Ranchers Cattlemen Action Legal Fund United Stockgrowers of America v. United States Department of Agriculture

FOR PUBLICATION UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT RANCHERS CATTLEMEN ACTION  LEGAL FUND UNITED STOCKGROWERS OF AMERICA, Plaintiff-Appellee, No. 05-35264 v. UNITED STATES DEPARTMENT OF  D.C. No. CV 05-006 RFC AGRICULTURE, Animal and Plant OPINION Health Inspection Service; MIKE JOHANNS, in his capacity as the Secretary of Agriculture, Defendants-Appellants.  Appeal from the United States District Court for the District of Montana Richard F. Cebull, District Judge, Presiding Argued and Submitted July 13, 2005—Seattle, Washington Filed July 25, 2005 Before: A. Wallace Tashima, Richard A. Paez, and Consuelo M. Callahan, Circuit Judges. Opinion by Judge Tashima 8731 RANCHERS CATTLEMEN v. USDA 8735 COUNSEL Mark B. Stern, Civil Division, U.S. Department of Justice, Washington, DC, for the defendants-appellants. Russell S. Frye, Frye Law PLLC, Washington, DC, for the plaintiff-appellee. Maureen E. Mahoney, Latham & Watkins LLP, Washington, DC, for amicus curiae Government of Canada. Gregory G. Garre, Hogan & Hartson LLP, Washington, DC, for amici curiae National Cattlemen’s Beef Association, American Farm Bureau Federation, National Pork Producers Council, 29 State Cattlemen’s Associations, 18 State Farm Bureaus, and 9 Individual Cattle Producers. Sarah Weinstein, Mayer, Brown, Rowe & Maw LLP, Palo Alto, California, for amicus curiae Alberta Beef Producers. Michael B. Gillett, McElroy Law Firm, PLLC, Seattle, Wash- ington, for amicus curiae Easterday Ranches, Inc. Joseph O. Click, Blank Rome LLP, Washington, DC, for amici curiae Canadian Cattlemen’s Association and Its Affili- ated Organizations. Jonathan L. Abram, Hogan & Hartson, Washington, DC, for amici curiae American Meat Institute, North American Meat Processors, Southwestern Meat Association, Eastern Meat 8736 RANCHERS CATTLEMEN v. USDA Packers Association, American Association of Meat Proces- sors, National Restaurant Association, and United Food and Commercial Workers. John O’Brien, Kerr Brosseau Bartlett O’Brien, LLC, Denver, Colorado, for amicus curiae Pioneer, Inc. Gregg Spyridon, Spyridon, Koch, Palermo, & Dornan, LLC, Metairie, Louisiana, for amici curiae the Camelid Alliance, et al. Alan Charles Raul, Sidley Austin Brown & Wood LLP, Washington, DC, for amicus curiae Tyson Foods, Inc. David A. Domina, Domina Law pc llo, Omaha, Nebraska, for amici curiae 67 National, State, and Local Consumer and Research Groups, Public Interest Organizations, Farm and Ranch Organizations, and Local and Private Organizations. Christian D. Tweeten, Chief Civil Counsel, Montana Attorney General, Helena, Montana, for amici curiae States of Mon- tana, Connecticut, Nevada, New Mexico, North Dakota, and South Dakota. OPINION TASHIMA, Circuit Judge: We must decide whether the district court erred in issuing a preliminary injunction prohibiting the implementation of a regulation of the United States Department of Agriculture (“USDA”) permitting the resumption of the importation of Canadian cattle into the United States. We conclude that it did and therefore reverse the district court. At the heart of this case lies a relatively new cattle disease caused by the practice of feeding cows, herbivores by nature, the brains and other central nervous system tissues of other RANCHERS CATTLEMEN v. USDA 8737 cows. Technically known as Bovine Spongiform Encephalop- athy (“BSE”), this disease, popularly known as mad cow dis- ease, has spread from farms in England to 25 countries around the world since its discovery in 1986. As BSE spread throughout the globe during the past 20 years, USDA instituted a policy of barring the importation of ruminants1 and ruminant products from countries where BSE was known to exist. In a final rule entitled Bovine Spongiform Encephalopathy: Minimal Risk Regions and Importation of Commodities; Final Rule and Notice, 70 Fed. Reg. 460 (Jan. 4, 2005) (the “Final Rule”), USDA relaxed this longstanding practice, allowing limited ruminant imports from Canada, despite the fact that two cases of BSE had been found in Can- ada at the time. Plaintiff-Appellee, Ranchers Cattlemen Action Legal Fund United Stockgrowers of America (“R-CALF”), successfully blocked the implementation of the Final Rule, convincing the court below to find the rule arbitrary and capricious under the Administrative Procedure Act (“APA”), 5 U.S.C. § 706(2), and to issue a preliminary injunction prohibiting its enforce- ment. See Ranchers Cattlemen Action Legal Fund United Stockgrowers of Am. v. United States Dep’t of Agric., 359 F. Supp. 2d 1058 (D. Mont. 2005) (“R-CALF I”). Because we conclude that the district court applied an incorrect legal stan- dard, we reverse.2 1 Ruminants are hoofed mammals generally defined by their four- chambered stomachs and their practice of chewing a cud consisting of regurgitated, partially digested food. Ruminants include cattle, sheep, goats, deer, giraffes, camels, llamas, and okapi, among others. 2 On July 14, 2005, after the completion of briefing and oral argument we issued a stay of the preliminary injunction pending the resolution of this appeal. See Fed. R. App. P. 8(a). 8738 RANCHERS CATTLEMEN v. USDA I. FACTUAL AND PROCEDURAL BACKGROUND A. Bovine Spongiform Encephalopathy BSE was first diagnosed in England in the late 1980s. This new disease spread rapidly, infecting thousands of English cattle and eventually reaching countries all over the globe. Although the disease has since been largely contained, it con- tinues to persist, and it resides at the center of the current law- suit. BSE is a species of Transmissible Spongiform Encephalop- athy (“TSE”), a family of degenerative neurological diseases that affects a wide range of animals, including sheep, goats, and deer, as well as humans. Although there remains some dispute, it is widely believed that BSE and other TSEs are caused by prions, abnormally shaped and extremely hardy proteins that were only recently discovered. TSEs have a debilitating neurological impact on their vic- tims. After an incubation period of months or years, the dis- eases create myriad tiny holes in the brain, slowly deteriorating their victims’ mental and physical abilities until death eventually results. In cattle, BSE has an incubation period of two to eight years, during which time the infected animal shows no outward sign of the illness. Once the disease progresses, however, infected cattle begin showing symptoms within two to three months. These symptoms can include ner- vousness or aggression, abnormal posture, impaired coordina- tion, decreased milk production, and loss of body condition despite continued appetite. At the height of the BSE epidemic in the United Kingdom, tens of thousands of cattle were confirmed to have the dis- ease, and by some estimates the number of infected cattle in the United Kingdom may have reached into the millions. All told, there have been more than 187,000 confirmed cases of RANCHERS CATTLEMEN v. USDA 8739 BSE in cattle worldwide, over 95 percent of which have occurred in the United Kingdom. Epidemiological investigations in England quickly deter- mined that BSE was likely spread through cattle feed that was infected with the BSE agent. The blame for the contaminated feed fell squarely on the practice, common in Europe at the time, of creating high-protein cattle feed through the “recy- cling” of otherwise unusable cattle parts. This process is known as “rendering,” and involves placing animal protein in large tanks and cooking at temperatures high enough to kill most microorganisms.3 Although the rendering process is able to eliminate most bacterial and viral diseases, the BSE agent is resistant enough to heat and other sterilization processes to withstand the conversion into feed. Infected tissue from a sin- gle infected cow, when rendered into cattle feed, could there- fore be fed to hundreds of cattle, exposing them all to the possibility of infection. Several years after the discovery of BSE, the disease became a matter of much more serious concern. In 1996, the British government announced that a new form of TSE in humans, variant Creutzfeldt-Jakob Disease (“vCJD”), was likely caused by human consumption of cattle products that were contaminated with the BSE agent. To date, only approx- 3 Rendering continues to this day in the United States, where approxi- mately 50 billion pounds of tissue from dead animals are converted into animal feed each year. The breadth of the practice at one Baltimore ren- dering facility has been reported to include: Bozeman, the Baltimore City Police Department quarter horse who died last summer in the line of duty. . . . A baby circus ele- phant who died while in Baltimore this summer. Millions of tons of waste meat and inedible animal parts from the region’s super- markets and slaughterhouses. Carcasses from the Baltimore zoo. The thousands of dead dogs, cats, raccoons, possums, deer, foxes, snakes, and the rest that local animal shelters and road-kill patrols must dispose of each month. Van Smith, What’s Cookin?, Baltimore City Paper, Sept. 27, 1995. 8740 RANCHERS CATTLEMEN v. USDA imately 150 cases of vCJD have been identified worldwide, the vast majority of which occurred in England during the height of its BSE epidemic. Although vCJD has been diag- nosed in two people in North America, in both cases the dis- ease is believed to have been contracted in England; no case of vCJD has ever been linked to North American beef.4 Because BSE is a relatively new disease, and because prions are a relatively recent scientific discovery, the state of knowledge surrounding BSE is somewhat incomplete. Efforts to understand the disease fully have been hampered because current testing methodology is not particularly effective in identifying it. No live animal test for BSE exists, meaning that cows must be slaughtered before they can be tested. In addi- tion, the tests that do exist are unable to detect the disease during the vast majority of the time a cow is infected. The earliest point at which current tests can detect the disease is two to three months before an animal starts showing clinical signs of infection. BSE has an incubation period that lasts for four to five years on average, however, during which the ani- mal carries the disease but shows no outward symptoms. Given these testing limitations, there remain a number of open public health questions surrounding BSE, in particular concerning the means through which the disease can be trans- mitted. The only documented method of BSE transmission is through the consumption of feed contaminated with the BSE agent. Some research involving both BSE and other TSEs, however, suggests that BSE may be transmitted through means other than contaminated feed. For example, in experi- ments on sheep, mice, and hamsters, both BSE and scrapie, a TSE disease that affects sheep, were transmitted through 4 The single known case of vCJD in the United States occurred in a Flor- ida woman who was born in England in 1979. It is believed she was exposed to BSE before she moved to the United States in 1992. Similarly, the single case of vCJD in Canada occurred in a man who had stayed in the United Kingdom on multiple trips. RANCHERS CATTLEMEN v. USDA 8741 whole blood transfusion. At least one case of vCJD is also believed to have been transmitted through human blood trans- fusion. Other studies have suggested that prions can be exchanged through saliva, while still others suggest that BSE may be transmitted maternally. Despite the highly infectious nature of the BSE agent, evi- dence suggests that meat from cows infected with BSE may be safely consumed by humans because BSE does not occur in all parts of its host. Specifically, the BSE agent appears not to exist in muscle tissue of cattle. Rather, the disease is gener- ally confined to the central nervous system — the brain, spi- nal cord, eyes, dorsal root ganglia, and trigeminal ganglia5 — although it has also been found in the tonsils and distal ilium, a part of the small intestine, of cattle. Research on other TSEs, however, calls into question whether the BSE agent is truly limited to these tissues. Specifically, some research has sug- gested that sheep infected with scrapie may have prions in their muscle tissue. Despite the fact that it has only been known to exist for 20 years, the geographic range of BSE is substantial. From England, it has spread to cattle in most of Europe, as well as in the Middle East, Japan, and Canada. 9 C.F.R. § 94.18(a)(1) (2003). As of the date of the district court’s opinion, however, BSE had never occurred in a cow native to the United States. That changed on June 24, 2005, when the Secretary of Agri- culture announced that a cow in Texas had tested positive for BSE. Statement by Dr. John Clifford Regarding the Epidemi- ological Investigation into the recently confirmed BSE case (June 29, 2005), available at http://www.aphis.usda.gov/lpa/ issues/bse/bse.html. A subsequent investigation revealed that the cow was born in the United States approximately 12 years ago. 5 Trigeminal ganglia are clusters of nerve cells connected to the brain that lie close to the exterior of the skull. Dorsal root ganglia are clusters of nerve cells attached to the spinal cord and contained within the bones of the vertebral column. 8742 RANCHERS CATTLEMEN v. USDA B. United States Regulation of BSE The federal government has implemented a number of safety measures to minimize the threat of BSE to U.S. citizens and livestock. These precautions consist of an interlocking regulatory framework overseen by three different federal agencies. First and foremost, since 1997, the Food and Drug Administration (“FDA”) has overseen a feed ban that prohib- its the feeding of ruminant protein to other ruminants. See 21 C.F.R. § 589.2000 (2005). Such feed bans are generally the first line of defense against the spread of BSE, and they have been highly effective in other countries. The prevalence of BSE in the United Kingdom, for example, dropped drastically after it implemented its feed ban. Critics, however, question whether the FDA feed ban is truly effective. See, e.g., Thomas O. McGarity, Federal Regu- lation of Mad Cow Disease Risks, 57 Admin. L. Rev. 289, 307 (2005). Given the highly infectious and resilient nature of the BSE agent, these critics argue that the FDA feed ban has “gaps” that could result in the use of feed derived from ren- dered cattle protein as feed for cattle. For example, cattle are allowed to be fed human “plate waste” from establishments such as amusement parks, despite the fact that this plate waste may contain beef products. In addition, the feed ban allows rendered cattle protein to be fed to non-ruminants, such as pigs and chickens. Thus, BSE could be spread through misla- beled feed or through misfeeding on a farm. Finally, waste from the floor of chicken coops is commonly scooped up and fed to cattle; uneaten chicken feed or chicken droppings that contain the BSE agent could therefore be fed to cattle via this procedure. An agency within USDA, Food Safety and Inspection Ser- vices (“FSIS”), oversees a second line of defense against BSE. FSIS promulgates regulations to ensure that the nation’s food supply of meat, eggs, and poultry is safe. See http:// www.fsis.usda.gov/About_FSIS/index.asp. These regulations RANCHERS CATTLEMEN v. USDA 8743 restrict certain cattle parts from being incorporated into the human food supply. For example, FSIS regulations prohibit the use of “downer” cattle6 as human food because inability to stand is a common BSE symptom. 9 C.F.R. § 309.2 (2005). FSIS regulations also prohibit those cattle parts that have demonstrated BSE infectivity, known as specified risk materi- als (“SRMs”), from being used in human food.7 9 C.F.R. § 310.22 (2005). Finally, FSIS regulations prohibit certain methods of slaughter and butchering thought to increase the risk of contaminating meat with central nervous system tissues.8 6 Non-ambulatory or “downer” cattle are cattle that “cannot rise from a recumbent position or that cannot walk.” 9 C.F.R. § 309.2(b) (2005). FSIS banned these cattle from the human food supply because “surveillance data from European countries in which BSE has been detected indicate that non-ambulatory cattle are among the animals that have a greater inci- dence of BSE than other cattle.” Prohibition of the Use of Specified Risk Materials and Requirements for the Disposition of Non-Ambulatory Dis- abled Cattle, 69 Fed. Reg. 1862, 1862 (January 12, 2004) (“FSIS SRM Rule”). 7 Because BSE infectivity spreads as a cow ages, current regulations define only the distal ilium and tonsils of all cattle to be SRMs. 9 C.F.R. § 310.22(a) (2005). The brain, spinal cord, and other central nervous sys- tem components are only considered to be SRMs in cattle of 30 months of age and older. Id. 8 Specifically, FSIS regulations prohibit the use of “air-injection captive bolt stunning,” a process through which a metal bolt and compressed air are driven into the cranium of cattle, because the practice poses a risk of contaminating edible meat with central nervous system tissue. See 9 C.F.R. § 310.13(a)(2)(iv)(C) (2005). The regulations also prohibit the use of “Advanced Meat Recovery” systems and the labeling of “mechanically separated beef” as meat. See FSIS SRM Rule, 69 Fed. Reg. at 1866. The former “is a technology that enables processors to remove the attached skeletal muscle tissue from livestock bones without incorporating a signif- icant amount of bone or bone product into the final meat product.” Id.; see also Meat Produced by Advanced Meat/Bone Separation Machinery and Meat Recovery (AMR) Systems, 69 Fed. Reg. 1874, 1876 (Jan. 12, 2004). The latter “is a paste-like and batter-like meat product produced by forcing [beef] bones with attached edible meat under high pressure through a sieve.” See http://www.fsis.usda.gov/oa/pubs/lablterm.htm. 8744 RANCHERS CATTLEMEN v. USDA Another branch of USDA, Animal and Plant Health Inspec- tion Services (“APHIS”), provides the final link in the regula- tory framework. APHIS promulgates regulations designed to protect the United States from the introduction of BSE from other countries. To achieve this goal, until the Final Rule was promulgated, APHIS banned the importation of all ruminants and ruminant products from countries where BSE was known to exist. See 9 C.F.R. §§ 93.401, 94.18 (2003). APHIS has also been actively involved in the development of international guidelines to fight the spread of BSE. In this role, APHIS works with the Office International des Epizoo- ties (“OIE”), the organization recognized by the World Trade Organization as responsible for the development and periodic review of standards, guidelines, and recommendations with respect to animal health and “zoonoses” (diseases that are transmissible from animals to humans). C. Factual Background Early this year, APHIS announced its decision to relax its ban on the importation of ruminants and ruminant products from countries where BSE was known to exist. The genesis of this policy change occurred on May 20, 2003, when a cow in Alberta was diagnosed with BSE. This represented not only the first case of BSE native to North America, but it wreaked havoc on the highly integrated beef market that exists between the United States and Canada. Shortly after the infected cow was announced, then Secretary of Agriculture Veneman issued an Emergency Order adding Canada to the list of regions where BSE was known to exist. Change in Dis- ease Status of Canada Because of BSE, 68 Fed. Reg. 31,939 (May 29, 2003). Under the regulations then in effect, all imports of live ruminants or ruminant meat products from Canada were prohibited. See 9 C.F.R. §§ 93.401, 94.18 (2003). Beginning in August 2003, the Secretary incrementally began moving to reopen the border to Canadian ruminants RANCHERS CATTLEMEN v. USDA 8745 and ruminant products and to reestablish the voluminous North American beef trade. On August 8, 2003, the Secretary announced that she would begin allowing certain “low-risk” ruminant products to be imported into the United States from Canada, the most significant of which was “[b]oneless bovine meat from cattle under 30 months of age.” See Final Rule, 70 Fed. Reg. at 536; USDA News Release No. 0281.03 (Aug. 8, 2003), available at http://www.usda.gov/wps/portal. On November 4, 2003, the Secretary published notice of a proposed rule, seeking to amend the regulations governing the importation of ruminants from countries where BSE is known to exist. Bovine Spongiform Encephalopathy; Minimal Risk Regions and Importation of Commodities, 68 Fed. Reg. 62,386 (Nov. 4, 2003). The proposed rule would have allowed the importation of ruminants from countries in a newly cre- ated category — “regions that present a minimal risk of intro- ducing [BSE] into the United States via live ruminants and ruminant products.” Id. The new regulation proposed to desig- nate only Canada as a minimal-risk region. Id. The comment period for the proposed rule was set to expire on January 5, 2004. Id. A month and a half after the Secretary published the notice of proposed rule, on December 23, 2003, a cow in Washing- ton State was diagnosed with BSE. Bovine Spongiform Encephalopathy; Minimal Risk Regions and Importation of Commodities, 69 Fed. Reg. 10,633 (Mar. 8, 2004). An investi- gation revealed that the cow was born in Canada and was imported into the United States in 2001. Id. at 10,634. Given that the cow was born before Canada’s feed ban went into effect in 1997, USDA determined that the likeliest cause of its BSE infection was contaminated feed. Id. Nevertheless, in response to this discovery USDA reopened the comment period for its proposed rule for an additional 30 days, extend- ing it until April 7, 2004. Id. at 10,633. 8746 RANCHERS CATTLEMEN v. USDA On April 19, 2004, USDA moved, without public notice, to expand the types of ruminant products eligible to be imported from Canada.9 Ranchers Cattlemen Action Legal Fund United Stockgrowers of Am. v. United States Dep’t of Agric., 2004 WL 1047837 (D. Mont. 2004) (“R-CALF TRO”). R-CALF sued to prevent this move, and the district court granted a temporary restraining order on April 26, 2004, barring the Secretary from proceeding with that plan. Id. On January 4, 2005, USDA published its Final Rule. The agency, after having considered 3,379 comments from inter- ested parties, proceeded with its plan to reopen the border to Canadian ruminants and ruminant products. Final Rule, 70 Fed. Reg. at 460, 469. Among other provisions, the Final Rule allowed the importation of Canadian cattle under 30 months of age provided the cattle were immediately slaughtered or fed and then slaughtered.10 Id. at 548. The Final Rule also per- mitted the importation of beef products from Canadian cattle of all ages. Id. at 461, 465. The rule was scheduled to go into effect on March 7, 2005. Id. at 460. 9 Specifically, USDA issued a memorandum stating that, effective April 19, 2004, all existing permits to import meat from Canada “will be deemed to cover all edible bovine meat products (bone-in, boneless, ground meat, further processed),” provided each shipment is accompanied by a statement that the meat was processed in “establishments that are cer- tified to FSIS as eligible for export to the United States.” R-CALF TRO, 2004 WL 1047837 at *2. USDA also published a table identifying “Low Risk Canadian Products.” That table included “boneless, bone-in, ground meat, and further processed bovine meat products,” bovine tongue, bovine hearts, kidneys, and tripe, and bovine lips.” Id. 10 According to amicus Pioneer, Inc., a family-owned feedlot, the cattle industry is generally comprised of three parts: ranchers, who breed cattle and grow them until they reach approximately 650 pounds; feedlots, which purchase cattle from ranchers and feed them high protein feed until they reach approximately 1,150 pounds; and meat packers, which pur- chase cattle from feedlots and process them for human consumption. Thus, the Final Rule allowed Canadian cattle either to be sold to a feedlot for feeding or to be sold directly to a meat packing company for slaughter. RANCHERS CATTLEMEN v. USDA 8747 At roughly the same time that USDA published its Final Rule, two additional cases of BSE were confirmed in Alberta — one on January 2, 2005, and another on January 11. Bovine Spongiform Encephalopathy; Minimal Risk Regions and Importation of Commodities; Finding of No Significant Impact and Affirmation of Final Rule, 70 Fed. Reg. 18,252, 18,254 (Apr. 8, 2005). One of these cows, like the two previ- ous Canadian cattle diagnosed with BSE, was born before Canada’s feed ban; the other, however, was born shortly thereafter. Id. at 18,258. Once again, USDA attributed the infections in both cows to contaminated feed manufactured before Canada’s feed ban went into effect. Id. at 18,255. Nonetheless, USDA indefinitely suspended the implementa- tion of the portion of its Final Rule that permitted the importa- tion of beef products from cattle over 30 months of age.11 Bovine Spongiform Encephalopathy; Minimal Risk Regions and Importation of Commodities; Partial Delay of Applicabil- ity, 70 Fed. Reg. 12,112 (Mar. 11, 2005). D. Procedural History Six days after USDA published the Final Rule, R-CALF filed this action, seeking to enjoin the rule’s implementation.12 In its complaint, R-CALF alleged that USDA’s rulemaking violated the Administrative Procedure Act (“APA”), the Reg- ulatory Flexibility Act (“RFA”), and the National Environ- mental Policy Act (“NEPA”). On February 1, 2005, three weeks after filing its complaint, R-CALF filed its application for a preliminary injunction to enjoin the Final Rule pendente lite. 11 As mentioned above, an additional cow in the United States tested positive for BSE on June 24, 2005. Because this cow was approximately 12 years old, USDA has attributed its infection to contaminated feed it was exposed to before the U.S. feed ban came into effect. 12 R-CALF describes itself as a non-profit cattle association that repre- sents U.S. “cattle producers, cattle backgrounders, and independent feedlot owners” on issues concerning international trade and marketing. 8748 RANCHERS CATTLEMEN v. USDA On March 2, 2005, the district court issued a preliminary injunction, barring USDA from implementing its Final Rule. See R-CALF I, 359 F. Supp. 2d at 1074. The district court’s primary reason for enjoining the Final Rule was its finding that the rule was arbitrary and capricious in violation of the APA. Id. at 1063-69; see also 5 U.S.C. § 706(2). The district court’s overarching concern was that USDA, “ignoring its statutory mandate to protect the health and welfare of the peo- ple of the United States, established its goal of re-opening the border to the importation of live beef from Canada and there- after attempted to work backwards to support and justify this goal.” R-CALF I, 359 F. Supp. 2d at 1066. Given the agency’s “preconceived intention, based upon inappropriate consider- ations, to rush to reopen the border regardless of uncertainties in the agency’s knowledge,” the district court found the Final Rule to be arbitrary and capricious. Id. at 1074. The district court specifically based its determination that the Final Rule was arbitrary and capricious under the APA on six independent grounds. First, the court found that USDA failed adequately to quantify the risk of Canadian cattle to humans, instead relying on a qualitative statement that the risk was “low” or “very low.” Id. at 1064-65. Without a quantita- tive assessment, the district court felt that it “ha[d] no way of assessing the merits of the USDA’s actions.” Id. at 1065. Second, the district court held that USDA had erroneously calculated the prevalence of BSE in the Canadian herd. Id. at 1065-66. USDA had divided the number of cases in the last 12 months (two) by the total size of the Canadian herd over 24 months of age (5.5 million) to arrive at a prevalence rate of approximately 0.4 cases per million head of adult cattle. Final Rule, 70 Fed. Reg. at 464. The district court rejected this calculation, however, and instead adopted R-CALF’s measure of 5.5 cases per million head.13 R-CALF I, 359 F. Supp. 2d at 1066. 13 To achieve the rate of 5.5 cases per million head, R-CALF calculated the prevalence of BSE among tested cattle in Alberta (one in 3,000) and RANCHERS CATTLEMEN v. USDA 8749 Third, the district court found that USDA’s reliance on the Canadian feed ban was unjustified. Id. at 1066-68. The court found that the science was uncertain in this area and that methods of BSE transmission other than consumption of con- taminated feed may exist. Id. at 1066. It also found that the feed ban had not been in place an adequate amount of time, and that it was not fully effective because it allowed both bovine blood and rendered animal fat in cattle feed. Id. at 1067-68. Fourth, the court found that USDA’s reliance on the removal of SRMs to protect human health was also unjusti- fied. Id. at 1068. According to the district court, evidence indicated that “it is no longer reasonable to presume that there is no risk of exposure to BSE infectious agents once an SRM removal requirement is in place.” Id. Fifth, the district court found that USDA’s failure to ban the importation of pregnant cows was arbitrary and capri- cious. Id. at 1069. According to the district court, BSE may be transmitted both maternally and through fetal bovine blood. Id. Thus, because the Final Rule did not require heifers to be pregnancy checked as a condition of entry into the United States, calves born to imported cattle could become “a vector for BSE infection in the U.S.” Id. Finally, the district court found that USDA had failed to respond adequately to comments recommending mandatory BSE testing for Canadian cattle. Id. Because testing can iden- tify a BSE infection up to three months before the cow shows outward signs of the disease, the court found that testing divided it by 60, the assumed amount by which tested cattle will have BSE over untested cattle (because tested cattle, which show outward signs of the disease, are more likely to have BSE than the population at large). The result is one infected cow per 180,000 head of cattle, or approximately 5.56 per million.” 8750 RANCHERS CATTLEMEN v. USDA would be useful because it would “detect some cases of BSE that would otherwise go undetected.” Id. In light of the “irrep- arable injury” that it believed a case of BSE would cause, the court viewed USDA’s actions as arbitrary and capricious. Id. In addition to finding the Final Rule arbitrary and capri- cious under the APA, the district court also relied on two other bases for enjoining its implementation. First, the court held that USDA had failed to satisfy NEPA’s procedural requirements, both by failing to make its environmental assessment available for public review and comment before the Final Rule was published, and by failing to prepare an environmental impact statement. Id. at 1069-71. Second, the court concluded that USDA had violated the RFA by failing to consider whether product labeling or voluntary BSE testing would have mitigated the Final Rule’s impact on small busi- nesses. Id. at 1071-73. Based on the above, the district court found that R-CALF had raised “very serious questions on the merits.” Id. at 1074. The district court also found that R-CALF, and the American public, would be irreparably harmed by allowing the importa- tion of Canadian beef. Id. at 1073-74. The court specifically found that the introduction of BSE into the United States would cause irreparable harm to the American public because of the increased risk of vCJD to consumers of beef. Id. at 1073. Further, it found that the association with Canadian beef would stigmatize all U.S. meat, causing a “serious, irrep- arable impact on ranchers in the U.S. and the U.S. economy.” Id. Finally, the district court found that the NEPA violation, in and of itself, would cause irreparable harm and warranted preliminary injunctive relief. Id. In light of its determination that R-CALF was likely to suc- ceed on the merits, and that the balance of hardships tipped in R-CALF’s favor, the district court issued a preliminary injunction barring implementation of the Final Rule. Id. at 1074. Two weeks later, USDA filed this timely appeal. RANCHERS CATTLEMEN v. USDA 8751 II. ANALYSIS “A district court’s order granting a preliminary injunction is subject to limited review.” Price v. City of Stockton, 390 F.3d 1105, 1109 (9th Cir. 2004). We will reverse “only where the district court abused its discretion or based its decision on an erroneous legal standard or on clearly erroneous findings of fact.” Id. A reviewing court should generally refrain from reviewing “the underlying merits of the case.” Southwest Voter Registration Educ. Project v. Shelley, 344 F.3d 914, 918 (9th Cir. 2003) (en banc). Rather, “[a]s long as the district court got the law right, it will not be reversed simply because the appellate court would have arrived at a different result if it had applied the law to the facts of the case.” Earth Island Inst. v. United States Forest Serv., 351 F.3d 1291, 1298 (9th Cir. 2003). “The standard for granting a preliminary injunction bal- ances the plaintiff’s likelihood of success against the relative hardship to the parties.” Clear Channel Outdoor, Inc. v. City of Los Angeles, 340 F.3d 810, 813 (9th Cir. 2003). This circuit has recognized two different sets of criteria for preliminary injunctive relief. Under the traditional test, a plaintiff must show: “(1) a strong likelihood of success on the merits, (2) the possibility of irreparable injury to plaintiff if preliminary relief is not granted, (3) a balance of hardships favoring the plaintiff, and (4) advancement of the public interest (in certain cases).” Save Our Sonoran, Inc. v. Flowers, 408 F.3d 1113, 1120 (9th Cir. 2005). The alternative test requires that a plain- tiff demonstrate “either a combination of probable success on the merits and the possibility of irreparable injury or that seri- ous questions are raised and the balance of hardships tips sharply in his favor.” Id. (emphasis in original). “These two formulations represent two points on a sliding scale in which the required degree of irreparable harm increases as the prob- ability of success decreases. They are not separate tests but rather outer reaches of a single continuum.” Id. 8752 RANCHERS CATTLEMEN v. USDA As we conclude below, the district court’s finding that R- CALF had a strong likelihood of success on the merits was premised on legal error. Further, we disagree with the district court’s assessment of the irreparable harm threatened by the Final Rule. Thus, we hold that a preliminary injunction was unwarranted in this case. A. Likelihood of Success on the Merits The district court identified three distinct grounds for its finding that R-CALF had a strong likelihood of success on the merits: (1) that the Final Rule was arbitrary and capricious under the APA; (2) that USDA had failed to satisfy NEPA’s procedural requirements; and (3) that USDA had failed ade- quately to consider the Final Rule’s effect on small busi- nesses, as required by the RFA. None of these grounds withstands scrutiny. 1. Administrative Procedure Act The APA provides that a court, when reviewing agency action, shall “hold unlawful and set aside agency action, find- ings, and conclusions found to be . . . arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706. An agency’s action violates this standard 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 expla- nation for its decision that runs counter to the evi- dence before the agency, or is so implausible that it could not be ascribed to a difference in view or the product of agency expertise. City of Sausalito v. O’Neill, 386 F.3d 1186, 1206 (9th Cir. 2004) (quoting Motor Vehicles Mfrs. Ass’n v. State Farm Mut. Auto Ins. Co., 463 U.S. 29, 43 (1983)). RANCHERS CATTLEMEN v. USDA 8753 [1] Regulations are presumed to be valid, and therefore review is deferential to the agency. Nat’l Ass’n of Home Builders v. Norton, 340 F.3d 835, 841 (9th Cir. 2003). All that is required is that the agency have “considered the relevant factors and articulated a rational connection between the facts found and the choices made.” Id. Further, “[t]he court is not empowered to substitute its judgment for that of the agency.” Ariz. Cattle Growers’ Ass’n v. United States Fish & Wildlife Serv., 273 F.3d 1229, 1236 (9th Cir. 2001). Deference to the informed discretion of the responsible federal agencies is especially appropriate, where, as here, the agency’s decision involves a high level of technical expertise. Id. While review is therefore deferential, it is not toothless; courts must conduct a “thorough, probing, in-depth” inquiry into the validity of regulations. Nat’l Ass’n of Homebuilders, 340 F.3d at 841. This inquiry must be “searching and careful” to ensure that the agency decision does not contain a clear error of judgment. City of Sausalito, 386 F.3d at 1206; Nat’l Ass’n of Homebuilders, 340 F.3d at 841. In performing this inquiry, the court is not allowed to uphold a regulation on grounds other than those relied on by the agency. Ariz. Cattle Growers’ Ass’n, 273 F.3d at 1236 (“The reviewing court may not substitute reasons for agency action that are not in the record.”). [2] The district court failed to abide by this deferential stan- dard. Instead, the district court committed legal error by fail- ing to respect the agency’s judgment and expertise. Rather than evaluating the Final Rule to determine if USDA had a basis for its conclusions, the district court repeatedly substi- tuted its judgment for the agency’s, disagreeing with USDA’s determinations even though they had a sound basis in the administrative record, and accepting the scientific judgments of R-CALF’s experts over those of the agency. For example, in assessing the prevalence of BSE in the Canadian herd, the district court rejected USDA’s calculation and accepted the 8754 RANCHERS CATTLEMEN v. USDA prevalence rate provided by R-CALF’s expert, completely without explanation. R-CALF I, 359 F. Supp. 2d at 1066. The district court’s lack of deference may be attributable to its misreading of the Animal Health Protection Act (“AHPA”), 7 U.S.C. § 8301 et seq., the statute under which the Final Rule was promulgated. Based on the AHPA’s state- ment of congressional findings, 7 U.S.C. § 8301, the district court appears to have imposed a requirement on USDA that its Final Rule present no additional risk to human or animal health.14 See R-CALF I, 359 F. Supp. 2d at 1065 (“The [AHPA] directs the Secretary of the USDA to protect the health and welfare of the people of the United States.”). The AHPA is, in fact, based upon congressional findings that “the prevention, detection, control, and eradication of diseases and pests of animals are essential to protect . . . animal health [and] the health and welfare of the people of the United States.” 7 U.S.C. § 8301(1). The provision of the Act under which the Final Rule was promulgated, however, states only that “the Secretary [of Agriculture] may prohibit or restrict . . . the importation or entry of any animal, article, or means of conveyance . . . if the Secretary determines that the prohibi- tion or restriction is necessary to prevent the introduction into 14 While the district court never explicitly stated that it was imposing such a “zero-risk” requirement, its reasoning suggests that it did. For example, the court faulted the agency for “presum[ing] that there is no risk of exposure to BSE infective agents once an SRM removal requirement is in place.” R-CALF I, 359 F. Supp. 2d at 1068. Similarly, the court found the Final Rule arbitrary and capricious because the agency refused to act to remove the “small probability” that BSE could be transmitted from a pregnant Canadian cow to its offspring. Id. at 1069. Indeed, the district court appears to have required USDA to disprove all scientific uncertainty associated with BSE. It noted, for example, that there is no “conclusive scientific proof” that cattle feed is the only method of BSE transmission. Id. at 1066. In other areas of the opinion, any level of scientific uncertainty surrounding a USDA decision rendered that deci- sion an “assumption.” E.g., id. at 1066, 1067, 1068; see also id. at 1074 (criticizing USDA for acting despite “uncertainties in the agency’s knowl- edge of the possible impacts on human and animal health”). RANCHERS CATTLEMEN v. USDA 8755 or dissemination within the United States of any pest or dis- ease of livestock.” 7 U.S.C. § 8303(a)(1). [3] The AHPA was only recently enacted, in 2002, and, as of yet, there are few reported cases interpreting its provisions. Nonetheless, the statute’s terms indicate a congressional intent to give the Secretary wide discretion in dealing with the importation of plant and animal products. More to the point, the AHPA does not impose any requirement on USDA that all of its actions carry no associated increased risk of disease. Indeed, the statute’s use of the word “may” suggests that the Secretary is given discretion over such decisions as whether to close the borders. See, e.g., United States v. George, 85 F.3d 1433, 1437 (9th Cir. 1996) (statute’s use of term “may” “indicates that we should review a district court’s decision . . . for abuse of discretion”). Although sparse, the AHPA’s legis- lative history also supports this view. See H.R. Conf. Rep. 107-424, reprinted in 2002 U.S.C.C.A.N. 141, 388 (in order to best protect against animal disease, “a regulatory definition of disease should be left to the discretion of the Secretary,” which will allow “the agency to have maximum flexibility to focus its resources and respond to new or emerging disease threats”). It is also notable that open borders are a default under the AHPA, and the Secretary can close them only if “necessary” to prevent livestock disease. See 7 U.S.C. § 8303. [4] The structure of the AHPA is therefore inconsistent with the district court’s strict requirement that the USDA reg- ulation remove all risk of BSE entering the United States. Because the district court interpreted the statute to contain such a requirement, its analysis of the Final Rule’s compli- ance with the APA was fundamentally flawed.15 15 We are in no way, of course, implying that the Secretary has unlimited powers to open and close the borders as he sees fit. As the AHPA’s struc- ture indicates, however, the Secretary has considerable discretion to decide when an open or closed border is appropriate. Absent a strong showing that the Secretary is not exercising that discretion consistent with the statutory requirements, his judgment should not be overturned. 8756 RANCHERS CATTLEMEN v. USDA Our own review of the Final Rule leads us to conclude that the Secretary had a firm basis for determining that the resumption of ruminant imports from Canada would not sig- nificantly increase the risk of BSE to the American popula- tion. In conducting this review, we believe it is appropriate to view the BSE prevention measures currently in place as part of a comprehensive system. Thus, rather than follow the “di- vide and conquer” strategy of analyzing each protective com- ponent of the regulatory system in isolation, we evaluate the cumulative effects of the multiple, interlocking safeguards. USDA’s comprehensive protections begin, first, with the low incidence of BSE in Canadian cattle. This assures that if any infected cattle are imported, the number will be relatively small.16 Next, Canada’s feed ban, which USDA considers effective, and its import restrictions on cattle from areas with high BSE rates, ensure that Canada’s prevalence rate will not rise dramatically. Canada also takes other measures, such as BSE testing and epidemiological investigations, that help it find and understand the source of BSE in its cattle population, which helps it further minimize the prevalence of BSE in its herd. These steps ensure, as USDA found, that Canada’s already low rate of BSE is decreasing. Final Rule, 70 Fed. Reg. at 464. From the already low prevalence rate in the Canadian herd as a whole, USDA permits the importation of only a subset of those animals that are extremely unlikely to have BSE — those under 30 months of age. In England, only 0.01 percent of those animals diagnosed with BSE were under 30 months of age. Meat Produced by Advanced Meat/Bone Separation Machinery and Meat Recovery (AMR) Systems, 69 Fed. Reg. 1874, 1875 (Jan. 12, 2004). In addition, USDA’s scientific evidence suggests that Canadian cattle under 30 months of 16 For example, assuming two million cattle enter the United States from Canada per year, less than one would be expected to have BSE based on Canada’s prevalence rate of 0.4 cases per million head of adult cattle. RANCHERS CATTLEMEN v. USDA 8757 age will be far less likely to be in the advanced stages of BSE, given that the incubation period of BSE depends on the amount of BSE agent to which an animal has been exposed. Based on Canada’s low BSE rate and its feed ban, Canadian cattle should have a much lower exposure than English cattle, resulting in a correspondingly greater incubation period. Thus, the age restriction further reduces the risk of introduc- tion of BSE from Canada’s herd. Inside the United States, the risk of dissemination of BSE is addressed by the requirement that Canadian cattle be imme- diately slaughtered or fed and then slaughtered before they reach the age of 30 months. Again, because of BSE’s lengthy incubation period, this age limit helps to ensure that BSE will not progress in any infected animals before they are slaugh- tered. Once they are slaughtered, the FDA’s feed ban ensures that they will not be fed to other cattle, preventing further dis- semination of the disease if, in fact, an imported cow were infected. As for human health, cattle slaughtered in the United States are subject to FSIS regulations designed to minimize the risk that any infectious material will enter the human food supply. These regulations largely prohibit parts of the central nervous system and other cattle parts that have shown BSE infectivity from contaminating human food. In addition, FSIS has placed restrictions on the manner in which cattle may be slaughtered — air compression devices are banned to protect against the possibility that they might inject parts of the brain into the bloodstream. FSIS regulations also require the removal of all SRMs from slaughtered cattle, and they restrict the use in human food of “mechanically separated beef” and meat obtained from “Advanced Meat Recovery” systems. The final defense against human BSE infection is biologi- cal. The limited nature of the vCJD outbreak indicates that there may be a substantial species barrier that prevents BSE from easily infecting humans. Indeed, the fact that there have 8758 RANCHERS CATTLEMEN v. USDA been only slightly over 150 confirmed cases of vCJD world- wide — orders of magnitude less than the number of cases of BSE in cattle — suggests that humans likely do not contract the disease easily. [5] This regulatory system, with its numerous overlapping and complementary safeguards, is designed to minimize the risk of BSE to American livestock and consumers. Thus, sub- stantial evidence supports USDA’s conclusion that these pro- tections will effectively achieve that goal. Further, a comprehensive study commissioned by USDA, known as the “Harvard-Tuskegee Study,” evaluated the likely effects of the introduction of BSE into the United States. The study con- cluded that, if 10 infected cows were imported into the United States from Canada, on average only three new cases of BSE would result and the disease was “virtually certain” to be eradicated from the United States within 20 years. Instead of evaluating the BSE safeguards as part of a larger system, the district court parsed the regulations and faulted USDA for any risk that a given step failed to remove. The dis- trict court listed six specific grounds as the bases for its find- ing that the Final Rule was arbitrary and capricious. We examine each of them seriatim and conclude that none of them supports its conclusion. a. Lack of quantitative standards The district court faulted USDA for “ma[king] assumptions of qualitative judgments,” rather than performing “a quantita- tive assessment of the risk of various options.” R-CALF I, 359 F. Supp. 2d at 1065. It concluded that, “[p]resented with the USDA’s conclusions that the risks to U.S. cattle and consum- ers are ‘low’ without any definition as to what that means and why the risks presented by the Final Rule are acceptable, this Court has no way of assessing the merits of the USDA’s actions.” Id. RANCHERS CATTLEMEN v. USDA 8759 [6] The district court’s imposition of such a bright-line pro- hibition on qualitative standards was incorrect. The Supreme Court has made clear that courts should not upset agency decisions, even those announced with “less than ideal clarity,” if “the agency’s path may reasonably be discerned.” Alaska Dep’t of Envtl. Conservation v. EPA, 540 U.S. 461, 496 (2004) (internal quotation marks omitted); see also Vigil v. Leavitt, 381 F.3d 826, 833 (9th Cir. 2004); Nat’l Ass’n of Homebuilders, 340 F.3d at 846. Moreover, the AHPA does not require the Secretary to quantify a permissible level of risk or to conduct a risk assessment. [7] Under this standard, the administrative record is an ade- quate basis for discerning USDA’s conclusions. For example, USDA’s conclusion that the prevalence of BSE in the Cana- dian herd is “very low” is supported by its observation that “Canada’s incidence rate of two infected cattle in 2003 out of a population of 5.5 million cattle over 24 months of age [is well below] OIE’s recommendation of less than two infected cattle per million during each of the last four consecutive 12- month periods within the cattle population over 24 months of age.” Final Rule, 70 Fed. Reg. at 464. Similarly, the “very low” risk of a consumer contracting vCJD is supported by its finding that “the removal of SRMs effectively mitigates the BSE risk to humans.” Id. at 465. Indeed, the Harvard- Tuskegee Study, one of the centerpieces of USDA’s rulemak- ing, concluded that SRM removal “would reduce . . . potential human exposure to BSE by 95 percent.” Id. at 467. The low risk of a human developing vCJD is also supported by USDA’s observation that “the number of cases of vCJD identified to date suggest a substantial species barrier that may protect humans from widespread illness due to BSE.” Id. at 462. It is also supported by anecdotal evidence of vCJD outbreaks in other parts of the world. In Switzerland, for example, the BSE rate in 1995 was 73.6 cases per million head of cattle, and has been above 20 for most of the past 10 years, see http://www.oie.int/eng/info/en_esbincidence.htm, 8760 RANCHERS CATTLEMEN v. USDA yet Switzerland has not identified a single case of vCJD. Finally, no case of vCJD has ever been attributed to Canadian beef or to the North American meat supply. b. Prevalence of BSE in Canada The district court concluded that “Canada has not con- ducted sufficient testing for BSE to accurately assess the rate of BSE infection in Canada.” R-CALF I, 359 F. Supp. 2d at 1065. It also concluded that the actual rate of BSE in Canada was “greater than 5.5 cases per million head of cattle . . . [put- ting] Canada on par with a number of European countries with a BSE problem.” Id. at 1066. Based on this number, the district judge found that the importation of “2-3 million head of cattle from Canada during the remainder of 2005” pre- sented a “potentially catastrophic risk of danger to the beef consumers in the U.S.” Id. [8] The district court, in this instance, impermissibly substi- tuted its judgment for that of the agency. The USDA, in its Final Rule, calculated Canada’s BSE prevalence rate to be between 0.3 and 0.4 per million head of cattle. Final Rule, 70 Fed. Reg. at 464. The district court gave no reason for depart- ing from this calculation and, instead, adopting the calculation of R-CALF’s expert wholesale. The district court did so even though R-CALF’s calculation contained the same type of unexplained assumptions that the court found fatal to the Final Rule. For example, R-CALF’s expert assumed that cat- tle with outward signs of BSE are 60 times more likely to have the disease than cattle with no symptoms, and assumed that the prevalence rate of BSE in Alberta was representative of the rate in Canada as a whole. [9] USDA, on the other hand, based its calculation of Cana- da’s BSE rate on OIE guidelines; indeed, the OIE website lists Canada’s 2003 incidence rate as 0.33 and its 2004 rate as 0.149. See http://www.oie.int/eng/info/en_esbincidence.htm. The district court erred by departing from USDA’s method of RANCHERS CATTLEMEN v. USDA 8761 calculation, which was supported by the administrative record, without providing any reason. Marsh v. Or. Nat. Res. Council, 490 U.S. 360, 378 (1989) (“When specialists express conflicting views, an agency must have discretion to rely on the reasonable opinions of its own qualified experts even if, as an original matter, a court might find contrary views more persuasive.”); Nat’l Wildlife Fed’n v. United States Army Corps of Eng’rs, 384 F.3d 1163, 1177 (9th Cir. 2004). c. Effectiveness of Canadian feed ban The district court also questioned USDA’s reliance on the Canadian feed ban. First, it found that there was “no conclu- sive scientific proof” that consumption of infected feed is the only method of BSE transmission, commenting that transmis- sion may occur through blood and saliva. R-CALF I, 359 F. Supp. 2d at 1066-67. Second, the court found evidence that the feed ban had not been effective, both because the ban had only been in place for seven years and because the 4.2-year average incubation period of BSE suggested that the infected Canadian cows had contracted BSE well after the feed ban was put in place. Id. at 1067. Finally, the court found gaps in the ban, finding that both bovine blood and rendered animal fat were allowed in animal feed and that both could transmit BSE. Id. at 1067-68. As to the first reason, the USDA explicitly considered sci- entific evidence on alternative theories of transmission and rejected them, finding that “oral ingestion of feed contami- nated with the BSE is the only documented route of field transmission of the disease.” Final Rule, 70 Fed. Reg. at 486; see also id. at 491 (discussing infectivity of blood). [10] The trial court’s criticisms of Canada’s feed ban are also baseless. The district court’s main criticism is that Cana- da’s feed ban had been in place for only seven and a half years, not the eight years recommended by OIE guidelines. Applying such a strict reading of OIE guidelines, however, 8762 RANCHERS CATTLEMEN v. USDA was incorrect. According to a declaration submitted by the Head of the International Trade Department of OIE, OIE rec- ommends that an importing country evaluate the exporting country’s risk mitigation measures as a whole, and “would not consider it appropriate for the importing country to apply each criterion as an item on a checklist.” Thus, “a deficiency in the length of time a feed ban has been effectively applied could be addressed through restrictions on the age of live cattle imported.” The Final Rule reveals that this is precisely the approach that USDA took. See, e.g., Final Rule, 70 Fed. Reg. at 463 (discussing multiple criteria used to evaluate a poten- tial minimal-risk region); id. at 548 (restricting imports of Canadian cattle to those under 30 months of age). Nor do we agree that the 4.2-year average incubation period demonstrates the ineffectiveness of Canada’s feed ban. USDA explained that the incubation period of BSE in cattle depends upon the level of exposure the cattle have to the BSE agent. The 4.2-year figure was obtained from analyzing cattle during the BSE epidemic in England, which represents the highest level of exposure to BSE in history. Cows in Canada can be expected to have a longer incubation period because of their significantly lower levels of BSE exposure. [11] Finally, the district court also erred in criticizing the Canadian feed ban based on its “gaps,” which allow blood and rendered animal fat in cattle feed. As discussed above, USDA considered BSE transmission through blood and deter- mined that the science did not support ingestion of blood as a means of transmission. Id. at 491. USDA also considered transmission through fat and concluded that, provided the fat is not impure, it poses no risk of transmission of BSE. Id. at 500-01 (discussing potential transmission of BSE through tal- low). Again, the district court gave no reason for rejecting USDA’s expert scientific opinion. d. Effectiveness of SRM removal [12] The district court also found that R-CALF had pres- ented sufficient evidence to establish that “it is no longer rea- RANCHERS CATTLEMEN v. USDA 8763 sonable to presume that there is no risk of exposure to BSE infectious agents once an SRM removal requirement is in place.” R-CALF I, 359 F. Supp. 2d at 1068. USDA’s conclu- sion that SRM removal is effective, however, had support in the administrative record. See Final Rule, 70 Fed. Reg. at 467 (discussing Harvard-Tuskegee Study, which concluded that SRM removal would reduce human exposure to BSE by 95 percent). e. Maternal transmission of BSE and fetal blood serum The district court also found the Final Rule arbitrary and capricious because it “does not prohibit cattle of breeding age from being bred either before or after entering the U.S.,” and “there is a small probability that BSE can be transmitted maternally.” R-CALF I, 359 F. Supp. 2d at 1069. In addition, the court found USDA’s prohibition of fetal blood serum to be inconsistent with the possibility of allowing pregnant cows to be imported into the United States. Id. [13] Contrary to the district court’s findings, however, USDA has made it abundantly clear that cattle may not be imported for breeding under the new regulations. Instead, they must be immediately slaughtered, or fed and slaughtered before they reach 30 months of age. Final Rule, 70 Fed. Reg. at 548-49. Furthermore, USDA discussed the concerns that the district court raised, and found that they were not suffi- cient to justify addressing. Id. at 515 (“Although some evi- dence suggesting maternal transmission exists, such transmission has not been proven, and, if it occurs at all, it occurs at very low levels not sufficient to sustain an epidem- ic.”). We also find that there is a basis for USDA’s disparate treatment of fetal blood serum. As the district court acknowl- edged, fetal blood serum is used for “bovine vaccine produc- tion” and “bovine embryo transfer.” R-CALF I, 359 F. Supp. 2d at 1069. Because the serum is injected directly into an ani- 8764 RANCHERS CATTLEMEN v. USDA mal’s bloodstream, it carries a higher risk of transmitting BSE, and “might pose a risk of livestock if used in” these applications. Final Rule, 70 Fed. Reg. at 502. Thus, any inconsistency in the USDA’s approach to offspring of imported Canadian cattle and fetal blood serum has an ade- quate explanation in the record. f. Mandatory testing of Canadian cattle Finally, the district court held that it was arbitrary and capricious for the agency not to require all Canadian cattle to be screened for BSE, because the screening test could identify some animals with BSE that would not otherwise be identi- fied. R-CALF I, 359 F. Supp. 2d at 1069. The Final Rule, however, contains a lengthy comment in which USDA responded to requests for testing of Canadian cattle. Final Rule, 70 Fed. Reg. at 475-76. USDA explained that, because testing can only detect the disease two to three months before a cow starts demonstrating clinical signs of the disease, a cow may be infected and thus produce a false negative on a test. Id. Because of the long incubation period of BSE, and the rel- atively short window in which non-targeted testing is effec- tive, the USDA did not consider testing to be a “food safety” measure. Id. Rather, testing was best used to determine if BSE exists in a country and to determine its prevalence — goals that can both be achieved by targeted testing of animals with clinical signs of BSE. Id. [14] Over the past few years, USDA’s policies regarding BSE testing have been subject to a high degree of criticism. See, e.g., Mad Beef Policy, Los Angeles Times, Jul. 1, 2005; McGarity, supra, at 337-40. These criticisms have generally focused on USDA’s refusal to allow voluntary testing of cat- tle, rather than its refusal to require mandatory testing of Canadian cattle. Although these criticisms are not without their valid points, we do not believe that they are so powerful as to render USDA’s testing policy invalid. USDA’s approach to BSE testing — that, until better tests are developed, pro- RANCHERS CATTLEMEN v. USDA 8765 phylactic measures such as the feed ban and SRM removal are the best methods of protecting human and animal health — is defensible. While its wisdom may be subject to debate on the merits, its choices are not so lacking support in the adminis- trative record as to be “arbitrary and capricious.” g. Conclusion In sum, USDA decided to reopen the border to Canadian ruminants after making a reasoned determination that the importation of a small number of BSE-infected cattle into this country would not pose a serious risk to humans or livestock. As part of its determination, USDA necessarily decided that the risks inherent in the uncertainty surrounding the current scientific understanding of BSE were insufficiently significant to justify the continued exclusion of Canadian cattle. Rather than criticizing USDA for allowing these risks as a part of its policy, the district court should have evaluated whether there was an adequate basis in the administrative record for USDA’s conclusion that the risks were acceptable. [15] Our review of the record leads us to conclude that the risks inherent in the Final Rule are small, and that the rule likely is supported by an adequate administrative record. We therefore conclude that the district court erred in finding that R-CALF has a strong likelihood of success on the merits of its APA claim. 2. Regulatory Flexibility Act We also conclude that the district court erred in concluding that R-CALF has a strong likelihood of success on its claim under the RFA. The RFA was passed in 1980 to “encourage administrative agencies to consider the potential impact of nascent federal regulations on small businesses.” Assoc. Fish- eries of Maine, Inc. v. Daley, 127 F.3d 104, 111 (1st Cir. 1997). In certain cases, it requires agencies to publish an “ini- tial regulatory flexibility analysis” at the time a proposed rule 8766 RANCHERS CATTLEMEN v. USDA is published, and a “final regulatory flexibility analysis” at the time a final rule is published. 5 U.S.C. §§ 603, 604. Judicial review is available only of the final analysis. 5 U.S.C. § 611. The RFA requires that a final analysis contain the follow- ing: (1) a succinct statement of the need for, and objec- tives of, the rule; (2) a summary of the significant issues raised by the public comments in response to the initial regulatory flexibility analysis, a summary of the assessment of the agency of such issues, and a statement of any changes made in the proposed rule as a result of such comments; (3) a description of and an estimate of the number of small entities to which the rule will apply or an explanation of why no such estimate is available; (4) a description of the projected reporting, record- keeping and other compliance requirements of the rule, including an estimate of the classes of small entities which will be subject to the requirement and the type of professional skills necessary for prepara- tion of the report or record; and (5) a description of the steps the agency has taken to minimize the significant economic impact on small entities consistent with the stated objectives of appli- cable statutes, including a statement of the factual, policy, and legal reasons for selecting the alternative adopted in the final rule and why each one of the other significant alternatives to the rule considered by the agency which affect the impact on small enti- ties was rejected. RANCHERS CATTLEMEN v. USDA 8767 5 U.S.C. § 604(a). The RFA imposes no substantive requirements on an agency; rather, its requirements are “purely procedural” in nature. United States Cellular Corp. v. FCC, 254 F.3d 78, 88 (D.C. Cir. 2001); see also Envtl. Defense Ctr., Inc. v. United States EPA, 344 F.3d 832, 879 (9th Cir. 2003), cert. denied, 541 U.S. 1085 (2004) (“Like the Notice and Comment pro- cess required in administrative rulemaking by the APA, the analyses required by the RFA are essentially procedural hur- dles; after considering the relevant impacts and alternatives, an administrative agency remains free to regulate as it sees fit.”). To satisfy the RFA, an agency must only demonstrate a “reasonable, good-faith effort” to fulfill its requirements. United States Cellular, 254 F.3d at 88; Alenco Communica- tions, Inc. v. FCC, 201 F.3d 608, 625 (5th Cir. 2000); Assoc. Fisheries, 127 F.3d at 114. The district court faulted USDA for considering only two alternatives in its final regulatory flexibility analysis: “leaving the regulations unchanged or modifying the import require- ments by either requiring that imported beef come from cattle slaughtered at less than 30 months of age or continuing to prohibit the entry of live ruminants.” R-CALF I, 359 F. Supp. 2d at 1072; see also Final Rule, 70 Fed. Reg. at 543. The dis- trict court held that the agency erroneously rejected the alter- natives of a country-of-origin labeling program and voluntary testing of slaughtered Canadian cattle. R-CALF I, 359 F. Supp. 2d at 1072. [16] The district court erred in concluding that USDA did not meet the RFA’s requirements. The Final Regulatory Flex- ibility Analysis, available at http://www.aphis.usda.gov/lpa/ issues/bse/risk_assessment/03-080-3_econ_analysis.pdf, reveals that USDA conducted a detailed economic assessment of the impact of its proposed rule on small businesses. It con- cluded that the majority of businesses affected by the pro- posed Final Rule would qualify as small businesses, and that 8768 RANCHERS CATTLEMEN v. USDA the effect of the Final Rule was likely to vary depending upon the sector of the cattle industry the business occupied, rather than the size of the business. The negative economic effects the rule would create would generally affect those on the sup- ply side of the beef industry — primarily ranchers — while those on the production side — feedlots and meat packers — would tend to benefit from the rule. In this respect, the alter- natives identified by the district court would not necessarily ease the burden on small businesses; rather, they would real- locate the rule’s burden to small businesses in different sec- tors of the beef industry. Cf. Assoc. Fisheries, 127 F.3d at 115 (where the majority of businesses affected by a rule are small businesses, “Congress’s desire to have agencies write rules that distinguish . . . between big and small businesses has diminished relevance.”). More importantly, the specific concerns the district court raised were considered by USDA in its response to comments on the rule. USDA rejected the first alternative — the imple- mentation of a country-of-origin labeling program — because it did not consider such a program to concern food safety or animal health. Final Rule, 70 Fed. Reg. at 533. USDA rejected the second alternative, voluntary BSE testing, because it does not consider such testing reliable enough to be used as a food safety measure, as discussed above. See Part II.A.1.f, supra. Given that USDA discussed and rejected these alternatives in the body of its Final Rule, the agency did not err in failing to consider them as alternatives in its final regu- latory flexibility analysis. See Assoc. Fisheries, 127 F.3d at 115 (“[S]ection 604 does not require that [a final regulatory flexibility analysis] address every alternative, but only that it address significant ones.”). 3. National Environmental Policy Act NEPA was enacted in 1970 to “promote efforts which will prevent or eliminate damage to the environment and bio- sphere.” 42 U.S.C. § 4321; see also Robertson v. Methow Val- RANCHERS CATTLEMEN v. USDA 8769 ley Citizens Council, 490 U.S. 332, 348 (1989) (“Section 101 of NEPA declares a broad national commitment to protecting and promoting environmental quality.”). Like the RFA, NEPA does not impose any substantive requirements on an agency’s decision; rather, it mandates only a process that the agency must follow. Id. at 350 (“NEPA itself does not man- date particular results, but simply prescribes the necessary process.”). Under NEPA’s procedural requirements, an agency must prepare a “detailed statement” on the environmental impact of a proposed rule when that rule is a “major Federal action[ ] significantly affecting the quality of the human environment.” 42 U.S.C. § 4332. NEPA provides no private right of action to enforce its requirements. Stratford v. FAA, 285 F.3d 84, 88 (D.C. Cir. 2002). Thus, to bring suit to vindicate NEPA’s requirements, a plaintiff must rely on the provisions of the APA that confer “standing to an ‘aggrieved party’ within the meaning of the substantive statute upon which the claim is based.” Id.; see also 5 U.S.C. § 702; Clarke v. Sec. Indus. Ass’n, 479 U.S. 388, 394-96 (1987). [17] To narrow the wide range of potential plaintiffs who may assert a “procedural injury” under this section of the APA, the Supreme Court has adopted a “zone of interests” test.17 See id. at 397 n.12 (stating that the purpose of the zone of 17 R-CALF incorrectly argues that the Supreme Court’s decision in Ben- nett v. Spear, 520 U.S. 154 (1997), drastically narrowed the applicability of the zone of interests test. In Bennett, the Court considered the specific question of standing under the Endangered Species Act’s citizen-suit pro- vision, not the APA. Id. at 161-62. It expressly found that the “ESA’s citizen-suit provision . . . negates the zone-of-interests test” based on its language and its purpose. Id. at 164-66. Thus, Bennett simply does not address actions under NEPA. Indeed, this court has continued to use the zone of interests test to evaluate the standing of NEPA plaintiffs after Ben- nett. See Save Our Sonoran, 408 F.3d at 1119; Cantrell v. City of Long Beach, 241 F.3d 674, 679 (9th Cir. 2001); see also Stratford, 285 F.3d at 88 (applying the zone of interest test in a NEPA action). 8770 RANCHERS CATTLEMEN v. USDA interests test is “to exclude those plaintiffs whose suits are more likely to frustrate than to further statutory objectives”). This test imposes the requirement, beyond constitutional standing requirements, that a plaintiff assert an interest “argu- ably within the zone of interests to be protected or regulated by the statute or constitutional guarantee in question.” Nev. Land Action Ass’n v. United States Forest Serv., 8 F.3d 713, 715-16 (9th Cir. 1993). Thus, to assert a claim under NEPA, a plaintiff must allege injury to the environment; economic injury will not suffice. Id. at 716 (“[A] plaintiff who asserts purely economic injuries does not have standing to challenge an agency action under NEPA.”); Stratford, 285 F.3d at 88 (“[A] NEPA claim may not be raised by a party with no claimed or apparent environmental interest.”); W. Radio Serv. Co. v. Espy, 79 F.3d 896, 902-03 (9th Cir. 1996) (“NEPA’s purpose is to protect the environment, not the economic inter- ests of those adversely affected by agency decisions.”) (inter- nal quotation marks omitted). A plaintiff can, however, have standing under NEPA even if his or her interest is primarily economic, as long as he or she also alleges an environmental interest or economic injuries that are “causally related to an act within NEPA’s embrace.” Port of Astoria, Or. v. Hodel, 595 F.2d 467, 476 (9th Cir. 1979). The injuries alleged in R-CALF’s complaint do not fall within NEPA’s zone of interests. R-CALF points to only one paragraph in its complaint to justify its standing under NEPA. Every allegation in this paragraph, however, concerns the eco- nomic interest of R-CALF members except the following: “R- CALF USA members will also be adversely affected by the increased risk of disease they face when Canadian beef enters the U.S. meat supply.” [18] We conclude that this alleged harm is insufficient to fall within NEPA’s zone of interests. As mentioned above, “NEPA’s purpose is to protect the environment.” W. Radio Serv. Co., 79 F.3d at 902; see also Stratford, 285 F.3d at 88 (“[A] NEPA claim may not be raised by a party with no RANCHERS CATTLEMEN v. USDA 8771 claimed or apparent environmental interest.”). More specifi- cally, NEPA is concerned with harm to the physical environ- ment: “If a harm does not have a sufficiently close connection to the physical environment, NEPA does not apply.” Metro. Edison Co. v. People Against Nuclear Energy, 460 U.S. 766, 778 (1983); cf. Cantrell, 241 F.3d at 679 (“In NEPA cases, we have described this ‘concrete interest’ test as requiring a ‘geographic nexus’ between the individual asserting the claim and the location suffering an environmental impact.”). R- CALF’s claimed interest, however, has no connection to the physical environment; rather, it is solely a matter of human health. While it is true that NEPA contains references to human health in its statement of policy, see 42 U.S.C. § 4321, as the Supreme Court has explained, those references are to the statute’s goals, not its means. Metro. Edison Co., 460 U.S. at 773 (“[A]lthough NEPA states its goals in sweeping terms of human health and welfare, those goals are the ends that Congress has chosen to pursue by means of protecting the physical environment.”). Here, R-CALF has failed to show any relationship between risks to human health and environ- mental harms. Cf. Port of Astoria, 595 F.2d at 476. [19] Because R-CALF has failed to allege any connection to injury to the physical environment, its injury falls outside of NEPA’s zone of interests. Even assuming R-CALF’s alleged injury could satisfy the zone of interests test, however, its NEPA claim must fail for the additional reason that R- CALF lacks organizational standing to assert a NEPA chal- lenge. An association has standing to bring suit on behalf of its members when its members would otherwise have standing to sue in their own right, the interests at stake are germane to the organization’s purpose, and neither the claim asserted nor the relief requested requires the participation of individual members in the lawsuit. 8772 RANCHERS CATTLEMEN v. USDA Friends of the Earth, Inc. v. Laidlaw Envtl. Serv. (TOC), Inc., 528 U.S. 167, 181 (2000) (citing Hunt v. Wash. State Apple Advertising Comm’n, 432 U.S. 333, 343 (1977)). R-CALF fails the second of these three elements. As mentioned above, R-CALF is a “non-profit cattle asso- ciation representing over 12,000 U.S. cattle producers on issues concerning international trade and marketing.” As is evident from the paragraph in its complaint that discusses standing, economic issues are highly relevant to its purpose. We do not see the connection, however, between the pur- ported environmental interest that R-CALF attempts to raise here and the “trade and marketing” interests it is organized to protect. [20] We therefore hold that R-CALF lacks standing to bring a NEPA challenge to the Final Rule. Thus, the district court erred in permitting R-CALF to proceed with its NEPA claim and in concluding that it had a likelihood of success on that claim.18 B. Balance of Hardships After finding that R-CALF had demonstrated a strong like- lihood of success on the merits, the district court found that the Final Rule carried a definitive risk of causing “significant irreparable harm.” R-CALF I, 359 F. Supp. 2d at 1073. The district court identified three ways in which the Final Rule would cause such harm: the increased risk of vCJD to Ameri- can beef consumers, unspecified environmental injury stem- ming from USDA’s failure to comply with NEPA, and injury to the U.S. beef industry and the U.S. economy that would result from a “stigma” that tainted Canadian beef would inflict upon the U.S. meat supply. Id. We believe the district 18 Given our holding that R-CALF lacks standing to bring a NEPA claim, we need not address the district court’s conclusion that the possibil- ity of environmental harm justifies its preliminary injunction. RANCHERS CATTLEMEN v. USDA 8773 court’s calculus overstated the harm that would result from the rule. [21] If the Canadian herd were to have a higher infection rate than the U.S. herd, the importation of Canadian cattle might pose some increased risk to the health of the U.S. popu- lation, however slight. Even assuming, however, that the introduction of a fatal disease into the United States would constitute irreparable harm, cf. Harris v. Board of Supervi- sors, 366 F.3d 754, 766 (9th Cir. 2004) (accepting as irrepara- ble harm “pain, infection, amputation, medical complications, and death”), the record does not justify the conclusion that the Final Rule makes such harm likely, or even probable. Rather, based on the low incidence of BSE in the Canadian herd, the numerous safeguards against BSE in this country, the lack of any Canadian cattle under 30 months of age found with BSE, and the lack of any case of vCJD attributable to Canadian beef, any increased risk to human and animal health created by the Final Rule is negligible. [22] In retrospect, the district court’s concern over the pos- sibility of “stigma” harming the American beef industry appears to be overstated. The record does not support the dis- trict court’s alarmist findings that the “irreparable economic harm” the district court foresaw from the stigma of Canadian beef will actually befall the American beef industry. Follow- ing the case of BSE diagnosed in a Washington State cow in 2003, consumer demand for, and confidence in, American beef remained strong. Final Rule, 70 Fed. Reg. at 522. According to USDA, American demand for beef in 2004 is estimated to have increased seven to eight percent over 2003 levels. Yet, Canadian beef was flowing into this country throughout 2004 under permits issued by USDA.19 This evi- 19 The district court’s April 26, 2004, temporary restraining order pre- vented USDA only from expanding the categories of Canadian beef that could be imported under existing importation permits. The court explicitly limited its order to “all edible bovine meat products beyond those autho- rized by USDA’s action of August 8, 2003 (boneless bovine meat, bone- less Veal (meat), and bovine liver) from cattle under the age of 30 months.” R-CALF TRO, 2004 WL 1047837 at *9. 8774 RANCHERS CATTLEMEN v. USDA dence belies the district court’s prediction of catastrophic injury to the U.S. beef industry.20 C. Preliminary Injunction [23] Contrary to the district court’s conclusion, we con- clude that the Final Rule will likely survive judicial scrutiny under the correct legal standard; thus, R-CALF has not shown a likelihood of success on the merits of its action. We also conclude that R-CALF has failed to make the requisite show- ing of irreparable harm. For these reasons, we must reverse the district court’s preliminary injunction. See Kootenai Tribe v. Veneman, 313 F.3d 1094, 1125-26 (9th Cir. 2002). III. CONCLUSION For the foregoing reasons, the district court’s grant of a pre- liminary injunction is REVERSED. 20 Indeed, the district court’s finding of irreparable economic harm is undermined by the industry itself. Numerous amici curiae briefs have been filed in this case by organizations representing large sectors of the American meat industry, all of whom seek reversal of the preliminary injunction. If the Final Rule posed a true risk of exposing American beef to an irreparable stigma one would not expect to see such a broad coalition of industry members supporting its implementation.