27, 2001 action establishing tolerances for pyme-trozine on cotton seed, cotton gin byproducts, the fruiting vegetables crop group, the cucurbit vegetables crop group, the leafy vegetables crop group (except Brassiea), head and stem Brassiea, leafy Brassiea, turnip greens, dried hops, and pecans. (66 FR 66786, December 27, 2001). Given pymetro-zine’s exposure pattern and toxicological characteristics, EPA determined that pymetrozine potentially presented acute, chronic, short-term, and cancer risks and EPA quantitatively assessed these risks in making its safety determination. (66 FR at 66791-66792). All of these risks were found to be below the Agency’s level of concern. (Id.). Although a DNT study was outstanding, EPA determined that the additional 10X chil*1052 dren’s safety factor could generally be reduced to 3X because the toxicological data showed no evidence of greater sensitivity to the young and there was no evidence of abnormalities in the development of the fetal nervous system. (64 FR 52438, 52444, September 29, 1999).
Final Order at 46,711.
The Final Order recites that the “toxicological data” showed no evidence of increased sensitivity for developing fetuses or the “young.” However, the Final Order does not explain the connection between the toxicological data and the 3x margin of safety selected by the EPA in place of the lOx standard. This same critique applies to the Final Order’s reasoning with respect to acetamiprid and mepiquat. As a result, it is entirely unclear why the EPA chose safety factors of 3x for pymetrozine and acetamiprid, and lx for mepiquat, as opposed to 4x or 5x or 8x or 9x. As far as we can tell from the record, it appears that the EPA chose these lower safety levels arbitrarily — to acknowledge certain concerns about each pesticide, but with no specific evidence that these lower safety factors would actually account for the risks to infants and children.
In sum, the Final Order does not provide enough information to demonstrate a rational connection between the factors that the EPA examined and the conclusions it reached. See State Farm, 463 U.S. at 43, 103 S.Ct. 2856. We are therefore unable to determine whether there was reliable data supporting the EPA’s reductions of the lOx child safety factor. Although on remand the EPA may be able to explain why a 3x or lx factor is appropriate, we cannot conclude it is appropriate on this record. To do so would abdicate our responsibility to ensure compliance with Congress’s expressed desire to have a presumptive lOx child safety factor. Accordingly, we remand the tolerance regulations for acetamiprid, mepiquat, and pyme-trozine to the EPA. See, e.g., State Farm, 463 U.S. at 57, 103 S.Ct. 2856 (remanding where the agency failed to supply the requisite “reasoned analysis”); Safe Food & Fertilizer v. EPA, 350 F.3d 1263, 1271 (D.C.Cir.2003) (remanding for the EPA to provide “further explanation” of an exemption level for chromium found in zinc fertilizers).
The EPA and the dissent argue that Petitioners did not raise this argument
CONCLUSION
Because the EPA failed to adequately explain the basis for its deviations from the lOx child safety factor for acetamiprid, mepiquat, and pymetrozine, we grant the petitions for review in part and remand to the EPA for further proceedings consistent with this opinion. On all other issues, we deny the petitions for review.
Each side to bear its own costs on appeal.
GRANTED IN PART, DENIED IN PART, AND REMANDED.
. This margin, of safety is "additional” to two safety factors that the EPA has long used to account for other uncertainties: a tenfold factor for interspecies differences (to account for the possibility that people are more susceptible than animals studied in laboratory experiments), and a tenfold safety factor for intra-species differences (to account for the wide range of sensitivities in the human population).
. For the EPA’s individual tolerance rulings, see Acetamiprid; Pesticide Tolerance, 67 Fed. Reg. 14,649 (Mar. 27, 2002); Fenhexamid; Pesticide Tolerance, 67 Fed.Reg. 19,114 (Apr. 18, 2002); Halosulfuron-methyl; Pesticide Tolerance, 66 Fed.Reg. 66,333 (Dec. 26,
. NRDC did not request an evidentiary hearing on its objections.
. Petitioners also initially challenged the tolerances for halosulfuronmethyl and zeta-cy-permethrin on these grounds. They now acknowledge, however, that subsequent agency actions have rendered those challenges moot.
. See FIFRA Scientific Advisory Panel, A Set of Scientific Issues Being Considered by the Agency in Connection with Estimating Drinking Water Exposure as a Component of Dietary Risk Assessment, 4-7 (Dec. 1997), available at http://epa.gov/scipoly/sap/meetings/ 1997/december/finaldec.pdf.
. Our conclusion, of course, is not meant to suggest that the EPA’s models are flawless. Petitioners have raised some concerns about the EPA's water exposure models that may warrant further scrutiny. To give one example, the EPA first performs a model-based water exposure test using conservative criteria; if the test results reveal unacceptable exposure levels, the EPA runs a second test to ensure that the first result was accurate. The EPA does not run a second test, however, if the initial test results reveal acceptable exposure levels. Although we are required to defer to the EPA’s technical expertise in this area, see, e.g., Envtl Def. Ctr., Inc. v. EPA, 344 F.3d 832, 869 (9th Cir.2003), we note that the EPA may wish to revisit this methodology.
. While we are aware that the scope of our review under the APA is narrow, we are also mindful that we should not endorse the removal of the statutorily presumed child safety factor if we cannot determine that the EPA's conclusions are rationally supported. As Judge J. Skelly Wright of the D.C. Circuit elucidated, there is an important difference between the depth of our review of an agency’s action and the scope of that review:
Judicial review of agency action requires that the reviewing court consider whether the [agency’s] decision was based on a consideration of the relevant factors and whether there has been a clear error of judgment. As noted by this court in International Ladies Garment Workers Union v. Donovan, 722 F.2d 795, 815 (D.C.Cir.1983), cert. denied, 469 U.S. 820, 105 S.Ct. 93, 83 L.Ed.2d 39 (1984), such review "is not merely perfunctory. We are to engage in a’searching and careful' inquiry, the keystone of which is to ensure that the [agency] engaged in reasoned decisionmaking."
This formulation points to an acknowledged disparity between the depth of our review and the ultimate scope of that review: Although the ultimate scope may be narrow, the depth must be sufficient for us to be able to comprehend the agency’s handling of the evidence cited or relied upon. The purpose of this in-depth review is to educate ourselves so that we can properly perform our reviewing function: determining whether the agency's conclusions are rationally supported. For, although data interpretation and analysis are functions that often lie within an agency's realm of expertise, it is our duty to review those functions to ascertain whether the agency's actions were complete, reasoned, and ade
Center for Auto Safety v. Peck, 751 F.2d 1336, 1373 (D.C.Cir.1985) (Wright, J., dissenting) (internal citations and quotation marks omitted). See also Riverkeeper, Inc. v. United States Environmental Protection Agency, 475 F.3d 83, 103-04 (2d Cir.2007) ("In a technical area of this sort, it is difficult for judges or interested parties to determine the propriety of the Agency's action without a justification for the action supported by clearly identified substantial evidence whose import is explained.”).