OPINION

The National Park Service and the United States Forest Service require backpackers who visit certain areas in the Sierras to store food in portable bear-resistant containers. Between 2001 and 2007, both the Park Service and the Forest Service required visitors to Yosemite National Park, Sequoia and Kings Canyon National Parks (“SEKI”), and the Inyo National Forest to use containers that had been tested and approved by the agencies. An informal body known as the Sierra Inter-agency Black Bear Group (“SIBBG”) tested privately manufactured bear-resistant containers and made recommendations to the Park and Forest Services regarding which containers to approve.

Plaintiff-appellant Ursack, Incorporated manufactures a bear-resistant container called the Ursack. Between 2001 and 2007, it urged SIBBG to recommend the Ursack for inclusion on the agencies’ lists of approved containers. Mostly it was unsuccessful, but in 2007, SIBBG recommended that the agencies grant conditional approval to the Ursack for the 2007 summer season. SIBBG recommended that the agencies withdraw approval if they determined that the container failed three or more times during the season. (We explain below what “failure” means in this context.) The agencies accepted this recommendation and granted conditional approval. At the end of the 2007 season, however, SIBBG determined that the Ur-sack had failed more than three times, and it recommended that the agencies withdraw conditional approval. The National Park Service accepted this recommendation and withdrew conditional approval, and to this day it refuses to permit backpackers to use the Ursack in the container-only areas of Yosemite and SEKI. The Forest Service, on the other hand, continues to allow backpackers to use the Ursack in Inyo National Forest.

Ursack and three individual users of the Ursack brought this action pursuant to the Administrative Procedure Act (“APA”) against SIBBG, the Park Service, the Forest Service, and the superintendents of the relevant national parks and forests, alleging that the decision to withdraw conditional approval of the Ursack was arbitrary and capricious and otherwise not in accordance with law. After reviewing the administrative record, the district court granted summary judgment to the agencies. Ursack and the three individuals appeal. We affirm.

I.

Wild bears that routinely obtain access to human food are known to cause problems in the wilderness. Among other things, bears that have been habituated to human food tend to associate humans with food and may become aggressive. To prevent these problems, the Park and Forest Services require visitors to securely store all food and trash. There are various secure storage methods, but the Park and Forest Services specify that in some areas of Yosemite, SEKI and Inyo — areas in which bear activity is particularly problematic — visitors must use portable bear-resistant food-storage containers. Almost all such containers consist of hard-sided, barrel-shaped canisters that can be left on the ground. With a proper container, a bear won’t be able to access or damage the stored food, nor will the bear be able to fit the container into its mouth and carry it off into the wilderness. However, canisters are relatively heavy and bulky by backpacking standards. Seeking to offer a light-weight alternative to hard-sided canisters, Ursack developed a portable bear-resistant container made out of bulletproof fabric. Ursack developed various iterations of this product over the years, but the product at issue in the present case is known as the Ursack S29.

The relevant national parks maintain lists of approved bear-resistant containers and do not allow visitors to use unapproved containers. See, e.g., http://www. nps.gov/yose/planyourvisit/containers.htm (approved container list for Yosemite as of April 11, 2011). For most of the summer backpacking seasons since 2001, the Ur-sack was not an approved container. The inability to use the Ursack in these parks — especially Yosemite — has substantially impacted Ursack’s sales. Ursack tells us that the national retailer REI refuses to stock the Ursack on account of Yosemite’s refusal to approve it.

Until October 21, 2007, the Forest Service maintained a similar list that likewise did not include the Ursack. On October 21, 2007, however, the Forest Service discontinued its practice of keeping an approved-container list and began allowing backpackers to use any bear-resistant container they wanted, including the Ursack. The Park Service, meanwhile, continues to withhold approval of Ursack’s products.

Ursack’s application for approval dates back to the spring of 2001, when members of the Park and Forest Services entered into a memorandum of understanding that created SIBBG, a group of biologists and wilderness and recreation managers from Yosemite, SEKI and Inyo. 2:ER:l-4. SIBBG’s goal was to coordinate bear-management policies for the lands under the control of its constituent agencies. To further this goal, SIBBG tested bear-resistant containers and made recommendations to the agencies. The group met on a regular basis until 2007, when it stopped accepting new products for testing.

In April 2001, SIBBG adopted a three-phase process for approving commercially available bear-resistant containers. 2:SER:224-29. First, the container had to pass a visual inspection. Second, SIBBG tested the container on captive black bears at California zoos. If the container passed both the visual inspection and captive-bear tests, SIBBG granted the container “conditional approval.” 2:SER:227. Once a container received conditional approval, members of the public were allowed to use it in the container-only areas of the parks and forests. During the first summer of conditional approval, SIBBG members also subjected the container to the final phase of the testing process — field tests to observe “ease of container use, durability of the container under field conditions, and the level of security from bears.” 2:SER:227. If the container passed the field tests, the container’s status was upgraded to “approved.” If the container failed the field tests, conditional approval was immediately revoked. SIBBG also had the option of approving a container with stated conditions (not to be confused with “conditional approval”) — that is, approving a container with conditions designed to improve the product’s reliability and durability and to lessen its impact on wilderness resources. Finally, SIBBG reserved the right to revoke even unconditional approval “due to persistent failures in the field.” 2:SER:227.

On May 21, 2001, SIBBG granted conditional approval to an early-model Ursack, the Ursack Ultra, which was designed to be tied to trees. 2:ER:118. Backpackers and SIBBG members then began using the product in the field. By July 24, 2001, however, SIBBG had determined that the Ultra had failed three times. Two of the three failures involved bears ripping into the containers and receiving food. 2:SER:278-79. In response to these failures, SIBBG revoked conditional approval of the Ultra.

Although Ursack disagreed with SIBBG’s findings regarding the Ultra, it returned to the drawing board and developed a new model, the Ursack TKO. SIBBG accepted the TKO for testing in early 2002 but did not grant it conditional approval because the bag failed the zoo test. Specifically, captive bears were able to puncture small holes in the bag and access honey that had been poured inside. 2:SER:286-92. Ursack again disagreed with SIBBG’s findings but decided to modify the TKO and resubmit it for approval.

SIBBG did not accept any Ursack product for testing in 2003, but in 2004 it agreed to test a modified version of the TKO. The modifications included, among other things, an aluminum insert designed to prevent food from leaking out of the bag. This time, SIBBG devised a detailed testing protocol specifically for the Ursack. 1:SER:161-71. In devising the protocol, SIBBG identified a list of issues to evaluate based on its past experience with Ur-sack’s products. l:SER:161-62. SIBBG then designed a series of tests to determine whether the modified TKO satisfactorily addressed those issues. 1:SER:163-71.

SIBBG tested the modified TKO over the summer of 2004 and prepared a detailed report explaining the results. 2:ER:41-65. The report stated that the modified TKO performed well, in that no bear was able to access a significant amount of food from the bag. Nonetheless, the report listed two major concerns. First, unlike with hard-sided canisters, bears were able to chew the bags, mutilating the food inside and rendering the food inedible due to the presence of bear saliva. In some instances, the aluminum insert was crushed and torn, resulting in small fragments of metal mixed into the smashed food. The ability of bears to destroy the food was problematic for three reasons. First, backpackers might be inclined to dump their spoiled food, allowing bears access to the food and also contributing to wilderness litter. Second, the presence of bear saliva on the food posed a small risk of rabies infection, and the possibility of a backpacker eating food containing shredded aluminum posed its own safety hazard. Third, a backpacker who saw a bear chewing the bag might attempt to frighten the bear away in order to save the food from spoilage, which would put the backpacker at risk of injury should the bear decide to attack. The second major concern was that when bags were tied to trees in accordance with the manufacturer’s directions and bears wrestled with the bags, testers noticed damage to the tree’s bark and the substrate (soil) around the trees.

On November 8, 2004, SIBBG informed Ursack that it had decided not to grant conditional approval to the modified TKO. 2:ER:130-31. The rejection letter cited damage to trees and the issues associated with bears chewing the bags as reasons for disapproval. Ursack again disagreed with the results, this time going so far as to retain a silviculture expert, who opined that use of the TKO would not result in significant tree damage. 2:ER:132-36. Nonetheless, Ursack elected to redesign its product to address SIBBG’s concerns.

Ursack’s redesign involved using a thicker aluminum insert, which was intended to both prevent bears from mutilating the food and eliminate the need to tie the bag to a tree. SIBBG was satisfied with the redesign and conditionally approved the bag — now known as the Ursack TKO Hybrid — for use in 2006. However, the bullet-proof material that Ursack used to make the bag, Spectra, was unavailable during the 2006 season. Ursack thus decided to use a different fabric, Vectran, for the TKO Hybrid. This fabric proved to be inadequate, however, in that bears were able to tear into the bags and access food. SIBBG thus revoked its conditional approval of the TKO Hybrid.

By the end of 2006, Spectra was once again available, and Ursack began production of the Ursack model at issue in this litigation — the Ursack S29. In November 2006, Ursack wrote to SIBBG and asked that it approve the S29 for the 2007 summer season. SIBBG then conferred and decided that although it would field-test the S29 “in house” during the 2007 season, it would not conditionally approve the product until the field test was over. This meant that the public wouldn’t be able to use the S29 until the 2008 season, at the earliest. Ursack discussed this decision with SIBBG members, and eventually it wrote a letter to the National Park Service stating that Ursack would commence litigation and seek a preliminary injunction unless SIBBG conditionally approved the S29 for 2007 season.

In response to Ursack’s letter, SIBBG conferred and decided to conditionally approve the S29 for 2007 rather than test it in house. SIBBG informed Ursack that this conditional approval would be revoked if the S29 failed three or more times during the season. This decision was conveyed to Ursack orally during a meeting on May 8, 2007. The administrative record does not contain written correspondence from SIBBG or the Park Service formalizing this decision. However, the record contains a “briefing statement” drafted by a member of SIBBG in anticipation of the May meeting stating that the S29 had been conditionally approved but that such approval would be withdrawn if the product failed three or more times. l:SER:96-97. The briefing statement defined “failure” as follows: “a container that is lost or abandoned, if the food inside is no longer edible after an encounter with an animal, if an animal receives a food reward/the fabric is compromised, and/or the container presents a safety hazard to humans or wildlife.” 1:SER:97. Ursack does not dispute that this briefing statement accurately describes SIBBG’s decision to conditionally approve the S29 and its definition of “failure.”

During the 2007 season, SIBBG received several reports of possible S29 failures. Ursack heard about at least three of these reports through the grapevine and wrote emails to SIBBG members attempting to disprove them. l:SER:42-43, 55-56, 127-29. SIBBG did not make any decisions about the S29 during the 2007 backpacking season but waited until its fall meeting to discuss the summer’s events. During the fall meeting, SIBBG concluded that the S29 had failed at least six times and decided to recommend withdrawal of conditional approval. 2:ER:27-28. The Park Service accepted this recommendation and notified Ursack of SIBBG’s decision in November 2007. 2:ER:155-57. However, as previously mentioned, the Forest Service did not withdraw its approval of the Ursack. In response to SIBBG’s decision, Ursack filed the present action against SIBBG and the Park and Forest Services.

II.

Before proceeding to the merits of the appeal, we clarify several issues relating to the identity of the proper defendants. First, although Ursack named SIBBG as an agency defendant, it now concedes that SIBBG is not an “agency” within the meaning of the APA. 5 U.S.C. § 701 (b)(1). Moreover, SIBBG did not make any final decisions that could be reviewed in an action under the APA. The agency that made the final decision was the Park Service, and the relevant decision-makers within that agency were the Superintendents of Yosemite and SEKI. However, the Superintendents’ decisions were based on SIBBG’s recommendations, and so, as a practical matter, SIBBG’s decision is the decision under review. Nonetheless, SIBBG itself is not a proper defendant.

Ursack also sued the Forest Service and the Supervisor of Inyo National Forest. However, as noted, the Forest Service and Inyo do not prohibit members of the public from using the Ursack and currently do not maintain lists of “approved” containers. The defendants contend that, for this reason, there is no live case or controversy against the Forest Service or Inyo’s supervisor. See, e.g., Kittel v. Thomas, 620 F.3d 949, 951 (9th Cir.2010) (“The Constitution limits the jurisdiction of the federal courts to live cases and controversies, and as such, federal courts may not issue advisory opinions.”). Ursack concedes that Inyo currently allows use of its products but points out that Inyo’s decisions regarding bear-resistant containers are reviewed annually and that “[a] final judgment in this case would influence any future decision by Inyo to ban Ursack based on SIBBG’s 2007 recommendation.” Reply Br. at 2. While that may be true, it does not create a live case or controversy between Ursack and the Forest Service. Thus, the Forest Service and the Supervisor of Inyo National Forest are not proper defendants.

Accordingly, we limit our review to the decisions of the Park Service and the Superintendents of Yosemite and SEKI.

III.

Turning to the merits, the central issue on appeal is whether the Park Service’s decision to revoke conditional approval of the S29 was arbitrary and capricious. See 5 U.S.C. § 706 . Although Ursack separately argues that the decision also violated equal protection principles, the equal protection argument can be folded into the APA argument, since no suspect class is involved and the only question is whether the defendants’ treatment of Ursack was rational (i.e., not arbitrary and capricious). See United States v. Weston, 255 F.3d 873, 880 (D.C.Cir.2001) (describing the rational basis test and the arbitrary and capricious standard as “analogues”). In addition to its arbitrary and capricious arguments, Ursack argues that the Park Service violated the “licensing” provisions of the APA, 5 U.S.C. § 558 . These provisions specify procedures that an agency must follow before it revokes a license. Ursack alleges that the conditional approval of the S29 was a license and that, in withdrawing conditional approval, the Park Service failed to provide Ursack with the necessary process. We address these issues below.

A.

Under the APA, a reviewing court may set aside agency actions that are “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706 (2)(A). Review under this standard is narrow, and the court may not substitute its judgment for that of the agency. Earth Island Inst. v. Carlton, 626 F.3d 462, 468-69 (9th Cir.2010). Rather, a decision may be reversed as arbitrary and capricious only if the agency relied on improper factors, entirely failed to consider an important aspect of the problem, or offered an explanation 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. Id. An agency action will not be reversed as arbitrary and capricious where the agency is able to demonstrate a “rational connection between the facts found and the conclusions made.” Native Ecosystems Council v. U.S. Forest Service, 418 F.3d 953, 960 (9th Cir.2005) (internal quotation marks omitted).

Ursack argues that SIBBG’s decision to revoke conditional approval of the S29 was arbitrary and capricious for several reasons: (1) in deciding to apply a three-strikes standard to the S29, SIBBG failed to consider an important aspect of the problem; (2) SIBBG’s decision to apply the three strikes standard to the S29 was arbitrary given SIBBG’s decision to not revoke the approval of a competing bear canister, the BearVault, after it failed twelve times; and (3) SIBBG’s decision to prohibit users from tying the S29 to trees was capricious.

1. Use of “three strikes” standard

Ursack contends that in deciding to revoke approval of the S29 based on three failures,-SIBBG failed to consider an important aspect of the problem. See Motor Vehicle Mfrs. Ass’n, Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 , 103 S.Ct. 2856 , 77 L.Ed.2d 443 (1983) (agency decision is arbitrary and capricious when the agency “entirely fail[s] to consider an important aspect of the problem”). In Ur-sack’s view, SIBBG failed to consider whether use of the Ursack — which is lighter and easier to carry than hard-sided canisters — might increase public compliance with park and forest food-storage requirements. Ursack believes that it’s better to have a handful of Ursack failures than to have a large number of incidents in which bears obtain food because a backpacker decided to store his or her food improperly. Thus, argues Ursack, SIBBG should have considered the extent to which use of the Ursack might increase visitor compliance before deciding that its approval would be revoked after only three failures.

However, the administrative record shows that SIBBG did not entirely fail to consider this aspect of the problem. In debating whether to conditionally approve the Ursack for the 2007 season, several members of SIBBG cited increased compliance as a reason favoring approval. During the debate, one SIBBG member, Harold Werner, stated the following in an email to other members of the group:

l:SER:70-75 (quote appears on page 74). Similarly, when Craig Axtell, the Superintendent of SEKI, suggested that SIBBG reconsider its denial of Ursack’s conditional approval for the 2007 season, he wrote:

2:SER:419-20. At least one other member of SIBBG, Peter Rowland, acknowledged this point and countered with an opposing viewpoint — specifically, that the compliance rate was aiready at 91% and that “attaining an extra few percent compliance” might not be “worth putting up with a product that currently appears to have a relatively high rate of failure compared to its competitors.” 2:SER:443-44.

The above exchange indicates that SIBBG did not entirely fail to consider the issue of increased compliance. To the contrary, SIBBG balanced the benefit of increased compliance against the risk of container failure during the process of deciding whether and on what terms the Ursack would be conditionally approved. In its reply brief, Ursack acknowledges that SIBBG did not entirely fail to consider this issue, but it argues that SIBBG approached the issue from the wrong angle. Ursack points out that the 91% compliance rate cited by Rowland refers to the percentage of visitors who carry canisters. Ursack’s position is that even though 91% of backpackers carry canisters, they often carry more food than they can fit into a single canister, which results in them storing “over-flow” food outside the canister. Thus, argues Ursack, SIBBG shouldn’t have asked whether approving the Ursack would have raised the compliance rate above 91%, it should have asked whether approving the Ursack would have resulted in better compliance even among the 91% of backpackers who already carry canisters.

The problem with this argument is that nothing in the administrative record indicates that the Ursack was such an obvious solution to the overflow food issue that the agency’s failure to consider the compliance issue from that angle was a failure to consider an important aspect of the problem. The record does not show that backpackers can fit more food into a single Ursack than into a single hard-sided canister. Thus, backpackers would still need to carry more than one Ursack (or a combination of Ursacks and canisters) in order to properly store all food. Further, the record does not suggest that backpackers would choose to carry multiple storage containers of any kind rather than store some overflow food improperly. Although Ursacks are lighter, they still take up pack space, and the record indicates that the aluminum lining would have to be removed from an Ursack before a backpacker would even consider carrying a second one. 2:ER:144, 148. Yet, as noted, SIBBG deemed the aluminum liner necessary to prevent food spoilage and tree damage. So Ursack is not an obvious solution to the overflow food problem, and for this reason the agencies did not entirely fail to consider an important aspect of the problem by approaching the issue from the standpoint of whether Ursack’s light weight would increase the number of backpackers who carried at least one food-storage container.

Moreover, the record does not show that SIBBG ignored the overflow food problem entirely. Although Rowland’s statement about compliance indicated that he was concerned about whether backpackers would carry food-storage containers at all, this does not mean that SIBBG did not also consider whether Ursaeks might reduce the number of bear incidents caused by overflow food. Both Werner and Axtell considered whether Ursaeks might increase “compliance,” and they did not say anything indicating that they were excluding overflow food from their definition of noncompliance. To the contrary, Werner was well aware that many bear incidents were caused by overflow food, 1:SER:62, as was SIBBG in general, 2:ER:144 (letter referring to SIBBG map informing backpackers that “most negative bear encounters occur due to overflow food”). Thus, although SIBBG members did not explicitly mention overflow food in the course of their debate over conditional approval of the Ursack, we cannot conclude that they ignored this aspect of the problem altogether.

2. Whether use of “three strikes” standard was violation of equal protection

Ursack argues that SIBBG’s decision to revoke approval of its product after three failures in 2007 amounted to a denial of equal protection because a competing product, the BearVault, failed twelve times in 2005 but did not have its approval revoked. Because SIBBG’s decision did not involve a suspect classification, rational basis scrutiny applies, see, e.g., United States v. Flores-Villar, 536 F.3d 990, 998 (9th Cir.2008), and thus the standard of review is identical to arbitrary and capricious review under the APA. See Weston, 255 F.3d at 880 .

The record indicates that SIBBG had a rational basis for revoking approval of the Ursack in 2007, even though it declined to revoke the BearVault’s approval in 2005 after it failed a dozen times — namely, SIBBG’s conclusion that almost all of the BearVault failures were likely caused by the same bear. As a SIBBG member explained to Ursack in an email at the end of the 2005 season:

2:ER:141. In contrast, SIBBG did not conclude that the six Ursack failures in 2007 were caused by the same bear.

In its reply brief, Ursack acknowledges this difference but speculates that the BearVault failures were caused by multiple bears and that most of the Ursack failures were caused by a single bear. But we must defer to the agency’s finding on these matters unless the record shows that the agency’s findings were not supported by substantial evidence — i.e., unless the evidence in the record “would compel a reasonable finder of fact to reach a contrary result.” Gebhart v. SEC, 595 F.3d 1034, 1043 (9th Cir.2010) (internal quotation marks omitted, emphasis in original). In the present case, Ursack tells us that, given the average home range of a black bear and the locations of the various container failures, it is “likely” that not all BearVault failures were caused by a single bear and that all but two of the Ursack failures in 2007 were caused by a single bear. Reply Br. at 6-8. However, the agency is better equipped than the court to determine how many bears were involved in the relevant incidents, and Ur-sack’s statement about the home range of a black bear (which is not supported by citations to the administrative record or anything else) does not show that a reasonable finder of fact would have been compelled to see things Ursack’s way. Thus, we must accept the agency’s findings on these matters and conclude that the distinctions SIBBG made between the BearVault and the Ursack were rational.

3. Whether refusal to allow users to tie Ursacks to trees was capricious

Ursack’s final arbitrary-and-eapricious argument involves SIBBG’s decision to approve Ursacks only if Ursack redesigned the product to eliminate the need to tie it to trees. Ursacks were originally designed to be tied to trees, and tying had three benefits: (1) the more a bear pulls on an Ursack, the tighter the cord cinching the opening becomes; (2) a bear can’t take the Ursack and abandon it in the woods; and (3) a bear can’t use its weight to crush the bag’s contents. In 2001, when SIBBG conditionally approved Ursack’s earliest model, it allowed users to tie Ursacks to trees. In fact, it conditioned its approval of the product on Ursack revising its instructions to clarify that the product had to be tied to a tree or rock. 2:ER:118-19. However, after SIBBG conducted its extensive reevaluation of the Ursack TKO in 2004, it discovered that tying Ursacks to trees resulted in damage to the bark and substrate around the trees. 2:ER:130-31. In its letter explaining its decision to deny approval of the Ursack for the 2005 season, SIBBG cited tree damage as one of its reasons. 2:ER:130-31.

In response to this letter, Ursack retained a silviculture specialist, Kevin O’Hara, who wrote a letter to SIBBG in which he opined that the tree damage caused by tying Ursacks to trees would be no more extensive than the tree damage caused by routine wild bear activity. 2:ER:132-34. Harold Werner, a member of SIBBG, responded to this letter. 2:SER:298-99. He explained that several of O’Hara’s assumptions were incorrect, and that he believed that the tree damage would be more extensive than damage caused by routine bear activity. Werner also explained that SIBBG would have wanted to prevent the damage even if it were no more extensive that the damage caused by routine bear activity, since the Park Service was obliged to minimize human impact on the backcountry. He explained that any tree damage caused by bears struggling with Ursacks would be attributable to human activity — tying Ur-sacks to trees — and thus even if the damage were no more extensive than the damage bears could be expected to generate in the absence of human activity, the fact that the damage was attributable to a human influence was reason enough to prevent it.

Ursack did not further challenge SIBBG’s reasoning as to tree damage and elected to redesign the product so that users did not need to tie it to trees. As noted, this redesign resulted in the thick aluminum insert, which was intended to both address the issue of food spoilage due to bear saliva and eliminate the need to tie Ursacks to trees. When SIBBG conditionally approved the Ursack S29 for use in 2007, it did so on the condition that backpackers would use it like any other bear container — by placing it on the ground without securing it to a tree. 2:SER:420. Ursack argues that this decision was capricious because (1) SIBBG initially required that Ursacks be tied to trees, and thus SIBBG acted capriciously (i.e., impulsively or unpredictably) by changing course, and (2) SIBBG’s concerns about tree damage are inconsistent with regulations allowing backpackers to hang food from trees in certain areas of the park.

Ursack’s first argument is easily disposed of. Although SIBBG initially had no objection to tying Ursacks to trees and asked Ursack to revise its instructions to make the need to tie Ursacks to trees more explicit, that was before SIBBG conducted the extensive tests of the Ursack in 2004 and noticed the tree-damage issue. Thus, SIBBG’s decision to change course was reasoned rather than capricious.

Ursack’s second argument stems from regulations that permit park and forest managers to require visitors to hang their food from tree branches. Pursuant to these regulations, food is suspended several feet from the trunk of the tree so that a bear can’t reach the food if it climbs the trunk. Although this method of storing food is permitted (and sometimes required) in certain areas of the parks and forests, it is not permitted in areas where food must be stored in bear-resistant containers. The reason for this is that in the container-only areas, bears have learned how to reach food that has been stored in trees.

Ursack argues that if visitors may hang food from trees in certain areas of the parks and forests — thereby causing some tree damage — then SIBBG cannot rationally prohibit visitors from tying Ursacks to trees in the parts of the parks and forests where tree storage is prohibited. Ursack’s position is that if the Park and Forest Services tolerate tree damage caused by food storage anywhere, they must tolerate it everywhere. In this regard, Ursack points to the fact that the prohibition on storing food in trees in the container-only areas was imposed because bears had learned how to reach food stored in trees, not because the Park and Forest Services wanted to prevent tree damage in those areas.

As far as record reveals, Ursack never raised this issue in its various communications with SIBBG members, and thus we don’t know what the agency would have said about it. However, a rational basis for SIBBG’s decision is readily apparent: Although the primary reason for prohibiting tree storage in container-only areas was that bears had learned how to obtain food stored in trees, the prohibition also had the beneficial effect of eliminating tree damage caused by human influences in those areas. In evaluating the Ursack for use in container-only areas, then, SIBBG members were rationally concerned about approving a food-storage container that might reestablish anthropogenic tree damage in areas where it had been eradicated. Accordingly, SIBBG’s tree-damage rationale was not arbitrary or capricious.

B.

Ursack’s remaining argument is that conditional approval of the Ursack was a “license” within the meaning of the APA, 5 U.S.C. § 551 (8), and that therefore SIBBG was required to follow the procedures in 5 U.S.C. § 558 (c) before revoking conditional approval. Under § 558(c), a licensee must receive the following process before an agency may revoke a license: (1) notice by the agency in writing of the facts or conduct which may warrant the revocation, and (2) opportunity to demonstrate or achieve compliance with all lawful requirements. The purpose of requiring this process is to provide a licensee threatened with the termination of its license an opportunity to correct its transgressions— i.e., to give the licensee a “second chance.” Buckingham v. Sec’y Dep’t of Agric., 603 F.3d 1073, 1085 (9th Cir.2010); Gallagher & Ascher Co. v. Simon, 687 F.2d 1067, 1074 (7th Cir.1982).

Under the APA, “license” is defined as “the whole or part of any agency permit, certificate, approval, registration, charter, membership, statutory exemption or other form of permission.” 5 U.S.C. § 551 (8). Ursack contends that because its products can be used in certain areas of the relevant parks only if the Park Service approves them, the Park Service’s “approval” amounts to a license. However, Ursack does not need the Park Service’s approval to manufacture or sell its products to the public (or to engage in any other activity). For this reason, the present case is unlike others in which we have determined that an agency had granted a license. In Anchustegui v. Department of Agriculture, for example, we found that a permit to graze sheep on federal land was an APA license. 257 F.3d 1124 , 1128-29 (9th Cir. 2001). And in Air North America v. Department of Transportation, we determined that an airline’s certificate of authority to provide air transportation was a license. 937 F.2d 1427 , 1437 (9th Cir.1991). In both of these cases the absence of agency approval prevented the purported licensee from engaging in a specific activity— grazing on federal land in Anchustegui and providing air transportation in Air North America. But in the present case, the only consequence to Ursack of SIBBG’s revocation of conditional approval is financial. Even without conditional approval, Ursack can manufacture and sell as many Ursacks as it pleases, but the lack of conditional approval will have an impact on the market for its products. Thus, the question is whether an agency decision that does not grant a form of permission to a member of the public nonetheless qualifies as a license due to the decision’s financial consequences.

The Seventh Circuit has answered this question in the negative. In Horn Farms, Inc. v. Johanns, 397 F.3d 472 (7th Cir.2005), the plaintiff argued that the Department of Agriculture’s decision to terminate his participation in a farm-subsidy program was a decision to revoke a “license” within the meaning of § 551(8). The decision did not prevent the plaintiff from farming anything; rather, it eliminated his eligibility for subsidy payments in connection with his farming activities. The plaintiff argued that the Department of Agriculture had to “approve” his participation in the subsidy program, and that therefore the revocation of approval was a revocation of a license within the meaning of § 558. In rejecting this argument, the court reasoned that an agency decision does not involve a license simply because it has financial consequences. Id. at 478-79. Although the court recognized that the APA’s definition of license is broad and “should be read so that it encompasses all situations in which federal approval is required to undertake some act,” the court concluded that the definition is not broad enough to encompasses situations in which “no agency stands as a gatekeeper to a proposed private activity.” Id. The court reasoned that since the plaintiff could farm as much or as little as he wanted with or without the Department of Agriculture subsidy, termination of the subsidy was not termination of a license.

In reaching its conclusion, however, the Seventh Circuit did not acknowledge contrary authority from the Second Circuit, New York Pathological & X-Ray Laboratories, Inc. v. INS, 523 F.2d 79 (2d Cir.1975). In that case, the INS adopted a regulation requiring aliens seeking permanent resident status to obtain a medical examination from a “selected civil surgeon” — i.e., a medical facility that had been approved by the INS. See 8 C.F.R. § 234.2 (1973). Even before it adopted this regulation, the INS required aliens to obtain medical examinations, but aliens were permitted to use a physician of their choice. The plaintiff was a medical facility that, prior to the adoption of the new regulation, had provided the needed medical examinations to aliens. When the INS adopted the new regulation, it refused to designate the plaintiff as a selected civil surgeon, and so aliens could not use the results of the plaintiffs examinations to satisfy the examination requirement. The plaintiff challenged the INS’s refusal, arguing that the designation of a medical facility as a selected civil surgeon was a “license” within the meaning of § 558. The Second Circuit agreed. New York Pathological, 523 F.2d at 82 ; see also Blackwell College of Business v. Attorney General, 454 F.2d 928, 932 (D.C.Cir.1971) (INS’s revocation of status as an approved school for attendance by nonimmigrant alien students was revocation of an APA license).

New York Pathological implies that agency approval can constitute a license within the meaning of the APA even if the lack of approval does not prohibit the purported licensee from engaging in any private activity. The lack of the “selected civil surgeon” designation did not prevent the plaintiff medical facility from treating or examining any particular patient, but because the lack of the designation prevented aliens from using the results of the plaintiffs exams as part of their applications for permanent residency, the agency’s decision had a substantial financial impact on the plaintiff, since no alien would pay for a medical exam if the results of the exam could not be used as part of an application. Similarly, although Ursack can manufacture and sell Ursacks with or without the Park Service’s approval, the decision to revoke Ursack’s conditional approval had a substantial impact on the market for Ursack’s products.

We conclude that we need not decide whether an APA license exists only where the agency stands as a gate-keeper to a proposed private activity, or whether it also extends to forms of agency approval carrying only financial consequences. This is because even if Ursack had been granted a license, it is not seeking the kind of process available to licensees under § 558(c). Ursack is not asking for notice and a chance “to demonstrate or achieve compliance with all lawful requirements.” 5 U.S.C. § 558 (c). Instead, Ursack wants to argue with SIBBG over its decision to adopt the relevant lawful requirements in the first place. Recall that SIBBG decided to revoke conditional approval on the ground that the Ursack had failed six times during the 2007 season. Ursack does not dispute that its products were involved in six incidents during the season. But Ursack believes that most of these incidents shouldn’t count as failures for various reasons, such as that the incident did not involve a bear obtaining food or that a bear obtained food but the failure was caused by user error. Br. at 28-29. However, in granting conditional approval, SIBBG defined “failure” in a way that makes all six incidents failures even under Ursack’s view of the physical evidence— i.e., even if no bear obtained food or the failure was caused by user error. 1:SER:97. Thus, to have convinced SIBBG not to revoke conditional approval, Ursack would have had to have convinced SIBBG to change its licensing criteria. Yet, challenges to licensing criteria are adequately handled through review under the arbitrary and capricious standard. 5 U.S.C. § 706 . As discussed, the process afforded by the APA’s licensing provisions is limited to providing the licensee with notice and a chance to either demonstrate compliance or cure its transgressions, and Ursack does not claim that it was injured by the Park Service’s failure to provide this process. Accordingly, we reject Ur-sack’s argument on the ground that § 558(c) does not entitle Ursack to the relief it requests, even if Ursack had been granted a license.

C.

Before concluding, we note that Ursack raised a new argument for the first time in its reply brief: that Yosemite and SEKI have arbitrarily and capriciously refused to consider granting conditional approval to Ursack’s latest product, the S29 “All White.” Reply Br. at 10-11. On April 7, 2010, Ursack wrote to the Superintendents of Yosemite and SEKI and asked them to approve the All White for use in the container-only areas of the parks. See Further Excerpts of Record (“FER”) at 8-6. The Assistant Field Solicitor for the Park Service responded on behalf of the Park Service as follows:

FER at 1. Ursack contends that the Park Service’s refusal to consider the All White or any other hard- or soft-sided container is arbitrary and capricious.

Although this argument was raised for the first time in Ursack’s reply brief, at the time Ursack filed its opening brief the events that gave rise to this argument had not yet occurred. Nonetheless, the final agency action under review in this lawsuit is the decision to revoke conditional approval of the S29 at the end of the 2007 season. The agencies’ decision to stop accepting food-storage containers for evaluation is a separate agency action. Moreover, the agencies have not yet had a chance to compile an administrative record for this decision or otherwise explain why their decision is not arbitrary and capricious. We are therefore in no position to review the decision. Should Ursack wish to challenge this decision, the proper procedure would be to file a fresh lawsuit under the APA.

IV.

For the foregoing reasons, we AFFIRM the judgment of the district court.

. Three individuals who desire to use the Ur-sack in the relevant national parks and forests joined Ursack as plaintiffs. However, none of plaintiffs’ arguments depends on the presence of the individual plaintiffs. Therefore, throughout the course of this opinion, we write as if Ursack were the only plaintiff. . This equal protection claim arises under the Fifth Amendment’s Due Process Clause. See, e.g., Buckley v. Valeo, 424 U.S. 1, 93-94 , 96 S.Ct. 612 , 46 L.Ed.2d 659 (1976). . Although Ursack separately challenges SIBBG's "tree damage" rationale, it does not challenge SIBBG’s conclusion that the aluminum liner was needed to prevent food spoilage. . SIBBG’s decision to revoke conditional approval of the S29 is best characterized as an informal agency adjudication, and the "substantial evidence” test normally applies only in formal agency proceedings. See 5 U.S.C. § 706 (2)(E). However, courts recognize that, as a practical matter, the arbitrary and capricious standard incorporates the substantial evidence test, and they therefore review factual findings made during the course of informal agency adjudications under the substantial evidence test. See Bonnichsen v. United States, 367 F.3d 864 , 879-80 & n. 19 (9th Cir.2004); Ass'n of Data Processing Serv. Orgs., Inc. v. Bd. of Governors of the Fed. Reserve Sys., 745 F.2d 677 , 683-84 (D.C.Cir.1984) (Scalia, J.). . Ursack also directs our attention to the minutes of a SIBBG meeting from October 2005 — a meeting that took place after the 2004 testing revealed the tree-damage issue. 2:SER:268. The minutes state that Ursack should revise its instructions to malte clear that Ursacks had to be tied to tree trunks rather than tree branches. Ursack suggests that this is evidence of capricious decision-making, since by this time SIBBG had supposedly decided that Ursacks could not be tied to trees at all. However, despite this reference in the minutes, SIBBG did not ask Ursack to revise its instructions. Instead, SIBBG informed Ursack that its products could not be tied to trees at all. 2:ER:301-02. Thus, SIBBG’s ultimate decision was consistent with the findings of its 2004 testing.