dissenting:
I respectfully but emphatically dissent.
The issue in this case is whether the Endangered Species Act (“ESA”) requires the U.S. Forest Service to consult with appropriate agencies of the federal government before approving a Notice of Intent (“NOI”) to conduct suction dredge mining in the Klamath National Forest. Section 7(a)(2) of the ESA requires that a federal agency consult with one or both of the Fish and Wildlife Service and the National Marine Fisheries Service to ensure that any “agency action” is “not likely to jeopardize the continued existence” of any endangered or threatened species or “to result in the destruction or adverse modification of habitat of such species.” 16 U.S.C. § 1536(a)(2). Consultation is required under Section 7(a)(2) whenever agency action “may affect listed species or critical habitat.” 50 C.F.R. § 402.14(a).
An NOI is required when suction dredge mining “might cause significant disturbance of surface resources.” 36 C.F.R. § 228.4(a). Mining is not allowed unless the NOI is approved by the Forest Service. “Surface resources” include underwater fisheries habitat. Id. at § 228.8(e). The Klamath River system is “critical habitat” for coho salmon, a listed species.
There are two questions before us.
The first is whether Forest Service approval of NOIs to conduct suction dredge mining in the Klamath National Forest is “agency action” within the meaning of Section 7(a)(2). Under our established case law, there is “agency action” whenever an agency makes a discretionary decision about whether, or under what conditions, to allow private activity to proceed. The record in this case shows that District Rangers in the Klamath National Forest made discretionary decisions about whether, and under what conditions, to allow suction dredge mining to proceed under NOIs.
The second is whether suction dredge mining under NOIs (which, by definition, is mining that “might cause significant disturbance” to fisheries habitat) “may affect” critical habitat of the listed coho salmon. The record in this case shows such mining satisfies the “may affect” standard.
I would therefore hold that the Forest Service must consult with the Fish and Wildlife Service and the National Marine *997Fisheries Service1 before allowing suction dredge mining to proceed under NOIs in the Klamath National Forest.
I. Background
The Karuk Tribe has inhabited what is now northern California since time immemorial. The Klamath River originates in southeastern Oregon, runs through northern California, and empties into the Pacific Ocean about forty miles south of the California-Oregon border. As it runs through northern California, the Klamath River passes though the Klamath National Forest. The Klamath River system is home to several species of fish, including coho salmon. Coho salmon in the Klamath River system were listed as “threatened” under the ESA in 1997. 62 Fed.Reg. 24588 (May 6, 1997). The Klamath River system was designated a “critical habitat” for coho salmon in 1999. 64 Fed.Reg. 24049 (May 5,1999).
The rivers and streams of the Klamath River system contain gold. Commercial gold mining in and around the rivers and streams of California was halted long ago due to the extreme harm to the environment caused by large-scale placer mining. See generally Charles N. Alpers et al., Mercury Contamination from Historical Gold Mining in California, U.S. Geological Survey Fact Sheet 2005-3014 (Oct. 2005); Green Versus Gold: Sources in California’s Environmental History (Carolyn Merchant ed., 1998); Scott Fields, Tarnishing the Earth, Environmental Health Perspectives 109:10 (Oct. 2001). However, small-scale recreational mining has continued. Some recreational miners “pan” for gold by hand, examining one pan of sand and gravel at a time. Others use mechanical suction dredging devices.
Suction dredge miners use gasoline-powered engines and hoses to suck rock, gravel and sand from streambeds. The material sucked from the streambed is discharged into a sluice box. As the material flows through the box, a small amount of the heavier material, including gold, is slowed by “riffles” and is then captured in the bottom of the box. The remaining material runs through the box and is deposited in a tailings pile in or beside the stream. The suction dredges at issue typically have intake hoses four or five inches in diameter. Dredging depths are usually about five feet, but can be as great as twelve feet.
The majority attempts to minimize the impact of suction dredge mining, stating it is “best described” as moving “a few cubic inches at a time” and “affect[ing] about one quarter of a cubic yard of the river.” Maj. Op. at 983, 986. A typical suction dredge picks up from the bottom of the stream and deposits in a tailings pile about one-quarter of a cubic yard of material per day. A cubic yard contains 11,664 cubic inches. Many square yards of stream bottom are scoured in order to obtain one-quarter of a cubic yard of movable material per day, but the record does not tell us how many.
The Karuk Tribe contends that suction dredge mining adversely affects fish, including coho salmon, in the Klamath River system. The Tribe brought suit in federal district court in 2004 seeking to limit suction dredge mining in the Klamath National Forest under the National Forest Management Act (“NFMA”), the National Environmental Policy Act (“ÑERA”), and the ESA. The Tribe alleged that the Forest Service defendants violated these statutes when they allowed suction dredge mining under Notices of Intent (“NOIs”) *998and Plans of Operation (“PoOs”). The district court granted judgment in 2005, but briefing on appeal was stayed by agreement of the parties until we decided a case involving suction dredge mining in the Siskiyou National Forest in Oregon. Siskiyou Regional Educ. Project v. U.S. Forest Service, 565 F.3d 545 (9th Cir.2009).
The Tribe prevailed in the district court in this case in its challenge to the Forest Service’s approval of large-scale suction dredge mining under PoOs. By stipulation filed in the district court in April 2005, the Forest Service defendants agreed that “each of the challenged PoOs were approved without compliance with the ESA, NEPA, and their implementing regulations.” That is, the Forest Service agreed that it had to prepare appropriate documents under NEPA and to consult with the appropriate federal agencies under the ESA before approving any PoO.
However, the district court agreed with the Forest Service that compliance with NEPA and the ESA was not required for suction dredge mining allowed under approved NOIs. On appeal, the Tribe does not contend that the Forest Service must comply with NEPA before approving an NOI. But it does contend that the Forest Service must consult with appropriate federal agencies under Section 7(a)(2) of the ESA before approving an NOI. For the reasons that follow, I strongly agree with the Tribe.
II. Regulation of Suction Dredge Mining
An approved NOI is required for any suction dredge mining that “might cause significant disturbance of surface resources.” 36 C.F.R. § 228.4(a) (emphasis added). An approved PoO is required for suction dredge mining that “will likely cause significant disturbance of surface resources.” Id. § 228.4(a)(3) (emphasis added). That is, an approved NOI is required for all suction dredge mining for which the likelihood of a “significant disturbance of surface resources” falls between “might cause” and “will likely cause.”
The Department of Agriculture defines “surface resources” as including underwater “fisheries ... habitat.” Id. § 228.8(e). See 70 Fed.Reg. at 32718 (“Section 228.8 characterizes fisheries habitat as a ‘National Forest surface resource[.]’ Fisheries habitat, of course, can consist of nothing other than water, streambeds, or other submerged lands.”).
The Department recognizes that the effects of suction dredge mining vary substantially from one site to another. It wrote in a 2005 commentary:
The environmental impacts of operating suction dredges, even small ones, are highly site-specific depending on the circumstances and resource conditions.... Given this variability, the Department believes that, insofar as suction dredge operations are concerned, the need for the prior submission of a notice of intent to operate or for the prior submission and approval of a proposed plan of operations must be evaluated on a site-specific basis.
70 Fed.Reg. at 32720.
The Department has made clear, in a response to a comment directed to 36 C.F.R. § 228.4(a), that an NOI for suction dredge mining is not a “regulatory instrument,” but rather “simply ... a notice given to the Forest Service by an operator which describes the operator’s plan to conduct operations on NFS lands.” Id. at 32728; 36 C.F.R. § 228.4(a)(2) (“The District Ranger will, within 15 days of a notice of intent to operate, notify the operator if approval of a plan of operations is required before the operations may begin.”). However, in that same response, the Depart*999ment also made clear that requirements for NOIs vary substantially depending on the site:
[TJhere can be no definitive answer to the question of what level of activity requires the submission of a notice of intent to conduct operations. As previously mentioned ..., given the variability of the lands within the NFS subject to the United States mining laws, identical operations could have vastly different effects depending upon the conditions of the lands and other surface resources which would be affected by those mining operations.... [I]n many cases the need for the submission of a notice of intent to operate must be determined based upon a case-by-case evaluation of the proposed operations and the hinds of lands and other surface resources involved.
70 Fed.Reg. at 32728 (emphasis added).
The majority writes that the Forest Service decision to allow mining to proceed under an NOI is “most accurately described as a decision not to act.” This is a critical point, and the majority is wrong. The Forest Service makes an actual decision whether to allow suction dredging to proceed pursuant to an NOI. As I will describe in detail below, there are seven NOIs in the record in this case. One was withdrawn. Of the remaining six, the Forest Service acted affirmatively to approve four and to deny two. There is no non-withdrawn NOI in the record that the Forest Service did not act affirmatively to approve or deny. The miners whose NOIs were approved each received a letter from the Forest Service District Ranger stating that “You may begin your mining operations when you obtain all applicable state and federal permits” (emphasis added). No miner was allowed to engage in suction dredge mining under an NOI unless that NOI had been explicitly approved by the Forest Service.
III. Notices of Intent
The term “Notice of Intent” is not specific to mining laws. It is a generic term used in a number of different statutory regimes. The discretion available to an agency in approving or denying an NOI varies depending on the statute and the implementing regulations under which the NOI is submitted. We described one such regime in Environmental Defense Center, Inc. v. Environmental Protection Agency, 344 F.3d 832, 853-54 (9th Cir.2003). We explained in our opinion that the EPA regulated stormwater discharges under the Clean Water Act. Some types of discharges were governed by a “general permit” that allowed applicants to submit short NOIs certifying that they would comply with the terms of the general permit. Id.; see also Texas Indep. Prod. & Royalty Owners Ass’n v. EPA 410 F.3d 964, 979 (7th Cir.2005). The EPA provided applicants with a simple form NOI for that purpose.2 Envtl. Def. Ctr., 344 F.3d at 853. We wrote that “because th[is] NOI represents no more than a formal acceptance of terms elaborated elsewhere,” the operator could begin discharges after submitting an NOI without waiting for a response from the EPA. Id.
But not all NOIs for stormwater discharges are ministerial and non-discretionary. Plaintiffs in Environmental Defense Center challenged a different sort of NOI from the one just described. The challenged NOI allowed discharges from small municipal storm systems. Each operator of these small systems was required to submit an NOI that included an “individu*1000alized pollution control program” addressing six criteria. Because the information required in this NOI was quite detailed, we held that this NOI was functionally identical to a permit application. Id. This NOI “crosse[d] the threshold from being an item of procedural correspondence to being a substantive component of a regulatory regime.” Id. at 855.
As is evident from Environmental Defense Center, the mere label “Notice of Intent” does not allow us to determine how much agency discretion is involved in allowing an operator to proceed under an NOI. To make that determination, we must examine the actual practice of the agency with respect to the particular NOI at issue.
IV. Consultation under the Endangered Species Act
Section 7(a)(2) of the ESA requires consultation prior to any “agency action” that “may affect” a listed species or its habitat:
Each Federal agency shall, in consultation with and with the assistance of the Secretary, insure that any action authorized, funded, or carried out by such agency (hereinafter in this section referred to as an “agency action ”) is not likely to jeopardize the continued existence of any endangered species or threatened species or result in the destruction or adverse modification of habitat of such species which is determined by the Secretary, after consultation as appropriate with affected States, to be critical[.]
16 U.S.C. § 1536(a)(2) (emphasis added).
Regulations implementing Section 7 provide:
Each Federal agency shall review its actions at the earliest possible time to determine whether any action may affect listed species or critical habitat. If such a determination is made, formal consultation is required[J
50 C.F.R. § 402.14(a) (emphasis added).
I discuss the “agency action” and “may affect” requirements in turn.
A. “Agency Action”
Congress intended the term “agency action” to have a broad definition. “[Tjhere is little doubt that Congress intended to enact a broad definition of agency action in the ESA[.] ... Following the Supreme Court’s lead in [Tennessee Valley Authority v. Hill, 437 U.S. 153, 98 S.Ct. 2279, 57 L.Ed.2d 117 (1978)], we have also construed ‘agency action’ broadly.” Pacific Rivers Council v. Thomas, 30 F.3d 1050, 1054, 1055 (9th Cir.1994) (statutory citations omitted); see also Western Watersheds Project v. Matejko, 468 F.3d 1099, 1108 (9th Cir.2006) (“[T]he term ‘agency action’ is to be construed broadly[.]”); Natural Res. Def. Council v. Houston, 146 F.3d 1118, 1125 (9th Cir.1998).
The regulations defining “agency action” make clear the breadth of the term:
Action means all activities or programs of any kind authorized, funded, or carried out, in whole or in part, by Federal agencies in the United States or upon the high seas. Examples include, but are not limited to:
(a) actions intended to conserve listed species or their habitat;
(b) the promulgation of regulations;
(c) the granting of licenses, contracts, leases, easements, rights-of-way, permits, or grants-in-aid; or
(d) actions directly or indirectly causing modifications to the land, water, or air.
50 C.F.R. § 402.02 (emphases added). “Section 7 and the requirements of this part apply to all actions in which there is discretionary Federal involvement or con*1001trol.” Id. at § 402.03 (emphasis added). The question before us is whether Forest Service approval of the NOIs at issue was an “action[ ] in which there is discretionary Federal involvement or control,” such that the Forest Service’s approval was “agency action” within the meaning of Section 7.
This circuit has a well-established body of law on discretion and agency action under Section 7 of the ESA. In Turtle Island v. National Marine Fisheries Service, 340 F.3d 969 (9th Cir.2003), we held that Section 7 required the Fisheries Service to consult within its own agency when issuing fishing permits under the High Seas Fishing Compliance Act (“the Compliance Act”). Because the Fisheries Service had discretion whether to issue the permits, the issuance of the permits was agency action. The Service was therefore required to consult under Section 7. We wrote, “Whether the Fisheries Service must condition permits to benefit listed species is not the question before this court, rather, the question before us is whether the statutory language of the Compliance Act confers sufficient discretion to the Fisheries Service so that the agency could condition permits to benefit listed species. We hold that the statute confers such discretion.” Id. at 977 (emphasis in original).
In National Wildlife Federation v. National Marine Fisheries Service, 524 F.3d 917 (9th Cir.2008), we reviewed a biological opinion prepared as part of the consultation process under Section 7. We wrote, ‘When an agency, acting in furtherance of a broad Congressional mandate, chooses a course of action which is not specifically mandated by Congress and which is not specifically necessitated by the broad mandate, that action is, by definition, discretionary and is thus subject to Section 7 consultation.” Id. at 929. In Washington Toxics Coalition v. Environmental Protection Agency, 413 F.3d 1024 (9th Cir.2005), we held that the EPA had to consult with the National Marine Fisheries Service under Section 7 before approving pesticides under the Federal Insecticide, Fungicide, and Rodentieide Act (“FIFRA”). We wrote, “EPA retains discretion to alter the registration of pesticides for reasons that include environmental concerns. Therefore, EPA’s regulatory discretion is not limited by FIFRA in any way that would bar an injunction to enforce the ESA.” Id. at 1033 (statutory citation omitted).
In Natural Resources Defense Council v. Houston, 146 F.3d 1118 (9th Cir.1998), we held that the Bureau of Reclamation had to consult with the National Marine Fisheries Service under Section 7 before renewing contracts with farmers for water from the federal Central Valley Project because “there was some discretion available to the Bureau during the negotiation process” leading up to the renewals. Id. at 1126. Finally, in Pacific Rivers Council v. Thomas, 30 F.3d 1050 (9th Cir.1994), we held that the Forest Service was required to consult under Section 7 before allowing projects under the Land and Resource Management Plans for particular national forests.
If an agency performs an act that does not involve the exercise of discretion, that act is not “agency action” within the meaning of Section 7. For example, in National Association of Home Builders v. Defenders of Wildlife, 551 U.S. 644, 127 S.Ct. 2518, 168 L.Ed.2d 467 (2007), the Supreme Court held that the EPA was required only to find that nine statutory criteria specified in the Clean Water Act (“CWA”) had been satisfied before transferring regulatory authority to a state. Under the CWA, the EPA had no discretion, once these criteria were satisfied, to take any action that would benefit or protect any listed species under the ESA. The Court *1002wrote, “[T]he ESA’s requirements would come into play only when an action results from the exercise of agency discretion. This interpretation [of the CWA and the ESA] harmonizes the statutes by giving effect to the ESA’s no-jeopardy mandate whenever an agency has discretion to do so, but not when the agency is forbidden from considering such extrastatutory factors.” Id. at 665, 127 S.Ct. 2518.
If an agency only has discretion that is unrelated to protecting a listed species, an act by that agency is not “agency action” within the meaning of Section 7. For example, in Sierra Club v. Babbitt, 65 F.3d 1502 (9th Cir.1995), the Bureau of Land Management (“BLM”) had entered into an agreement granting a logging company the right to build new logging roads on BLM land subject to BLM approval under specified criteria. None of the criteria was relevant to the protection of protected species under the ESA. Therefore, there was no “agency action” under Section 7: “[W]e conclude that where, as here, the federal agency lacks the discretion to influence the private action, consultation would be a meaningless exercise; the agency simply does not possess the ability to implement measures that inure to the benefit of the protected species.” Id. at 1509; see also Envtl. Prot. Info. Ctr. v. Simpson Timber Co., 255 F.3d 1073, 1081 (9th Cir.2001) (“[N]owhere in the various permit documents did the FWS retain discretionary control to make new requirements to protect species that subsequently might be listed as endangered or threatened.”).
Sometimes an earlier act dictates later agency actions such that a later act involves no discretion and therefore does not require consultation. For example, in Western Watersheds Project v. Matejko, 468 F.3d 1099 (9th Cir.2006), private parties had been granted vested rights to divert water for irrigation long before the passage of the ESA. The Bureau of Land Management (“BLM”) announced that it would not interfere with those previously vested rights. We held that so long as the private parties limited their activities to those consistent with their vested rights they did not have to notify the BLM of their activities, and the BLM did not have the ability to regulate their activities. Under these circumstances, we concluded that the BLM had not undertaken any discretionary “agency action” that would have required it to consult under Section 7. Id. at 1107-08.
An out-of-circuit example is Texas Independent Producers and Royalty Owners Ass’n v. EPA, 410 F.3d 964 (7th Cir.2005), in which the EPA consulted under the ESA before exercising its discretion to grant a “general permit” authorizing private operators to discharge stormwater under the Clean Water Act. Id. at 979. The operators then filed individual NOIs to discharge in accordance with the conditions of the general permit. Id. at 968. The Seventh Circuit held that the EPA did not have to consult on the individual NOIs because it had already consulted under the ESA before granting the general permit. The terms of the general permit dictated the manner in which stormwater would be discharged, thereby eliminating any discretion by the EPA in approving or denying an individual NOI.
B. “May Affect”
An agency is required to consult when its action “may affect” listed species or designated critical habitat. 50 C.F.R. § 402.14(a). An agency can avoid the obligation to consult only if it determines that its action will have “no effect” on listed species or designated critical habitat. Thomas, 30 F.3d at 1054 n. 8. Once an agency has determined that its action “may affect” listed species or critical habi*1003tat, the agency may proceed with formal consultation or may choose instead to consult informally with the appropriate agency. If the consulting agency determines during informal consultation that the proposed action is “not likely to adversely affect any listed species or critical habitat,” formal consultation is not required and the process ends. 50 C.F.R. § 402.14(b)(1). Thus, actions that have any chance of affecting listed species or critical habitat— even if it is later determined that the actions are “not likely” to do so — require an agency at least to consult informally.
We have previously explained that “may affect” is a “relatively low ... threshold” for triggering consultation. California ex rel. Lockyer v. U.S. Dep’t of Agric., 575 F.3d 999, 1018 (9th Cir.2009). “ Any possible effect, whether beneficial, benign, adverse or of an undetermined character, triggers the formal eonsultation[.]’ ” Id. at 1019 (quoting 51 Fed.Reg. 19926, 19949 (June 3, 1986)) (emphasis in Lockyer). The Secretaries of Interior and Commerce have explained that “the threshold for formal consultation must be set sufficiently low to allow Federal agencies to satisfy their duty to ‘insure’ ” that their actions do not jeopardize listed species or critical habitat under section 7(a)(2). 51 Fed.Reg. at 19949.
In response to concerns that the “may affect” standard is too burdensome, the Secretaries explained that the availability of informal consultations mitigates any burden on the affected agencies. Id. at 19950. The Secretaries therefore rejected the suggestion that the consultation requirement should be triggered on a higher showing than the low “may affect” threshold. Id. at 19949.
V. Discussion
A. Challenged Notices of Intent
Four NOIs are challenged in this appeal. All four are for suction dredge mining in the Happy Camp District of the Klamath National Forest. As noted above, an approved NOI is required for all suction dredge mining for which the likelihood of a “significant disturbance of surface resources” falls between “might cause” and “will likely cause.” 36 C.F.R. § 228.4(a). “Surface resources” includes “fisheries habitat.” Id. § 228.8(e). The Klamath River system is critical habitat for the listed coho salmon.
Before the 2004 dredging season, the Forest Service had issued a two-page generic handout requiring information from operators who sought to engage in suction dredge mining pursuant to an NOI:
Describe what you plan to do. Include when and how you will be operating, the proposed start-up date, and the expected duration of the activities. List other details such as the number of people involved in the operation, equipment you intend to use (sizes, capacity, frequency of use), depth of proposed suction dredging or excavation, how waste material will be handled, what vegetation will be removed, the size of area to be disturbed, quantity of material to be removed, housing or camping facilities to be used, and the method for sewage and waste disposal.
In preparation for the 2004 season, Happy Camp District Ranger Alan Vandiver decided that he needed more information than required by the handout. He was particularly concerned with the effect of suction dredge mining on the critical habitat of listed coho salmon. Vandiver consulted with biologists Bill Bemis and Jon Grunbaum, who are employees of the Forest Service, not the Fish and Wildlife Service or the National Marine Fisheries Service.
Vandiver wrote the following memorandum on May 24, 2004:
*1004On April 20th a meeting was held in Orleans to discuss possible fisheries issues relating to dredging. A number of opinions were shared on the possible effects....
Following the Orleans meeting I asked our District Fisheries biologists, Bill Bemis and Jon Grunbaum, to develop recommendations, for my consideration, for the upcoming dredging season. They were not able to come to agreement on a list of fisheries recommendations. Their opinions varied widely on the effect of dredge operations on fisheries. I identified three key fisheries issues specific to the Happy Camp District[:] cold water refugia areas in the Klamath River, the intensity of dredge activities and the stability of spawning gravels in some portions of Elk Creek. These issues I used to help develop a threshold for determining a significant level of surface disturbance. I felt it was important from a cumulative effects standpoint to determine a threshold of dredge density on the streams, as well as identify the critical cold water refugia areas....
... I discussed at length with Bill [Bemis] and Jon [Grunbaum] the effect on fisheries if the dredge activity was concentrated or dispersed over the length of the river. Concentrated use would result in longer river stretches without dredge activity and therefore less possible impacts to fisheries in the longer stretches. Distributed use would result in dispersed possible effects over the entire length of the river.... Considering the limited dredge operations in cold water refugia areas and the limited dredge access, I developed a threshold of 10 dredges per mile on the Klamath River and 3 dredges per mile on the Klamath tributaries. My thinking was the larger Klamath River, excluding the cold water refugia, could accommodate more dredge density with less impact than the smaller tributaries....
On May 17, 2004 I met with members of the New 49'ers, the Karuk Tribe and our District fisheries biologists to discuss the upcoming dredge season. We discussed the key issues with respect to fisheries including cold water refugia areas in the Klamath River, the intensity of dredge activities and the stability of spawning gravels in the portion of Elk Creek from the East Fork of Elk Creek to Cougar Creek. See notes for May 17th for more detail.
The first of the NOIs challenged in this appeal was submitted by a recreational mining group called the “New 49'ers.” The New 49'ers own numerous mining claims in the Happy Camp District. On May 17, 2004, District Ranger Vandiver met with two representatives of the New 49'ers. Based on his earlier consultation with Bemis and Grunbaum, Vandiver instructed the New 49'ers on “three primary issues.”
First, Vandiver instructed the New 49'ers that cold water refugias must be maintained within 500 feet of the mouths of twenty-two named creeks that fed into the Klamath River. Second, he instructed them that tailings piles must be raked back into the “dredge holes in critical spawning areas” of Elk Creek “in a timely manner as operations proceed, but no later than the end of the season.” Third, he instructed them that there could be no more than ten dredges per mile on the Klamath River, and no more than three dredges per miles on Klamath tributaries.
On May 24, 2004, a week after their meeting with Vandiver, the New 49'ers submitted a detailed eight-page single-spaced NOI for suction dredge mining in the Happy Camp District during the 2004 season. The NOI proposed mining on ap*1005proximately 35 miles of the Klamath River and its tributaries. The NOI estimated that each dredge would move an average of one quarter of a cubic yard of material per day. In accordance with Vandiver’s instructions, the NOI specified that no dredging would occur in specified cold water refugia in the summer and early fall, that dredging holes would be filled in coho salmon spawning grounds on Elk Creek, and that dredge density would not exceed ten dredges per mile on the Klamath River and three dredges per mile on its tributaries.
On May 25, Vandiver sent the New 49'ers a letter approving their NOI. On May 26, Bemis sent a “Note to the File” stating:
The Notice of Intent (NOI) for the new 49'ers this year has an intensity of approximately 40 dredges over the 35 miles of the Klamath covered by their claims. They have agreed to a density of no more than 10 dredges in any one-mile at anytime. The new 49'ers have agreed to avoid the area around tributaries to the Klamath Rivers. The club has agreed to pull back dredging tailings in a critical reach within Elk Creek. These agreements and others explained in the NOI should reduce the impacts to anadromous fisheries on the Happy Camp Ranger District.
The second NOI was submitted by Nida Johnson, an individual miner who planned to mine thirteen claims. She submitted the NOI on May 29, 2004, noting that it was the “result of a meeting at the Happy Camp U.S.F.S. May 25, 2004.” She explained that she was processing ore with dredges with four and five inch intake pipes. She wrote that “[djredge tailings piles in Independence Cr[eek] will be leveled.” In an attachment, she wrote:
As recommended by the Forest Service, no dredging will be conducted on the Klamath River within 500 feet above and below the mouth of Independence Creek between June 15th and October 15th. I totally disagree with these distances and believe that dredging is actually beneficial to fish survival, but -I am willing to follow these recommendations in order to continue with my mining operations.
Vandiver approved the NOI on June 14.
The third NOI was submitted by Robert Hamilton, an individual miner who planned to mine on four claims. He submitted his NOI on May 11, 2004. He stated that he planned to use a four-inch suction dredge for about two weeks during July. Under the heading “precautions,” he wrote that he would limit dredge density to three per mile, and that “Mailings will be returned to dredge hole if possible in shallow areas or spread over large area in deep areas.” Vandiver approved the NOI on June 15.
The fourth NOI was submitted by Ralph Easley, an individual miner who planned to mine on a single claim. He submitted his NOI on June 14. He stated that he planned to use a four inch suction dredge from the beginning of July to the end of September. He stated that the “[d]redge tailings will be raked back into dredge holes.” Vandiver approved the NOI on June 15.
In addition to the four NOIs specifically at issue in this appeal, the record contains information about NOIs for suction dredging in two other districts of the Klamath National Forest — the Orleans and the Scott River Districts. Examination of these two NOIs provides important information about the Forest Service’s practices with respect to section dredge mining pursuant to NOIs.
First, on April 26, 2004, the New 49'ers submitted a detailed eight-page single-spaced NOI for suction dredge mining in the Orleans District. On May 13, Acting *1006Forest Supervisor William Metz refused to approve the NOI. Metz wrote:
There is an important cold water refugia at the mouth of Wooley Creek that was discussed on the April 23, 2004 field trip as needing protection. This was not mentioned in your NOI. Protection of this refugia is critical to the survival of migrating anadromous fish.
Metz wrote further:
Due to the anadromous fisheries in the lower Salmon River the stability of spawning gravels for fish redds [spawning nests] is a major concern. Redds can be lost if loose tailings piles erode away by stream course action while eggs are still present. Your NOI and the California Fish and Game Suction Dredge regulations fall short of addressing mitigations for this issue.
On May 24, the New 49'ers submitted a revised NOI for mining in the Orleans District. Dave McCracken, General Manager of the New 49'ers, wrote in a cover letter to the NOI, “If this Notice does not adequately address your concerns than [sic] I would suggest that we arrange an on-the-ground meeting at the earliest possible time.” Then, anticipating that Metz would still not approve the NOI, the New 49'ers withdrew the revised NOI on May 29. McCracken wrote to Metz:
From the substantial amount of dialog we have had with your office, other District offices, the Supervisor’s office, Karuk Tribal leaders, active members of the Salmon River Restoration Council and others within local communities over the past several months, it has become increasingly clear that there are too many sensitive issues for us to try and manage a group mining activity along the Salmon River at this time.
Second, on April 28, 2004, the New 49'ers submitted a detailed seven-page single-spaced NOI for suction dredge mining in the Scott River District. The NOI proposed an estimated fifteen dredges along fifteen miles of “stream course,” with “[densities of above five dredges per 100 yards ... not anticipated.” The NOI for the Scott River District made a general commitment concerning mining in cold water refugias at the mouths of tributaries. After giving an example of a refugia, the NOI stated, “The 49'ers are committed to working with the Forest Service and [Department of Fish and Game] to identify these areas ... and to adjust their operation to prevent disturbance and stress to these fish during critical time periods.” Unlike the NOIs for mining in the Happy Camp and Orleans Districts, the NOI for the Scott River District made no provision for raking tailings piles back into dredge holes. On May 10, District Ranger Ray Haupt refused to approve the NOI, but for reasons unrelated to protection of fisheries. Haupt wrote,
I am unable to allow your proposed mining operations for the SRRD [Scott River Ranger District] under a NOI because of your bonded campsite which allows your club members to camp (occupancy) longer than the 14 day camping limit. Your current Plan of Operations allows for extended camping (longer than 14 days) for your members, while they are actively engaged in mining. I am approving your mining operations for 2004 under a Plan of Operations with the following conditions ....
None of the conditions in the Plan of Operations related to specific cold water refugia or tailings piles.
In total, there are seven NOIs in the record. Four of them are for suction dredge mining in the Happy Camp District. All four of these NOIs were approved by the Forest Service because they complied with the criteria formulated by *1007District Ranger Vandiver for the protection of the critical habitat of the listed coho salmon. A fifth NOI was submitted for suction dredge mining in the Orleans District. That NOI was denied by the Forest Service because it did not comply with criteria for the protection of critical fisheries habitat. A revised NOI was then submitted, but it was withdrawn in anticipation of its being denied. Finally, a seventh NOI was submitted for suction dredge mining in the Scott River District. That NOI was denied by the Forest Service for reasons unrelated to fisheries habitat.
The Forest Service took affirmative action on all of the six NOIs that were not withdrawn. The Forest Service approved four of them and denied two of them. In no case did the Forest Service take “no action,” as the majority opinion erroneously contends.
B. Consultation under Section 7(a)(2)
As noted above, two criteria must be met before consultation is required under Section 7(a)(2) of the ESA. Those criteria are: (1) there must be a proposed “agency action,” and (2) the proposed agency action “may affect” a listed species or its habitat. I conclude that each of these criteria have been satisfied.
1. Agency Action
The Forest Service takes “agency action” under Section 7(a)(2) of the ESA in deciding whether to approve or deny NOIs for suction dredge mining if it exercises discretion in making that decision. 16 U.S.C. § 1536(a)(2); 50 C.F.R. § 402.03 (Section 7 “applies] to all actions in which there is discretionary Federal involvement or control”).
I conclude that the Forest Service exercised discretion in three ways in approving or denying NOIs for suction dredge mining in the Klamath National Forest. Because the Forest Service exercised discretion in approving or denying these NOIs, it took “agency action” within the meaning of Section 7(a)(2).
First, the Forest Service exercised discretion in formulating criteria for the protection of critical habitat of listed coho salmon. Those criteria governed the approval or denial of NOIs for suction dredge mining. As described in detail above, District Ranger Vandiver of the Happy Camp District prepared for the 2004 mining season by meeting with Forest Service biologists Bemis and Grunbaum. After consulting with them, Van-diver formulated criteria for protecting critical habitat from the effects of suction dredge mining conducted pursuant to NOIs. He specified by name each of the tributaries to the Klamath River that provided cold-water refugias that should be protected; he specified the maximum number of dredges per mile on the river and on its tributaries; and he required that tailings be raked back into dredge holes.
Once Vandiver had exercised his discretion to formulate these specific criteria, they became conditions with which any would-be miner submitting an NOI in the Happy Camp District had to comply. For example, Nida Johnson’s NOI indicated that she would respect a cold-water refugia by refraining from dredging within 500 feet of the mouth of Independence Creek. But she made clear that she was doing so only because of the condition imposed by Vandiver, and that, absent compliance with that condition, she would not be allowed to engage in mining:
I totally disagree with these distances and believe that dredging is actually beneficial to fish survival, but I am willing to follow these recommendations in order to continue with my mining operations.
*1008Similarly, a week after Vandiver had communicated the criteria to the New 49'ers, that group submitted an eight-page single-spaced NOI for suction dredge mining in the Happy Camp District that complied with the criteria. Vandiver approved the NOI the next day.
In one sense, Vandiver is to be commended. He recognized the danger that suction dredge mining posed to the critical habitat of coho salmon, and he consulted with Forest Service biologists Bemis and Grunbaum in formulating protective criteria for approving mining under NOIs. The problem is that Vandiver failed to consult with employees of the required agencies. The ESA requires Vandiver consult with the Fish and Wildlife Service and the National Marine Fisheries Service, not merely within his own agency. Therefore, Vandiver’s consultation with Forest Service biologists Bemis and Grunbaum did nothing to comply with Section 7.
Second, the Forest Service exercised discretion in refusing to approve a detailed NOI submitted by the New 49'ers for suction dredge mining in the Orleans District. Acting Forest Supervisor Metz refused to approve the NOI because, in his view, it provided insufficient protection of fisheries habitat: first, a cold-water refugia at the mouth of a particular creek was not mentioned in the NOI; second, there was insufficient mitigation of the dangers posed by loose tailings piles left by the dredges. The New 49'ers submitted a new NOI, but then withdrew it five days later. The New 49'ers’ representative wrote that despite a “substantial ... dialog,” the Forest Service’s protective conditions meant that “there are too many sensitive issues for us to try and manage a group mining activity along the Salmon River at this time.”
Third, the Forest Service exercised discretion when its employees applied different criteria for the protection of fisheries habitat in different districts of the Klamath National Forest. District Ranger Vandiver developed and applied very specific protective criteria for granting or denying NOIs in the Happy Camp District. Different protective criteria for NOIs were developed and applied in the Scott River District. There is nothing in the record to tell us how the criteria were developed in the Scott River District. But it is clear from the record that those criteria were different, at least in their application, from those in the Happy Camp District. The New 49'ers submitted an NOI to District Ranger Haupt in the Scott River District that complied in full with one of the criteria applied in the Happy Camp District by specifying the maximum number of dredges per mile. The NOI complied, to some degree, with a second Happy Camp criterion by committing to “work[ing] with” the Forest Service to identify cold-water refugia. But the NOI did not promise to observe any particular cold-water refugia and did not promise to stay a specified distance from any creek mouth. Finally, the NOI did not comply at all with the third Happy Camp criterion, for it did not mention raking tailings piles back into dredge holes. Scott River District Ranger Haupt denied the NOI for reasons unrelated to these three criteria, and he did not include these criteria in the Plan of Operations.
A discretionary decision is one that is not dictated or controlled by precise rules or regulations. District Rangers Vandiver and Haupt each formulated and applied their own, differing criteria in deciding whether to grant or deny NOIs for suction dredge mining in their districts. In neither district were those criteria dictated or controlled by precise rules or regulations. See 70 Fed.Reg. at 32720, 32724 (explaining that NOIs must be evaluated on a site-*1009specific basis, and that there is no “universal definition” of “significant disturbance”). This difference in formulating and applying criteria is the very definition of the exercise of discretion.
In every instance in the record before us, except one in which the NOI was withdrawn, the Forest Service affirmatively acted. In each of those instances, it either approved or denied the NOI in which suction dredge mining was proposed. In each instance, the Forest Service took some kind of discretionary action. Those actions were “agency actions” within the meaning of Section 7 of the ESA.
2. “May Affect” Listed Species or Habitat
Section 7 and an implementing regulation require consultation whenever an agency action “may affect ... critical habitat” of a listed species. 50 C.F.R. § 402.14(a). An NOI is required whenever proposed suction dredge mining “might cause significant disturbance of surface resources.” 36 C.F.R. § 228.4(a). “Surface resources” include fisheries habitat. Id. at § 228.8(e). The Klamath River system is a “critical habitat” for listed coho salmon.
Whether suction dredge mining under NOIs “may affect” “critical habitat” can almost be resolved as a textual matter, without the necessity to consult the factual record. That is, by definition, suction dredge mining under an NOI “might cause significant disturbance” of fisheries habitat in the Klamath River system. If the phrase “might cause significant disturbance” of “fisheries habitat” is given an ordinary meaning, it follows almost automatically that suction dredge mining pursuant to an NOI “may affect” critical habitat of the coho salmon. Indeed, the Forest Service does not dispute that suction dredge mining in the Klamath River system pursuant to NOIs “may affect” the listed coho salmon and its critical habitat.
However, the New 49'ers contend that the record “is devoid of any evidence whatsoever that the four challenged suction dredge mining activities ‘may affect’ the coho salmon ‘species’ listed in Northern California.” The New 49'ers make two arguments in support of their contention. Neither argument withstands scrutiny.
First, the New 49'ers argue that there is no evidence “that even a single member of any listed species would be ‘taken’ by reason” of the suction dredge mining at issue. “Take” has a particular definition under the ESA. 16 U.S.C. § 1532(19) (“The term ‘take’ means to harass, harm, pursue, hunt, shoot, wound, kill, trap, capture, or collect, or to attempt to engage in any such conduct.”); Babbitt v. Sweet Home Chapter of Cmties. for a Great Or., 515 U.S. 687, 691, 115 S.Ct. 2407, 132 L.Ed.2d 597 (1995). Even if it is true (which I will assume arguendo) that suction dredge mining does not effectuate a “taking” of coho salmon under the ESA, this has no bearing on whether such mining “may affect” the salmon or its critical habitat under 50 C.F.R. § 402.14(a).
Second, the New 49'ers argue that Van-diver’s consultation process within the Forest Service, and its resulting guidelines, “assured” that there would be “no impact whatsoever on listed species.” This argument cuts against rather than in favor of the New 49'ers. The fact that District Ranger Vandiver formulated his own criteria to mitigate effects of suction dredging on the coho salmon and their critical habitat does not mean that the “may affect” standard was not met. Indeed, the fact that Vandiver consulted with Forest Service biologists in an attempt to reduce any adverse impact on coho salmon and their habitat suggests exactly the opposite.
*1010A review of the record reveals abundant evidence that suction dredging under NOIs in the Happy Camp District “may affect” coho salmon and their critical habitat. Coho salmon in the Klamath River system were listed as “threatened” in 1997, and the river was listed as “critical habitat” two years later. 62 Fed.Reg. 24588, 24588 (May 6, 1997); 64 Fed.Reg. 24049 (May 5, 1999). In listing the salmon, the National Marine Fisheries Service noted that its population was “very depressed.” 62 Fed.Reg. at 24588. The Fisheries Service concluded that “human-induced impacts,” including overharvesting, hatchery practices, and habitat modification including mining had played a significant role in the decline, and had “reduced the coho salmon populations’ resiliency” in the face of natural challenges. Id. at 24591-92. The Fisheries Service also concluded that “existing regulatory mechanisms are either inadequate or not implemented well enough to conserve” the salmon. Id. at 24588.
The record also includes information that Forest Service biologist Grunbaum provided on the effects of suction dredge mining at a meeting of Forest Service personnel on April 20, 2004. Grunbaum wrote that relatively few studies of suction dredging had been performed, but “the majority ... showed that suction dredging can adversely affect aquatic habitats and biota.” The effects varied across ecosystems; in some, “dredging may harm the population viability of threatened species.” Grunbaum summarized specific potential adverse effects. First, “[e]ntrainment by suction dredge can directly kill and indirectly increase mortality of fish — particularly un-eyed salmonid eggs and early developmental stages.” Second, disturbance from suction dredging can kill the small invertebrates that larger fish feed on, or alter the invertebrates’ environment so that they become scarce. Third, destabilized streambeds can “induc[e] fish to spawn on unstable material,” and fish eggs and larvae can be “smothered or buried.” Fourth, because the streams the salmon occupy are already at “near lethal temperatures,” even “minor” disturbances in the summer can harm the salmon. Fifth, juvenile salmon could be “displaced to a less optimal location where overall fitness and survival odds are also less.” Finally, a long list of other factors — disturbance, turbidity, pollution, decrease in food base, and loss of cover associated with suction dredging — could combine to harm the salmon.
I therefore conclude that the suction dredge mining challenged in this case “may affect” the listed coho salmon and its critical habitat.
C. Burden on the Forest Service
The burden imposed upon the Forest Service by the obligation to consult under Section 7 of the ESA is not great. Indeed, District Ranger Vandiver has already consulted with Forest Service biologists Bemis and Grunbaum in formulating the detailed criteria for suction dredge mining NOIs in the Happy Camp District of the Klamath National Forest. That consultation could not satisfy Section 7 because Bemis and Grunbaum work for the Forest Service rather than the Fish and Wildlife Service or the National Marine Fisheries Service. But if Vandiver had consulted with employees of those agencies, that consultation could have satisfied Section 7. If, after engaging in that consultation, Vandiver had formulated sufficiently detailed coho-protective criteria based on the views of the Fish and Wildlife Service and the National Marine Fisheries Service, any NOIs approved using those criteria would not have required the exercise of further discretion and therefore would not have required further consultation. See Texas Indep. Producers, 410 F.3d at 979; Envtl. *1011Def. Ctr., 344 F.3d at 853. Of course, Vandiver formulated his criteria for NOIs only for the Happy Camp District. But there is no reason why the Forest Service could not consult with the Fish and Wildlife Service and the National Marine Fisheries Service to formulate comparable criteria for all of the districts in the Klamath National Forest, with the result that any individual NOI approved under those criteria would not require further consultation.
Conclusion
By definition, suction dredge mining pursuant to an NOI is mining that “might cause” “significant disturbance of surface resources,” including the surface resource of “fisheries habitat.” The Forest Service does not dispute that such mining “may affect” critical habitat of coho salmon in the Klamath River system within the meaning of Section 7 of the ESA. The Forest Service therefore has an obligation under Section 7 to consult with the relevant agencies at some point in the process of allowing such mining.
The Forest Service had several available choices. It could have consulted under Section 7 when it promulgated the regulation for dredge mining under NOIs. That is, it could have consulted when it set the threshold criterion for an NOI as mining that “might cause significant disturbance of surface resources” including fisheries habitat. Or it could have consulted under Section 7 when it formulated habitat-protective criteria for approving NOIs. That is, it could have consulted when District Ranger Vandiver formulated his criteria for approving the NOIs for the Happy Camp District. Or, finally, in the absence of criteria such as those formulated for the Happy Camp District, it could have consulted under Section 7 with respect to each individual NOI.
The one choice that was not available to the Forest Service was never to consult. Yet that is the choice the Forest Service made. In making that choice, the Forest Service violated Section 7 of the ESA.
I respectfully but emphatically dissent from the conclusion of the majority to the contrary.
. The parties appear to assume that if consultation is required under Section 7(a)(2), it is required with both agencies. Without deciding the question, I will also so assume.
. The current general permit and form NOI are available at http://www.epa.gov/npdes/ pubs/cgp2008_finalpermit.pdf. The NOI form and instructions are found at Appendix E.