MEMORANDUM DECISION RE CROSS MOTIONS FOR SUMMARY JUDGMENT (DOCS. 430, 435, 446, 474, 477)

TABLE OF CONTENTS

I. INTRODUCTION.............................................................812

II. BACKGROUND .............................................................813

A. The Listed Species........................................................813

1. Sacramento River Winter-Run Chinook Salmon..........................813

2. Spring Run Chinook..................................................814

3. Central Valley Steelhead..............................................815

4. Green Sturgeon......................................................816

5. Southern Resident Killer Whale .'.......................................816

B. The 2009 Salmonid BiOp and RPA..........................................816

III. STANDARD OF DECISION .................................................818

IV. BASIC LEGAL FRAMEWORK...............................................819

A. Review under the APA....................................................819

1. Record Review.......................................................819

2. Deference to Agency Expertise........................................819

B. General Obligations Under the ESA.........................................820

1. Best Available Science................................................821

2. Best Available Science Standards and the Application of Analytical/ Statistical Methodologies.................................. 824

V. EXPORT PLAINTIFFS’ & DWR’S CLAIMS....................................824

A. Alleged Clear Scientific Errors Pertaining to Delta Operations..................824

1. Challenged Statistical Methodologies ...................................825

2. Failure to Perform a Population-Level Quantitative Analysis ..............834

B. Baseline Analysis Challenges...............................................845

1. Failure to Distinguish Between Discretionary and Nondiscretionary Actions ...........................................................845

2. Treatment of Available Data on Ocean Harvest and Ocean Conditions.....852

C. Effects Analysis Challenges................................................858

*812 1. Use of a 100-Year Timeframe..........................................858

2. Winter-Run Viability Analysis.........................................861

3. . Orea Analysis........................................................864

4. Interior Delta Mortality as an Indirect Effect............................866

D. Critical Habitat Analysis...................................................871

1. There Is No Requirement that NMFS Identify a Numerical Threshold for Adverse Modification............................................871

2. Significance of Impacts to Critical Habitat...............................872

E. Use of Surrogates..........................................................880

F. Challenges to the Reasonable and Prudent Alternative.........................882

1. RPA Action IV.2.1....................................................882

2. RPA Action IV.2.3....................................................898

3. Action IV.3..........................................................909

G. Compliance with 50 C.F.R. § 402.02 .........................................911

1. Consistency with Purposes of the Action ................................914

2. Consistency with the Action Agency’s Legal Authority and Jurisdiction.....917

3. Economic and Technical Feasibility.....................................919

4. Avoidance Jeopardy and/or Adverse Modification.........................922

5. DWR’s Feasibility Challenges to Action IV.4.2...........................922

VI. STANISLAUS RIVER PLAINTIFFS’ CLAIMS ................................928

A. Relevant Factual Background..............................................928

1. The New Melones Project.............................................928

2. The Stanislaus River Plaintiffs.........................................928

3. The Status of Steelhead in the Stanislaus River..........................928

B. Inclusion of the New Melones Unit in the Proposed Action Subject to Consultation ...........................................................930

C. Effects Analysis Challenges................................................931

1. New Melones Operations v. Baseline Effects.............................931

2. Challenge to Critical Habitat Adverse Modification Finding................934

D. Stanislaus River RPA Challenges...........................................937

1. Challenge to the Assumptions Used to Model New Melones Project Operations........................................................937

2. Do Actions III.1.2, III.1.3, and IV.1.2 Improperly Require Reclamation to Infringe Upon OID and SSJID’s Prior Right to Stanislaus River Water in violation of 50 C.F.R. § 402.02 ?..............................938

3. Use of the San Joaquin River Temperature Model........................941

4. Exceptions Built into Action III.1.2.....................................947

5. Does the Record Support the Finding that Action III.1.3 Will Avoid Jeopardy to or Adverse Modification of CV Steelhead or Critical Habitat?..........................................................948

6. DFG Salmon Population Model ........................................950

7. SR Plaintiffs’ “Impermissible Major Changes” Argument..................951

8. Challenges to the BiOp’s Feasibility Analyses?...........................951

9. Are Actions III.1.3, III.2.2 Consistent with the Purposes of the Project?.....952

10. Waste and Unreasonable Use of Water (California Constitution Article X, Section 2).......................................................953

VII. RECLAMATION’S LIABILITY UNDER THE ESA............................954

VIII. CONCLUSION............................................................955

I. INTRODUCTION

These consolidated cases arise out of continuing efforts to protect several species listed under the Endangered Species Act (“ESA”), namely the endangered Sacramento River winter-run Chinook salmon (Oncorhynchus tshawytscha) (“winter-run”), threatened Central Valley spring-run Chinook salmon (O. tshawytscha) (“spring-run”), threatened Central Valley *813 steelhead (O. mykiss) (“CV steelhead”), threatened Southern Distinct Population Segment (“DPS”) of North American green sturgeon (Acipenser medirostris), and endangered Southern Resident killer whales (Orcinus orea) (collectively, “Listed Species”); and associated impacts to the water supply for more than half the State of California.

Plaintiffs, San Luis & Delta Mendota Water Authority and Westlands Water District; State Water Contractors (“SWC”); Kern County Water Agency and Coalition for a Sustainable Delta; and Metropolitan Water District of Southern California (“MWD” or “Metropolitan”) (collectively “Export Plaintiffs”) move for summary judgment on their claims that the United States National Marine Fisheries Service’s (“NMFS”) June 4, 2009 Biological Opinion, addressing the impacts of the coordinated operations of the federal Central Valley Project (“CVP”) and State Water Project (“SWP”) (collectively the “Project”) on the Listed Species (“2009 Salmonid BiOp” or “BiOp”) and its Reasonable and Prudent Alternative (“RPA”), violates the ESA and the Administrative Procedure Act (“APA”). Doc. 430. 1 Plaintiffs Stockton East Water District, Oak-dale Irrigation District, and South San Joaquin Irrigation District (“Stanislaus River Plaintiffs” or “SR Plaintiffs”) filed a separate motion for summary judgment, raising unique challenges to the BiOp. Doc. 435. Plaintiff-in-Intervention, the California Department of Water Resources (“DWR”) filed a separate motion for summary judgment on narrower grounds. Doc. 446.

Federal Defendants, the United States Department of Commerce (“DOC”), the National Oceanic and Atmospheric Administration (“NOAA”), the agency within DOC of which NMFS is a part, NMFS, the United States Department of the Interior (“DOI”), and its sub-agency the United States Bureau of Reclamation (“Reclamation”), oppose and cross move for summary judgment on all remaining claims, Doc. 477, as do Defendanb-Intervenors California Trout, Friends Of The River, Natural Resources Defense Council, Northern California Council of the Federation of Fly Fishers, Pacific Coast Federation of Fishermen’s Associations/Institute for Fisheries Resources, Sacramento River Preservation Trust, San Francisco Baykeeper, The Bay Institute, and the Winnemem Wintu Tribe, Doc. 474. All parties filed replies. Docs. 487, 492, 513, 515. These cross motions, which included over 700 pages of briefing and thousands of pages of supporting declarations and exhibits, came on for hearing on December 16 and 17, 2010.

II. BACKGROUND

A. The Listed Species.

1. Sacramento River Winter-Run Chinook Salmon.

Sacramento River winter-run Chinook salmon (Oncorhynchus tshawytscha) (“winter-run”) is listed as “endangered” under the ESA. 70 Fed.Reg. 37,160 (June 28, 2005). Historical winter-run population estimates were as high as approximately 230,000 fish in the 1960s, BiOp at 82, but declined to under 200 fish in the 1990s. Id. at 81 . In recent years, population surveys of winter-run estimated a high of 17,344 fish in 2006, followed by a decline in 2007 (2,542 fish) that persisted into 2008 (2,830 fish). Id. 2

*814 Adult winter-run Chinook salmon migrate upstream from the Pacific Ocean through the Bay-Delta estuary during November through July, moving upstream past Red Bluff Diversion Dam (“RBDD”) from mid-December through early August, with peak passage occurring in mid-March. BiOp at 80. Spawning typically occurs in the mainstream Sacramento River downstream of Keswick Dam during April through August, with the greatest spawning activity typically taking place during May and June. Id.

Winter-run fry begin to emerge from the gravel beds where eggs are laid in late June and early July, continuing through October. Id. Juvenile rearing and emigration typically occurs between July and February in the upper Sacramento River, with juvenile migration downstream past RBDD beginning as early as mid-July, peaking in September, and continuing through March in some years. Id. at 80-81 .

Juvenile winter-run occur in the Delta from November through May. Id.; Pac. Coast Fed’n of Fisherman’s Ass’ns. v. Gutierrez (“Gutierrez II’), 606 F.Supp.2d 1195, 1216-17 (E.D.Cal.2008). Winter-run juveniles typically remain in the Delta until they reach a fork length of approximately 118 millimeters and are from 5 to 10 months of age. BiOp. at 81. Juveniles begin exiting to the ocean as early as November and continue to do so through May. Id.

Designated critical habitat for winter-run includes the Sacramento River, the Delta, and downstream bays to the Golden Gate Bridge. 58 Fed.Reg. 33,212 (June 16, 1993). Gutierrez II, 606 F.Supp.2d at 1217 . The following physical and biological features are identified as essential for the conservation of winter-run:

BiOp at 90. Currently, the value of winter-run critical habitat is “degraded,” by, among other things, the presence of dams, temperature control issues on the upper Sacramento River, unscreened diversions, and degraded spawning and riparian habitat. Id. at 93.

2. Spring Run Chinook.

Central Valley spring-run Chinook salmon (O. tshawytscha) (“spring-run”) is listed as “threatened” under the ESA. 71 Fed. Reg. 834 (June 5, 2005); 70 Fed.Reg. 37160 (June 28, 2005) (critical habitat designated). There are three “independent” populations of spring-run, located on Butte, Deer and Mill Creeks, several “dependent” populations (which rely on the three independent populations for continued existence), and a population of hatchery fish from the Feather River Hatchery (FRH). BiOp at 93-94.

Spring-run Chinook have been declining over recent years. The Central Valley as a whole is estimated to have supported spring runs as large as 600,000 fish between the late 1880s and 1940s. Id. at 94 . *815 The 2007 escapement was 7,819 for all tributary populations (all independent and dependent populations, excluding those fish returning to FRH). Id. at 97 .

Adult spring-run enter freshwater in the spring, beginning in late January, entering the Sacramento River between March and September, primarily in May and June, and entering spawning grounds between mid-April and mid-June. Id. at 93 . Adults hold over the summer in cool, high elevation streams while they sexually mature, and then spawn in the fall, between September and October, depending on water temperatures. Id. at 93 .

Juveniles typically spend a year or more in freshwater before emigrating to the ocean. Id. at 93 . The emigration period for spring-run extends from November to June and is highly variable. Id. at 94 .

Designated critical habitat for spring-run includes the Sacramento River, tributaries supporting spring-run, the Delta, and downstream bays to the Golden Gate Bridge. Gutierrez II, 606 F.Supp.2d at 1217 . The value of spring-run critical habitat currently is “degraded.” BiOp at 101, 104.

3. Central Valley Steelhead.

Central Valley steelhead (O. mykiss) (“CV steelhead”) is listed as “threatened” under the ESA. 71 Fed.Reg. 834 (Jan. 5, 2006). Wild CV steelhead are confined mostly to the upper Sacramento River and its tributaries. BiOp at 107. Recent surveys also have detected small, self-sustaining populations on the Stanislaus, Mokelumne, and Calaveras Rivers, as well as observations of juvenile steelhead on the Tuolumne and Merced Rivers. Id. These small populations make up the remaining representatives of the Southern Sierra Nevada Diversity Group (“SSNDG”) of CV Steelhead. Id. at 198.

While there is limited information on population size, NMFS estimates that the current population for the entire distinct population segment (“DPS”) 3 (including the SSNDG as well as all other populations) in the Central Valley is less than 3,628 spawning females, compared with 40,000 spawners in the 1960s. BiOp at 106. The CV Steelhead population has shown a pattern of negative growth since the late 1960s, and there is no indication that the trend has changed. BiOp at 108-09 & Figures 4-4 & 4-5.

CV steelhead generally leave the ocean from August through April and spawn from December through April in small streams and tributaries where cool, well-oxygenated water is available year-round. Id. at 104. Unlike Pacific salmon, steel-head are capable of spawning more than once before death. Although one-time spawners are the great majority, approximately 17.2 percent in California streams are repeat spawners. Id. at 103-104.

Steelhead eggs hatch approximately 30 days after spawning, and fry emerge from the gravel four to six weeks later into shallow areas where they feed. Id. at 105. Steelhead rear during the summer and emigrate “episodically” from their natal streams during fall, winter, and spring high flows. Id. at 106. Emigrating CV Steelhead use the lower reaches of the Sacramento River and Delta for rearing and as a migration corridor to the ocean. Id. Juvenile CV steelhead typically emi *816 grate through the Delta from late September through June. Id. at 105 (Table 4-6).

Approximately 80% of historical CV Steelhead range is blocked by dams. Id. at 109. CV steelhead critical habitat is degraded. Id. at 113.

4. Green Sturgeon.

The southern distinct population segment of the North American green sturgeon (“green sturgeon”) (Acipenser medirostris) is listed as “threatened” under the ESA. 71 Fed.Reg. 17757 (Apr. 7, 2006); 73 Fed.Reg. 52,084 (critical habitat designated).

Green sturgeon are anadromous fish that spawn and rear in freshwater rivers and estuaries but spend most of their lives in the ocean. See BiOp at 114-15. They are a long-lived, slow-growing species. 68 Fed.Reg. 4,433, 4,436 (Jan. 29, 2003). Juvenile green sturgeon are present in the Delta year round. BiOp at 119.

There are no definitive population counts or figures for the Southern DPS green sturgeon. Evidence available at the time the BiOp was written suggests that the population in the Delta watershed is “relatively small,” ranging from several hundred to a few thousand adults. Id. at 124.

Critical habitat for the Southern DPS of green sturgeon was proposed on September 8, 2008, 73 Fed.Reg. 52,084, but had not been adopted as of the issuance of the BiOp. Proposed critical habitat included “approximately 325 miles of riverine habitat and 1,058 square miles of estuarine habitat in California, Oregon, and Washington, and 11,927 square miles of coastal marine habitat off California, Oregon, and Washington within the geographical area presently occupied by the Southern DPS of green sturgeon.” BiOp at 126. In addition, approximately 136 square miles of habitat within the Yolo and Sutter bypasses, adjacent to the Sacramento River, are proposed for designation. Id. The BiOp concluded that the current condition of proposed critical habitat for the Southern DPS of green sturgeon is “degraded over historical conditions.” Id. at 134.

5. Southern Resident Killer Whale.

The Southern Resident DPS of killer whale (Orcinus orea) (“Southern Residents”) was listed as “endangered” under the ESA on November 18, 2005, 70 Fed. Reg. 69,903 (Nov. 18, 2005), and the DPS is designated as “depleted” under the Marine Mammal Protection Act. BiOp at 158— 59. Southern Residents are found throughout the coastal waters off Washington, Oregon, and Vancouver Islands and are known to travel as far south as central California. Id. at 159 .

The BiOp addresses the impact of Project operations on Southern Residents and concludes that extinction of winter-run and spring-run Chinook salmon, as well as reductions in fall-run 4 Chinook salmon populations, “would reduce prey availability and increase the likelihood for local depletions of prey in particular locations and times,” which would, in turn, increase the risk of extinction of the Southern Residents. BiOp at 573-74.

B. The 2009 Salmonid BiOp and RPA.

The 2009 Salmonid BiOp, prepared pursuant ESA § 7, 16 U.S.C. § 1536 (a)(2), concluded that “the long-term operations of the CVP and SWP are likely to jeopardize the continued existence” of the Listed Species and “destroy or adversely modify” critical habitat for winter-run, spring-run, and CV steelhead. BiOp at 575. As required by law, the BiOp includes an RPA designed to allow the projects to continue *817 operating without causing jeopardy to the species or adverse modification to its critical habitat. Id. at 575-671 . The RPA “is composed of numerous elements for each of the various project divisions and associated stressors,” which, according to the BiOp, “must be implemented in its entirety to avoid jeopardy and adverse modification.” Id. at 578 . The BiOp provides a succinct overview of the RPA:

Id. at 576-78 .

III. STANDARD OF DECISION

Summary judgment is appropriate when the pleadings and the record demonstrate that “there is no genuine dispute as to any material fact and the movant is entitled to judgment as a matter of law.” Fed.R.Civ.P. 56(c). The claims in this case involve NMFS’s issuance of a biological opinion, final agency action subject to judicial review under the APA, 5 U.S.C. § 702 . Nat’l Wildlife Fed’n v. Nat’l Marine Fisheries Serv., 524 F.3d 917, 925 (9th Cir.2008) (“NWF v. NMFS II”). A court conducting APA judicial review may not resolve factual questions, but instead determines “whether or not as a matter of law the evidence in the administrative record permitted the agency to make the decision it did.” Sierra Club v. Mainella, 459 F.Supp.2d 76, 90 (D.D.C. 2006) (quoting Occidental Eng’g Co. v. INS, 753 F.2d 766, 769 (9th Cir.1985)). “[I]n a case involving review of a final agency action under the [APA] ... the standard set forth in Rule 56(c) does not apply because of the limited role of a court in reviewing the administrative record.” Id. at 89. In this context, summary judgment becomes the “mechanism for deciding, as a matter of law, whether the agency action is supported by the adminis *819 trative record and otherwise consistent with the APA standard of review.” Id. at 90.

IV. BASIC LEGAL FRAMEWORK

A. Review under the APA.

APA invalidation of a biological opinion requires Plaintiffs to prove that NMFS’s action was “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706 (2)(A).

1. Record Review.

APA review of a biological opinion is “based upon the evidence contained in the administrative record.” Arizona Cattle Growers’ Ass’n v. U.S. Fish and Wildlife, 273 F.3d 1229, 1245 (9th Cir. 2001). Judicial review under the APA must focus on the administrative record already in existence, not some new record made initially in a reviewing court. Parties may not use “post-decision information as a new rationalization either for sustaining or attacking the agency’s decision.” Ass’n of Pac. Fisheries v. EPA 615 F.2d 794 , 811-12 (9th Cir.1980). Exceptions to administrative record review for technical information or expert explanation make such evidence admissible only for limited purposes, and those exceptions are narrowly construed and applied. Lands Council v. Powell, 395 F.3d 1019 , 1030 (9th Cir.2005).

Here, as evidentiary rulings explained, see, e.g., Docs. 387, 392 (10/19/09 Hearing Transcript (“Tr.”)), 406, 407, 462, 740 (7/8/10 Tr.), 750, expert testimony has been considered solely for explanation of technical terms and complex scientific subject matter beyond the Court’s knowledge; and to understand the agency’s explanations, or lack thereof, and the parties’ arguments.

2. Deference to Agency Expertise.

A court must defer to the agency on matters within the agency’s expertise, unless the agency completely failed to address some factor, consideration of which was essential to making an informed decision. Nat’l Wildlife Fed’n v. Nat’l Marine Fisheries Serv., 422 F.3d 782, 798 (9th Cir.2005) (“NWF v. NMFS I”). A court “may not substitute its judgment for that of the agency concerning the wisdom or prudence of the agency’s action.” River Runners for Wilderness v. Martin, 593 F.3d 1064, 1070 (9th Cir.2010):

Id.

Although deferential, judicial review under the APA is designed to “ensure that the agency considered all of the relevant factors and that its decision contained no clear error of judgment.” Arizona v. Thomas, 824 F.2d 745, 748 (9th Cir.1987) (internal citation and quotation omitted). “The deference accorded an agency’s scientific or technical expertise is not unlimited.” Br ower v. Evans, 257 F.3d 1058, 1067 (9th Cir.2001).

Motor Vehicle Mfrs. Ass’n of U.S. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 , 103 S.Ct. 2856 , 77 L.Ed.2d 443 (1983); see also Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 416 , 91 S.Ct. 814 , 28 L.Ed.2d 136 (1971) (reviewing court may overturn an agency’s action as arbitrary and capricious if the agency failed to consider relevant factors, failed to base its decision on those factors, and/or made a “clear error of judgment”), overruled on other grounds by Califano v. Sanders, 430 U.S. 99, 105 , 97 S.Ct. 980 , 51 L.Ed.2d 192 (1977)).

More generally, “[u]nder the APA ‘the agency must examine the relevant data and articulate a satisfactory explanation for its action including a rational connection between the facts found and the choice made.’ ” Humane Soc. of U.S. v. Locke, 626 F.3d 1040, 1048 (9th Cir.2010) (quoting Motor Vehicle Mfrs. Ass’n, 463 U.S. at 43 , 103 S.Ct. 2856 ). “The reviewing court should not attempt itself to make up for an agency’s deficiencies: We may not supply a reasoned basis for the agency’s action that the agency itself has not given.” Id.

B. General Obligations Under the ESA.

ESA Section 7(a)(2) prohibits agency action that is “likely to jeopardize the continued existence” of any endangered or threatened species or “result in the destruction or adverse modification” of its critical habitat. 16 U.S.C. § 1536 (a)(2).

To “jeopardize the continued existence of’ means “to engage in an action that reasonably would be expected, directly or indirectly, to reduce appreciably the likelihood of both the survival and recovery of a listed species in the wild by reducing the reproduction, numbers, or distribution of that species.” 50 C.F.R. § 402.02 ; see also NWF v. NMFS II, 524 F.3d 917 (rejecting agency interpretation of 50 C.F.R. § 402.02 that in effect limited jeopardy analysis to survival and did not realistically evaluate recovery, thereby avoiding an interpretation that reads the provision “and recovery” entirely out of the text). An action is “jeopardizing” if it keeps recovery “far out of reach,” even if the species is able to cling to survival. NWF v. NMFS II, 524 F.3d at 931 . “[A]n agency may not take action that will tip a species from a state of precarious survival into a state of likely extinction. Likewise, even where baseline conditions already jeopardize a species, an agency may not take action that deepens the jeopardy by causing additional harm.” Id. at 930 .

To satisfy this obligation, the federal agency undertaking the action (the “action agency”) must prepare a “biological assessment” that evaluates the action’s potential impacts on species and species’ habitat. 16 U.S.C. § 1536 (c); 50 C.F.R. § 402.12 (a). If the proposed action “is likely to adversely affect” a threatened or endangered species or adversely modify its designated critical habitat, the action agency must engage in “formal consultation” with NMFS 5 to obtain its biological opinion as to the impacts of the proposed action on the listed species. See 16 U.S.C. § 1536 (a)(2), (b)(3); see also 50 C.F.R. § 402.14 (a), (g). Once the consultation process has been completed, NMFS must give the action agency a written biological opinion “setting forth [NMFS’s] opinion, and a summary of the information on which the opinion is based, detailing how the agency action affects the species or its critical habitat.” 16 U.S.C. *821 § 1536 (b)(3)(A); see also 50 C.F.R. § 402.14 (h).

If NMFS determines that jeopardy or destruction or adverse modification of critical habitat is likely, NMFS “shall suggest those reasonable and prudent alternatives which [it] believes would not violate subsection (a)(2) of this section and can be taken by the Federal agency or applicant in implementing the agency action.” 16 U.S.C. § 1536 (b)(3)(A). “Following the issuance of a ‘jeopardy’ opinion, the agency must either terminate the action, implement the proposed alternative, or seek an exemption from the Cabinet-level Endangered Species Committee pursuant to 16 U.S.C. § 1536 (e).” Nat’l Ass’n of Home Builders v. Defenders of Wildlife, 551 U.S. 644, 652 , 127 S.Ct. 2518 , 168 L.Ed.2d 467 (2007).

1. Best Available Science.

Under the ESA, an agency’s actions must be based on “the best scientific and commercial data available.” 16 U.S.C. § 1536 (a)(2); 50 C.F.R. § 402.14 (g)(8) (“In formulating its Biological Opinion, any reasonable and prudent alternatives, and any reasonable and prudent measures, the Service will use the best scientific and commercial data available----”). A failure by the agency to utilize the best available science is arbitrary and capricious. See Gutierrez II, 606 F.Supp.2d at 1244 .

“The obvious purpose of the [best available science requirement] is to ensure that the ESA not be implemented haphazardly, on the basis of speculation or surmise.” Bennett v. Spear, 520 U.S. 154, 176 , 117 S.Ct. 1154 , 137 L.Ed.2d 281 (1997).

Id. at 176-77, 117 S.Ct. 1154 .

A decision about jeopardy must be made based on the best science available at the time of the decision; the agency cannot wait for or promise future studies. See Ctr. for Biological Diversity v. Rumsfeld, 198 F.Supp.2d 1139, 1156 (D.Ariz.2002) (the best scientific and commercial data available standard “recognizes that better scientific evidence will most likely always be available in the future”). The “best available science” mandate of the ESA sets a basic standard that “prohibits the [agency] from disregarding available scientific evidence that is in some way better than the evidence [it] relies on.” Am. Wildlands v. Kempthorne, 530 F.3d 991, 998 (D.C.Cir.2008) (internal quotation omitted).

What constitutes the “best” available science implicates core agency judgment and expertise to which Congress requires the courts to defer; a court should be especially wary of overturning such a determination on review. Baltimore Gas & Elec. Co. v. Natural Res. Defense Council, 462 U.S. 87, 103 , 103 S.Ct. 2246 , 76 L.Ed.2d 437 (1983) (a court must be “at its most deferential” when an agency is “making predictions within its area of special expertise, at the frontiers of science”). As explained in the en banc decision in Lands *822 Council v. McNair, 537 F.3d 981, 993 (9th Cir.2008), courts may not “impose on the agency their own notion of which procedures are best or most likely to further some vague, undefined public good.” In particular, an agency’s “scientific methodology is owed substantial deference.” Gifford Pinchot Task Force v. U.S. Fish & Wildlife Serv., 378 F.3d 1059 , 1066 (9th Cir.2004).

“When specialists express conflicting views, an agency must have discretion to rely on the reasonable opinions of its own qualified experts even if, as an original matter, a court might find contrary views more persuasive.” Lands Council, 537 F.3d at 1000 (quoting Marsh v. Oregon Natural Res. Council, 490 U.S. 360, 378 , 109 S.Ct. 1851 , 104 L.Ed.2d 377 (1989)). Mere uncertainty, or the fact that evidence may be “weak,” is not fatal to an agency decision. Greenpeace Action v. Franklin, 14 F.3d 1324, 1337 (9th Cir. 1992) (upholding biological opinion, despite uncertainty about the effectiveness of management measures, because decision was based on a reasonable evaluation of all available data); Nat’l Wildlife Fed’n v. Babbitt, 128 F.Supp.2d 1274, 1300 (E.D.Cal.2000) (holding that the “most reasonable” reading of the best scientific data available standard is that it “permits [NMFS] to take action based on imperfect data, so long as the data is the best available”). NMFS “must utilize the best sci entific ... data available, not the best scientific data possible.” Building Indus. Ass’n v. Norton, 247 F.3d 1241, 1246 (D.C.Cir.2001), cited with approval in Kern County Farm, Bureau v. Allen, 450 F.3d 1072, 1080-81 (9th Cir.2006) (“Absent superior data occasional imperfections do not violate” the ESA best available data standard); see also Defenders of Wildlife v. Babbitt, 958 F.Supp. 670, 680 (D.D.C.1997) (best available science standard does not require “conclusive evidence,” only that agency use best science available and not ignore contrary evidence).

The deference afforded under the best available science standard is not unlimited. For example, Tucson Herpetological Society v. Salazar, 566 F.3d 870, 879 (9th Cir.2009), held that an agency may not rely on “ambiguous studies as evidence” to support findings made under the ESA. There, in the context of an ESA § 4 listing determination, NMFS “affirmatively relief] on ambiguous studies as evidence of persistence (i.e., stable and viable populations), and in turn argue[d] that this ‘evidence’ of persistence ... proves that the lizard’s lost range is insignificant for purposes of the ESA.” Id. The Ninth Circuit found this conclusion to be unreasonable because “[t]he studies do not lead to the conclusion that the [species] persists in a substantial portion of its range, and therefore cannot support [NMFS’s] conclusion.” Id. 6 ; see also Rock Creek Alliance v. U.S. *823 Fish & Wildlife Service, 390 F.Supp.2d 993, 1008 (D.Mont.2005) (rejecting section 7 biological opinion’s reliance on a disputed scientific report, which explicitly stated its analysis was not applicable to the small populations addressed in the challenged opinion).

Alternatively, the presumption of agency expertise may be rebutted if the agency’s decisions, although based on scientific expertise, are not reasoned, Greenpeace v. NMFS, 80 F.Supp.2d 1137, 1147 (W.D.Wash.2000), or if the agency disregards available scientific evidence better than the evidence on which it relies, Kern County Farm Bureau, 450 F.3d at 1080 .

Courts routinely perform substantive reviews of record evidence to evaluate the agency’s treatment of best available science. The judicial review process is not one of blind acceptance. See, e.g., Kern County, 450 F.3d at 1078-79 (thoroughly reviewing three post-comment studies and FWS’s treatment of those studies to determine whether they “provide[d] the sole, essential support for” or “merely supplemented” the data used to support a listing decision); Home Builders Ass’n of N. Cal. v. U.S. Fish and Wildlife Serv., 529 F.Supp.2d 1110, 1120 (N.D.Cal.2007) (examining substance of challenge to FWS’s determination that certain data should be disregarded); Trout Unlimited v. Lohn, 645 F.Supp.2d 929 (D.Or.2007) (finding best available science standard had been violated after thorough examination of rationale for NMFS’s decision to withdraw its proposal to list Oregon Coast Coho salmon); Oceana, Inc. v. Evans, 384 F.Supp.2d 203, 217-18 (D.D.C.2005) (carefully considering scientific underpinnings of challenge to FWS’s use of a particular model, including post decision evidence presented by an expert to help the court understand the complex model, applying one of several record review exceptions articulated in Esch v. Yeutter, 876 F.2d 976, 991 (D.C.Cir.1989), which are similar to those articulated by the Ninth Circuit).

Courts are not required to defer to an agency conclusion that runs counter to that of other agencies or individuals with specialized expertise in a particular technical area. See, e.g., Am. Tunaboat Ass’n v. Baldrige, 738 F.2d 1013, 1016-17 (9th Cir.1984) (NMFS’s decision under the Marine Mammal Protection Act was not supported by substantial evidence because agency ignored data that was product of “many years’ effort by trained research personnel”); Sierra Club v. U.S. Army Corps of Eng’rs, 701 F.2d 1011 , 1030 (2d Cir.1983) (“court may properly be skeptical as to whether [the conclusions of an environmental impact statement prepared under the National Environmental Policy Act] have a substantial basis in fact if the responsible agency has apparently ignored the conflicting views of other agencies having pertinent experience! ]”). A court should “reject conclusory assertions of agency ‘expertise’ where the agency spurns unrebutted expert opinions without itself offering a credible alternative explanation.” N. Spotted Owl v. Hodel, 716 F.Supp. 479, 483 (W.D.Wash.1988) (citing Am. Tunaboat Ass’n, 738 F.2d at 1016 ).

In Conner v. Burford, 848 F.2d 1441, 1453-54 (9th Cir.1988), the agency attempted to defend its biological opinions by arguing that there was a lack of sufficient information to perform additional analysis. In rejecting this defense, the Ninth Circuit held that “incomplete information' ... does not excuse the failure to comply with the statutory requirement of a comprehensive biological opinion using the best information available,” and noted that FWS could have completed more analysis with the information that was available. Id. at 1454 .

Id.

2. Best Available Science Standards and the Application of Analytical/Statistical Methodologies.

These above-described standards apply with equal force to the use and interpretation of statistical methodologies. As the D.C. Circuit in Appalachian Power Co. v. EPA 135 F.3d 791 (D.C.Cir.1998), explained in reviewing a challenge to a decision of the Environmental Protection Agency (“EPA”) under the “arbitrary and capricious” standard of review:

Id. at 802 .

The model must fit the available data. See Nat’l Wildlife Fed’n v. EPA 286 F.3d 554 , 565 (D.C.Cir.2002) (“NWF v. EPA ”) (a court will only reject the choice of a model “when the model bears no rational relationship to the characteristics of the data to which it was applied”). For example, Oceana, 384 F.Supp.2d at 220 , rejected a challenge to NMFS’s use of a particular analytical model that used data drawn from existing literature, even though experts “suggested that reliable take limits cannot be established without quantitative data gathered from ‘in-water’ surveys.” Although NMFS conceded “a thorough quantitative analysis based on empirical estimates of population size would be a superior way to analyze the impact [ ] on [the species],” it was undisputed that “given the paucity of information on sea turtles and the difficulties of using the data that does exist, [a] different or more complex model [than that used by NMFS] was not available and could not even be constructed.” Id. (internal quotations omitted). Likewise, “the fact that a given model has some imperfections does not prevent it from constituting the ‘best scientific information available.’ ” Oceana v. Evans, 2005 WL 555416 , *16-*17 (D.D.C. Mar. 9, 2005) (citing 16 U.S.C. § 1851 (a)(2)) (approving NMFS’s use of a model despite known limitations, where it was the only model available and the agency supplemented its analysis with other sources to address areas where the model was unable to make accurate predictions).

V. EXPORT PLAINTIFFS’ & DWR’S CLAIMS.

A. Alleged Clear Scientific Errors Pertaining to Delta Operations.

A major premise of the BiOp is that pumping “causes reverse flows, leading to loss of juveniles migrating out from the Sacramento River system in the interior Delta and more juveniles being exposed to the State and Federal pumps, where they are salvaged 7 at the facilities.” BiOp at *825 577. The effects analysis also concluded “that juvenile steelhead migrating out from the San Joaquin River Basin have a particularly high rate of loss due to both project and non-project related stressors.” Id. at 577-78 . To mitigate for these impacts, the RPA “prescribes Old and Middle River flow levels to reduce the number of juveniles exposed to the export facilities and prescribes additional measures at the facilities themselves to increase survival of fish.” Id. at 577 . In addition, “to improve survival of San Joaquin steelhead smolts,” the RPA prescribes “both increased San Joaquin River flows and export curtailments.” Id. at 578 .

Plaintiffs strenuously argue that NMFS made certain “clear” scientific errors in reaching the conclusion that exports adversely affect juvenile salmonid survival.

1. Challenged Statistical Methodologies.

a. Use of Raw Salvage in Figures 6-65 and 6-66.

NMFS relied on salvage data provided by Plaintiff-Intervenor, DWR, presented in Figures 6-65 and 6-66 of the BiOp:

[[Image here]]

*826 [[Image here]]

Id. at 361-62 . These figures were cited to demonstrate that “[l]oss of older juveniles at the CVP and SWP fish collection facilities increase sharply at [OMR] flows of approximately -5,000 cfs and depart from the initial slope at flows below this.” Id. at 361 . Federal Defendants’ cross motion explains that NMFS used this data to “help evaluate where along the spectrum of OMR flows any significant change in salvage could be observed.” Doc. 477-1 at 53.

These figures, which are based upon average salvage figures over many years, use “raw” salvage numbers that are not scaled to reflect the size of the population from which the fish were salvaged at the time the particular sample was taken. Previous rulings in this and the related Consolidated Delta Smelt Cases have discussed at length why the use of such data is not consistent with standard practice in the fields of fish biology and population dynamics. See San Luis & Delta-Mendota Water Authority v. Salazar, 760 F.Supp.2d 855, 885-90 (E.D.Cal.2010). The May 18, 2010 Findings of Fact and Conclusions of Law Re Plaintiffs’ Request for Preliminary Injunction (“PI Decision”) in this case found:

Consol. Salmonid Cases, 713 F.Supp.2d 1116, 1142-43 (E.D.Cal.2010) (emphasis added).

Federal Defendants attempt to explain their use of these figures in two ways. First, Jeffrey Stuart, NMFS Fisheries Biologist and the primary author of the Delta section of the BiOp, opines that “the general trend in fish loss should still be apparent regardless of scaling.” Fourth Stuart Deck, Doc. 485 at ¶ 72. He insists that this data “indicates that additional loss of fish occurs with increasing export levels as measured by the OMR flow values.” Id. This explanation simply defies common sense. When trying to discern trends from data points that range across many years, an obvious “confounding variable” is population size. A trend observed in data that is unsealed for population size may change or completely disappear when scaled for population size. See Deriso Deck, Doc. 440 at ¶¶ 13 — 31. Federal Defendants had the information needed to perform such a simple analysis of the available data, but did not do so, and unjustifiably relied on the unsealed data to form a quantitative conclusion that salvage rates increased sharply above — 5,000 cfs OMR. This was a clear scientific error.

DefendanWlntervenors argue that the agency cannot violate the ESA by failing to take an additional step to scale the salvage data to salmonid population abundance. They maintain that all the ESA requires is that NMFS consider the “available” evidence not that it create new data or “follow scientific practices defined by Export Plaintiffs experts.” Doc. 484 at 51. It is true that the best available science standard does not require NMFS to create new data or apply new models to existing data, Building Indus. Ass’n, 247 F.3d at 1246 . However, here, NMFS put data to a use for which it is not appropriate, as it produces unreliable results. Cf. NWF v. EPA, 286 F.3d at 565 (a court may reject agency’s choice of model when it “bears no rational relationship to the characteristics of the data to which it was applied”). Every biostatistics expert who presented evidence in this and related fish cases has agreed that it is wholly inappropriate and scientifically unreasonable to draw management conclusions from a plot comparing unsealed salvage data to OMR flows collected over a period of years when population varied. The agency is required to apply generally recognized and accepted biostatistical principles, which constitute best available science, in reaching its decisions.

*828 b. Was the BiOp’s Reliance on Figures 6-65 and 6-66 Hamiless?

Alternatively, Federal Defendants argue that any such error was harmless given the other record evidence that supports the BiOp’s conclusions. No party has provided authority that a harmless error rule applies when the agency commits a substantive error under the ESA. 8 Arguendo, Defendants’ alternative evidence is considered.

(1) Record Citations Provided by Mr. Stuart.

Mr. Stuart states: “[additional assessment of the effects of the OMR flow levels on salmon loss was derived from data provided by NMFS staff for the BDCP. NMFS 79238-239 ; 79240-83808; 90852-98.” Fourth Stuart Decl., Doc. 485 at ¶ 73. These are several thousand pages of documents. Mr. Stuart does not explain how the voluminous referenced information was used or analyzed. It is impossible to determine whether these referenced pages provide a sufficient alternative basis for the BiOp’s conclusions. NMFS has provided no explanation for an alternative to its scientifically unreliable conclusions. Humane Soc. v. Locke, 626 F.3d at 1048 (holding NMFS did not offer a satisfactory explanation for its findings); Am. Tunaboat Ass’n, 738 F.2d at 1016 (finding that despite broad discretion afforded NOAA, where record evidence detracts from that relied upon by the agency, a court may find the agency’s decision arbitrary and capricious). Citing this information and then failing to explain it is arbitrary and capricious.

(2) Particle Tracking Model Results.

The BiOp also relied on outputs from computer model runs utilizing the so-called Particle Tracking Model (“PTM”). Export Plaintiffs’ and DWR’s 2010 PI Motion challenged the use of PTM, arguing that while PTM is useful to track the fate of neutrally buoyant particles, it does not accurately reflect the behavior of salmonids, which are strong, volitional swimmers. These challenges were rejected in the May 18, 2010 PI Decision on the grounds that the BiOp acknowledged the limitations of PTM and reasonably relied on PTM studies to support certain conclusions:

Consol. Salmonid Cases, 713 F.Supp.2d at 1141 . These challenges are discussed in other sections of this decision. Here, it is sufficient to note that the PTM results are not a complete replacement for Figures 6-65 and 6-66, as they only explain how particles, not actual fish, would respond to increased OMR flow. Nor do Federal Defendants point to any other information in the record that delineates “where along the spectrum of OMR flows any significant change in salvage could be observed,” the *829 purposes for which Federal Defendants use 6-65 and 6-66.

The same reasoning applies to acoustic tag studies and other data the BiOp relies upon. This other data was treated in the May 18, 2010 decision and below in greater detail. It is undisputed that none of these studies or additional data pinpoint for management purposes at what point negative OMR flows must be controlled. Doc. 347 ¶¶ 128-138.

It is not appropriate to speculate how NMFS’s analysis and/or conclusions would have changed had the data used in figures 6-65 and 6-66 been scaled to population size. This must be done on remand.

Whether there is sufficient data to scale CV steelhead salvage information to population size is unclear. Compare BiOp at 107 (discussing existing population data and difficulties posed by “lack of monitoring program”) with Doc. 431 at 11 (citing Burnham Deck, Doc. 439 at ¶ 42 (“NMFS has access to the number of hatchery-produced salmon each year”), for the proposition that NMFS had “readily available” data regarding how many CV Steelhead were released from hatcheries each year)). That scaling for population size may not be possible for all species may limit NMFS’s efforts. If population data is unavailable for certain species, the agency must nevertheless explain how it can make management conclusions without such information. The extent to which any such limitations mitigate NMFS’s failure to scale raw salvage data cannot be discerned from the present record.

(3) Figures 6-71, 6-72, and 6-73.

The BiOp also relies on a series of plots, taken directly from Reclamation’s Biological Assessment (“BA” or “OCAP BA”), of monthly juvenile salmonid “loss” against average exports. The first set of plots, Figure 6-71, depicts loss of juvenile Chinook salmon versus average monthly CVP and SWP exports for a period from 1995 through 2007.

[[Image here]]

*830 BiOp at 370. This plot was specifically cited as quantitative evidence of a relationship between exports and loss at the pumps:

Id. at 368-69. In this way, the BiOp used Figure 6-71 as quantitative evidence of a statistically significant connection between loss of older juvenile Chinook salmon and export levels. These plots suffer from the same flaw of misusing raw salvage data. They must be re-analyzed and explained on remand if they are to be used as scientific justification for the BiOp’s conclusions.

Figures 6-72 and 6-73 present a more difficult issue, as they concern CV steel-head salvage, for which much less data is available. Figure 6-72 plots monthly CV Steelhead salvage (both clipped/hatchery and unelipped/wild 9 ) against exports:

[[Image here]]

Id. at 371. The BiOp indicates that these “regressions resulted in significant rela *831 tionships between exports and salvage of steelhead at the facilities, more so for the SWP than the CVP.” Id. at 369. Figure 6-73 plots monthly CV Steelhead salvage against the monthly average export to inflow ratio, finding significant relationships:

[[Image here]]

Id. at 370-71. The 2010 PI Decision found that "[a]s to the CV steelhead, for which no population numbers are available, it is less clear whether the use of raw salvage numbers is always inappropriate.” Consol. Salmonid Cases, 713 F.Supp.2d at 1143 . Export Plaintiffs do not explain how it would be possible to scale to overall population size the wild CV steelhead salvage data, referenced in Figure 6-73 as “undipped” CV steelhead. However, plaintiffs point out that the exact number of hatchery CV steelhead released each year is known, and therefore argue that NMFS could have scaled the hatchery or “clipped” CV Steelhead data to population size. Doc. 431 at 11. Federal Defendants do not respond to this assertion. There appears to be no reason to distinguish between the clipped CV steelhead analyses and Chinook salmon analyses, for which population scaling is the best available scientific methodology.

More importantly, Federal Defendants do not explain how these figures, even if valid, serve the same purpose as Figures 6-65 and 6-66, 66, which were cited to demonstrate that “[l]oss of older juveniles at the CVP and SWP fish collection facilities increase sharply at [OMR] flows of approximately -5,000 cfs and depart from the initial slope at flows below this.” Id. at 361.

None of the alternative bases offered by Federal Defendants are sufficient to render NMFS’s reliance on Figures 6-65 and 6-66 “harmless error.” The significance of *832 Mr. Stuart’s voluminous record citations is unexplained. Neither the PTM Modeling Results, nor Figures 6-71, 6-72, or 6-73 provide alternative bases for NMFS’s conclusions regarding the negative OMR flows below which loss of juvenile salmonids “increases sharply.” Export Plaintiffs’ motion for summary judgment that Federal Defendants acted unlawfully by relying on raw salvage analyses is GRANTED; Federal Defendants’ and Defendant-Intervenors’ cross motions are DENIED.

c. Separate Challenges to Statistical Analyses in Figures 6-71, 6-72, and 6-73.

Export Plaintiffs also maintain that all three sets of graphs misrepresent the statistical significance of the data because “the decision to divide the data by month created an arbitrarily large sample size.” Doc. 431 at 10. According to Export Plaintiffs, this “produce[d] facially incredible P-values (the standard statistical measure of significance) that misrepresented the validity of the models that were fit to the data.” Id. A P-value “represents the probability that the result obtained in a statistical test is due to chance rather than a genuine relationship between the variables.” Burnham Decl., Doc. 439 at ¶ 43. Regression analysis is generally considered statistically significant when the P-value is smaller than 0.05. Id.

The upper right graph from Figure 6-72, which plots monthly SWP exports for January through May of 1998 through 2006 against the raw salvage of hatchery steelhead, presents a P-value of 0.000000000136. This is several million times smaller than 0.05. Dr. Burnham opines that while “[sjuch a P-value is not impossible ... it would be surprising and especially so with this graph” because “[t]he scatter of the data points indicates that the regression is not a very good fit.” Id. at ¶ 44.

Dr. Burnham describes the R2 value of 0.62 as not very strong. The R2 value is a statistical measure of how well the regression explains the data. “Roughly speaking an R2 of 0.5 means that the regression accounts for 50% of the variation observed in the data, while the other 50% is explained by other, unknown factors. Generally speaking an R2 of 0.5 is considered weak, while an R2 of 0.8 or above is considered strong.” Id. at ¶ 36.

Dr. Burnham opined that this high level of apparent “significance” is the result of treating each of the monthly export points as an independent data point, which increases the sample size and influences the P-value. Id. at ¶ 45. Because the monthly data is “highly correlated with each individual year, due to the unique natural conditions that characterize each year, such as the abundance of the salmon cohort, patterns of flow, changes in temperature, etc.” Id. at 47. Dr. Burnham offers a helpful explanation of why this is a problem:

Id. at ¶ 46. In the present case, Dr. Burn-ham opines that NMFS’s approach caused the relationship between exports and salvage to “appear more significant than it really is,” when in fact “the high salvage levels in [certain] years may have actually been primarily caused by one independent factor, such as a large hatchery release *833 that year.” Id. at ¶ 49. The results of the “January Only” data depicted in red on Figure 6-73, are different, showing much higher P-values, with only one of the four graphs showing statistical significance. Id. at 51.

Mr. Stuart responds to these critiques in his Fourth Declaration:

Doc. 485 at ¶ 75. Mr. Stuart now argues that the trend lines have biological significance even if the statistical analyses were incorrectly performed. Yet, the BiOp specifically emphasizes how these graphs demonstrated “significant relationship[s]” between the variables. See, e.g., BiOp at 369.

Mr. Stuart also provides a substantive rebuttal to Dr. Burnham’s critique of the statistical analyses, premised on Dr. Burn-ham’s argument that it was clear error to use monthly data points in the correlation analysis because they are highly correlated within years. Mr. Stuart opines that this premise is flawed because “[m]onthly exports do not necessarily correlate with each other between months or between years.” Fourth Stuart Deel., Doc. 485 at ¶ 76.

Id. This provides a partial explanation for the statistical analyses.

This is an area of disagreement among experts. The agency is due deference in such circumstances, unless its experts’ opinions are unsupported or wrong. Cactus Corner, LLC v. U.S. Dept. of Agric., 346 F.Supp.2d 1075, 1113 (E.D.Cal. 2004).

Marsh v. Oregon Natural Resources Council, 490 U.S. 360, 378 , 109 S.Ct. 1851 , 104 L.Ed.2d 377 (1989) (explaining in the context of an agency’s decision not to supplement a NEPA environmental impact statement that “courts should not automatically defer to the agency’s express reliance on an interest in finality without carefully reviewing the record and satisfying themselves that the agency has made a reasoned decision based on its evaluation *834 of the significance-or lack of significance-of the new information” and noting that “[a] contrary approach would not simply render judicial review generally meaningless, but would be contrary to the demand that courts ensure that agency decisions are founded on a reasoned evaluation “of the relevant factors” ”).

This is a close call. Dr. Burnham’s opinion suggests that it was unreasonable for NMFS to rely on the statistical analyses it performed. Mr. Stuart offers some explanation to counter certain aspects Dr. Burn-ham’s critique, but does not satisfactorily explain the anomalous statistical results. Because the BiOp’s reliance on these graphs must be remanded for other reasons, the agency must explain the flaws in its approach to the statistical analyses on remand.

2. Failure to Perform a Population-Level Quantitative Analysis.

Plaintiffs maintain that the BiOp violated the best available science requirement because it failed to employ a “population-level quantitative analysis.” Doc. 431 at 15. It is undisputed that quantitative population dynamics models, or life cycle models, are the most reliable method to evaluate the impacts of various stressors on a fish population. This has been indisputably established in these related cases. Plaintiffs’ expert Dr. Hilborn opined:

Hilborn Deck, Doc. 443 at ¶ 5. The BiOp concedes that “[i]deally, a life cycle approach, in which the effects on individual life stages on the life cycle could be estimated independent of the effects on other stages, would be implemented to assess the relative impacts on abundance.... ” BiOp at 66.

However, such models only qualify as “best available science” where an appropriate model is available. 10 In the Consolidated Delta Smelt Cases, an appropriate smelt population dynamics was not available at the time the biological opinion in dispute in that case was issued:

San Luis v. Salazar, 760 F.Supp.2d at 884-85 .

The 2010 PI Decision in this case found that Plaintiffs’ were not likely to succeed on the merits of their life cycle modeling claim because they had not “presented] evidence that they, or anyone else[,j developed or made available to NMFS an appropriate life cycle model or the results of an appropriate life cycle analysis prior to the issuance of the BiOp.” Consol. Salmon-id Cases, 713 F.Supp.2d at 1132 . However, the PI Decision is not law of the case. S. Or. Barter Fair v. Jackson County, Or., 372 F.3d 1128 , 1136 (9th Cir.2004) (“[Djecisions on preliminary injunctions are just that — preliminary—and must often be made hastily and on less than a full record.”) (citing Univ. of Tex. v. Camenisch, 451 U.S. 390, 395 , 101 S.Ct. 1830 , 68 L.Ed.2d 175 (1981)).

Plaintiffs insist that NMFS did have access to “several fully featured quantitative life-cycle models[,j which had been specifically designed for use in the Delta____” Doc. 431 at 15. Plaintiffs focus on

the Interactive Object-Oriented Salmon simulation (“IOS”) model, developed by Bradley Cavallo, and the Oncorhynchus Bayesian Analysis (“OBAN”) model, which was developed in coordination with NMFS. Id. at 19

a. IOS.

The BiOp discussed its decision not to use IOS, which was designed to evaluate the influence of different Central Valley water operations on the life cycle of winter-run using simulated historical flow and water temperature inputs.” BiOp at 65.

Id. at 65-65 (italics in original). It is ironic that NMFS’s reverence for “rigorous scientific methodology” is honored in the breach by the agency’s failure to utilize *837 the most rigorous method possible in the disputed BiOp.

Garwin Yip, supervisor for the Water Operations and Delta Consultations branch of NMFS’s Sacramento Area Office, elaborated on the first explanation provided in the BiOp — that IOS estimates did not match actual historic winter run population levels — by pointing out that even Plaintiffs’ expert Dr. Burnham stated that: “If the data does generally not match the model, then we know that our assumption is somehow incorrect and needs refinement.” Yip Decl., Doc. 481 at ¶ 9 (citing Burnham Decl., Doc. 489, at ¶ 24). Mr. Yip points out that “[h]ad NMFS based our conclusion for winter-run on the quantitative approach of the IOS life cycle model results contained in the BA, we would have erroneously concluded that the proposed CVP/ SWP operations would help a great deal in recovering the species, rather than jeopardizing it.” Id. at ¶ 10.

Second, Mr. Yip points out that the CalFed Science Panel, in its review of the December 11, 2008, draft BiOp, discussed NMFS’s decision not to use IOS in the BiOp and acknowledged that “the IOS model is relatively new and has not been extensively vetted and published, but all of these types of models are flexible and the Panel wonders if, with sufficient time and with some adjustments and modifications, whether a new version of the IOS could be used.” Id. at ¶ 11 (citing AR 11 00108178 (Anderson, et al.2009)). The CalFed Science Panel did not recommend or encourage the use of IOS in its current state at the time the BiOp was being developed.

Mr. Yip points out that the NMFS Central Valley Office requested that NMFS’s Southwest Fisheries Science Center (“NMFS-SWFSC”) review various models, including IOS, for overall assumptions and limitations. According to Yip, NMFS-SWFSC “did not have the considerable staff resources and time it would take to adequately review and comment on the IOS model, it did have previous experience with similar models developed by Cramer Fish Sciences, and therefore, offered some comments.” Id. at ¶ 12.

AR 00101045-6 (emphasis added).

NMFS contends that the IOS 'model was not available in a reliable, appropriate, and usable form at the time the BiOp was developed, and denies that IOS could have been adapted in a reasonable amount of time for use in the BiOp. NMFS knows all experts agree a life cycle model is the best methodology for measuring population effects. The agency continues to dodge seri *838 ous questions the parties are entitled to have answered by refusing employ existing models. At the same time, NMFS continues to plead poverty by describing this case as a “data-poor situation.” If the data is so sparse that a workable model cannot be formulated, when does NMFS’s failure to ensure appropriate data collection is taking place constitute bad faith? After more than five years of dispute, future pleas of data poverty will no longer be accepted.

As to the IOS model, although it is a close call, Federal Defendants’ contention that IOS was not available in a reliable form at the time the BiOp issued has not been rebutted. Plaintiffs’ motion for summary judgment that NMFS violated the ESA by failing to apply IOS in the BiOp is DENIED; Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED.

b. OBAN.

Plaintiffs further argue that NMFS acted unlawfully by failing to apply the OBAN model in the BiOp. The BiOp mentions OBAN, but does not specifically address it, opting instead to generally explain that because existing life-cycle models only address population abundance, leaving out other aspects of the Viable Salmonid Population (“VSP”) framework (e.g., spatial structure and genetic and life history diversity), the BiOp’s analysis is superior because it encompasses these other factors:

Id. at 67-68.

NMFS staff from the Seattle office collaborated with Dr. Nobel Hendrix, the author of the OBAN model, during the model’s development. AR 00050578. Throughout 2008, NMFS communicated with Dr. Hendrix about the model, AR 00023869-70, and NMFS staff scheduled and attended meetings about the model, see, e.g., AR 00050874. NMFS requested that its own Science Center review the OBAN model. AR 00046767-69.

Plaintiffs argue that Dr. Brian Wells, an NMFS fisheries biologist, described the OBAN model as “a great approach,” AR 00050825, “a superior model design” and “well thought out,” a model that “is the best approach laid out yet and deserves full attention,” AR 00103798. Plaintiffs selectively quote Dr. Wells. For example, in a November 13, 2008 letter to Bruce Oppenheim, a NMFS biologist working on the BiOp, Dr. Wells did compliment the OBAN model:

AR 000103798. However, his next paragraph raises a concern about the available data set:

Id.

The model’s own developer agreed with the Science Center’s concern that OBAN contains “a lot of factors that are hypothesized to affect winter run relative to the *840 amount of data,” and suggested that a “goal of the modeling effort is to identify some of the most important places to collect additional data.” AR 00054082 (emphasis added). As of February 2009, NMFS still had not received a clear response about whether appropriate and sufficient data were available to reliably model population dynamics using OBAN. AR 00070672 (indicating it “is uncertain as to whether appropriate and sufficient data are available to reliably model the population dynamics of winter-run Chinook salmon and spring-run Chinook salmon. If appropriate data are not available, application of the OBAN and IOS models to inform risk analysis may lead to erroneous management decisions. Before utilizing the OBAN or IOS models, it is necessary for PRD to gain further confidence in the various results each model can produce.”).

. Plaintiffs also do not address the 2010 PI Decision’s finding that the application of population or life cycle models is not feasible for any analyses applicable to the CV steelhead, for which no population indices are available. Consol. Salmonid Cases, 713 F.Supp.2d at 1164 .

Federal Defendants also remonstrate that even if sufficient data had been available, email communications in late 2008 and early 2009, while the BiOp was being prepared, describe the model as still under development. See AR 00060571 (January 21, 2009 email discussing need to develop temperature metrics). The software needed to run the model was not available until late April 2009. AR 00086362 (April 20, 2009 workshop demonstrating new software for OBAN model).

Plaintiffs rejoin that “Defendants’ principal criticism of the OBAN model ... that the data were incomplete ... suggests a fundamental misunderstanding of life-cycle modeling generally.” Doc. 48 at 15. Plaintiffs characterize Defendants’ position as based on an erroneous premise that a life cycle model can never be used unless and until it is fully and finally perfected, peer reviewed, and populated with perfect data.” Id. Plaintiffs complain that this is an impossible standard that ignores the reality that perfect data does not exist and that modeling is an iterative process. For example, Dr. Hilborn opines that life cycle models are “always evolving and many times being challenged by models that make alternative assumptions.” Hilborn Reply Deck, Doc. 496 at ¶ 26.

Id. Dr. Hilborn pointed out that preliminary results from OBAN indicated that water temperature and harvest were the dominant factors affecting salmonid populations and that water exports was “not one of the most powerful explanatory factors.” Id. at ¶ 25.

The record reflects that NMFS was working in late 2008 to integrate OBAN into the consultation process, see AR 00060572-73; AR 00052306-07, but that the agency elected not to make use of the model in the final BiOp approximately three months before the BiOp was issued, AR 0065191-94. Plaintiffs assert that this was unreasonable because OBAN was “fully functional and ready to be integrated into the BiOp.” Doc. 487 at 16. Agency experts concluded that there was not enough data to reliably apply OBAN. 12

*841 This is more disassembling by NMFS. Having not devoted necessary attention to OBAN, it rationalizes its doubts about the reliability of application of OBAN in the BiOp as not clearly erroneous. NMFS remains in the position where it can raise doubt about all conflicting science and hide behind the rubric it cannot be compelled to collect data or develop a model. Plaintiffs’ motion for summary judgment that NMFS violated the ESA by failing to apply the OBAN model in the BiOp is DENIED and Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED, but this is the last time NMFS will be permitted to avoid studying, analyzing, and applying a life cycle model. NMFS’s chronic failure to do so now approaches bad faith in view of the undeniable importance of the information to resolve the perennial dispute over population dynamics. At some point, this diminishes the agency’s credibility.

c. Ricker and/or Beverton-Holt Models.

Plaintiffs argue that, even if use of the IOS and/or OBAN models was not legally required, NMFS violated the ESA because it did not make use of certain “basic tools of fisheries management,” such as the Ricker or Beverton-Holt models, two mathematical models developed in the mid-1900s. See Doc. 431 at 26. According to Plaintiffs’ expert Dr. Burnham, these models make use of “simple mathematical expressions, based solely on past observations [that] combine all life history and environmental information into the period between parent spawners and the resultant returning spawners in the next generation.” Burnham Decl., Doc. 439 at ¶ 18. According to another of Plaintiffs’ experts, Dr. Ray Hilborn, these models are part of a “standard set of population dynamics models ... that form the core of the body of knowledge” among fisheries biologists. Hilborn Reply Decl., Doc. 493 at ¶ 6. Dr. Hilborn concludes:

Id. at ¶ 30.

NMFS’s biologist Garwin Yip submitted a detailed response:

Third Yip Decl., Doc. 518. The application of these “simple” models is not as straightforward as Plaintiffs claim. The law relegates their use and application to the discretion of the agency unless clearly erroneous. Plaintiffs have not demonstrated that NMFS’s use or non-use of these models is more than scientific dispute, which is resolved in favor of the agency. A court cannot lawfully second-guess the agency, unless clear scientific error or bad faith is so manifest that the agency’s judgments can no longer be trusted. Plaintiffs’ motion for summary judgment that NMFS acted unlawfully by failing to use the Ricker and BevertonHolt models is DENIED; Federal Defendants’ and Defendan-Intervenors’ cross motions are GRANTED. 13

d. Does the Absence of a Quantitative Analysis Render the BiOp’s Quantitative Limits Per Se Unjustified?

Plaintiffs argue that even if the BiOp’s failure to apply quantitative life cycle modeling was not per se unlawful, the absence of such analyses “necessarily cripple[s] the specific quantitative limits imposed by the RPAs.” Doc. 431 at 31. The Peer Review Panel addressed this issue:

AR 00089620-21 (emphasis added). Although the Peer Review certainly warned of the pitfalls of attempting to formulate RPA’s without “quantitative integrative tools,” it acknowledged that the “long-term solution” was further development of modeling tools. The Peer Review did not outright advise the abandonment of quantitative RPAs in the short term. It did caution that the benefits cannot be quantified, nor can the fiscal and social costs.

For the purposes of this challenge to the BiOp and its RPAs, it is not appropriate or justified to find all the RPA Actions unlawful simply because Plaintiffs were not satisfied with the quantitative analyses performed in the BiOp. Each challenged RPA must be analyzed in light of the record evidence.

B. Baseline Analysis Challenges.

1. Failure to Distinguish Between Discretionary and Nondiscretionary Actions.

Plaintiffs opening brief advances an elaborate argument based on National Association of Home Builders v. Defenders of Wildlife, 551 U.S. 644 , 127 S.Ct. 2518, 168 L.Ed.2d 467 (2007), which held that the ESA’s consultation requirement was not triggered when an agency undertook nondiscretionary actions, because the agency has no choice. Plaintiffs argue that NMFS erred by failing to distinguish between the discretionary and nondiscretionary aspects of CVP and SWP project operations. Doc. 431 at 48-55. Although it is undisputed that Reclamation operates the project to fulfill certain mandatory water delivery obligations, Plaintiffs’ argument that Home Builders should be extended to require NMFS to segregate discretionary from non-discretionary operations, placing non-discretionary ones in the “baseline” for purposes of evaluating the action’s effect on the Listed Species, was rejected in a December 14, 2010 Memorandum Decision issued in the related Consolidated Delta Smelt Cases:

San Luis v. Salazar, 760 F.Supp.2d at 947-48 . Any voluntary efforts by NMFS to segregate discretionary from non-discretionary actions in the BiOp does not alter the fact that Home Builders imposes no legal obligation to do so.

Plaintiffs advance a related argument that even if Home Builders does not require segregation of discretionary from non-discretionary project operations in the BiOp, the ESA otherwise requires NMFS to consider only discretionary operations when evaluating the “effects of the action” vis-a-vis the environmental baseline. Doc. 431 at 55-58. This is based in part on the Science Panel’s recommendation that NMFS model a baseline that represents a “hypothetical situation in which physical project infrastructure exists, but no project operations are performed except those mandated by prior agreements or those that are not part of the proposed actions.” AR 00108175. The Panel offered:

Id. (emphasis added).

NMFS addressed this recommendation in the BiOp:

BiOp at 57-60.

NWF v. NMFS II, 524 F.3d 917, 929 , applies to whether the BiOp’s baseline rationale is reasonable for the proposed action. There, NMFS’s 2004 biological opinion for the Federal Columbia River Power System (“FRCPS”) “included in the environmental baseline for the proposed action the existing FCRPS, various supposedly nondiscretionary dam operations, and all past' and present impacts from discretionary operations.” Id. at 926 . In addition, NMFS “adopted a novel ‘reference operation’ approach ... purportedly in order to account for the existence of the FCRPS dams.” Id.

Id. The Ninth Circuit evaluated this “reference operation” approach:

Id. at 928-29.

NWF v. NMFS found it inappropriate for NMFS to treat as “non-discretionary” activities undertaken to achieve “broad mandates which do not direct agencies to perform any specific nondiscretionary actions, but rather, are better characterized as directing the agencies to achieve particular goals.” The opinion does not address the converse situation, present here, where it is alleged that NMFS included non-discretionary aspects of Project operations in the effects of the action, rather than in the environmental baseline.

Whether or not a particular aspect of project operations is “non-discretionary” is a complex legal inquiry that may take years of litigation to resolve. See, e.g., Natural Resources Defense Council v. Kempthorne, 627 F.Supp.2d 1212 (E.D.Cal. 2009) (resolving lengthy cross motions for summary judgment on the issue of whether Sacramento River Settlement Contracts constituted non-discretionary water delivery obligation). Practically speaking, in all but the most obvious of situations or for obligations that have previously been determined to be “mandatory,” it is not feasible for the action agency to finely parse the legal determinations required to distinguish discretionary aspects of the action from non-discretionary ones in the prepa *851 ration of a biological opinion, especially in a system as complex as the joint Project operations. This is reflected in the 2009 Salmonid BiOp’s explanation that NMFS could not reliably propose a “no project operations” scenario to characterize the environmental baseline.

Plaintiffs’ ultimate argument is that if non-discretionary project operations are backed out of the “effects of the action” and instead are included in the “baseline” the effects of the action will no longer be “appreciable.” This identical argument was explicitly rejected by NWF v. NMFS II's holding that comparison of the effects of the action against a hypothetical “reference operation” is not appropriate because the jeopardy analysis must focus on “whether the action[’s] effects, when added to the underlying baseline conditions, would tip the species into jeopardy.” 524 F.3d at 930 . The Ninth Circuit reasoned:

Id. at 929-31 (emphasis added).

Nothing in the law requires NMFS to segregate discretionary aspects of coordinated Project operations from non-discretionary ones in the manner Export Plaintiffs demand. If feasible, it could have made sense for NMFS to do this to better document the relationship between the requirements of the species and the action agency’s statutory authority to implement the RPA. But, NMFS disclaims the capacity to undertake appropriate modeling and related analysis. Export Plaintiffs have not demonstrated that NMFS’s disclaimer is unreasonable.

Plaintiffs’ motion for summary judgment that NMFS acted unlawfully by failing to segregate discretionary aspects of Project operations from non-discretionary ones is DENIED; Federal Defendants’ and DefendanUntervenors’ cross motions are GRANTED.

2. Treatment of Available Data on Ocean Harvest and Ocean Conditions.

Plaintiffs next argue that NMFS acted unlawfully by failing to quantitatively analyze available data on ocean conditions and ocean harvest. Plaintiffs assert that these failures resulted in an “improper jeopardy finding and invalid RPA.” Doc. 431 at 34.

a. Consideration of Ocean Conditions Data.

Plaintiffs argue that the BiOp should have used available data to quantitatively analyze the impact of ocean conditions on the Listed Species. Doc. 43 at 40-45. Specifically, Plaintiffs assert that NMFS should have performed quantitative analyses using data measuring the Pacific Decadal Oscillation (“PDO”), a recognized index of ocean conditions, so that the “major role ocean conditions play in determining abundance levels of salmonids ... could be compared with other stressors, such as the effects caused by water exports.” Deriso Decl., Doc. 440 at ¶ 41. Record evidence suggests ocean conditions play a substantial role in salmon abundance. See id. (citing Hare & Mantua (1997), AR 00120076-84); see also AR 00084001 (Reclamation arguing to NMFS that “[o]cean conditions likely amount to 99.999% of the cause of the status of Central Valley species” and complaining that NMFS “isn’t acknowledging this overwhelming stressor”).

*853 In a four and a half page section, the BiOp discusses the importance of natural environmental cycles, including those affecting ocean productivity:

BiOp at 149-53.

Plaintiffs do not argue that the BiOp entirely failed to consider ocean conditions and/or the PDO. 14 Rather, they argue that NMFS should have evaluated the impact ocean conditions have on salmon populations quantitatively, so that the effect of ocean conditions can be compared to the effects of project operations. In support of this argument, Plaintiffs quote the Peer Review: “[T]he possibility exists that we may be analyzing effects that occur within the system that ultimately are overshadowed by dynamics and effects in the marine phase.” Doc. 487 at 36 (citing AR 0089603). Plaintiffs’ quotation is incomplete; the whole paragraph provides:

AR 0089603. Although the Peer Review expressly recognizes a need to consider “the dynamics and effects in the marine phase,” as possibly significant to the species, this is not a pronouncement that standard scientific practice demands a quantitative analysis of ocean conditions. Plaintiffs cite no legal requirement that NMFS perform a quantitative analysis to determine the relative impact of ocean conditions on salmon populations. The ESA does not require such an analysis. The caselaw affirmatively decries such a relativistic approach. See NWF v. NMFS II, 524 F.3d at 930 (“even where baseline conditions already jeopardize a species, an agency may not take action that deepens the jeopardy by causing additional harm ”). The relevant question is whether or not the record supports NMFS’s conclusion that Project operations appreciably diminish those species’ likelihood of survival and recovery in light of all preexisting natural and manmade conditions. The appropriate focus, under NWF v. NMFS II, is not on the Projects’ relative *857 contribution to harm compared to ocean conditions, but rather, whether Project operations cause separate harm, including by making the species more vulnerable to adverse ocean conditions.

The BiOp concludes that because the natural cycles that drive ocean conditions are “highly variable,” it makes it “very difficult, if not impossible, to accurately predict what they will be like in the future,” and because “the potential for poor ocean conditions exists in any given year, and there is no way for salmon managers to control these factors, any deleterious effects endured by salmonids in the freshwater environment can only exacerbate the problem of an inhospitable marine environment.” BiOp at 152-53.

The BiOp cites Lindley (2009) for the proposition that deterioration in ocean conditions has “acted on top of a long-term, steady degradation of the freshwater and estuarine environment.” Id. at 149 (citing Lindley (2009), AR 00123514-631). Plaintiffs are correct that Lindley (2009) found that ocean conditions and fishery management played roles in the low escapement of 2007. AR 00123517-18. Plaintiffs quote Lindley (2009)’s conclusion that “unfavorable ocean conditions were the proximate cause” of declines to the 2004 and 2005 broods. Doc. 487 at 38. Plaintiffs take these statements out of context. Before discussing impacts to salmon populations caused by human effects on the freshwater environment, Lindley (2009) emphasized the difference between “proximate” and “ultimate” causation:

AR 00012355. The paper concluded that human manipulation of the freshwater environment likely “played a significant role in making this stock susceptible to collapse during periods of unfavorable ocean conditions.” AR 00123551.

The law does not require a quantitative, comparative fault type analysis. If the species is in decline and one of the causes is Project operations, the agency has discretion to address and mitigate the resulting harm. The extent to which the record affirmatively demonstrates that Project operations cause separate harm is examined below in connection with Plaintiffs’ challenges to the effects analysis.

Plaintiffs’ motion for summary judgment that NMFS violated the ESA by failing to quantitatively analyze ocean conditions is DENIED; Federal Defendants’ and DefendanNIntervenors’ cross motions are GRANTED.

b. Consideration of Ocean Harvest Impacts.

Plaintiffs also argue that NMFS acted unlawfully by failing to quantify the effect of ocean harvest on the Listed Salmonids. Their argument is that: (1) NMFS has sufficient quantitative data to analyze the effects of ocean harvest on the Listed Salmonids because it manages the ocean harvest; (2) that data, if it had been quantitatively analyzed in the BiOp, would have revealed that the losses caused by Project operations are miniscule in comparison to losses caused by ocean harvest. Doc. 431 at 37-40.

The ESA requires NMFS to evaluate to what extent the losses are caused by the proposed action, here the operation of the CVP and SWP. The action in question does not include ocean harvest, which in part is the result of separate government activity. NMFS quantitatively evaluated the impacts of ocean harvest on the Listed Species in a separate biological opinion. *858 The Salmonid BiOp acknowledges that ocean -harvest is a part of the environmental baseline affecting species viability, see BiOp, at 144-46 (discussing ocean commercial and ocean and inland sport harvest as “factors responsible for the current status” of the Listed Species), but does not quantitatively integrate the impact of ocean harvest into the analysis of Project-related impacts on the species.

NMFS’s obligation under the ESA is to evaluate how Project operations affect the Listed Species, in light of a depleted population impacted by ocean harvest and other conditions. It is inexplicable that these species are being managed in a piecemeal fashion, without considering all aspects of their life cycle in the same analysis, which would facilitate description of the true effect Project operations have on the species in light of other conditions. What population is available to be affected by Project operations is entirely relevant, as all Defendants have sought to attribute the species’ decline to Project operations. Nonetheless, under NWF v. NMFS, the analytical focus is not on the relative contribution "of the Projects to the species’ condition, but whether the Projects cause additional, independent harm. Plaintiffs’ motion for summary judgment that NMFS acted unlawfully by failing to quantitatively analyze ocean harvest impacts to determine whether, relatively speaking, they overwhelm Project impacts is DENIED; Federal Defendants’ and Defendanb-Intervenors’ cross motions are GRANTED.

C. Effects Analysis Challenges.

1. Use of a 100-Year Timeframe.

The BiOp evaluated how the proposed action would impact the species’ risk of extinction over a 100-year time period. BiOp at 51. The BiOp explains that the jeopardy standard has been interpreted in the Joint Consultation Regulations as “a requirement that Federal agencies ensure that their actions are not likely to result in appreciable reductions in the likelihood of both the survival and recovery of the species in the wild by reducing its numbers, reproduction, or distribution.” Id. at 42 (citing 50 C.F.R. § 402.02 ). This means:

BiOp at 42. The VSP framework estimates the viability of salmonid populations by defining a viable salmonid population as one that “has a negligible probability of extinction over a 100-year time frame.” Id. at 51 . More specifically, the BiOp sets the threshold for jeopardy as the point at which the effects of the action, in the context of the baseline, result in a risk of extinction of greater than five percent over 100 years. The threshold combines two types of information: a probability of extinction expressed as the percentage likelihood of extinction and a timeframe within which that probability may come to pass, expressed in years. NMFS utilizes a NMFS technical memorandum by McElhany et al. (2000), AR 00124576, and a 2007 article by Lindley et al., AR 00123475, as “a bridge to [this] jeopardy standard.” BiOp at 42-43, 51-53.

*859 The five percent probability of extinction component of the BiOp’s standard is derived Lindley (2007), which opines: “We assume a 5% risk of extinction in 100 years is an acceptably low extinction risk for populations (Thompson, 1991).” AR 00123477 (emphasis added). Lindley (2007) describes specific criteria for assessing the risk of extinction, and “assumefs] that a 5% risk of extinction in 100 years is [] acceptably low....” AR 00123477. Lindley (2007) characterizes a risk of extinction of less than five percent within 100 years as “low,” greater than five percent within 100 years as “moderate,” and greater than 20% within 20 years as “high.” AR 00123478.

The BiOp appears to derive the 100-year timeframe from McElhany (2000). See BiOp at 51. McElhany (2000) describes a viable salmonid population as “an independent population of any Pacific salmonid [] that has a negligible risk of extinction due to threats from demographic variation (random or directional), local environmental variation, and genetic diversity changes (random or directional) over a 100-year time frame.” AR 00124594. Regarding the selection of the 100-year time frame, McElhany (2000) states: “While it is ultimately an arbitrary decision, the 100-year time scale was chosen to represent a ‘long’ time horizon for evaluating extinction risk.” Plaintiffs claim that neither the BiOp nor McElhany provide a reasoned basis for the decision to choose a time frame of 100 years, as opposed to any other, shorter, timeframe. When the Mc-Elhany (2000) sentence is read in context, an explanation is provided for the 100-year time scale emerges:

AR 00124595. Plaintiffs identify no record evidence suggesting that this explanation is irrational.

Plaintiffs also argue that the 100-year timeframe is arbitrary in light of the fact that NMFS used a 24-year time frame just a year earlier in the biological opinion for the Federal Columbia River Power System (“FCRPS BiOp”). AR 00130923. The FCRPS BiOp addressed critiques suggesting that it use a 100-year extinction risk period as follows:

AR 00130937.

Plaintiffs maintain that these paragraphs' from the FCRPS BiOp demonstrate that NMFS adopted a “prior practice” of using a 24-year extinction period and that NMFS failed to supply a reasoned basis for departing from that prior practice. See River Runners for Wilderness v. Martin, 593 F.3d 1064, 1075-76 (9th Cir.2010) (“Part of the discretion granted to federal agencies is the freedom to change positions.... [A]n agency’s view of what is in the public interest may change, either with or without a change in circumstances. But an agency changing its course must supply a reasoned analysis.”) (internal citations and quotations omitted).

Plaintiffs suggest there is conflict between these two biological opinions. The FCRPS BiOp utilizes a 24-year timeframe to quantitatively evaluate short-term extinction risk where sufficient data was available to do so. AR 00131546. That was only possible for six of the 13 species covered by that biological opinion. NMFS did not have sufficient data to perform a 24-year analysis for the remaining seven species, so NMFS used a qualitative analysis of the VSP factors that considers a 100-year timeframe. See FCRPS BiOp, Chapter 8. 15 Plaintiffs have not established that the 2009 Salmonid BiOp is a marked departure from prior and/or contemporaneous practice for the risk of extinction assessment. No evidence shows the shorter time span represents the best available science. This is another dispute that ends by default, with NMFS claiming the absence of data to permit it to engage in its preferred analysis. What has not been explained is whether or not a 100-year *861 period introduces bias toward an extinction finding.

Based on limited precedent, the agency’s partial justification, and the lack of any evidence demonstrating the agency’s approach was irrational, the law defers to the agency. Plaintiffs’ motion for summary judgment that NMFS acted unlawfully by failing employing a 100-year timeframe is DENIED; Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED.

2. Winter-Run Viability Analysis.

Plaintiffs contend that the BiOp’s determination that winter-run are at a “high risk of extinction” is not based on the best available science because that determination is an unexplained departure from a “nearly contemporaneous classification” to the contrary by Lindley (2007). AR 00123478. In addition, Lindley (2007) incorporates assessments of spatial distribution, as well as genetic and life history diversity. Plaintiffs maintain that Lindley classified the winter-run as “low risk” in 2007 and that the BiOp’s reclassifying the species as being at “high risk” of extinction is unexplained. Doc. 431 at 64-66.

This argument is unconvincing for two reasons. First, Lindley (2007) did not unequivocally classify the winter-run as “low risk.” Lindley (2007) assesses a population’s viability by examining criteria relating to: (1) population size, (2) population growth rate, (3) the occurrence of catastrophic declines, and (4) the degree of hatchery influence. AR 000123478. In Table 1 of Lindley (2007) the thresholds for finding “high,” “moderate,” or “low” risk as to each of these four criteria are defined.

Risk of Extinction

Criterion High Moderate Low

Extinction risk > 20% within > 5% within < 5% within

from PVA 20 years - or any ONE of - 100 years - or any ONE of - 100 years - or ALL of -

Population size a < 50 Ne -or-N < 250 50 < Ne < 500 -or-250 < N < 2500 Ne > 500 -or-N > 2500

Population decline Precipitous decline b Chronic decline or depression c No decline apparent or probable

Catastrophe, rate and effect d Order of magnitude decline within one generation Smaller but significant decline e not apparent

Hatchery influence f High Moderate Low

*862 AR 00123478.

Lindley (2007) concluded that, at the time the paper was published, winter run “easily satisfie[d] the low-risk criteria for population size, population decline, and catastrophe, but hatchery influence [was] a looming concern.” AR 000123486. Lindley (2007) also factors in spatial distribution, as well as genetic and life-history diversity as part of an overall assessment of viability. AR 00123481. Applying these additional criteria to winter-run, Lindley (2007) concluded:

AR 00123487.

Lindley (2007) advocated that an alternative assessment, population viability analysis (“PVA”), be applied where possible and that the results of the PVA be compared to the “simpler” criteria described in Lindley (2007). The authors opined that, at the time the paper was published, winter run were at a “moderate extinction risk” according to the PVA. AR 00123486.

Federal Defendants accurately described the Lindley (2007) findings and identified more recent information, including the 2007 population crash, that render Lindley (2007)’s specific conclusions outdated. NMFS first focused on the catastrophe criteria:

BiOp at 86.

Plaintiffs argue that the BiOp’s conclusion that the almost seven-fold decrease in population in 2007, and resulting conclusion that winter run were at “high risk” of extinction based on population size is without support in the record, because, according to Lindley, even the 2007 population decline does not meet the “high risk” criteria (see Table 1 above). The population never fell to or below 500 spawners, nor *863 did the 2007 decline meet or exceed the 90% “order of magnitude” decline definition. Cramer Decl., Doc. 448 at ¶¶ 42, 44. 16 Federal Defendants do not attempt to refute this criticism, and it appears that the record does not support a high risk finding in light of Lindley (2007)’s definition of a “high risk” designation based on population. 17

Plaintiffs also challenge this determination on the ground that Lindley (2000) defines a “catastrophe,” as an event occurring within the last 10 years that caused “an order of magnitude decline within one generation,” which “is created by a 90% decline in population size” over that generation. AR 00123478. Plaintiffs point out, and Federal Defendants do not dispute, that the 2007 population decline of 76% in one generation, while significant, did not meet this standard. Doc. 487 at 46. A 76% decline arguably meets the standard for “moderate” catastrophe, which is described as one that is “smaller” than a high-risk catastrophe, but “still [a] significant decline.” NMFS’s conclusion that the three subsequent years of drought caused a “high-risk” catastrophe is not supported by the record. It is at most a “moderate-risk” catastrophe.

Federal Defendants point out that in order for a population to be considered viable, it “must meet all the low-risk thresholds.” Doc. 477-1 (citing BiOp at 84). Whether the drought was a “high” or “moderate” risk catastrophe or whether the population should actually have been classified as “low-risk” based on population size, does not change the fact that the winter-run are “not viable,” because a classification of “moderate” is justified as to at least one criteria: catastrophe.

In addition, the BiOp found that winter-run are at a high-risk of extinction based on spatial structure. BiOp at 86-87. Although “spatial structure” was not one of Lindley (2007)’s primary criteria for population viability, it was considered. AR 00123481, 00123487. Lindley (2007) concluded that the winter-run “does not currently satisfy the representation and redundancy rule because it has only one population and that population spawns outside of the ecoregion where it evolved.” AR 00123487. To satisfy this rule at least two populations would need to be re-established. Id. Plaintiffs emphasize that this situation is “entirely attributable to baseline conditions (i.e., dams).” Doc. 487 at 48. Lindley acknowledges this, noting that establishment of additional winter-run populations “may require passage past Shasta and Keswick dams.” AR 00123487. But, this does not render spatial structure irrelevant to the BiOp’s analysis. “[A]n agency may not take action that will tip a *864 species from a state of precarious survival into a state of likely extinction. Likewise, even where baseline conditions already jeopardize a species, an agency may not take action that deepens the jeopardy by causing additional harm.” NWF v. NMFS II, 524 F.3d at 930 .

Is there a practical implication in the BiOp of NMFS’s unsupported description that the winter-run are at “high-risk,” as opposed to the less serious classification of “not viable”? Plaintiffs cite Steven Cram-er’s Reply Declaration to support their argument that the BiOp used the “high-risk” designation to “justify its failure to do a careful, scientific analysis of the RPA because immediate radical action supposedly is necessary.” Doc. 487 at 44 (citing Cramer Reply Deck, Doc. 487 at ¶¶ “20-17[sic]”). Cramer opines:

Id. at ¶ 20. Cramer’s accusations are troubling, but are not reflected in the record. Mr. Cramer cites pages 8-9 of Federal Defendants’ memorandum in support of their cross motion for summary judgment. Nowhere on those pages do Federal Defendants even mention the “high-risk” rating, let alone rely upon it to justify the RPAs in any way.

Plaintiffs have identified areas of NMFS’s analysis that are completely unsupported by the record, constituting “clear error.” The extent to which they undermine the viability determination is properly addressed on remand. This aspect of the BiOp must be remanded for correction.

3. Orea Analysis.

The BiOp concluded that the Southern Resident population was so diminished that “the loss of a single individual, or the decrease in reproductive capacity of a single individual, is likely to reduce the likelihood of survival and recovery of the DPS.” BiOp at 573. The BiOp also concluded that any reduction in the Southern Resident’s prey base may have adverse physiological effects on Southern Residents. Id.

Plaintiffs point to yet another separate biological opinion issued May 5, 2009, evaluating the effects of the Pacific Coast Salmon Plan, which governs management of commercial and recreational salmon fishing off the west coast of the United States, on the Southern Residents (“Orea Salmon Harvest BiOp”). See AR 00131721 — 802. According to Plaintiffs, the Orea Salmon Harvest BiOp “produced an extraordinarily detailed quantitative analysis of the effect of decreases in the adult [C]hinook population on Southern Residents ... that incorporated data on factors such as orea abundance, size, and kilocalorie requirements, which NMFS used to project the percent changes in prey availability at different locations in the oreas’ range, different times of the year, and different levels of quality in yearly [C]hinook salmon production.” Doc. 431 at 35. The Orea Salmon Harvest BiOp concluded that planned ocean harvest of salmon would not jeopardize the Southern Resident Killer Whales. AR 00131781.

*865 The crux of Plaintiffs’ complaint is that both the conclusions reached and the methodologies used in the 2009 Salmonid BiOp are inconsistent with those of the earlier-issued, more comprehensive and focused Orea Salmon Harvest BiOp. Plaintiffs argue:

Doc. 431 at 35-36.

Federal Defendants maintain that the two biological opinions are not inconsistent because they address impacts over different time frames and from different actions. The Orea Salmon Harvest BiOp describes short-term impacts to prey availability in specific months during high abundance Chinook years. The 2009 Salmonid BiOp considered impacts to Southern Residents caused by long-term increase in the risk of extinction for winter-run and spring-run Chinook, in addition to long-term impacts to fall-run. BiOp at 573. The Orea Salmon Harvest BiOp concluded the long-term impact of ocean harvest is not likely to appreciably reduce the survival and recovery of the listed Chinook and other salmon affected by harvest, in part because the fishery is managed to adjust harvest levels annually according to the actual salmon population available for harvest, thereby *866 avoiding harm to the species. AR 00131776-81. The 2009 Salmonid BiOp concluded that Project operations would increase the risk of extinction of winter-run and spring-run, which “increases the risk of a permanent reduction in prey available to Southern Residents, and increases the likelihood for local depletions of prey in particular locations and times.” Id. at 574.

Although these biological opinions facially consider different time frames and different actions, it is undeniable that they are temporally and factually interrelated. The Salmonid BiOp specifically concludes that Project operations will reduce the abundance of naturally produced CV fall run Chinook salmon, a source of prey to the Southern Residents. BiOp at 574. As a result, the Salmonid BiOp concludes “Southern Residents would likely experience nutritional, reproductive, or other health effects from reduced prey as a result of the proposed action.” Id. In contrast, the Orea Salmon Harvest BiOp concludes that, even in the long run, implementation of the Pacific Coast Salmon Plan will not have long-term deleterious effects on Chinook salmon. AR 00131776-77. It is true that the Pacific Coast Salmon Plan is designed to manage commercial and recreational salmon harvest to meet salmon recovery goals and requires conservation measures, including suspension of all harvest if necessary, when Chinook stocks are doing poorly. AR 00131777. This amounts to a “do no harm” approach to managing the fishery. However, under such a management approach, it is plausible that any impact to fall-run Chinook, and any related impact to orea, caused by Project operations could be automatically mitigated by reduced harvest in the ocean. How these two sets of human actions (Project operations and harvest restrictions) interplay, and how this interplay might impact the likelihood that Project operations would harm the Southern Residents, is not discussed in the Salmonid BiOp, which postdates the Orea Salmon Harvest BiOp, albeit by only one month. NMFS’s own findings in the Orea Salmon Harvest BiOp are certainly “relevant factors” NMFS should have taken into consideration before issuing the Salmonid BiOp. Federal Defendants, through counsel, provide a partial, but insufficient, post hoc explanation.

Plaintiffs’ motion for summary judgment that the Orea jeopardy analysis is unlawful is GRANTED; Federal Defendants’ and Defendant-Intervenors’ cross motions are DENIED. On remand, NMFS must explain how the findings of these two biological opinions can be reconciled.

4. Interior Delta Mortality as an Indirect Effect.

Plaintiffs assert that the BiOp unlawfully classifies mortality from predators, pollution, and other adverse conditions in the interior delta, as “indirect effects” caused by Project operations. Doc. 431 at 66-72.

a. Applicable Legal Standard.

The Joint Consultation Regulations promulgated by FWS and NMFS explain that “effects of the action” refers to “the direct and indirect effects of an action on the species or critical habitat... that will be added to the environmental baseline.... ” 50 C.F.R. § 402.02 . “Indirect effects are those that are caused by the proposed action and are later in time, but still are reasonably certain to occur." Id. (emphasis added). The ESA’s definition differs from NEPA’s definition of indirect effects of an action: “[ijndirect effects, which are caused by the action and are later in time or farther removed in distance, but are still reasonably foreseeable.” 40 C.F.R. § 1508.8 (b) (emphasis added). In the preamble of the Final Rule adopting the ESA regulations, NMFS and FWS explained that it intended a narrower regula *867 tory definition of indirect effects under the ESA than applied in the NEPA context (i.e., compare “reasonably certain to occur” with “reasonably foreseeable”). 51 Fed. Reg. 19,926 (June 3, 1986). NMFS and FWS distinguished the ESA from NEPA and expressly explained the intent and rationale for adopting the more narrow “reasonably certain to occur” standard for indirect and cumulative effects under the ESA:

51 Fed.Reg. at 19,933.

Shortly after adoption of the ESA regulations, the Ninth Circuit confirmed “ ‘[t]he reasonably certain to occur’ standard applies to ‘indirect effects ... caused by the proposed action.’ ” Sierra Club v. Marsh, 816 F.2d 1376, 1388 (9th Cir.1987); Ctr. for Biological Diversity v. U.S. Dept. of Hous. & Urban Dev., 541 F.Supp.2d 1091, 1100-01 (D.Ariz.2008) (dismissing a suit alleging federal agencies had violated the ESA by failing to analyze the indirect effects of providing federal funding to local development projects, concluding that the link between such financial assistance and groundwater depletion that could harm listed species was “too attenuated” to meet the standards of 50 C.F.R. § 402.02 ).

The December 14, 2010 summary judgment Decision in the Consolidated Delta Smelt Cases found that the “reasonably certain to occur” standard controlled the asserted causes of indirect mortality to the smelt in the interior Delta. San Luis v. Salazar, 760 F.Supp.2d at 946-47 . Here, NMFS resists such a finding, arguing that Plaintiffs (and by implication the December 14, 2010 MSJ Decision in the Consolidated Delta Smelt Cases) confuse the BiOp’s discussion of “indirect mortality” with the regulatory term “indirect effect.” Doc. 477-1 at 55. Federal Defendants argue that the “reasonably certain to occur” standard does not refer to the certainty of the effect on the species, but rather to the certainty of whether a future activity (i.e. the activity that may have an effect on the species) will occur. The federal register notice promulgating the relevant regulations explains that NMFS considers “effects to listed species from such future activities that are reasonably certain to occur under the analysis of ‘indirect effects.’” 51 Fed.Reg. 19,926, 19,932 (June 3, 1986)(emphasis added). Indirect effects are further defined as “those that are caused by the action and are later in time but are still reasonably certain to occur.” Id. (emphasis added). Federal Defendants point out that the kinds of “indirect mortality” discussed in the BiOp are not “future activities.” Rather, they are a category of effects that are purportedly occurring all the time.

Plaintiffs rejoin by citing a single sentence from the Final ESA Section 7 Consultation Handbook, jointly prepared by FWS and NMFS, which explains that “[ijndirect effects may include other Federal actions that have not undergone section 7 consultation but will result from the action under consideration.” AR 00217743 (“Consultation Handbook”) at 4-29 (emphasis added). 18 Plaintiffs argue that the *868 use of the word “include” suggests “that NMFS considers effects from future activities to be only a subset of possible indirect effects, and that indirect effects are not limited to future activities.” Doc. 487 at 57. Plaintiffs do not mention the very next sentence of the Consultation Handbook. The entire paragraph reads:

Id. (emphasis added). Here, the indirect mortality findings challenged by Plaintiffs do not constitute “indirect effects.” The indirect mortality discussed in the BiOp is caused by the action subject to consultation, not by some other action that is the subject of work plans, permits, or budgeting. The emphasized language specifies actions “reasonably certain to occur,” not those that have occurred. This suggests but does not explicitly reference actions other than the action under consultation.

The “reasonably certain to occur” standard does not apply to the indirect mortality analysis in the BiOp. 19 However, this does not immunize the indirect mortality findings from review. “Jeopardize” means to “engage in an action that reasonably would be expected, directly or indirectly, to reduce appreciably the likelihood of both the survival and recovery of a listed species in the wild but reducing the reproduction, numbers and distribution of that species.” 50 C.F.R. § 402.02 . The BiOp finds project operations cause indirect mortality. Whether such findings are reasonable must be addressed. 20

a. Does the Record Support a Finding that Project Operations Can Reasonably Be Expected to Cause More Salmonids to Enter the Interior Delta?

Plaintiffs concede that the mortality rate of migrating salmonids is generally higher for fish traveling through the interior Delta than for fish that remain in the main-stem Sacramento River. Doc. 431 at 68. *869 Plaintiffs argue, however, that the record does not support the BiOp’s conclusion that project operations cause more salmon-ids to take the more dangerous routes through the interior Delta. Id. at 67-70 . Plaintiffs fault the BiOp for not providing any “analysis or articulation whatsoever of [what] additional fraction of emigrating salmonids — above the baseline number that will enter the Delta irrespective of the projects — will be induced to enter the interior Delta solely as a result of proposed project operations.” Doc. 431 at 68. Plaintiffs incorporate by reference the arguments made by DWR in its challenge to Action IV.2.1. Id. at 69-70 . As Plaintiffs’ challenge turns on the merits of DWR’s challenge, which is thoroughly discussed below in the context of Action IV.2.1, there is no need to separately discuss them here,

b. Does the Record Support a Finding that Project Operations Can Reasonably Be Expected to Cause Indirect Mortality from Exotic Species, Pollution, and/or Food Limitations in the Interior Delta?

Plaintiffs also contend that there is no record evidence to support the BiOp’s implied conclusion that project operations cause indirect mortality from exotic species, pollution, and other adverse environmental conditions in the interior Delta. Doc. 431 at 70-71.

(1) Exotic Species.

Among other things, the BiOp concludes that Project operations create conditions that favor exotic over native species:

BiOp at 382 (emphasis added). Elsewhere, the BiOp concludes:

Id. at 433 .

Plaintiffs do not directly contest the conclusion that the altered hydrologic conditions are favorable for invasive species. Nor do Plaintiffs challenge the BiOp’s conclusion that CVP and SWP operations contribute to this ecosystem alteration. Rather, they argue that the operators of the CVP and SWP did not release the exotic predators or introduce the exotic weeds, nor can the operators of the projects control these alien species. Doc. 431 at 71.

This is not disputed. The BiOp does not assert, as it cannot, that the Projects were the oñginál cause of these problems. The BiOp concludes that the hydrologic conditions created by the projects favor the continued presence of these exotics and that proposed project operations are likely to make this situation worse. See BiOp at 382 (“Continued operations of the CVP and SWP are unlikely to benefit the health of the Delta, and increases of the facility operations are likely to degrade the system beyond their current conditions, rather than return the Delta to a more natural condition, with more functional hydraulics conducive to a healthy ecosystem.”). The BiOp cites recent studies, including Brown and Michniuk (2007), see BiOp at 382, to support its conclusion that this “new” Delta hydrology favors exotic species over native ones. Plaintiffs do not challenge the BiOp’s reliance on these studies.

However, assuming the BiOp properly found a Project-exotics connection, NMFS failed to adequately consider this factor in its jeopardy analysis. What effect do these exotics have on the Listed Species? To what extent does the contribution of the Projects to the continued presence of these exotics contribute to the jeopardy finding? Could altered project operations reduce the presence of exotics? NMFS’s logic taken to the extreme means the Projects cannot operate, as no analysis has been done to evaluate the impact on the Listed Species from this indirect effect at varying pumping levels. It may be that there is insufficient information to answer these questions, but this is pure speculation, as the sufficiency of information is not discussed. This is another example of the need for a realistic analysis of relative effect from Project operations on conditions that are not related to pumping.

The BiOp’s analysis of the influence of Project operations on the continued presence of exotic species, and how this relates to indirect mortality to the Listed Species, must be explained. Plaintiffs’ motion for summary judgment on this issue is GRANTED. Federal Defendants’ and Defendant-Intervenors’ cross motions are DENIED.

(2) Pollution and Food Limitation.

Plaintiffs also argue that the BiOp unlawfully “blames the project for pollution and food limitation by labeling them effects of the action.” Doc. 431 at 72. This is the logical inference drawn from the focus on predators and contaminants, which are mentioned throughout the BiOp. The agency does not explain how the projects influence contaminants or cause food limitations. Plaintiffs point to a statement in the biological assessment that “there is no direct evidence of food limitation for salmon in the delta or lower estuary,” AR 00143672. It is not clear that the BiOp actually asserts that there is a food limitation in the lower estuary. This imprecision contributes to the inadequacy of the BiOp. There is no way to understand the BiOp’s attribution of adverse indirect effects to the Projects.

Plaintiffs’ motion for summary judgment that the record does not support the BiOp’s conclusions about the connection *871 between Project operations and pollution and food limitation, causing indirect mortality to the Listed Species is GRANTED. Federal Defendants’ and Defendant-Intervenors’ cross motions are DENIED.

D. Critical Habitat Analysis.

1. There Is No Requirement that NMFS Identify a Numerical Threshold for Adverse Modification.

Destruction or adverse modification is defined by regulation to mean “a direct or indirect alteration that appreciably diminishes the value of critical habitat for both the survival and recovery of a listed species.” 50 C.F.R. § 402.02 . Previous rulings in related cases have held “that NMFS and FWS have interpreted the term ‘appreciably diminish’ to mean ‘considerably reduce.’ ” Findings of Fact and Conclusions of Law Re the Existence of Irreparable Harm, PCFFA v. Gutierrez, 1:06-cv-245 OWW GSA, Doc. 367 at 24:6-9 (citing Consultation Handbook at 4-34).

Plaintiffs demand that NMFS set a threshold for adverse modification and directly analyze whether the action “appreciably diminishes” the capability of habitat to support survival or recovery vis-a-vis this threshold. Id. at 75 . This demand was rejected in the December 14, 2010 MSJ Decision in the Consolidated Delta Smelt Cases:

San Luis v. Salazar, 760 F.Supp.2d at 945 . NMFS is not required to set a numeric threshold for adverse modification.

2. Significance of Impacts to Critical Habitat.

Plaintiffs argue that the adverse modification findings are unlawful because the BiOp explicitly declines to apply the regulatory definition of adverse modification found in 50 C.F.R. 402.02. Doc. 431 at 75. The BiOp states:

BiOp at 43. Plaintiffs maintain that this reads the word “appreciably” out of the regulatory definition of adverse modification. Doc. 431 at 75-76. The record provides a reasoned basis for this statement in the BiOp and demonstrates that NMFS has not read the term “appreciably diminish” out of the definition.

In 2005, after Gifford Pinchot invalidated FWS’s application of the regulatory definition in 50 C.F.R. § 402.02 because FWS had not evaluated whether the amount of habitat anticipated to be lost would impact recovery, NMFS issued a guidance memo on how to conduct “destruction or adverse modification” determination. See AR 00005204-209. That memo explicitly directs NMFS to identify the current condition of the Primary Constituent Elements (“PCE”) 21 of each critical habitat designation before examining how the proposed action will affect the function and conservation role of each PCE. Id. Federal Defendants do not assert that this guidance has invalidated the “appreciably diminishes” aspect of the critical habitat regulation. Doc. 515 at 29. Rather, the guidance memo, which instructs NMFS to “discuss the significance of anticipated effects to critical habitat,” is sufficient to implement an “appreciably diminish” standard. AR 00005208 22 (emphasis added). The guidance memo’s requirement of “significant” impacts to critical habitat is consistent with the regulatory definition of adverse modification to include only those alterations that “appreciably diminish[ ] the value of critical habitat.”

Because an agency’s rationale must be upheld if it can “reasonably be discerned,” see Modesto Irr. Dist. v. Gutierrez, 619 F.3d 1024, 1035 (9th Cir.2010), there is no requirement that the agency use “magic words” in its analysis. The key question is whether the record supports the adverse modification findings in the BiOp. In other words, does the record demonstrate that Project operations will have a significant *873 (i.e., appreciable or considerable) impact on the critical habitat of each of the listed species for which adverse modification was found.

The BiOp examines impacts to critical habitat at length. For each species, the BiOp describes the PCEs of that species’ critical habitat, examines the current status of the critical habitat and describes factors responsible for the current status, evaluates the impacts of current and future non-project (i.e., baseline) impacts, and describes the anticipated impacts of proposed project operations on that habitat.

a. Winter-Run Habitat Analysis.

The evaluation of winter-run critical habitat provides a representative example. There are seven PCEs of Chinook critical habitat: (1) access from the Pacific Ocean to appropriate spawning areas in the Upper Sacramento River; (2) clean gravel for spawning; (3) adequate river flows for spawning, egg incubation, fry emergency, and juvenile downstream migration; (4) appropriate water temperatures for spawning, egg incubation, and fry development; (5) uncontaminated habitat and food sources; (6) riparian habitat for juvenile development and survival; and (7) downstream migration access to the Pacific Ocean. BiOp at 90. The BiOp evaluates the current status of each of these PCEs. Id. at 90-92. In addition, the BiOp contains a lengthy section describing the factors responsible for the current status of the species, many of which also affect the species’ habitat. See id. at 134-142. The BiOp concludes that the current condition of critical habitat is degraded and has low value for the conservation of the species. Id. at 93.

Id. (emphasis added).

In the environmental baseline analysis, NMFS concluded climate change will negatively affect all of the Central Valley critical habitat designations at issue. Id. at *874 173. With respect to upstream habitat, NMFS evaluated the current and future environmental baseline of winter-run Chinook critical habitat in the Shasta and Sacramento Divisions of the CVP, and concluded that the current baseline is “degraded, and has low value for the conservation of the species,” and future baseline habitat impacts will “affect the fitness... of the critical habitat....” Id. at 181-83, 187-91. For the Delta Division, NMFS concluded that the migratory function of this critical habitat is degraded, id. at 203-OS, and that the future environmental baseline included continued “ongoing habitat modifications” and adverse habitat impacts from levees, predation, non-native species, contaminants, entrainment, dredging, recreational boating, and temporary irrigation barriers, id. at 215-16.

(1) Project Impacts to Winter-Run Spawning Habitat.

In addition to these past, current, and future non-project adversities, NMFS found that proposed project operations in the Sacramento River constrain spawning habitat by providing relatively less cool water temperatures below Keswick Dam and by stranding or dewatering redds and juveniles. See id. at 273. The BiOp’s section on the “Effects of the Action on Critical Habitat in the Sacramento River” in particular on “Spawning Habitat” provides:

Id. at 273. Spawning habitat has been impacted by baseline conditions (such as the presence of Shasta and Keswick Dams) and climate change. The BiOp provides explanation for its conclusion that additional Project operations will add to those baseline impacts. As to winter-run spawning habitat, the section references an earlier discussion of “the impacts of operations on cold water;” addressing CALSIM II modeling runs, comparing temperature conditions (and resulting egg mortality) between baseline operations and operations under the proposed action. Figure 6-14, which depicts winter-run egg mortality by water year type, permits comparison of the baseline (Study 7.0), near future project operations (Study 7.1) and future project operations (Study 8.0).

[[Image here]]

Id. at 259. These results show that in critical years, which are 15 percent (15%) of the years modeled, egg mortality more than doubles between Study 7.0 and Study 7.1, and increases by 50 percent between Study 7.0 and Study 8.0, under past and future operations. (No explanation is given for why study 7.1 shows higher mortality than Study 8.0.) Because egg mortality is a direct result of temperature conditions in winter-run spawning habitat, this demonstrates that Project operations will significantly reduce spawning habitat in critical years.

(2) Project Impacts to Rearing and Migratory Habitat.

Information to support NMFS’s finding of significance for winter-run rearing and migratory habitat is less apparent. In Section 6.3.8 (“Effects of the Action on Critical Habitat in the Sacramento River”), the BiOp reviews impacts to rearing and migratory habitat very generically:

BiOp at 273-74.

The BiOp’s “Synthesis of Effects” provides the following additional discussion of rearing habitat:

Id. at 469-70. The next sub-section assesses risk to winter-run critical habitat.

Id. at 470.

All of these discussions of impacts to rearing and migratory habitat on the Sacramento River focus on the operation of Red Bluff Diversion Dam (“RBDD”) and Anderson Cottonwood Irrigation District (“ACID”) diversion dam, which obstruct passage and alter large areas of habitat. For example:

Id. at 469. Although the BiOp does not offer a numerical analysis of what percentage of the designated rearing and/or migratory habitat is disturbed by these operations, at least for those fish that must pass these structures (the entire winter and spring-run populations) the significance of such barriers is obvious.

Similar evidence of significant impacts for other aspects of critical habitat exists for each of the species. E.g., id. at 260 (demonstrating significant impacts to the spring-run spawning habitat); id. at 501-503, 504 (summarizing project impacts to spring-run habitat), id. at 549-53 (same as to steelhead); id. at 570-71 (same as to green sturgeon proposed critical habitat, noting that “[w]hen the gates are down, RBDD precludes access to 53 miles of spawning habitat for 35-40 percent of the spawning population of green sturgeon”).

Plaintiffs’ argument is simply that Federal Defendants acted unlawfully by failing to directly articulate that project operations have “appreciable” or “significant” impacts on critical habitat. The test is that the agency’s reasoning should reasonably be discerned from the BiOp. A num *880 ber of evident causes are identified, which adversely impact the Listed Species. NMFS provided no quantification other than year-to-year population fluctuations. Data for CV Steelhead and green sturgeon are sparse. The record reflects a number of adverse modifications of the species’ critical habitat. Although the BiOp does not show what proportion of the population will be affected, this is not required. The explanation of the adverse effects on habitat and how these changes have the ability to effect harm to the species is sufficient.

Plaintiffs’ motion for summary judgment that the critical habitat analysis is unlawful because NMFS did not apply the proper standard for adverse modification is DENIED; Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED.

E. Use of Surrogates.

In the effects analysis, the BiOp utilized fall-run Chinook salmon as a surrogate for steelhead, and hatchery Chinook salmon as a surrogate for wild Chinook salmon. Plaintiffs argue that NMFS violated the best available science standard by failing to “validate” its use of surrogate species. Doc. 431 at 81.

Plaintiffs’ expert Kenneth Cummins opines that there is a consensus in the scientific community that, whenever possible the use of surrogates should be avoided. Cummins Decl., Doc. 445 at ¶ 8. Surrogates should be a “tool of last resort.” Id. This is undisputed.

Dr. Cummins further opines that “for a surrogate to be appropriate, it should share the same key ecological or behavioral traits, that make the target ... sensitive to environmental disturbance and the relationship between population vital rates (for example, survival) and level of disturbance should match that of the target.” Id. at ¶ 11 (citing .Caro et al. (2005)). Dr. Cummins maintains that because “all species are different to some degree in regards to their life history strategies, ecological relationships with other species, and selection and use of habitat, substituting data from one species to draw inferences about another for purposes of conservation planning without validating that decision a priori is not justified.” Id. at ¶ 14. He continues: “since no two co-occurring species are biologically identical, that would seem to rule out management planning for one species that is informed using biological information that is available for another unless use of a surrogate species for the target species is validated.” Id. Dr. Cummins cites a study by Favreau, et al. (2006), which found that “in less than 2 percent of the cases examined did a surrogate represent the target species better than a random selection of potential surrogates. Further, in less than 4 percent of the cases could the surrogate be considered as effective in representing the target species.” Id. From this, Dr. Cummins concludes:

Id. Dr. Cummins describes “various approaches to validation that scientists may employ before relying on surrogate data.”

Id. at ¶ 15. The problem with Plaintiffs validation argument, and Dr. Cummins’ related opinions, is that they require that NMFS conduct new experiments to justify reliance on existing experimental data. For this reason, those portions of Dr. Cummins’ declarations that opine NMFS should have conducted validation experiments were stricken from the record. See Doc. 536 at ¶ 15; see also S.W. Ctr. for Biological Diversity v. Babbitt, 215 F.3d 58, 60 (D.C.Cir.2000) (best available science standard does not impose an obligation to conduct independent studies). The record does not support Plaintiffs’ validation requirement argument. 23

To the extent Plaintiffs advance a more generic challenge to NMFS’s use of surrogates, NMFS explained its use of surrogates and addressed the limitations of surrogate data:

BiOp at 62. NMFS maintains that the use of surrogates “minimizes the amount and extent of take associated with tagging or capturing listed species to monitor take.” Id. at 62-63 . Appendix 3 of the BiOp contains a comparison of delta survival rates between hatchery and wild Chinook. BiOp App. 3, at 10-11.

One of the draft BiOp peer reviewers considering the BiOp’s analyses of winter- and spring-run Chinook noted: “where information was lacking, reasonable surrogates are used.” AR 00061498. Plaintiffs’ own experts, e.g., Mr. Cramer, Dr. Hanson, and Mr. Cavallo, used data from experiments utilizing surrogates without independently validating the surrogates. See Second Yip Decl., Doc. 481 at ¶ 33; Third Yip Decl., Doc. 518 at ¶ 16. It is undisputed that in many circumstances unverified surrogate data was the only data available for use by NMFS to evaluate the impact of project operations on the Listed Species. Eliminating the surrogate data would have “considerably limit[ed] the util *882 ity of any biological analyses undertaken” in the BiOp. Id. at ¶ 14.

Plaintiffs’ motion for summary judgment that the BiOp’s use of surrogates violated the ESA’s best available science standard is DENIED; Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED. The agency recognized there were shortcomings in using surrogates. This is a dispute among scientists.

F. Challenges to the Reasonable and Prudent Alternative.

1. RPA Action IV.2.1.

Action IV.2.1 limits export pumping from April 1 through May 31 and has two requirements. First, the Action requires a minimum flow, as measured at Vernalis, based on an index of storage at New Mel-ones Reservoir (“New Melones Index”). BiOp at 642; BiOp App. 5 at 71. The Vernalis flow requirement is not challenged.

The second requirement of Action IV.2.1 restricts combined CVP and SWP export pumping based on the flows at Vernalis, with the permissible level of exports rising in relation to increased flows at Vernalis. BiOp at 642; BiOp App. at 71-72. The action is phased. Phase I governs operations during 2010 and 2011, when combined CVP and SWP exports were restricted as follows:_

BiOp at 642. Under Phase I, the baseline export rate is set at 1,500 cfs, deemed an operational minimum required to address health and human safety needs. Id. at 74. Flood warning stage at Vernalis is 21,750 cfs. Id. at 71 n. 2.

During Phase II, which operates from 2012 on, combined exports are governed by the following table from April 1 through May 31:

Id. at 643-14. Action IV.2.1 includes an exception for multiple dry years and a health and safety exception. Id. at 644.

Action IV.2.1 is designed primarily to “reduce the vulnerability of emigrating CV steelhead within the lower San Joaquin River to entrainment into the channels of the South Delta and at the pumps caused by the diversion of water by the export facilities in the South Delta, by increasing the inflow to export ratio.” BiOp at 641. A secondary purpose of Action IV.2.1 is to more generally “enhance the likelihood of salmonids successfully exiting the Delta at Chipps Island by creating more suitable hydraulic conditions in the main stem of the San Joaquin River for emigrating fish, including greater net downstream flows.” Id.

Both the Export Plaintiffs and DWR have twice previously sought injunctive relief against the imposition of Action IV.2.1. On May 18, 2010, Action IV.2.1 was addressed in Findings of Fact and Conclusions of Law, granting in part and denying in part Plaintiffs’ motion for preliminary injunction:

Doc. 347 ¶¶ 99-102 (internal paragraph numbers omitted from quotation to avoid confusion). The injunction decision found likely success on the merits, but requested additional information on the status of the species before ordering injunctive relief:

Id. at 133-34.

After receiving additional evidence about the status of the species, Action IV.2.1 was enjoined for a limited period of time (from May 26 through May 31, 2010), in part because only a small percentage of the population of concern, the SSNDG of CV Steelhead, remained in the area that *884 would be impacted by the injunction. Doc. 380.

A second motion for injunctive relief was filed against Action IV.2.1 in February 2011, Doc. 538, and then withdrawn in light of wet hydrologic conditions that obviated the need to implement the challenged aspects of the Action in this water year. Doc. 625, filed March 30, 2011.

Export Plaintiffs and DWR again challenge the scientific basis for Action IV.2.1. Export Plaintiffs’ and DWR’s briefs on the issue substantially overlap,

a. Does the Record Support NMFS’s Imposition of an Flow.Export Ratio Requirement?

(1) Studies Cited by DWR.

DWR’s principle argument is that the last twenty years of San Joaquin River fisheries studies have not produced any statistically significant evidence of a negative relationship between salmonid survival and project pumping. Doc. 446-1 at 11. DWR’s expert, Bradley Cavallo, refers to various statistical analyses of San Joaquin River salmonid experiments that reveal either no statistically significant relationship, or a positive one. His citation to a study by Kjelson, Loudermilk, Hood, and Brandes, “The Influence of San Joaquin River Inflow, Central Valley and State Water Project Exports and Migration Route on Fall-Run Chinook Smolt Survival in the Southern Delta During the Spring of 1989,” published in 1990, is representative of these critiques. Kjelson, et al. (1990) concluded:

AR 00122358-59 (cited in Cavallo Decl., Doc. 452 at ¶ 8a).

Mr. Stuart, the lead author of the Delta section of the BiOp, asserts that Mr. Cavallo has selectively quoted from the relevant studies. For example, as to the Kjelson, et al. study:

Fourth Stuart Decl., Doc. 485 at ¶ 13. The BiOp specifically discussed Kjelson, et al. (1990)’s conclusions in Appendix 5:

BiOp App. 5 at 5-6.

DWR correctly rejoins that Mr. Stuart does not contest that Kjelson, et al. (1990) concluded that survival Was lower during low exports than high exports. DWR is also correct that Mr. Stuart does not explain how the study “affirmatively supports the United States’ claim that a relationship exists between project exports and smolt survival sufficient to justify the Inflow/Export ratio.” Doc. 495 at 16. Mr. Stuart never opined that Kjelson’s study provides such affirmative support. The BiOp considered the study, its caveats, and acknowledged the study’s “surprising” conclusion that survival was not higher during low export conditions.

Mr. Cavallo quotes from five more studies, Cavallo Decl., Doc. 452 at ¶ 8:

Mr. Stuart now submits alternative explanations to support his opinion why each of these studies does not definitively rule out a relationship between exports and survival:

The best that can be said from all these studies is that they do not affirmatively support the purported relationship between exports and survival NMFS uses to justify Action IV.2.1’s flow:export ratio. However, without more, DWR has not established that these studies were not properly evaluated. NMFS relies on additional record evidence to support imposition of Action IV.2.1’s flow:export ratio limitation.

(2) Studies Cited by NMFS in Support of a Flow:Export Ratio.

(a) VAMP Data.

The BiOp concedes that analyses of the evidence gathered during the Vernalis Adaptive Management Program (“VAMP”) 24 are equivocal regarding the impact of exports on survival. BiOp at 373. The BiOp also recognizes that the VAMP experiments may have resulted “in weak to negligible” associations because of the “correlation between flow and export rates during VAMP.” Id. Mr. Stuart explains the VAMP experimental design was not implemented in full, in that not all of the planned relationships have been tested, with overrepresentation at certain combinations of flow and exports. Fifth Stuart Decl., Doc. 516 at ¶ 6. Mr. Stuart opined: “Newman (2008) concluded that the testing of the extremes of combinations is necessary to increase the precision of the experiments and allow discrimination of differences between the parameters.” Id. at ¶ 50. The 2010 PI Decision found that the BiOp considered the VAMP evidence and its limitations and did not disregard any important conclusions generated from the VAMP data. Consol. Salmonid Cases, 713 F.Supp.2d at 1132-34 .

Notwithstanding the lack of statistical significance, the BiOp relied on the following Figure copied from the 2006 VAMP Technical Report to demonstrate that, during times when the Head of Old River Barrier (“HORB”) was in place, as the ratio between Vernalis flow and exports increased, survival increased. BiOp App. 5 at 20.

*888 [[Image here]]

BiOp App. 5 at 20. The relationship was not statistically significant, but the BiOp states that this may have been due to the narrow range of export rates tested. Id. The 2010 PI Decision found NMFS’s reliance on this data was not arbitrary. Consol. Salmonid Cases, 713 F.Supp.2d at 1133-34 .

(b) Escapement Data.

NMFS includes the following chart from the 2006 VAMP annual report that showed a positive relationship between the spring Vernalis flow/export ratio and adult escapement (i.e. return from the ocean to freshwater) two and a half years later, based on data from 1951 through 2003. BiOp App. 5 at 21.

[[Image here]]

*889 The 2010 PI Decision found it not unreasonable for NMFS to consider the analysis depicted in Figure 11. Consol. Salmonid Cases, 713 F.Supp.2d at 1134 .

DWR argues that NMFS’s reliance on Figures 10 and 11 to support the conclusion that there is a correlation between exports and survival is unfounded. Doc. 446-1 at 14. DWR’s expert, Bradley Cavallo, compares Figures 10 and 11 to Figures 8 and 9, which plot the impact of San Joaquin River (“SJR”) flow against exports:

[[Image here]]

[[Image here]]

BiOp App. 5 at 19. Mr. Cavallo opines that “inspection of the degree of scatter in [Figures 10 and 11] relative to [Figures 8 and 9] suggests SJR inflow to export ratio *890 provides a poorer fit to observed data than does SJR inflow alone.” Cavallo Decl., Doc. 452 at ¶ 12. For example, “the model describing smolt survival in relation to SJR flows alone (Exhibit 1, bottom) has an r2 value of 0.73 while the comparable model with the ratio of SJR flows to exports has an r2 value of only 0.26 (Exhibit 2, bottom). [¶] An r2 value closer to 1 signifies that salmon survival is better explained by SJR flows (r2 = 0.73) than by the ratio of SJR flows to exports (r2 = 0.26).” Id. at ¶¶ 12-13. Although SJR flows better explain salmon survival than the ratio of SJR flows to exports, Mr. Cavallo does not opine that there is no relationship between salmon survival and the ratio of SJR flows to exports depicted in Figures 10 and 11. Although NMFS overstates and over-relies on questioned data, this is a scientific dispute among experts that does not involve error of the magnitude that rises to unlawfulness.

(c) CDFG (2005).

DWR also criticizes NMFS’s treatment of a 2005 Department of Fish and Game (“CDFG”) study, which provides a “description of the process [CDFG] used to develop, and apply, its Model in the formulation of spring Vernalis flow objectives that were submitted to the [State Water Resources Control Board].” AR 00212414. Mr. Stuart opined that the report “clearly shows that while flows are the primary driver, exports play a role, albeit less than that attributable to flows.” Fourth Stuart Decl., Doc. 485 at 14. DWR responds that NMFS “misrepresents the [report’s] flow and export conclusions.” DWR focuses on several statements from the report, including the conclusion that “Delta export level, relative to Delta inflow level, does not influence juvenile salmon survival on a regular, normal, or repetitive pattern.” AR 000212423. CDFG determined that non-flow parameters, such as exports, ocean conditions, “have little, or no, relationship to fall-run Chinook salmon population abundance in the SJR and that spring flow magnitude, duration, and frequency all had significant influence upon SJR fall-run Chinook salmon abundance in the SJR.” AR 00212413. CDFG excluded consideration of project exports, ocean conditions, and/or density dependence from its model because of “the lack of substantial cause and effect relationships” between these non-flow factors and abundance. AR 00212426.

DWR’s contextual approach requires examination of the entire section:

Delta Exports

*892 AR 00212423-24 (footnotes omitted)(emphasis added). 25 Although the CDFG report supports Mr. Cavallo’s assessment that flows are the primary driver of salmon abundance in the SJR, the report acknowledges that as the exportrflow ratio “decreases” (i.e., as the flow:export ratio increases) escapement and cohort production increases. This supports NMFS’s use of a flowrexport ratio. NMFS’s minimization of the CDFG study was scientifically undesirable, but the law does not prevent it, by extending discretion to be mistaken. A candid appraisal of the true effect of flows, without masking the lack of significance in a flow:export ratio, would be welcome.

(d) Delta Action 8 Studies.

The BiOp considered data from the so-called “Delta Action 8 studies,” which compared the relative survival rates of coded-wire tagged salmon released at (a) Ryde on the Sacramento River and (b) Georgiana Slough, a channel that splits off of the Sacramento River at Walnut Grove and leads to the interior Delta, joining the South Fork of the Mokelumne River just before it meets the San Joaquin River. The 2010 PI Decision discussed NMFS’s treatment and critiques of these studies in detail:

Evaluating the data from the Delta Action 8 studies, Newman (2008) first explained that there was a high level of environmental variation in the data. [3/30/10 Tr.] at 78:18-23. Dr. Newman performed further analysis to reduce the amount of environmental variation and subsequently found a 98% probability that a negative relationship between exports and survival is present. Id. at 79:5-7. Mr. Stuart stated the significance of Newman’s finding is that as exports increased, survival decreases for those salmonid smolts that are moving down into the San Joaquin River, where they would be exposed to the influences of the export pumps. 4/2/10 Tr. 32:8-34:12. For those fish released into Georgiana Slough, survival was better when exports were lower.

This study is relevant to assessing the impacts of export pumping on fish migrating through the San Joaquin River, because fish released into Georgiana Slough must exit into the San Joaquin River, where they are subject to the influence of the pumps. 3/31/10 Tr. 76:20-23. The Georgiana Slough fish share a common migratory pathway with fish that exit the San Joaquin River basin. Id at 76:21-77:6. Regardless *893 of their origin, once the fish are in this common migratory pathivay, they are subject to the same hydraulic conditions. Id. at 78:1-17.

Mr. Cavallo stated that his interpretation of the Newman (2008) study is that there is a weak relationship between exports and survival in the interior Delta, but conceded that there was some relationship. 4/1/10 Tr. 98:24-99:4. Mr. Stuart testified that Newman’s studies are the best available and the fact that Newman could find a relationship given the considerable amount of “environmental noise” and the very low signal to noise ratio “shows that the relationship is probably very real.” Id. at 159:6-10. Whether this opinion is entitled to weight is disputed by Plaintiffs.

A September 26, 2008 paper prepared by Dr. Newman with Patricia L. Brandes entitled “Hierarchical Modeling of Juvenile Chinook Salmon Survival as A Function of Sacramento-San Joaquin Delta Water Exports” (“Newman and Brandes 2008”) examined the Delta Action 8 data concerning the relative survival rates for Ryde and Georgiana Slough releases and declared: what “we cannot conclude is that exports are the cazise of this lower relative survival.” 4/1/10 Tr. 67:20-23 (emphasis added); DWR Ex. 507 at 22. Newman and Brandes 2008 reached this conclusion because “the evidence for an association between exports and survival is somewhat weak” and because of the study’s inability to randomize export levels within a given outmigration season. 4/1/10 Tr. 68:1-12; DWR Ex. 507 at 22-23. A later version of this study, dated 2009, omitted this language from the conclusion. 4/2/10 Tr. 28:2-13.[FN 6]

Consol. Salmonid Cases, 713 F.Supp.2d at 1134-35 (emphasis added).

DWR again challenges NMFS’s reliance on Newman’s review of the Delta Action 8 studies to demonstrate that increasing exports will negatively affect salmonids migrating to the ocean through the San Joaquin River and its tributaries. Mr. Cavallo opines that it is inappropriate to rely on the Delta Action 8 studies to reach conclusions about San Joaquin basin salmonids. Cavallo Deck, Doc. 452 at ¶¶ 19-25. He also opines that the overall effect even on migrating salmon smolts in the Sacramento River is relatively small. Id. at ¶23. NMFS acknowledged the limitations of these studies in the BiOp, yet relies on them to support Action IV.2.1.

The Delta Action 8 studies marginally support Action IV.2.1. Newman’s analysis of the Delta Action 8 studies revealed that for those fish passing through Georgiana slough and the interior Delta, survival was negatively impacted by exports. Those fish share a common migratory pathway with all of the fish exiting the San Joaquin Basin. Fourth Stuart Deck, Doc. 485 at ¶¶ 28-29. It continues to be marginal logic to apply the admittedly weak correlative results of Newman’s analysis to San Joaquin salmonids. Mr. Cavallo’s criticism that Georgiana Slough, which is not a tidally influenced watercourse and which nev *894 er experiences reverse flows, is distinct from the SJR, which is tidally influenced and regularly experiences reverse flows, Cavallo Reply Decl., Doc. 497 at ¶ 55, does not abrogate the use of Delta 8 studies. Once fish exit Georgiana Slough, they must travel into the Mokelumne River system and the lower SJR, where they are influenced by tidal movements and exports. Fourth Stuart Decl., Doc. 485 at ¶28. Mr. Cavallo’s criticisms represent another dispute among experts, to which the agency is due deference, even recognizing that DWR and Mr. Cavallo have no apparent incentive to reach objective opinions contrary to NMFS, while Mr. Stuart makes every call in favor of the species, no matter how questionable the basis.

(3) Treatment of Data Related to the Use of Bubble Curtains at HORB.

Export Plaintiffs argue that NMFS ignored record evidence demonstrating the effectiveness of non-physical barriers, such as bubble curtains, which use sound, lights, and air bubbles, to guide fish. Doe. 431 at 91. A May 18, 2009 transmittal of “preliminary data” from Reclamation Biologist Dr. Mark Bowen showed that a bubble curtain kept a substantial percentage of fish in the San Joaquin River, rather than allowing them to move into the Old River toward the pumps. AR 00093348. Export Plaintiffs’ contention that NMFS did not consider this data in developing Action IV.2.1 is incorrect. NMFS considered Dr. Bowen’s preliminary data, Fourth Stuart Decl., Doc. 485 at ¶ 91, and recognized Dr. Bowen’s warning that the results were preliminary and was “NOT to be cited!” AR 00093348 (emphasis and punctuation in original). Export Plaintiffs’ objection on this ground is not valid,

b. Does the Record Support the Specific Flow/Export Ratios Imposed?

The 2010 PI Decision discussed NMFS’s rationale for the 4:1 Ratio:

Consol. Salmonid Cases, 713 F.Supp.2d at 1135-36 (emphasis added) (footnote omitted).

The PI Decision addressed the Phase I flow:export ratio, which operated through May 31, 2011. Phase II, which will control operations starting next spring (from April 1 through May 31), imposes the following flow:export ratios:

BiOp at 643-44.

DefendaniAIntervenors offer the following record-based justification for these ratios:

NMFS 00106725 (BO at 645) (emphasis added); see also NMFS 00107220-21 (BO, App. 5, at 74-75). Figure 11 in Appendix 5 of the BiOp depicts data of flow/export ratios over a 50 year period (1951 to 2003) and reveals that increasing the flow/export ratio was positively correlated with increased escapement of fall-run Chinook salmon 2 1/2 years later. See NMFS 00107166-67 (BO, App. 5, at 21-22).

The BiOp’s rationale for the 4:1 flow/export ratio is likewise clearly set forth and logical:

NMFS 00107168 (BO, App. 5, at 22). NMFS went on to explain that:

Id. (emphasis added). These flow and export levels were then assessed through computer modeling. Id. See NMFS 00107169-00107214 (BO, App. 5, at 23-68).

NMFS acknowledged and responded to DWR’s objections to Action IV.2.1: “Both the Bureau of Reclamation and DWR have strong initial opposition to the proposed RPA. DWR has indicated that the RPA is unfeasible as it [is] currently written. They have proposed alternative actions that NMFS has investigated.” NMFS 00107214 (App. 5 at 68). Among the alternative proposals made by DWR was “real time monitoring at Mossdale utilizing additional Kodiak trawling.” NMFS responded, reasonably, by stating that:

NMFS 00107214 (BO, App. 5, at 68).

The analysis discussed above amply demonstrates that, rather than base Action IV.2.1 solely on feasibility concerns, as Export Plaintiffs incorrectly argue, see Export Br. at 94-96, NMFS used the VAMP fish experiments as a starting point for the agency’s analysis of a flow/export ratio. See NMFS 00107168 (BO, App. 5, at 22). The results of those VAMP studies established that a 2:1 ratio (involving a 1,500 cfs export limit) resulted in “low survival indices,” and that a larger ratio was preferable when possible to “increase survival” and adequately protect the species. Id. After modeling results showed that it would be difficult to increase this ratio in dry years, NMFS reasonably set the Phase II ratios at 1:1 for critically dry years and 2:1 for dry years. NMFS 00107219 (BO, App. 5, at 73). In below normal and above normal years, however, NMFS reasonably concluded that more water would be available to meet the more protective ratios, thus allowing a 3:1 ratio in below normal years and a 4:1 ratio in above normal years. Id.

In its PI ruling, this Court questioned whether Action IV.2.1 was “protective enough” or whether a “5:1 or higher” ratio was necessary. PI Findings, CoL ¶ 51; id. FoF ¶97. As NMFS explained in Appendix 5 of the BiOp, there are no conclusive studies of flow/export ratios greater than 2:1, but the best available data supports a minimum feasible flow of 6,000 cfs in most years, along with a ratio somewhat higher than the historically tested 2:1 ratio, which had proved insufficiently protective as discussed above. See NMFS 00107164-68 (BO, App. 5, at 18-22). Specifically, the 6,000 cfs target minimum flow was determined based on (1) water reasonably available based on historical flow patterns since 1922; and (2) flow-to-escapement relationships indicating that flows over 5,000 to 6,000 cfs “were required to move into the linear phase of increasing fish escapement.” NMFS 00107167-68 (BO, App. 5, at 21-22). Unless NMFS reduced exports to even less than 1,500 cfs — the minimum believed necessary to protect human health and safety — simple math reveals that it would be impossible to achieve a 5:1 or higher ratio, assuming 6,000 cfs as the target minimum flow.

*897 As to whether a less restrictive ratio (e.g., 3:1) was biologically appropriate, NMFS did adopt a 3:1 ratio for below normal years. Only in “above normal” or “wet” years is a 4:1 ratio required. Given the parameters for a flow/export ratio that could feasibly be implemented using a 6,000 cfs target for flows — somewhere between 2:1 and 4:1 — and the lack of data on any of the ratios in between, NMFS reasonably adopted a sliding scale that allows the absolute minimum ratio during drought conditions but increases protections for species when it is feasible to do so, as determined by hydrological conditions.

Of course, NMFS cannot prove with absolute certainty that the 4:1 ratio is protective enough for the species. The existing data simply does allow it. But, given the record evidence of harm to salmon and the need to modify the flow/export ratio, the best available science standard does not require that NMFS stand by and do nothing, paralyzed by a lack of perfect data. To the contrary, NMFS had to act and reasonably exercised its expertise by adopting a flow/export ratio that is both feasible and more protective of the species than the status quo. See Greenpeace Action, 14 F.3d at 1337 (upholding biological opinion even though FWS admitted that it was “uncertain about the effectiveness of its management measures” because it “premised these measures on a reasonable evaluation of available data, not on pure speculation”).

Doc. 484 at 68-71.

This explanation for the basis for the ratios imposed by Action IV.2.1 is supported by record references, which now explain in part NMFS’s choice of ratios, aided by hindsight and judicial review. This is another close call. DWR opposes the flow ratings as infeasible and arguably unnecessary. NMFS uses VAMP flow data to corroborate its position that a 2:1 ratio is insufficiently protective. It justifies the use of a 3:1 ratio when possible (e.g., in below normal years) as necessarily more protective than a 2:1 ratio. The consequences of imposing a 4:1 ratio in above normal and wet years demand a clearer explanation of NMFS’s rationale for imposing a 4:1 ratio, rather than a 3:1 ratio in above normal and wet years. The ESA Handbook requires “a thorough explanation of how each component of the [RPA] is essential to avoid jeopardy and/or adverse modification.” ESA Handbook at 4-43 (emphasis added). 26 This is not to be *898 done by attorneys, post hoc, in litigation. The importance of this requirement is heightened in light of the weak (arguably equivocal) evidence supporting the imposition of any ratios at all. 27

Plaintiffs’ and DWE’s challenge to Action IV.2.1 is valid and their MSJ is GRANTED IN PART AND DENIED IN PART, as are Federal Defendants’ and Defendant-Intervenors’ cross motions. Although there is marginal record support for the imposition of some form of flowrexport ratio, the Action must be remanded for further explanation of the necessity of a 4:1 ratios in above normal and wet years.

2. RPA Action IV. 2.3.

Action IY.2.3 operates from January 1 through June 15 or until the average daily water temperature at Mossdale is greater than 72° F, and limits OMR flows to no more negative than -2,500 to -5,000 cfs, depending on juvenile entrainment levels. BiOp at 648-52. At the first level of increased juvenile loss, exports must be reduced to achieve an average net flow of - 3,500 cfs for a minimum of five days, and at the second level, a more positive OMR average of -2,500 cfs must be achieved for at least five days. Id. For each trigger, OMR averages can return to -5,000 cfs only after three consecutive days of not meeting the higher-density juvenile loss trigger. Id.

Action IV.2.3 is meant to:

Id. at 648.

NMFS utilized several sources of data to determine that export flow limitations would achieve the objectives of RPA Action IY.2.3, including the relationship between OMR flows and salvage, particle tracking model simulations, and other studies evaluating survival of fish within the central and southern Delta. Export Plaintiffs and DWR challenge the scientific basis for NMFS’s determination that an export limitation should be part of the RPA. Export Plaintiffs’ general arguments largely overlap with and incorporate the specific arguments presented in DWR’s briefs, this discussion focuses on DWR’s briefs.

a. Challenge to the Use of the Particle Tracking Method.

DWR argues that the record does not support NMFS’s use of the Particle Tracking Model (“PTM”). A similar argument was addressed in the 2010 PI Decision, which provides the starting point:

Consol. Salmonid Cases, 713 F.Supp.2d at 1138-41 . DWR raises several additional arguments regarding the use of PTM.

(1) DWR’s Argument that NMFS Failed to Address PTM Limitations Described by Kimmerer and Nobriga.

DWR argues that NMFS did not adequately address PTM limitations described in Kimmerer and Nobriga’s 2008 article “Investigating Particle Transport and Fate in the Sacramento San Joaquin Delta using a Particle Tracking Model.” AR 00122246-71.

First, Kimmerer and Nobriga cautioned that PTM “was a useful predictor of entrainment probability if the model were allowed to run long enough to resolve particles’ ultimate fate” and that “model accuracy varies depending on the length of the simulation.” AR 00112246, 00122250. DWR argues that NMFS disregarded these “words of caution.” Doc. 446-1 at

20. This is inaccurate. NMFS convened a group of State and federal scientists to discuss PTM simulations, including DWR representatives, who raised some of the same criticisms asserted here. AR 00106021-25 (June 3, 2009 Memo Re: PTM “results for [OMR] flow manipulation” (“PTM Memo”)). NMFS considered the length of simulation at multiple meetings, AR 00106021-22 (PTM Memo discussing meetings from January through March 2009 and durations of PTM tracking);. AR 00061290-91 (communication between Jeffrey Stuart and Tracy Hinojosa at DWR regarding PTM simulations), AR 00060023-24 (agenda items for discussion at “Modeling work group), and addressed concerns raised in the PTM Memo, see AR 00105025-27.

Second, DWR highlights that Kimmerer and Nobriga (2008) notes that the PTM model “has not been calibrated.” AR 00122262. DWR argues that NMFS should not have used an un-calibrated model. Doc. 446-1 at 20. This ignores the fact that DWR and others have in fact performed validation and calibration on the PTM to ensure that it accurately depicts hydrodynamics. Fourth Stuart Deck, Doc. 485 at ¶ 47.

Finally DWR argues that NMFS’s reliance on this study to “analyz[e] the potential ‘zone of effects’ for entraining emigrating juvenile and smolting salmonids,” BiOp at 361, conflicts with the recommendations of Kimmerer and Nobriga, Doc. 446-1 at 21. Specifically, Kimmerer and Nobriga stated: “[w]e are ... not inclined to define a ‘zone of influence’ of the pumps on the basis of our results.” AR 00122263. The entire paragraph goes on:

AR 00122263. Kimmerer and Nobriga (2008) addresses both larval delta smelt and juvenile salmonids. Their reluctance to define a “zone of influence” is focused on the difficulties posed by modeling larval delta smelt, which may metamorphose to a more complex state within the time horizon of the PTM simulation. This apologetic does not suggest there are not problems with using PTM to define a of zone of influence, it is simply a statement that endeavors to explain uncertainty. Although NMFS’s interpretation and use of Kimmerer and Nobriga (2008) was not accurate, again it is the agency’s spin on the science. It is not unlawfully erroneous.

(2) DWR’s Argument that NMFS Failed to Address Evidence in the Record Critical of the Use of PTM to Explain Salmonid Behavior.

DWR revisits the issue of whether NMFS gave adequate consideration to record evidence critical of the use of PTM to explain salmonid behavior. DWR specifically cites a 2001 article by Baker and Morhardt, AR 00108384-403, and an analysis conducted by DWR included in DWR’s April 24, 2009 comments on the draft BiOp. See Doc. 446-1 at 22-23. Baker and Morhardt (2001) demonstrated that the fate of particles in the PTM was different from actual salmon behavior. AR 00108394 (“for the hydraulic simulations available to us ... 77% of the tracer [PTM] particles ended up at the export pumps, while only 13% of the smolts arrived there”). Likewise, the DWR (2009) analysis concluded there “is no correlation” between coded wire tagged (“CWT”) Chinook recoveries and PTM particle behavior. AR 00105430.

These additional studies do not undermine the reasoning of the 2010 PI Decision:

Consol. Salmonid Cases, 713 F.Supp.2d at 1139 . DWR has acknowledged that this is a permissible PTM use. It has cast doubt on the efficacy of NMFS’s reliance on the PTM, even for this narrow purpose, but has not shown it to be substantially unreasonable.

There is more to DWR’s critique. To validly rely on the PTM results to impose management measures designed to aid salmonid survival, the movement of water described by the PTM must be reasonably related to the movement of salmonids. Citing the Baker and Morhardt (2001) study and DWR (2009) comments, Mr. Ca *904 vallo opines that this “has been shown to be incorrect.” Cavallo Decl., Doc. 452 at ¶ 54.

NMFS justifies reliance on PTM simulations as a proxy for salmonid behavior:

BiOp 366-67. The BiOp does not explain the basis for NMFS’s “belief’ that salmon-ids within a body of water with a higher probability of particle entrainment will themselves “also experience a higher probability of entrainment by the export pumping”; its “belief’ that salmonids within a body of water “where the eventual fate of injected particles indicate a high probability of successfully exiting the Delta at Chipps Island” will themselves have a “higher probability of exiting the Delta successfully”; nor its conclusion that conditions which delay movement of particles out of the Delta yet don’t result in increased entrainment at the export facilities suggest conditions that might delay migration through the Delta. What support for these conclusions does the record contain?

(3) Salvage Data.

The BiOp relies in part on the plots of juvenile loss versus monthly OMR flows in Figures 6-65 and 6-66 discussed above to link the PTM results to salmonid behavior.

BiOp at 361. The BiOp’s reliance on figures 6-66 and 6-66 has been found unlawful. Without scaling for population size, the trends seen in Figures 6-65 and 6-66 are meaningless, because data points indicating greater salvage may simply be the result of a greater absolute number of individuals present in the entire Delta. The number of individuals lost to salvage could go up simply because the volume of water pumped through the salvage facility increases, not because increasing exports causes a greater percentage of the population to make its way toward the salvage facilities than would otherwise be present there.

*906 (4) Other Studies.

In addition to Figures 6-65 and 6-66, NMFS relied on other studies, namely Vogel (2004), Perry & Skalski (2008), Newman (2008), and Newman and Brandes (2009), to conclude that as exports increase, greater numbers of salmonids are drawn into the interior Delta.

(a) Vogel (200k).

The 2010 PI Decision addressed the BiOp’s reliance on Vogel (2004):

Consol. Salmonid Cases, 713 F.Supp.2d at 1144 . Defendant-Intervenors attempt to justify NMFS’s reliance on Vogel (2004):

Doc. 484 at 48-49. Vogel’s call for further experiments or statement that he believes reduced exports, in conjunction with other factors, “may be” beneficial to migrating salmonids, does not change the fact that Vogel’s own 2004 work does not explain ivhy the fish studied moved off the main-stem San Joaquin. Reliance on this study to demonstrate that Action IV.2.3’s negative OMR flow limitations will reduce the vulnerability of juvenile salmonids to entrainment in the Delta is unreasonable.

(b) Perry and Skalski (2008).

The BiOp also relied upon a 2008 study by Perry and Skalski, which was previously addressed in the 2010 PI Decision:

Consol. Salmonid Cases, 713 F.Supp.2d at 1143-44 .

Mr. Stuart’s Fifth Declaration clarifies that the BiOp cited Perry and Skalski (2008) in reference to “the risk that individual salmon smolts face to entrainment into either the channel of the Delta Cross Channel (when it is open) or into the channel of Georgiana Slough as they migrate downstream in the Sacramento River and not to the probability of ending up at the export facilities.” Fifth Stuart Deck, Doc. 519 at ¶ 19. The BiOp did not rely on Perry and Skalski to justify its conclusion that the PTM is a valid proxy for salmon-ids or to demonstrate that exports cause salmonids to move toward the export facilities. Perry and Skalski does not support PTM as a proxy and should not have been cited.

(c) Newman (2008).

Newman (2008), which concluded that salmonids passing through Georgiana slough into the interior delta had slightly reduced survival when exports were higher relative to times when exports were lower, has been discussed. This lends marginal support to NMFSs conclusion that increasing exports negatively impacts salmonids moving through the interior delta.

*908 (d)Newman and Brandes (2009).

DWR cites a 2008 draft of a study by Newman and Brandes analyzing the Delta Action 8 studies, “Hierarchical Modeling of Juvenile Chinook Salmon Survival as a Function of Sacramento-San Joaquin Delta Water Exports.” In that draft, discussing the lower rate of survival of smolts traveling through Georgiana slough and the interior delta under high export conditions relative to low export conditions, the authors opined: “what we cannot conclude is that exports are the cause of this lower relative survival.” NMFS 00127347 . However, that sentence and several following paragraphs were eliminated by Newman and Brandes from the final, published version of the study. See AR 00089883. Like Newman (2008), Newman and Brandes (2009) found negative relationships between exports and survival. Fifth Stuart Decl., Doc. 519 at ¶ 33 (citing AR 00089884).

(e)Brandes & McLain (2001).

DWR also cites Brandes and McLain, “Juvenile Chinook Salmon Abundance, Distribution, and Survival in the Sacramento-San Joaquin Estuary” (2001). This study found “mixed results”:

AR 00109602. That the study’s authors question the validity of the results from the years showing a positive relationship between exports and survival and do not critique the results from the year showing a negative relationship, lends minimal support to NMFS’s conclusion that exports influence salmonid survival.

(f)Kimmerer 2008.

DefendanWIntervenors point to a Kim-merer 2008 that unequivocally found that “[t]he estimated proportion of migrating fish salvaged at the export facilities increased with increasing export flow.” AR 00122236. DWR and Export Plaintiffs do not contest this conclusion.

(g)SJRGA 2007

Finally, DefendanU-Intervenors point to the San Joaquin River Group Authority (“SJRGA”) 2007 review of VAMP data, which found evidence of a negative relationship between exports and survival. See AR 00134423 (“The CDRRs [combined differential recovery rates] measured for the first group released in 2006, under low exports, appeared higher than those obtained in 2003-2005 and for the 2006 group released under higher exports and higher temperature.”).

This entire record shows that the science is conflicting and often equivocal. Most of the evidence does not show a negative relationship between levels of exports and salmon survival. It is impossible to discern the effect of exports on salmon behavior from the record. Despite these numerous criticisms, NMFS chose to use PTM as a modeling tool for salmonid behavior. DWR’s own staff biologist, *909 Sheila Greene, testified in a related case by declaration that “Given the insufficiency of behavioral data, particle tracking is the best available science to estimate the proportions of juvenile Chinook salmon that emigrate through the Delta.” AR 00118803. DWR is bound by this statement. Whether to use PTM modeling then becomes a matter of agency discretion.

b. Justification for the Specific Flow Prescriptions in Action IV.2.3.

Existence of record support for NMFS’s reliance on the PTM does not end the inquiry. The BiOp applied the PTM to generate flow prescriptions using the following approach:

BiOp at 652. Even if the PTM is the best available mechanism for modeling salmonid behavior, NMFS has failed to justify this leap of logic, which in essence assumes that salmonids will be drawn toward the export facilities to the same extent as neutrally buoyant particles. In light of undisputed record evidence discussed above demonstrating the many deficiencies in the PTM and that it is far from a perfect proxy for salmon behavior, NMFS has not provided “a thorough explanation of how each component of the [RPA] is essential to avoid jeopardy and/or adverse modification.” ESA Handbook at 4-43 (emphasis added). Conclusory explanations of the value of RPA actions are insufficient. PCFFA v. U.S. Bureau of Reclamation, 426 F.3d 1082, 1093 (9th Cir.2005).

Plaintiffs’ and DWR’s motion challenges to Action IV.2.3 is GRANTED IN PART AND DENIED IN PART, as are Federal Defendants’ and Defendant-Intervenors’ cross motions. There is nominal record support for the imposition of some form of OMR flow restriction, but the Action must be remanded for further explanation of the necessity the specific flow prescriptions imposed, which are derived primarily from PTM simulations, a method that is undisputedly an imperfect predictor of salmon behavior.

3. Action IV.3.

From November 1 through December 31, Action IV.3 restricts combined CVP and SWP exports to 6,000 cfs or 4,000 cfs when certain salvage thresholds are met. BiOp at 653. The 6,000 cfs export limit is triggered when daily SWP and CVP older juvenile loss density is greater than 8 fish per thousand acre feet (“TAF”), daily loss is greater than 95 fish per day, or the Coleman National Fish Hatchery coded wire tagged late fall-run Chinook salmon (“Coleman CWT fish”) or Livingstone Stone National Hatchery coded wire tagged winter-run (“L-S CWT fish”) cumulative loss is greater than 0.5%. Id. The *910 more restrictive 4,000 cfs export limit is triggered when the daily older juvenile loss density is greater than 15 fish per TAF, daily loss is greater than 120 fish per day, or the Coleman CWT or L-S CWT fish cumulative loss is greater than 0.5%. Either export restriction remains in place for three days, or until daily older juvenile loss density is less than 8 fish per TAF. Id. Action IV.3 also establishes an “alert,” which signals that export restrictions may need to be altered when either the Knights Landing or Sacramento catch index is greater than 10 fish captured per day from November 1 to February 28, or greater than 15 fish captured per day from March 1 to April 30. Id. at 652.

The objective of Action IV.3 is to “[r]educe losses of winter-run, spring-run, CV steelhead, and Southern DPS of green sturgeon by reducing exports when large numbers of juvenile Chinook salmon are migrating into the upper Delta region, at risk of entrainment into the central and south Delta and then to the export pumps in the following weeks.” Id.

DWR does not challenge this RPA Action. Export Plaintiffs, however, raise several objections. First, Export Plaintiffs challenge NMFS’s underlying assumption for Action IV.3 that “[ejxport pumping changes flow patterns and increases residence time of ... diverted fish in the central Delta, which increases the risk of mortality from [other factors], as well as the likelihood of entrainment at the pumps.” Id. at 653. They argue that this assumption is not supported by the best available science, reiterating the argument that the studies relied upon by NMFS do not “connect mortality in the interior delta to export levels,” and that NMFS’s use of PTM studies to show “potential vulnerabilities of fish” is not appropriate. Doc. 431 at 106-108. Although the evidence supporting a survival effect of increased exports weak and disputed, NMFS’s conclusion that export pumping negatively impacts salmonid survival has marginal support in the record and is not unlawfully erroneous. The BiOp’s related use of PTM as a modeling tool for salmon-ids is a highly disputed scientific choice, described by DWR as the best available science in at least one application.

Export Plaintiffs also challenge the specific triggers used in Action IV.3, on the grounds that: (1) as was the case with the plots of raw salvage used to justify other Actions, the triggers do not account for the relative size of the various salmonid populations; and (2) “NMFS nowhere explains how it arrived at these thresholds. Doc. 431 at 108-09.

Federal Defendants respond that Plaintiffs’ concern that Action IV.3 fails to scale the loss triggers against the population size is “a red herring” because “NMFS did not conclude that Action IV.3 alone is avoiding jeopardy. Rather, the measure simply reflects NMFS’s conclusion that it is important to have increased protections when you have more fish at the salvage facilities.” Doc. 477-1. This is a total abdication from NMFS’s self-imposed requirement that the RPA provide “a thorough explanation of how each component of the [RPA] is essential to avoid jeopardy and/or adverse modification.” ESA Handbook at 4-43 (emphasis added). Federal Defendants cannot impose a complex and burdensome RPA, damaging of other interests, without specifically justifying each of its components. Adoption of Federal Defendants’ “trust me” approach would mean that the more complex an RPA, the more obscured it is from judicial review. Each component need not eliminate jeopardy on its own, but that does not excuse NMFS from separately justifying individual Actions.

While there is record explanation for an action designed to prevent large numbers *911 of fish from being killed or harmed at the export pumping facilities, Export Plaintiffs raise serious questions related to the need to scale the triggers to the overall size of the salmonid populations they aim to protect. Given previous findings about the use of raw salvage figures, the best available science calls for an index related to population size, rather than a fixed number. More importantly, even if this were not a problem, Federal Defendants have entirely failed to provide any record explanation for why the specific triggers were chosen. NMFS must address and correct this failure on remand. Plaintiffs’ motion is GRANTED on this issue; Federal Defendants’ and Defendant-Intervenors’ cross motion is DENIED.

G. Compliance with 50 C.F.R. § 402.02 .

ESA section 7(b)(3)(A) provides that, “[i]f jeopardy or adverse modification is found, the Secretary shall suggest those reasonable and prudent alternatives which he believes would not violate subsection (a)(2) of [Section 7] and can be taken by the Federal agency ... in implementing the agency action.” Id. “Reasonable and prudent alternatives refer to alternative actions identified during formal consultation [1] that can be implemented in a manner consistent with the intended purpose of the action, [2] that can be implemented consistent with the scope of the Federal agency’s legal authority and jurisdiction, [3] that is economically and technologically feasible, and [4] that the Director believes would avoid the likelihood of jeopardizing the continued existence of listed species or resulting in the destruction or adverse modification of critical habitat.” 50 C.F.R. § 402.02 (the “four RPA requirements”).

NMFS and FWS’s joint Consultation Handbook explains that during the formal consultation period, NMFS should “meet or communicate with the action agency ... to gather any additional information necessary to conduct the consultation.” Consultation Handbook at 4-6. Among other things, the formal consultation period should be used to “develop reasonable and prudent alternatives to an action likely to result in jeopardy or adverse modification....” Id.

Consultation “should be undertaken cooperatively with the action agency and any applicant, thus allowing the Services to develop a better understanding of direct and indirect effects of a proposed action and any cumulative effects in the action area. Action agencies also have the project expertise necessary to help identify reasonable and prudent alternatives, and reasonable and prudent measures. Other interested parties (including the applicant, and affected State and tribal governments) should also be involved in these discussions .... These cooperative efforts should be documented for the administrative record.” Id.

The Handbook contains a section on RPAs, which provides as follows:

Reasonable and prudent alternatives

Consultation Handbook, 4-41 — 4-42.

San Luis & Delta-Mendota Water Authority v. Salazar (Consolidated Delta Smelt Cases), 666 F.Supp.2d 1137 *913 (E.D.Cal.2009), discussed section 402.02 in considering a facial challenge to a related biological opinion, filed before the administrative record was completed. Plaintiffs argued that FWS acted unlawfully by failing to discuss the four § 402.02 factors on the face of the biological opinion. The decision found that FWS was only required to make explicit findings in the biological opinion on the fourth factor, namely whether the RPA will avoid the likelihood of jeopardy or adverse modification. “[Wjhether FWS properly promulgated the RPA [consistent with the requirements of § 402.02] must be decided on the basis of the entire record.” Id. at 1158-59.

Export Plaintiffs now bring a record-based challenge to NMFS’s alleged noncompliance with § 402.02, asserting that entire RPA is invalid because the record does not support a finding that NMFS complied with the four § 402.02 factors. Doc. 431 at 109-118. DWR joins this aspect of Export Plaintiffs’ motion. Doc. 446-1 at 27.

There is scant authority to aid interpretation of § 402.02. The text of the Federal Register Notice promulgating § 402.02 provides limited guidance:

51 Fed.Reg. 19,926, 19,937 (June 3, 1986).

1. Consistency with Purposes of the Action.

The BiOp reasons that because the operational changes demanded by the RPA do not preclude continued operation of the CVP and SWP, the RPA is consistent with the purpose of the action:

BiOp at 724. The BiOp also discussed the various purposes of the CVP:

Id. at 724-25.

Export Plaintiffs challenge NMFS’s finding that the RPA is consistent with the multiple purposes of the Projects. First, *915 Export Plaintiffs argue that NMFS ignored warnings about the water costs of the RPA. For example, DWR commented that “the average combined water supply impact to the SWP and the CVP of the NFMS proposed RPA is roughly 900 [thousand acre feet (“taf’) ] to 1.1 [million acre feet (“Maf’) ] (or about 16% to 19%).” AR 00086760. DWR’s estimate continues:

Id. Plaintiffs also point to Reclamation’s comments on the RPA, which express general concerns over water supply impacts. AR 00210461-69, 00210473-76, 00105273.

That there are water supply impacts does not necessarily render the RPA inconsistent with the purposes of the action. However, NMFS is absolutely obligated do more than simply check to ensure that the proposed operational changes do not “preclude operation of the Projects.” See BiOp at 724. Assumedly, if the Projects delivered ten AF of water in a water year, the Projects would be “operating.” An RPA that effectively eliminates Project water deliveries to parts of the CVP’s service area is inconsistent with one of the coequal purposes of that project. 28 What is *916 the ultimate impact of the salmonid BiOp RPA? The BiOp does not provide explicit answers to these questions. The BiOp predicted a lower estimate of water supply costs than DWR:

BiOp at 720-21 (footnote omitted).

The agency abandoned its legal duties and said in effect: “we can’t model, we won’t do it.” However, much of the Defendants’ support for the BiOp and its RPA actions is based upon the same highly variable and questionable modeling of species populations and effects from exports. As this agency practices, what is “science” for the “goose” is clearly not “for the gander.”

Export Plaintiffs argue that NMFS has failed to explain why its estimate is more reliable than the 900,000 — 1,100,000 AF water cost estimate (16%-19% of the Projects’ combined water supply) provided by DWR, one of the Project co-operators. Doc. 431 at 113-14. In particular, Export Plaintiffs challenge the BiOp’s consideration of only the impacts of export curtailments “associated with the NMFS San Joaquin River Ratio actions in the RPA,” presumably a reference to Action IV.2.1. The BiOp explains that many of the OMR restrictions in both the Salmon and Smelt BiOps “tend to result in export curtailments of similar quantities at similar times of year,” but does not explain why it is appropriate to entirely ignore the effects of those curtailments that may overlap with those mandated by the Smelt BiOp. This requires further clarification and revision in light of competent and meaningful impact studies. 29

*917 Even assuming, arguendo, the BiOp’s water cost prediction is correct, is such a reduction “consistent” with the irrigation purpose of the CVP? How should the RPA analyze the extent of water supply reductions that are consistent with the co-equal legislative irrigation purpose? The ESA provides no guidance, nor do the joint ESA regulations or any other authority identified by any party. It cannot simply be said that if an ESA-listed species’ protection is at stake, the “no balancing of hardships” principle excludes such consideration. This would impermissibly rewrite Reclamation law to eliminate the regulatory requirement that NMFS consider the RPA’s effect on the co-equal statutory purpose of irrigation. Federal Defendants’ examination of this factor is insufficient. Export Plaintiffs’ motion for summary judgment on this issue is GRANTED; Federal Defendants’ and Defendant-Intervenors’ cross motions are DENIED.

2. Consistency with the Action Agency’s Legal Authority and Jurisdiction.

Export Plaintiffs dispute the BiOp’s conclusion that the RPA can be implemented in a manner consistent with the legal authority and jurisdiction of Reclamation and DWR. The BiOp reasons that “[t]he CVPIA gives Reclamation broad authority to mitigate for the adverse effects of the projects on fish and wildlife, and nothing in the Rivers and Harbors Act of 1937 requires any set amount of water delivery.” BiOp at 724-25. The BiOp also recognizes that the CVPIA contains a goal of doubling the natural production of anadromous fish in Central Valley rivers and streams, and that this goal has not yet been met. Id. at 725. As Federal Defendants well know, the CVPIA dedicates a finite 800,000 AF of annual CVP yield to the fish-doubling objective.

The BiOp reasons that Reclamation has broad powers to restore anadromous fish populations:

(Cummins et al.2008 at 11)

Id. The CVPIA contains prescriptives; it does not elevate the ESA over all other statutory purposes for use of Project water. Although specific provisions of the law may authorize finite increase in fish protection water appropriation, there is no indefinite, unlimited power for NMFS to take whatever Project water it deems essential for the species.

The BiOp also finds that “state law gives DWR authority to provide for needs of fish and wildlife independent of the connection of the two water projects.”

Id. at 726.

Although § 402.02’s RPA requirements demand that NMFS engage in an evaluation of its legal authority to implement the RPA Actions, NMFS’s interpretations of these laws set forth in the BiOp are not entitled to deference, as they were neither promulgated through notice and comment rulemaking procedures, see Chevron v. NRDC, 467 U.S. 837 , 104 S.Ct. 2778 , 81 L.Ed.2d 694 (1984), nor contained within an agency policy statement, manual, enforcement guideline, or other document entitled to limited deference, see Skidmore v. Swift & Co., 323 U.S. 134 , 65 S.Ct. 161 , 89 L.Ed. 124 (1944); Christensen v. Harris County, 529 U.S. 576, 587 , 120 S.Ct. 1655 , 146 L.Ed.2d 621 (2000).

State and federal law impose upon Reclamation and DWR a nondiscretionary duty to comply with state water rights law. See 43 U.S.C. § 383 ; California v. United States, 438 U.S. 645, 675 , 98 S.Ct. 2985 , 57 L.Ed.2d 1018 (1978). Export Plaintiffs point to the “obligation imposed upon both Reclamation and DWR by D-1641 to comply with the reasonable and beneficial use requirements and prohibition against waste [of water] set forth in Article X, section 2 of the California Constitution, in their respective operations of the CVP and SWP.” Doc. 431 at 117. The argument continues:

Id.

It is undisputed that California law identifies the preservation of fish and wildlife as a beneficial use of water. Cal. Water Code § 1243 . In addition to requiring compliance with California’s reasonable and beneficial use standard, D-1641 imposes a condition upon both Reclamation’s and DWR’s water rights requiring both to “meet[ ] all requirements of the applicable Endangered Species Act for the project authorized under [their respective] permit[s]/license[s].” D-1641 at 148. When jeopardy is found, the ESA requires implementation of a RPA. D-1641 authorizes Reclamation’s implementation of lawful RPAs. 30

However, several of the specific RPA prescriptions have failed to demonstrate compliance with the Handbook’s requirement that every RPA be “essential to avoid jeopardy and/or adverse modification.” Obviously, to the extent that any RPA Action has been found unlawful, Federal Defendants cannot establish that implementation of that RPA is consistent with Reclamation’s legal authority.

Export Plaintiffs’ motion for summary judgment that NMFS failed to demonstrate the RPA’s consistency with Reclamation and DWR’s legal authority is GRANTED IN PART AND DENIED IN PART, as are Federal Defendants’ and Defendant-Intervenors’ cross motions. To the extent individual RPA Actions are otherwise lawful, Export Plaintiffs’ argument that Federal Defendants’ failed to demonstrate Reclamation’s authority to implement those Actions is belied by D-1641, which expressly requires implementation of lawful RPA Actions. Correlatively, to the extent individual RPA Actions are unlawful, Federal Defendants cannot find authority for their implementation.

3. Economic and Technical Feasibility.

The BiOp contains a lengthy discussion of economic and technical feasibility. Export Plaintiffs attack the discussion as insufficient in several respects. First, Export Plaintiffs argue that NMFS ignored objections and evidence submitted by Reclamation and DWR suggesting that the RPA was not technologically or economically feasible. A March 23, 2009 letter from Reclamation to NMFS details a number of concerns with the proposed RPA. See AR 00105277-84. A March 20, 2009 letter from DWR to NMFS describes some specific feasibility concerns and a general objection that several of the RPA actions were not economically feasible. See AR 00105285-99.

Export Plaintiffs object generally that NMFS failed to articulate a reasoned response to DWR and Reclamation’s objections. Doc. 431 at 114. The BiOp explained that the relevant state and federal agencies engaged in a back-and-forth exchange of information regarding feasibility of the RPA and adjustments were made:

Id. at 719-20 .

NMFS viewed adaptive management as another tool to address feasibility issues:

Id. at 720 .

Export Plaintiffs do not identify any specific technological feasibility objection *921 that was not addressed by NMFS’s adjustments to the draft RPA. Export Plaintiffs do argue that NMFS ignored “extensive evidence” submitted by DWR about economic feasibility. DWR informed NMFS of its opinion as to the economic impact of the RPA:

AR 00113831-32. Export Plaintiffs argue that according to DWR’s figures, the net cost of the NMFS RPA over a 20-year implementation period could exceed $8 billion dollars. Doc. 431 at 115. Based on its own figures, DWR urged NMFS to find that the RPA did not meet the standard for economic feasibility. AR 00113831-32.

Does section 402.02 contemplate consideration of economic costs to third parties or just to the action agencies? Without any analysis or legal authority, the district court in Kandra concluded: “Read in context ... the RPAs must be economically and technically feasible for the government to implement.” Id. at 1207 (emphasis added). The regulation itself does not specify whether feasibility should be limited to the economic impact on the action agencies or on others affected by the agency action. Defendants contend the regulation must be interpreted in a manner that does not violate TVA v. Hill, 437 U.S. 153, 184 , 98 S.Ct. 2279 , 57 L.Ed.2d 117 (1978), which concluded that Congress enacted the ESA to “halt and reverse the trend toward species extinction, whatever the cost.” (Emphasis added.) This language directs the conclusion that the economic feasibility requirement refers only to the costs to the action agency, requiring analysis of whether the corrective measures required by an RPA can be implemented from a purely budgetary perspective.

NMFS engaged in such an analysis. Starting with its 330,000 AF water supply impact projection, which has been remanded for other reasons, NMFS examined the impact of water supply reductions on Reclamation and DWRs costs:

*922 BiOp at 723. NMFS also evaluated direct Project Costs.

Id. at 723-24.

Even assuming DWR’s higher water costs figures (approximately three times NMFS’s estimate), no party suggests that the costs to the agency would be prohibitive, given the reimbursement structure. Export Plaintiffs’ motion for summary judgment that NMFS failed to demonstrate economic and technological feasibility is DENIED. Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED. DWR’s specific challenge to the feasibility of Action TV.4.2 is addressed separately below.

4. Avoidance Jeopardy and/or Adverse Modification.

Export Plaintiffs incorporate by reference their substantive challenges to the RPA, arguing that for all those reasons, NMFS failed to comply with the fourth requirement of section 402.02. Consistent with and incorporating the rulings on the merits of the challenges to RPA Actions IV.2.1, IV.2.3 and IV.3, Export Plaintiffs’ motion regarding the fourth section 402.02 requirement is GRANTED IN PART AND DENIED IN PART and Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED IN PART AND DENIED IN PART. While there is some record support for the general approaches used in these RPA Actions, the specific prescriptions imposed are not sufficiently justified. As a result, NMFS did not reasonably conclude that Actions IV.2.1, IV.2.3 and IV.3 were essential to avoid jeopardy to the continued existence of the Listed Species and/or destruction or adverse modification of the species’ critical habitat.

5. DWR’s Feasibility Challenges to Action IVA.2.

The stated objective of Action IV.4.2, entitled “Skinner Fish Collection Facility Improvements to Reduce Pre-Screen Loss and Improve Screening Efficiency,” is to “[ijmplement specific measures to reduce pre-screen loss and improve screening efficiency at state facilities.” BiOp at 655. The Action requires DWR to undertake the following actions at the Skinner Fish Collection Facility:

Id. at 655-56.

a. Is Action IV.4.2 Inconsistent with Action IV.4.

DWR argues that Action IV.4.2 is arbitrary and capricious because it is inconsistent with Action IV.4 (“Modifications of the Operations and Infrastructure of the CVP and SWP Fish Collection Facilities”), which provides:

Id. at 653-54. DWR suggests that Action IV.4 defines 75% salvage efficiency as a “performance goal,” rather than a requirement, and therefore that Action IV.4.2’s “requirement” of 75% efficiency is inconsistent with Action IV.4. Action IV.4 does not relegate the 75% target to the status of a “performance goal.” The action sets a requirement for the Bureau and DWR “shall each achieve a whole facility salvage efficiency of 75 percent.” Id. at 653. There is nothing equivocal about this language and no inconsistency between Actions IV.4 and IV.4.2.

DWR’s motion for summary judgment that Action IV.4.2 is unlawful because it is inconsistent with IV.4 is DENIED; Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED,

b. DWR’s Argument that the Record Does not Support the Conclusion that Action IV.4.2 (Subpart(l)) is Economically and Technologically Feasible.

DWR next complains that the record does not support the conclusion that the first subpart of Action IV.4.2, which requires 75% salvage efficiency at the Skinner Fish Facility for Chinook salmon, CV steelhead, and Southern DPS of green sturgeon by December 31, 2012, is technologically or economically feasible. DWR maintains that NMFS arbitrarily took the “goal” of Action IV.4, namely achieving a *924 75% salvage efficiency for the Skinner Fish Facility, turned it into an “action,” and “slapped a date for compliance on the goal.” Doc. 446-1. DWR’s first premise — that the 75% efficiency target in Action IV.4.2 is a “goal”' — is mistaken. Nor is it inherently illogical for NMFS to impose a compliance deadline for the 75% target. DWR has been studying salvage efficiency and ways to improve the salvage process for many years, see, e.g., AR 00109712-31, and NMFS was warned by a Reclamation biologist with experience working with individuals at the facility that without a deadline, improvements might never take place, AR 00105052. The deadline of December 31, 2012 provided DWR with approximately three and a half years from the adoption of the BiOp in June 2009. DWR has not demonstrated that imposing this deadline was irrational or arbitrary.

DWR next argues that the record does not support the conclusion that the standard of 75% efficiency at the Skinner Fish Facility is technologically or economically feasible. DWR points to its assertion in a March 20, 2009 letter to NMFS that it might not be possible to meet the December 31, 2012 deadline.

AR 00078204. That DWR’s expressed “uncertainty” to NMFS that it could not meet the higher target by the end of 2010 does not mean the action is “infeasible.” One and one-half years remain to perform.

DWR also maintains that NMFS had information indicating that DWR could not even complete the necessary studies on the current facilities’ efficiency by the deadline. Steelhead studies had been ongoing since 2005, AR 00003660, 3642, 4105, 4128-29. DWR maintains that “[tjhere are no similar studies as to sturgeon and salmon and “[f]rom the steelhead facility study, NMFS was aware that a study on facility efficiency would take at least three years to perform.” However, DWR provides no record citation to support this three-year timeframe. To the contrary, the methods described in a 2008 technical study plan for the Tracy Fish Facility suggest that the actual experiments would be run over a period of only several months. AR 00078557, 00078563 (explaining that efficiency experiments for fiscal year 2009 will be “completed during the months of March-June,” with results available by August 2010).

*925 DWR next argues “the standard for efficiency imposed by NMFS seems to have changed from 90%, to 80%, to 75%” without explanation as to whether one or any of the standards was economically or technologically feasible. Doc. 446-1 at 6. Federal Defendants emphasize that a study cited by DWR found that the Skinner Fish Collection Facility is already operating at an estimated 74% efficiency for steelhead. AR 00113798 (cited in BiOp at 346). Federal Defendants further explain the reasoning behind the 75% efficiency standard:

Doc. 477 at 104-106.

DWR argues that Brown, et al. (1996) is based upon obsolete data collected at the louvers approximately 40 years ago, in 1970 and 1971. AR 00109725. The Brown study recognized that changes to the Skinner Fish Facility had been made between the time the data was collected and the article’s publication. AR 00109728. Therefore, DWR argues that the data relied upon in Brown, et al. (1996) does not reflect the current or potential efficiency at the Skinner Fish Facility. Doc. 495 at 10. Nor does the data reflect the entire process by which DWR protects fish at Skinner, which involves handling, trucking, and releasing entrained fish. The focus of Action 4.2.1(1) is the overall efficiency of the Skinner Facility, not just the louver facility. Id.

Federal Defendants do not respond to these critiques of the obviously outdated Brown (1996) study, but instead focus on the fact that DWR’s own 2007 study predicted that efficiency at the entire Skinner Facility was estimated to be “74 ± 5% (mean ±95% Confidence Interval) for the 2007 study period. AR 00113798. DWR objects that “even this data does not support Federal Defendants’ conclusion that an improvement of 1% to 75% overall efficiency standard for operating the Skinner Fish Facility as to Chinook salmon, steel-head trout, and green sturgeon, is economically and technologically feasible by the date imposed under the BiOp.” Doc. 495 at 10-11. Based on the information before NMFS, was it unreasonable to conclude that a 1% improvement was technologically or economically feasible? NMFS justifies its conclusion on the ground that DWR “did not say it could not possibly reach a one percent higher efficiency target by 2012.” Doc. 477-1 at 107 (emphasis added). Rather, DWR’s comment letter stated that DWR had “uncertainty” as to whether the efficiency improvements were feasible.

To uphold an agency’s decision, its rationale must “reasonably be discerned,” from the record. See Modesto Irr. Dist., 619 F.3d at 1035 . Here, NMFS has failed to cite any record evidence indicating that the efficiency improvement, albeit a minor one, is economically or technologically feasible. DWR’s own 2007 study indicates the efficiency is close to the target, from which it could be inferred that the technological changes may be possible, but the record lacks affirmative support for a finding of feasibility.

DWR’s motion for summary judgment that the record lacks support for a finding that Action IV.4.2(1) is feasible is GRANTED; Federal Defendants’ and Defendant Intervenors’ cross motions are DENIED,

c. DWR’s Argument that the Record Does not Support the Conclusion that Action IV4-2 (Subpart(2)) is Economically and Technologically Feasible

DWR maintains that the record does not support NMFS’s economic and feasibility determination as to subpart 2 of Action IV.4.2, which requires reduction of predation at Clifton Court Forebay to 40% by March 31, 2014.

There is undisputed record evidence that at DWR’s facility, most loss of fish occurs in the Clifton Court Forebay. AR 00117410-441 (Gingras 1997). DWR con *927 cedes that pre-screen loss at the Forebay is estimated at between 63 and 99% for juvenile Chinook salmon. Doc. 446-1 at 7 (citing BiOp at 348 and AR 00106736).

It is also undisputed that reducing predation would improve survival across the Forebay. See AR 00113817 (DWR (2008)). However, DWR argues that the record does not support Action IV.4.2(2)’s imposition of the specific requirement that DWR reduce “predation” in the Forebay to 40%. In support of Action IY.4.2(d), NMFS cites a 2008 DWR study that in turn cites a 1952 study by Ricker. Doc. 477-1 (citing AR 113817). Ricker concluded that when survival rates are below 25%, a reduction of predator numbers to below 50% can double the survival rate of the prey. But, DWR points out that Ricker’s finding that predator numbers should be reduced to below 50% is distinct from whether predation should be reduced to below 50%. 31 In response, Federal Defendants disclaim reliance on Ricker, asserting that NMFS considered DWR (2008) as support for the proposition that predator removal is a method of reducing pre-screen loss. Doc. 515 at 44. Rather than rely directly on Ricker’s work, NMFS “reasoned by simple math that if predation was reduced by half to no more than 40%, giving 60% survival rather than 20-25% survival, overall survival through the Skinner Facility would reach approximately 39%, roughly equivalent to the current CVP survival efficiency.” Id. Federal Defendants provide no record citation related to this imprecise guestimate.

Even if the 40% target is scientifically justified, whether predation in the Fore-bay is a problem and/or whether a reduction to 40% is a reasonable goal is an entirely different inquiry from whether reducing predation to 40% is feasible. NMFS mentions numerous examples of methods of reducing predators on juvenile salmon and steelhead, but nothing in the record indicates whether the 40% target can be met, or whether it could be met by the deadline imposed by the BiOp. 32 It cannot be determined from the existing record whether NMFS’s feasibility determinations are supportable. There is no explanation.

DWR’s motion for summary judgment that the record lacks support for a finding that Action IV.4.2(2) is feasible is GRANTED; Federal Defendants’ and Defendant Intervenors’ cross motions are DENIED,

d. DWR’s Argument that Action IV.Jf£ is Arbitrary and Capricious Because it Fails to Explain How the Action Will Avoid Jeopardy and!or Adverse Modification.

DWR also argues that the record does not support NMFS’s finding that Action IV.4.2 is essential to avoid jeopardy and/or adverse modification. There is record evidence to support NMFS’s findings that *928 pre-screen loss and loss due to salvage are significant and that reducing these sources of loss will improve survival. However, the record does not explain why increasing the existing salvage efficiency by 1% and/or reducing predation to 40% “is essential to avoid jeopardy and/or adverse modification.” ESA Handbook at 4-43 (requiring a “thorough explanation of how each component of the [RPA] is essential to avoid jeopardy and/or adverse modification”). The RPA is not lawful without the required through explanation, which shall be provided on remand.

VI. STANISLAUS RIVER PLAINTIFFS’

CLAIMS.

A. Relevant Factual Background.

1. The New Melones Project.

The New Melones Project was approved as the last unit of the CVP in 1962. Pub.L. No. 87-874, 76 Stat. 1173, 1191-92 (1962). The New Melones Project includes a dam and 2.4 million acre-foot reservoir on the Stanislaus River. USBR AR 007570. The New Melones Reservoir is “operated primarily for purposes of water supply, flood control, power generation, fishery enhancement, and water quality improvements in the lower San Joaquin River. The reservoir and river also provide recreation benefits.” Id. The United States holds appropriative water rights issued by the SWRCB for the New Melones Project, conditioned by Water Rights Decisions 1422, 1616 and Revised Decision 1641 (“D-1641”). See generally USBR AR 007571-73.

2. The Stanislaus River Plaintiffs.

Plaintiffs Oakdale Irrigation District (“OID”), and South San Joaquin Irrigation District (“SSJID”) hold pre-1914 water rights to Stanislaus River water. OID and SSJID receive water from New Melones under a 1988 Agreement with the United States designed to fulfill their prior rights. USBR AR 007571-72; USBR AR 011751. That agreement requires Reclamation to provide to OID and SSJID:

Plaintiff Stockton East Water District (“SEWD”) is one of only two “Eastside Contractors” that receive a CVP supply from New Melones pursuant to Reclamation water service contracts. SWED’s contract provides for up to 75,000 AF of water annually. See USBR AR 011728-29. (Collectively, these three plaintiffs are referred to as the “Stanislaus River Plaintiffs” or “SR Plaintiffs.”)

3.The Status of Steelhead in the Stanislaus River.

The OCAP BA summarizes the history and status of Steelhead in the Stanislaus River:

USBR AR 007670-71.

The BiOp describes the impacts of proposed operation of New Melones on survival of CV steelhead and its critical habitat, *930 BiOp at 296-313, and imposes a number of RPAs that affect the New Melones Unit.

SR Plaintiffs raise a number of challenges to the treatment of New Melones in the BiOp, its effects analysis, and RPAs related to New Melones. 33

B. Inclusion of the New Melones Unit in the Proposed Action Subject to Consultation.

SR Plaintiffs challenge NMFS’s decision to include the New Melones Unit in the action subject to consultation. The ESA’s consultation requirement applies to “agency actions.” See 16 U.S.C. 1536(a)(2). The ESA implementing regulations define “action” to mean “all activi *931 ties 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.” 50 C.F.R. § 402.02 . No regulation defines the scope of the action to be considered. The question is whether NMFS’s definition of the scope of the action is reasonable in light of the record. See PCFFA, 426 F.3d at 1090 (“Even when an agency explains its decision with less than ideal clarity, a reviewing court will not upset the decision on that account if the agency’s path may reasonably be discerned.”).

SR Plaintiffs make two attacks on NMFS’s decision to include New Melones in the Project Description. They first argue that the touchstone of inclusion in the action appears to be “coordination” of the Unit in question with Project operations. For example, the SWP is included in the action because its operations are closely coordinated with those of the CVP through the Coordinated Operating Agreement. BiOp at 31; USBR AR 007495 (BA 1-4) (“SWP operations are coordinated with CVP operations and, as such, are consulted on as part of the proposed action described in the BA.”). In contrast, the Friant Unit was deliberately excluded from the action because it operates separately from the rest of the CVP and is not integrated into the CVP OCAP.” BiOp App. 1 at 79. ' SR Plaintiffs argue that while “New Melones is an element of the CVP, it is also clear that operation of New Melones is not coordinated with the operation of the rest of the CVP and/or SWP.” Doc. 454 at 37. This assertion is belied by the record. Stanislaus Plaintiffs admit that New Mel-ones is one of the major reservoirs in the CVP system and releases from it are needed to meet non-consumptive downstream purposes, such as water quality and the preservation and enhancement of fish and wildlife in both the Stanislaus and San Joaquin rivers. Id. at 37-38; see also Sixth Milligan Decl., Doc. 517 at ¶ 10 (explaining that “[rjeleases from New Mel-ones down the Stanislaus River affect Reclamation’s ability to comply with Vernalis flow and water quality requirements.”).

SR Plaintiffs also argue that inclusion of New Melones in the action subject to consultation will lead to “absurd results” because the Incidental Take Statement provides that the RPAs must be implemented as a whole and if Reclamation and/or DWR fail to comply with the terms of the ITS, they may no longer comply with the ESA. See BiOp at 728. SR Plaintiffs maintain that this is “absurd given the lack of coordination” between New Melones and other operations of the CVP and SWP. Doc. 454 at 38. “Simply put, what happens on the Sacramento River as a result of actions taken by the CVP and/or SWP has nothing to do with the listed species contained in the Stanislaus River, and vice-versa.” Id. This statement is incorrect. Mr. Milligan opines: “during balanced conditions, releases from New Melones down the Stanislaus River affect overall Delta conditions, which potentially play a role in determining how Reclamation operates the rest of the CVP____Therefore [Reclamation] typically eoordinate[s] operations of the various Delta facilities and CVP reservoirs, including New Melones Reservoir ... with DWR in its operation of the SWP, on a daily basis.” Sixth Milligan Deck, Doc. 517 at ¶ 10. SR Plaintiffs have presented no contrary evidence.

SR Plaintiffs’ motion for summary judgment that Federal Defendants erred by including New Melones in its coordinated Project description is DENIED. Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED.

C. Effects Analysis Challenges.

1. New Melones Operations v. Baseline Effects.

SR Plaintiffs argue that the BiOp is unlawful because NMFS improperly *932 identified as “effects of the action,” effects caused by the existence of New Melones Dam. Doc. 454 at 18-21. Specifically, SR Plaintiffs point to the BiOp’s conclusions that the action (1) altered flows, which impact habitat conditions and survival at various life history stages, and (2) modified the hydrograph to dampen peak flood flows, mute flow variability, and reduce or eliminate channel forming flows. Id. at 19. SR Plaintiffs maintain that “[t]hese are effects associated with the fact that the dam exists, rather than effects associated with the operational plan.” Id. at 20. SR Plaintiffs’ seminal argument is that NMFS has “failed to identify with any clarity how continued operations, as opposed to the basic existence of the dam itself, cause any additional incremental harm to steelhead, deepen their jeopardy, or otherwise ‘tip’ them into extinction.” Doc. 454 at 20.

The record does explain how continued operations will cause additional incremental harm. The BiOp compares effects of the action both to pre-dam “unimpeded” conditions and to the “future baseline” which includes the existence of the dams:

BiOp at 300-301 (emphasis added). The subsequent pages provide more specific support for these conclusions. Id. at 302-309.

SR Plaintiffs specifically challenge only one aspect of this analysis; NMFS’s reliance on a 2001 Kondolf, et al. study to support the assertion that available steel-head spawning gravel habitat decreased 40% since 1994. BiOp at 308. Kondolf, et al. (2001) concluded that spawning gravel habitat decreased 40% between 1972 and 1993, and thereafter decreased by a smaller percentage, within the study’s margin of error, between 1993 and 2000, excluding gravel augmentation efforts. The significance of this minor error to SR Plaintiffs’ argument is unclear. The BiOp does not specifically attribute this 40% loss to Project operations. Rather, later in the same paragraph, NMFS explains with specificity the ongoing impact of dam operations:

BiOp at 308.

SR Plaintiffs do not dispute the science underlying this conclusion, nor do they suggest that the impacts of operations, per se, are de minimis. Rather, they argue that the real issue is whether the amount of New Melones water within Reclamation’s discretion is significant enough to cause appreciable harm to CV Steelhead and/or appreciable diminishment of its critical habitat. See Doc. 492 at 4-6. The median historical unimpaired runoff in the Stanislaus River Basin is 1.1 million acre feet per year (“MAFY”). BiOp App. 1 at 69. OID and SSJID are legally entitled to the first 600,000 AF. USBR AR 011751-53. In addition, Reclamation must release between 98,300 and 302,000 AF for fish pursuant to its agreement with CDFG. BiOp App. 1 at 71. Additional releases may be required to meet dissolved oxygen criteria and D-1641. BiOp App. 1 at 72-73, 76-77. 34

*934 Federal Defendants concede that these mandatory delivery requirements do exist, but emphasize that Reclamation nonetheless possesses discretion over how those releases are made. See Doc. 515 at 49. For example, while OID and SSJID have an entitlement to 600,000 AF, past water use data demonstrates that this full amount is not always requested, which in turn changes the amount of water available for other beneficial uses. See Hilts Decl., Doc. 480, Ex. 1 (showing that during the 1987-1992 drought, OID and SSJID never requested full allocation). In addition, Reclamation has the ability to request temporary exemptions from SWRCB conditions such as for dissolved oxygen and Vernalis flow objectives when warranted, and assumptions to reflect this option were inserted into the CalSim II modeling for the Stanislaus River. Fifth Milligan Decl., Doc. 479 at ¶¶ 7-8; Hilts Decl., Doc. 480 at ¶ 12. Defendants do not suggest these are not Home Builders non-discretionary obligations on Reclamation.

SR Plaintiffs argue that NMFS must independently demonstrate that discretionary operations alone satisfy the jeopardy/adverse modification standard. This contention was rejected above. The ESA does not require the agency to segregate discretionary from non-discretionary impacts for the purposes of the effects analysis. (Whether an agency can implement an RPA within its legal authority if an insufficient amount of discretionary water is available is a different question.) Given that there is some discretionary water in the New Melones system and that Reclamation authority over how make discretionary deliveries, is there enough discretionary project water to cause appreciable harm to the species?

The BiOp identifies several negative impacts caused by Project operations, including increasing the likelihood that CV Steelhead will be exposed to unfavorable temperatures at various life stages and, by lowering instream flows, the amount of energy juveniles and smolts must expend to avoid predation is increased. BiOp at 301. SR Plaintiffs do not challenge these underlying findings. The BiOp does not have to demonstrate that these negative effects, alone, satisfy the jeopardy standard by “reducing] appreciably the likelihood of both the survival and recovery of [the] listed species in the wild by reducing the reproduction, numbers, or distribution of that species.” Rather, the jeopardy analysis must determine the overall impact on the species of the entire project, not just the New Melones unit. See 50 C.F.R. § 402.14 (NMFS’s obligation during formal consultation is to determine “whether the action, taken together with cumulative effects, is likely to jeopardize the continued existence of the listed species or result in the destruction or adverse modification of critical habitat”).

SR Plaintiffs’ motion for summary judgment that the BiOp’s effects analysis is unlawful because it does not properly distinguish between baseline effects and effects of the action is DENIED; Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED on this issue.

2. Challenge to Critical Habitat Adverse Modification Finding.

SR Plaintiffs complain that NMFS has not specifically identified how the proposed action will cause “adverse modification” to the steelhead’s critical habitat. Doc. 492 at 7-9. 35

*935 a. Spaumable Area.

SR Plaintiffs first challenge aspects of the BiOp’s treatment of the “spawnable area” aspect of CV Steelhead critical habitat. Doc. 492 at 8. The BiOp found that steelhead spawning habitat would be “maximized” if instream flows were maintained at 200 cfs. However, operations for the protection of fall-run Chinook require higher flow rates may “conflict” with the needs of steelhead. BiOp at 311. SR Plaintiffs argue that habitat “maximization” is not a requirement of the ESA and is not relevant to the effects analysis. Doc. 492 at 8. NMFS does not explain why it set the benchmark for evaluating project impacts at the spawning habitat “maximum.” To ascertain whether project operations will impact the likelihood of CV steelhead survival and recovery, the more appropriate benchmark is that amount of habitat that is “essential” for survival and recovery. The BiOp does not identify the extent of this “essential” habitat or how it relates to the “maximum” habitat. The use of the “maximum” habitat benchmark necessarily resulted in a finding of adverse modification to this aspect of CV habitat. That finding is not justified.

SR Plaintiffs motion for summary judgment that the record does not support NMFS’s findings regarding spawnable area is GRANTED; Federal Defendants’ and Defendant-Intervenors’ cross motions are DENIED.

b. Spawning Gravel Quality and Quantity.

SR Plaintiffs dispute the BiOp’s findings related to “Spawning Gravel Quality and Quantity.” The entire section on this topic provides:

BiOp at 311. SR Plaintiffs contest the BiOp’s reliance on the 2001 Kondolf, et al. study to find that an increase in fine material in spawning beds since the construction of New Melones impairs egg incubation and survival. Id. at 311 . Federal Defendants acknowledge that the loss of gravel recruitment from upstream sources is not the result of the proposed action. See Doc. 515 at 51. However, Federal Defendants argue that “continuing discretionary flow releases eliminates the variability which replenishes spawning riffles,” id., citing page 301 of the BiOp, which discusses how “[t]he dampening of flood events and freshets eliminates the geomorphic processes that are important to CV steelhead to replenish and rejuvenate *936 spawning riffles and to inundate floodplain terraces to provide nutrients and rearing habitat for juvenile salmonids.” Page 301 includes this finding, but the section of the BiOp challenged by SR Plaintiffs, at page 311, specifically discusses gravel recruitment (i.e. the volumes of gravel present), not “riffle rejuvenation.” There is no record evidence that loss of gravel recruitment is an effect of the action. This effect is completely without support and the purported impact of any changed analysis on the overall critical habitat discussion must be addressed by NMFS on remand.

SR Plaintiffs motion for summary judgment that the record does not support NMFS’s finding that New Melones operations effect gravel recruitment is GRANTED; Federal Defendants’ and DefendantIntervenors’ cross motions are DENIED on this issue.

c. Challenge to Temperature Requirements for Spawning Habitat.

SR Plaintiffs also purport to challenge the BiOp’s finding regarding “degradation of rearing habitat conditions,” but actually advance arguments about temperature requirements for spawning habitat. See Doc. 492 at 9. The BiOp explains that “[bjecause CV steelhead are unable to reach their historical spawning areas above Goodwin Dam, they are dependent on East Side Division operations maintaining temperatures suitable for spawning below the dam...” and concludes that appropriate temperature conditions likely cannot be met for April and May for future operations. BiOp at 310. SR Plaintiffs argue “NMFS fails to explain whether or not these ‘temperature conditions’ will be met with the proposed operations or whether there is any evidence that temperatures, as the result of existing operations, have been detrimental to steelhead.” Doc. 492 at 9. This argument entirely ignores the four-and-a-quarter page discussion of temperature at BiOp pages 302 through 306, discussing results of computer modeling showing that project operations will result in temperature exceedances that will have detrimental effects on certain life stages of CV steelhead in the Stanislaus. This challenge is without merit.

SR Plaintiffs motion for summary judgment that the record does not support NMFS’s findings regarding New Melones operations’ impacts on temperature conditions in spawning habitat is DENIED; Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED on this issue.

d. Freshwater Migration Corridors.

SR Plaintiffs challenge the BiOp’s finding that proposed operations will negatively affect upstream and downstream migration corridors. The relevant section of the BiOp provides:

BiOp at 312-13. SR Plaintiffs argue “there is nothing in the AR that indicates that existing operations have negatively affected upstream or downstream migration to begin with, let alone that future operations will ‘continue’ to do so.” Doc. 492 at 9. SR Plaintiffs’ argument continues:

Doc. 492 at 9. The pages cited by SR Plaintiffs are from the section of the BiOp discussing the need for the RPA Actions. SR Plaintiffs do not challenge the clearly explained conclusions of the effects section. Project operations reduce spring pulse flows to levels that are below normal migratory flows, and the flow regime implemented by Reclamation under the action results in channel incision, which reduces connection to floodplain areas necessary for steelhead to rear to large enough size to begin the smolting process. BiOp at 312-13.

SR Plaintiffs motion for summary judgment that the record does not support NMFS’s finding that New Melones operations effect downstream migration corridors is DENIED; Federal Defendants’ and DefendanNIntervenors’ cross motions are GRANTED.

D. Stanislaus River RPA Challenges.

1. Challenge to the Assumptions Used to Model New Melones Project Operations.

SR Plaintiffs’ claim that NMFS used a flawed project description for New Melones and that this “fundamental error renders the entire consultation for the New Melones unit erroneous.” Doc. 454 at 17. This objection concerns the assumptions used to represent New Melones operations in computer modeling.

In 1997, Reclamation and FWS adopted an Interim Plan of Operations to guide the annual operations for New Melones (“NMIPO”). Although the NMIPO was only a two-year plan, it is still used today as an operational guide. USBR AR 0007573-74. Reclamation has deviated from NMIPO in recent years to provide more water to meet State Water Resources Control Board (“SWRCB”) conditions and fulfill CVP Contracts, USBR AR 0007575. The 2008 OCAP BA described *938 the modified operating plan that was the subject of consultation as a “Transitional Operating Plan” (“TOP”). See USBR AR 0007513. The TOP differs from the NMI-PO in several ways, which are described in Table 2-11 of the BA. USBR AR 0007576. SR Plaintiffs note that under the NMIPO, allocations to CVP contractors are capped at 90,000 AF, while under the TPO Reclamation provides for the full 155,000 AF allocation in “high allocation years.” Id.

It is undisputed that NMFS used the TOP as the basis for its effects analysis. See Doc. 477-1 at 119; Doc. 492 at 4. SR Plaintiffs object instead to NMFS’s decision to use the NMIPO assumptions to model RPA options. NMFS elected to use the NMIPO assumptions after it concluded that the TPO would not provide sufficient water for fishery needs in 59% of years. BiOp at 306. The BiOp explained that a 1993 study by Aceituno applied the so-called “instream flow incremental methodology” to the Stanislaus River between Riverbank and Goodwin dam and “determined that 155 TAF was needed to maximize weighted usable habitat area for salmon, not including outmigration flows or fall attraction flows.” Id. The BiOp then determined that the proposed allocation strategy for the East Side Division under the TPO only commits to providing this level of water for fisheries in 41 percent of years (meaning insufficient supplies would be present in 59% percent of years). Id.

SR Plaintiffs assert in a footnote that this conclusion is “bogus” because NMFS did not explain to “[wjhich fisheries” it was referring, nor how much water is “sufficient.” Doc. 492 at 4 n. 4. More importantly, the BiOp nowhere explains why it is “essential” to achieve flows designed to “maximize” steelhead habitat area. Is the status of the species so dire that improvement to 60, 70, 80, or 90% of the “maximum” would be insufficient, even if that marginal difference from the maximum saved large amounts of water? The record provides no explanation of the decision to aim for “maximum” habitat in a system of limited resources. This must be specifically addressed and explained on remand.

NMFS admits that the modeling used to support the RPA builds upon this unexplained decision to set a “maximum habitat” goal. Doc. 477-1 at 119. The agency’s own internal guidance requires an explanation why operating to this goal is “essential.” None is provided. It is impossible to determine how a change in this goal impacts the overall rationale for the RPA. This too must be addressed on remand.

SR Plaintiffs’ motion for summary judgment that Federal Defendants erred by modeling RPA actions based on inappropriate assumptions is GRANTED. Federal Defendants’ and Defendant-Intervenors’ cross motions are DENIED.

2. Do Actions III. 1.2, III.1.3, and IV.1.2 Improperly Require Reclamation to Infringe Upon OID and SSJID’s Prior Right to Stanislaus River Water in violation of 50 C.F.R. § 402.02 ?

“Reasonable and prudent alternatives refer to alternative actions identified during formal consultation [1] that can be implemented in a manner consistent with the intended purpose of the action, [2] that can be implemented consistent with the scope of the Federal agency’s legal authority and jurisdiction, [3] that [are] economically and technologically feasible, and [4] that the Director believes would avoid the likelihood of jeopardizing the continued existence of listed species or resulting in the destruction or adverse modification of critical habitat.” 50 C.F.R. § 402.02 (the “four RPA requirements”).

SR Plaintiffs claim that Actions III.1.2, III.1.3, and IV.1.2 exceed Reclamation’s *939 legal authority because they require Reclamation to infringe upon OID and SSJID’s prior (superior) water rights in the Stanislaus River. It is undisputed that OID and SSJID hold perfected water rights to Stanislaus River water that are senior to Reclamation’s rights to divert from the Stanislaus. OID and SSJID receive water from New Melones under a 1988 Agreement with the United States designed to fulfill their prior rights. USBR AR 0007571-72; USBR AR 0011751. That agreement requires Reclamation to provide to OID and SSJID:

SR Plaintiffs point to studies in the record that they claim indicate the RPAs will require Reclamation to short OID and SSJID 13,000 AF on average. AR 00219154. They maintain that this is actually an under-estimate of the amounts they will be shorted under the Stanislaus River RPA Actions because of certain of NMFS’s modeling assumptions. Specifically, the modeling assumed:

Doc. 492 at 13.

Neither the underlying study purportedly demonstrating that water rights will be shorted, nor the inclusion of the listed modeling assumptions require Reclamation to short senior water rights or demonstrated that it is likely they will be unable to comply with the RPA without doing so. The study cited by SR Plaintiffs was restricted to modeling two years, 2010-2011, when the Phase I requirements of Action IV.2.1 were in place. AR 00219154. Reclamation complied with the RPA during this period and there is no indication that Reclamation shorted senior water rights. More to the point, neither NMFS nor the Bureau has discretion to violate these water rights. It is inappropriate to speculate they will break the law.

As for the modeling assumptions, each is justified based on past practice and experience and has long been included in the CALSIM II modeling process. The Calsim II model inputs do not assume that OID and SSJID’s rights will be shorted. They cannot be. Rather, the RPAs assume that demand from these districts will be reduced under certain circumstances, based upon land use projections developed by the California Department of Planning and Local Assistance. Hilts Decl., Doc. 480 at ¶ 6 (“hydrology-land-use-demand input [data] set ... was best available... [and] suggest that OID and SSJID will not use their full entitlement in most years”); Fifth Milligan Decl., Doc. 479 at ¶ 5 (SR Plaintiffs’ expert Mr. Steiner participated in 2005 update of land use demand assumptions, which were used in the BA and relied upon in the BiOp). In addition to the land-use based assumptions, the relevant modeling included assumptions designed to reasonably reflect water usage *940 by the Stanislaus basin stakeholders during sustained dry periods. Hilts Deck, Doc. 480 at ¶ 10.

SR Plaintiffs cite a number of cases in which mitigation measures were deemed unsatisfactory to satisfy an agency’s burden to insure against jeopardy because those measures were not “reasonably specific, certain to occur, [ ] capable of implementation, [and] subject to deadlines or otherwise-enforceable obligations.... ” Ctr. for Biological Diversity v. Rumsfeld, 198 F.Supp.2d 1139, 1152-54 (D.Ariz.2002). But, the eases cited are distinguishable, and it is unclear whether the “reasonably certain to occur” language should be applied to RPAs. Rumsfeld concerned a biological opinion’s “no jeopardy” finding that relied upon the action agency to mitigate groundwater impacts of its activities through participation in a regional plan to protect groundwater resources, despite the fact that the action agency had no authority to ensure the regional plan was implemented. Nor did the biological opinion set any goals or deadlines regarding groundwater protection. Rumsfeld reasoned that necessary mitigation measures designed to prevent adverse impacts to groundwater must be identified and included either in the proposed action or as RPAs. Id. at 1154 . Without these adjustments there was no rational basis for the “no jeopardy conclusion.” Id. Rumsfeld address the requirements for mitigation measures, not RPAs. See also NWF v. NMFS, 254 F.Supp.2d 1196, 1213-14 (D.Or.2003) (requiring reasonable certainty when NMFS relied upon off-site federal actions to conclude that jeopardy will not occur).

Rumsfeld relied upon Sierra Club v. Marsh, 816 F.2d 1376 (9th Cir.1987), which addressed whether an agency was required to reinitiate formal consultation after failing to acquire certain mitigation lands. Those lands were considered a “vital” RPA by FWS in its biological opinion concerning the agency’s action. Id. at 1378 . Marsh explained that the “reasonably certain to occur” standard applies to “[i]ndirect effects ... caused by the proposed action,” not to RPA actions. See id. at 1388 (citing 50 C.F.R. § 402.02 ). Rather, Marsh applied the regulatory criteria from 50 C.F.R. § 402.16 to determine whether the action agency unlawfully failed to reinitiate consultation. Id. at 1388-89 .

Even if reasonable certainty is the benchmark, it is satisfied here. The RPAs in question here require Reclamation to use its own water resources for particular purposes. Reclamation has reasonably examined past patterns of Project water use by third parties and concluded that water will be available to implement the RPAs. See S.W. Ctr. for Biological Diversity v. U.S. Bureau of Reclamation, 143 F.3d 515 , 518-19 (9th Cir.1998) (upholding generalized RPA requiring agency to protect 1,400 acres without identifying the particular location or timeframe). SR Plaintiffs have not demonstrated that reliance on past practice is unreasonable. If, however, Reclamation’s predictions prove incorrect and make the RPAs’ implementation infeasible, the burden cannot be imposed on senior water rights holders. Rather, Reclamation must then re-initiate consultation.

Federal Defendants have reasonably explained the remaining modeling assumptions about acquisition of waivers from the SWRCB regarding dissolved oxygen and D-1641 flow and salinity requirements. Fifth Milligan Decl., Doc. 497 at ¶ 7-8; Hilts Decl., Doc. 48 at ¶ 12 (explaining it is “reasonable to assume the SWRCB will take a holistic approach and grant such petitions” under relevant conditions). This is speculation and may be mistaken, however the law does not require more. If no Petitions are granted, absent available existing water, NMFS must reinitiate consul *941 tation. SR Plaintiffs have not demonstrated that these assumptions were clearly erroneous.

SR Plaintiffs’ final challenge to the modeling assumptions is based on a March 10, 1982 injunction imposed in United States v. California, purportedly requiring Reclamation to limit non-flood flows to no more than 1,500 cfs. It is undisputed that Action III.1.3 calls for spring pulse flow releases as high as 5,000 cfs, BiOp at 623, Fig. 11-1, and Action IV.1.2 requires releases from New Melones to meet higher Vernalis flow rates, BiOp at 642.

The Ninth Circuit’s March 10, 1982 injunction “pending determination of appeal,” required the United States to provide the State of California with a plan to protect downstream property from damage caused by inundation or seepage. SR Plaintiffs’ Request for Judicial Notice (“SRJN”), Doc. 453-7, Ex. 7 at 2. That plan, set forth in a February 1982 memo drafted by the Bureau, indicated that flows above 1,500 cfs would “create water tables high enough to have the potential to damage the almond and walnut orchards adjacent to the [Stanislaus] river.” Id., Ex. 8, at 1. 36 But, the injunction, by its own terms, was limited to the time period pending appeal. The appeal was decided nine months later on December 20, 1982. United States v. California, 694 F.2d 1171 (9th Cir.1982). The Ninth Circuit’s re manded with instructions that “[t]he injunction previously issued by the court may be modified or amended by the district court as it deems necessary and appropriate in view of this opinion and the present circumstances of the dam and its storage facility,” id., but there is no evidence in the record that the district court ever imposed a similar 1,500 cfs ceiling on non-flood flows. NMFS reasonably concluded that the limitation no longer applies and could be omitted from Stanislaus River modeling. 37 This is not the appropriate forum for SR Plaintiffs to attempt to enforce a nineteen-year-old injunction, which has no continuing validity.

SR Plaintiffs’ motion for summary judgment that the RPA improperly requires Reclamation to infringe on OID and SSJID’s prior rights to Stanislaus River water is DENIED; Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED.

3. Use of the San Joaquin River Temperature Model.

Federal agencies must use “the best scientific and commercial data available” in developing reasonable and prudent alternatives. 16 U.S.C. § 1536 (a)(2); 50 C.F.R. § 402.14 (g)(8). SR Plaintiffs assert that Federal Defendants did not use the best available science in formulating *942 the Stanislaus River RPAs because they did not model the feasibility of the RPAs using the San Joaquin River Water Temperature Model (“SJRWTM”).

There is no dispute that temperature modeling is critical to the management of the Stanislaus River and to implementation of the Stanislaus River RPAs. NMFS relied on Reclamation’s “USBR Temperature Model,” to run an operational scenario involving a draft RPA for new minimum flow releases on the Stanislaus. AR 00105890.

SR Plaintiffs’ expert, Avery Dotan, opines that no reasonably prudent modeler would choose to use the USBR Temperature Model, which can only simulate the mean monthly vertical temperatures, to assess the feasibility of meeting a seven-day average daily maximum temperature requirement, such as Action III.1.2. See Dotan Decl., Doc. 442 at ¶¶ 53-58

The agencies had numerous discussions throughout 2009 regarding temperature modeling, including some specific requests to look into the use of the SJRWTM. See, e.g., AR 00065939 (Feb. 6, 2009 email regarding modeling), 00070965, 00074969 (requesting use of a different model), 00077217 (Feb. 18, 2009 email asking questions about “Derek’s model run”), 00077613 (Mar. 20, 2009 inquiry regarding application of the SJRWTM), 00078887 (Mar. 29, 2009 email containing information about SJRWTM), 00079052 (Mar. 27, 2009 email containing information about SJRWTM), 00085078 (Apr. 10, 2009 email asking for assistance from “Don Smith and Avry Dotan” as the “SJR-Basin wide temperature modelers”). CalFed’s Science Program held a special workshop on temperature modeling for the BO in April 2008 and advised the agencies to utilize the “latest technology” in temperature modeling, including adopting models with “smaller time-steps to better assess biological effects.” AR 00038723. The CalFed Science Review Panel, in reviewing the draft BiOp, specifically recommended that Federal Defendants utilize the SJRWTM, a sub-daily temperature model developed for the Stanislaus River by Avry Dotan and Resources Management Associates. See AR 00219651. Several federal agencies, including NMFS, FWS, and Reclamation, participated in its development. Do-tan Deck, Doc. 442 at ¶¶ 6, 23-34. The model was funded by CalFed and peer reviewed by CalFed scientists. Id. at ¶¶ 24, 21, 26 .

The SJRWTM could have modeled temperatures on a seven-day average daily maximum basis, a more appropriate time scale according to Mr. Dotan. Id. at ¶¶ 53, 80 . SR Plaintiffs assert that the SJRWTM was the best available science and should have been used to evaluate the feasibility of the RPA actions.

The model runs in the AR using the USBR Temperature Model predict that the new flow requirements in Action III.1.3 will occasionally cause temperatures to exceed the objectives set forth in Action III.1.2. Dotan Deck, Doc. 442 at ¶¶ 73-77. Mr. Dotan opines that these results are unreliable because the model could only predict monthly mean temperatures and was not capable of determining when the seven-day average daily maximum temperature was or was not met. Id. at ¶¶ 52-72 . To demonstrate that this error is material, Mr. Dotan repeated the analysis using the SJRWTM. The results of this analysis are depicted in Figure 7 to his Declaration, which shows that in February, March, April, May, June, July, August, and September the Bureau’s model estimates fewer exceedances than does the SJRWTM. Doc. 441-15. (Mr. Dotan does not discuss the fact that this figure also shows that in October and November, the SJRWTM indicates fewer exceedances than the Bureau’s model. Id.)

*943 The SJRWTM model also predicts that the water cost associated with meeting Action III.1.2 vary between 22,000-190,000 AF per year with an average cost of 84,000 AF, Dotan Decl., Doc. 442 at ¶ 87; that operating for temperature control will deplete the volume of water in New Melones by as much as 717,000 AF during 1987-1995, id. at ¶¶ 86 ; and that this successive operation for temperature control will eventually cease to be effective as New Melones’ pool of cold water is depleted, id. at ¶¶ 89 .

Federal Defendants do not dispute the superiority of the SJRWTM. Rather, they strenuously object that contemporaneous documents in NMFS administrative record demonstrate that the model was not “available” to the agencies during the consultation. Doc. 477-1 at 131. Although employees of the Federal Defendants were trained to use the model between 2001 and 2009, Dotan Decl., Doc. 442 at ¶¶ 35-42, and Mr. Dotan answered specific questions posed by NMFS and Reclamation regarding the use of model, see, e.g., NAR 00093319 & 00094138, there were concerns that documentation of the complex model was insufficient to allow others to run it, AR 00089101 (May 1, 2009 email indicating “NMFS has the model” but discussing problems with the contract for technical support); AR 00089027 (model in public domain but difficult to run unassisted). The BiOp explains why it did not use the SJRWTM:

AR 00105884.

Record evidence demonstrates that the model was not self-explanatory, even for staff with background in a related model used as the basis for the SJRWTM: HEC-5Q. NMFS had to seek outside help to use the model, and encountered implementation issues. See AR 00077320 (NMFS discussing hiring Tetra Tech to help run SJRWTM), 00092267 (NMFS seeking assistance from Bureau engineer with' model), 00093101 (NMFS obtaining promise of documentation from FWS), 00093319 (Dotan answering NMFS’s questions about application of model), 00093538 (discussing “numerous issues (above and beyond downscaling) with bringing CalSim data into the SJ temp model”), 00094138 (Dotan answering NMFS’s questions about application of model). Once NMFS engaged a consultant to run the modeling, the model’s other co-developer, Don Smith, resisted providing the assistance necessary to run the model. See AR 00089096. In addition, the material provided with the then-available version of the SJRWTM did not allow the model to be correctly utilized. Reed Decl., Doc. 482 at ¶¶ 10-13.

SR Plaintiffs concede that Federal Defendants are only required to use the best science available, and not the best science possible. See Doc. 492 at 17 (citing R.W. Ctr. for Biological Diversity v. Babbitt, 215 F.3d 58, 60-61 (D.C.Cir.2000)). How *944 ever, SR Plaintiffs assert it “stretches credulity” to accept that the Government was unable to run this model because:

Id. This argument continues:

Id. at 17-18.

Ms. Reed, an NMFS employee with significant involvement in the development of the BiOp disagrees with Plaintiffs’ assessment of the circumstances. She declares that the SJRWTM was not functionally available during the consultation:

Reed Decl., Doc. 482.

NMFS claims it did not have the expertise and could not get Dotan to respond. This is a factual dispute over whether NMFS could use the model. There is no dispute using the shorter time step was the best science. The Supreme Court has “repeated time and again, an agency has broad discretion to choose how best to marshal its limited resources and personnel to carry out its delegated responsibilities.” Massachusetts v. EPA, 549 U.S. 497, 527 , 127 S.Ct. 1438 , 167 L.Ed.2d 248 (2007). Although the record suggest that the resources required to run this model properly would be modest, that the model results would be preferable to those presented in the BiOp, and that NMFS had knowledge of the model for over 8 years, a court does not have the authority to order the agency how to direct and allocate its resources. Congress has chosen to partially immunize such agency “mis-performance.” SR Plaintiffs’ motion for summary judgment that Federal Defendants acted unlawfully by failing to utilize the SJRWTM is DENIED; Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED.

4. Exceptions Built into Action III. 1.2.

Even if, arguendo, Federal Defendants’ failure to employ the SJRWTM was unlawful, Federal Defendants alternatively argue that that exceptions built into Action III.1.2 render any dispute over the model used irrelevant. Action III.1.2 was developed to address the impacts of adverse temperatures on the species. The temperature compliance schedule is purportedly based on the species’ biological and physiological needs. Reed Deck, Doc. 482 at 3-4. Because the modeling indicated these temperatures could not always achieved, the RPA action has a built-in exception, which can be exercised any time temperature requirements of Action III.1.2 will be exceeded on a three-day average daily maximum temperature. BiOp at 621. Operational adjustments to address such exceptions will be coordinatthrough the SOG and WOMT. Id. NMFS concluded:

Reed Decl., Doc. 482 at ¶ 29. Because the exception provision “has no limitations,” Federal Defendants argue it is immaterial whether the SJRWTM would have shown more instances of exceeding the RPA Action’s seven-day average daily maximum temperatures. Doc. 477-1 at 133.

This argument presents a conflict between the adaptive management scheme and the ESA Regulations’ explicit demand that Federal Defendants demonstrate the necessity and feasibility of implementing every RPA Action. Flexibility is the essence of adaptive management, a tool that is indisputably beneficial both to the species and impacted stakeholders. But, Federal Defendants describe an exception that “has no limitations.” 38 How often can the exception be triggered without rendering the Action ineffectual? This is not examined. Without such an analysis, the extent *948 to which this RPA is essential” to avoiding jeopardy cannot be evaluated. This makes the RPA unlawful and it must be addressed on remand.

5. Does the Record Support the Finding that Action III.l.S Will Avoid Jeopardy to or Adverse Modification of CV Steelhead or Critical Habitat?

The objective of Action III.1.3 is to operate the East Side Division 39 dams to optimize CV steelhead habitat for all life history stages and to incorporate habitat maintaining geomorphic flows in a flow pattern that will provide migratory cues to smolts and facilitate out-migrant smolt movement on [the] declining limb of pulse.” BiOp at 622. Specifically, the Action requires the Bureau to achieve a minimum flow schedule prescribed in Appendix 2-E and generally described in Figure 11-1, copied below:

[[Image here]]

Id. at 623 . SR Plaintiffs take issue with the requirement of releases as high as 5,000 cfs in the spring of wet years, which represents a drastic change from the prior flow regime and such high flows are unjustified.

NMFS explains in a May 31, 2009 memo from Rhonda Reed to Maria Rea that the flow requirements of Action III.1.3 are based on a 1993 study by Aceituno, which uses “In-stream Flow Incremental Methodology” (“IFIM”). AR 00105879. NMFS then conferred with CDFG and *949 FWS biologists regarding CV Steelhead’s need for pulse flows. AR 00105881-82. These consultations revealed that a fall attraction pulse was needed. This is included in Action III.1.3 and is not challenged by SR Plaintiffs.

NMFS also assessed whether CV steel-head needed a spring pulse flow:

AR 00105882. SR Plaintiffs object that this language is equivocal and that a life stage of the species “may” need a particular pulse flow is not sufficient justification for requiring one.

Although this passage from the Reed Memo does admit that “early smolts may not need a spring pulse,” AR 00105882, Defendants point to other record evidence supporting the imposition of a spring pulse flow requirement. Spring pulse flows cue more smolts to migrate, protecting the anadromous form. BiOp at 306-307; AR 00105882 (variability in flow triggers important to anadromy), AR 00105883 (flow variability important to anadromy).

SR Plaintiffs do not directly challenge this rationale. Rather, they argue that even if some form of spring pulse is justified, nothing in the BiOp justifies a 5,000 cfs pulse flow. SR Plaintiffs point out that Aceituno’s 1993 IFIM study called for flows ranging from between 50-500 cfs. Aceituno’s study focused on instream needs, and did not include an assessment of water needed for spring pulse flows to convey steelhead to delta. BiOp at 307 (“IFIM analysis did not include an assessment of the volume of water needed for a spring pulse flow to convey CV steelhead or fall run from the Stanislaus River into the Delta”); AR 00107828 (Aceituno (1993) explicitly acknowledging that “[t]his study did not directly provide information on flows needed for smolt emigration in the spring”).

CDFG’s initial draft recommendation for the RPA Action called for a spring pulse flow of 3,500 cfs. AR 00105882. CDFG’s highest recommendation was for a pulse of 4,000 cfs. AR 00061652. NMFS raised the pulse to 5,000, reasoning that this would provide “minimum channel forming flows.” AR 00105887. In support of providing such “channel forming flows,” NMFS cites Kondolf (2001), which provides an analysis of pre- and post-New Melones flood frequency rates at Knights Ferry on the Stanislaus River. Kondolf (2001) concludes: “flows in excess of 5,000 to 8,000 cfs are needed to mobilize the bed and thereby maintain channel form and gravel quality.” AR 00122645. Such flows are “important to rejuvenate spawning beds and floodplain rearing habitat and to recruit allochthonous nutrients and large wood into the river.” BiOp at 308.

According to Kondolf (2001), channel-forming flows occurred every 1.4-1.8 years prior to the construction of New Melones, but only once every 5 to 20 years since construction of New Melones. Id. Kondolf further explains:

AR 00122626. Kondolf (2001) then evaluated the post-dam flood frequency, and concluded that the two-year return flow is 3,070 cfs, meaning that such a flow returns every two years. AR 00122714. A 5,000 cfs flow has a return rate of just over three years. Id. Kondolf (2001) supports a regime that would provide for high pulse flows to maintain gravel quality. In general, this is what Action III.1.3 attempts to achieve.

However, in light of Kondolf (2001)’s conclusion that “flows in excess of 5,000 to 8,000 cfs are needed to mobilize the bed and thereby maintain channel form and gravel quality,” SR Plaintiffs challenge whether RPA Action III.1.3, which calls for peak flows of 3,000 cfs in above normal years and 5,000 in wet years, would maintain channel form and gravel quality. Kondolf (2001) provides the only record support for flows above the 3,000 suggested by CDFG, yet Action III.1.3 does not actually implement the flow regime suggested by Kondolf (2001). The record provides no support for the conclusion that the regime imposed by Action III.1.3 is sufficient to maintain gravel quality. Particularly in light of the potentially high water costs of these pulse flows, the rationale for Action III.1.3 must be lawfully explained and justified on remand.

SR Plaintiffs’ motion for summary judgment that the record does not support the imposition of Action III.1.3’s 5000 cfs spring pulse flow is GRANTED; Federal Defendants’ and Defendant-Intervenors’ cross motions are DENIED.

6. DFG Salmon Population Model a. Use of the Model to Set Out-Migration Flows.

In calculating the flows CV Steel-head need for outmigration, NMFS relied upon the “SJR salmon model (V.1.0) (output for doubling salmon and calculating the Stanislaus flow contribution....)” AR 00105883. It is undisputed that this is a reference to a CDFG model used to determine flows needed to double salmon in the San Joaquin River. SR Plaintiffs complain that NMFS’s use of the model was inappropriate because: (1) steelhead are not salmon; and (2) the “doubling” goal is distinct from the goal of “avoiding jeopardy.” Doc. 454 at 29.

The former argument is identical to the surrogate challenges raised by Plaintiffs and rejected above. Salmon are the best available surrogates for CV steelhead, for which the available data is inadequate for modeling purposes.

Nothing in the record explains why it is appropriate to use a model designed to double the existing salmon population to set numeric flow targets to avoid jeopardy to the CV steelhead. The BiOp must explain why each aspect of the RPA is essential to avoid jeopardy or adverse modification. The facial disconnect between the goal of the salmon-doubling model and the goal of ESA section 7 consultation requires explanation on remand.

SR Plaintiffs’ motion for summary judgment that the BiOp unlawfully utilized the CDFG salmon doubling model is GRANTED IN PART AND DENIED IN PART; and Federal Defendants’ and Defendant Intervenors’ cross motions are GRANTED IN PART AND DENIED IN PART. That the model used salmon as a surrogate for CV Steelhead was not inappropriate, but the record does not support the use of a model designed to double salmon to set flow targets to avoid jeopardy.

b. NMFS’s Reliance on Draft Model Runs from Outdated Version 1.

The record further suggests that NMFS relied on runs from an outdated version of the CDFG Salmon Model. When it pro *951 duced the data to NMFS, CDFG explained that the results were preliminary and based upon version 1.0, AR 00061644, which was subsequently subject to peer review and further clarification, AR 00103255-58. CDFG specifically warned NMFS that the results would need to be confirmed through the performance of several checks. AR 00061644. The record reveals no evidence that such corrections were made. The need for confirmation must be addressed on remand.

SR Plaintiffs’ motion for summary judgment that the BiOp unlawfully and unreasonably relied upon an outdated version of the CDFG salmon model is GRANTED; Federal Defendants’ and Defendant-Intervenors’ cross motions are DENIED.

7. SR Plaintiffs’ “Impermissible Major Changes’’ Argument

SR Plaintiffs originally advanced the argument that the Stanislaus River RPAs were unlawful because they constituted “impermissible major changes” to the New Melones Project. This argument was based on SR Plaintiffs citation to 50 C.F.R. § 402.14 (i)(2) which provides that “[r]easonable and prudent measures, along with the terms and conditions that implement them, cannot alter the basic design, location, scope, duration, or timing of the action and may involve only minor changes.” However, reasonable and prudent measures (“RPMs”) are those measures “necessary or appropriate to minimize [the] impact” of incidental take. 50 C.F.R. § 402.02 . No RPMs are imposed upon Stanislaus River operations. There is no “impermissible major changes” language associated with the imposition of RPAs. SR Plaintiffs’ motion for summary judgment on this ground is DENIED; Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED. 40

8. Challenges to the BiOp’s Feasibility Analyses?

SR Plaintiffs argue that it is impossible to determine from the record whether certain of the Stanislaus River RPA Actions are feasible.

a. General Objection that Feasibility Modeling Employed Erroneous Assumptions.

First, SR Plaintiffs argue that the feasibility modeling employed erroneous assumptions, such as assumptions that constrain allocations to OID and SSJID below their entitlements. Doc. 492 at 12-13. This objection has been rejected, as the modeling captures the actual operation of these districts with reasonable accuracy.

SR Plaintiffs’ motion for summary judgment that the BiOp’s feasibility modeling employed erroneous assumptions is DENIED; Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED.

b. Objection that Action III.1.2’s Exception Procedure Renders the BiOp’s Feasibility Analysis of that Action Arbitrary and Capricious.

Second, SR Plaintiffs argue that the RPA’s exception procedures render any feasibility analysis irrational. Id. at 13 . SR Plaintiffs do not specifically identify which Action, III.1.2, III.1.3 or IV.2.1, they assert is rendered infeasible by its exception procedures; but they must be refer *952 ring to Action III.1.2, which contains the broad exception procedure discussed above, see BiOp at 621. Action III. 1.3 contains no exception procedure, BiOp at 622-26, and Action IV.2.1’s exception procedure is narrowly limited, BiOp at 644. SR Plaintiffs succeeded on their argument that Action III.1.2’s exception procedure is so broad that it has the potential, without further refinement, to render the RPAs ineffectual. Relatedly, an exception procedure without any guarantees as to whether the exception may be successfully invoked when necessary renders any feasibility analysis impossible. Although Federal Defendants’ feasibility analysis need not be perfect, it must be rational. Federal Defendants must reconsider their approach to the feasibility analysis in light of the numerous problems with the exception process identified above.

SR Plaintiffs’ motion for summary judgment that the BiOp’s feasibility analysis for III.1.2 is arbitrary and capricious is GRANTED; Federal Defendants’ and Defendanb-Intervenors’ cross motions are DENIED.

c. Feasibility of Action III.2.2.

SR Plaintiffs also challenge the feasibility of Action III.2.2, which calls for Reclamation to confer with the SOG to develop an operational strategy to meet the purpose of achieving floodplain inundation flows on a one to three year schedule. 41 See Doe. 454 at 32. As this RPA defines no action per se, it is impossible to perform a feasibility analysis of it. Federal Defendants cannot escape the requirement of a feasibility analysis simply because they delay the design of this RPA. Before implementation, Federal Defendants must ensure that any action implemented under RPA Action III.2.2 complies with the requirements of law.

SR Plaintiffs’ challenge to this feasibility analysis is correct to the extent there is no validly formulated RPA Action.

9. Are Actions III.1.3, III.2.2 Consistent with the Purposes of the Project?

SR Plaintiffs also argue that implementation of Actions III.1.3 and III.2.2 conflict with one of the express project purposes of New Melones, namely flood control, in violation of 50 C.F.R. § 402 .02’s requirement that an RPA be “consistent with the intended purpose of the action.” See Doc. 454 at 32. As to Action III.2.2, which calls for a plan to provide flows large enough to “inundate floodplains” in the winter or spring, no action has yet been defined. SR Plaintiffs’ challenge to this feasibility analysis is valid to the extent there is not yet a validly formulated RPA Action.

Action III.1.3 imposes certain pulse flows to benefit CV Steelhead, including the 5,000 cfs pulse flows in wet years discussed above. SR Plaintiffs suggest that these pulse flows, designed to be “channel forming,” will conflict with New Melones’ flood control purpose. Doc. 454 at 32. However, the BiOp specifically explains that Action III.1.3 is to be implemented for ten days or less in order to limit seepage impacts to nearby landowners. BiOp at 624. SR Plaintiffs fail to *953 acknowledge the short duration of the pulse flows, nor do they otherwise explain how flows of this magnitude and limited duration conflict with the flood control purpose of New Melones.

SR Plaintiffs’ motion for summary judgment that RPA Action III.1.3 conflicts with the flood control purpose of the New Melones Dam is DENIED; Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED.

10. Waste and Unreasonable Use of Water (California Constihition Article X, Section 2).

Finally, the SR Plaintiffs argue that implementation of the Stanislaus River RPAs would require water waste and unreasonable use in violation of Article X, Section 2 of the California Constitution. The Bureau must comply with non-conflicting state water law. Reclamation Act of 1902, Pub.L. No. 57-161, 32 Stat. 288, at § 8 (June 17, 1902); California v. United States, 438 U.S. 645, 675 , 98 S.Ct. 2985 , 57 L.Ed.2d 1018 (1978).

The California Constitution states that the right to water is limited to reasonable use, and does not extend to waste or unreasonable use:

Cal. Const. art. X, § 2.

SR Plaintiffs concede that release of water for fish is a beneficial use of water in California. However, they argue that the spirit of Article X, Section 2 dictate that any releases must be carefully tailored to “just what is needed to avoid jeopardy so that the remaining water can be reserved for other equally important beneficial uses.” Doc. 454 at 35. SR Plaintiffs then argue that Actions III.1.2, III.1.3, III.2.2 and IV.2.1 violate Article X, Section 2 “absent record evidence to support a finding that these RPAs-use only as much water as is reasonable and necessary to avoid jeopardy.” Id. SR Plaintiffs underestimate the complexity of the waste and unreasonable use standard and the process by which they must establish waste and unreasonable use is occurring.

SR Plaintiffs cite no caselaw to support their assertion that the California Constitution’s reasonable use doctrine demands that an RPA “be carefully tailored to just what is necessary to avoid jeopardy.” To the contrary, the reasonable use doctrine protects a broad range of interests, including fish protection interests that go far beyond prevention of jeopardy. See, e.g., Nat’l Audubon Soc’y v. Superior Court, 33 Cal.3d 419, 443 , 189 Cal.Rptr. 346 , 658 P.2d 709 (1983) (use of water to maintain scenic and recreational values consistent with the reasonable use doctrine).

Nor do Federal Defendants bear any burden to affirmatively demonstrate that the RPA’s comply with the California Constitution. The ESA’s implementing regulations specifically enumerate in 50 C.F.R. § 402.02 the analyses NMFS and the Bureau must undertake when promulgating an RPA. It is Plaintiffs who bear the burden in a challenge based upon Article X. State Water Resources Control Board Cases, 136 Cal.App.4th 674, 762 , 39 Cal.Rptr.3d 189 (2006) (rejecting allegation that releases of water pursuant to D-1641 constituted waste and unreasonable use because the plaintiffs failed to demonstrate that the releases “necessarily result[] in an unreasonable use of water.”). The reasonableness of a use of water is a question of fact that depends on the particular cir *954 cumstanees of each case. Id. Any such claim arises under state law, not the APA, and is not limited to the administrative record. The briefing in this case has not addressed in any analytic respect the unreasonable use issue.

SR Plaintiffs motion for summary judgment that the Stanislaus River RPA Actions violate Article X, Section 2 of the California Constitution is DENIED WITHOUT PREJUDICE; Federal Defendants’ and Defendant-Intervenors’ cross motions are PREMATURE.

VII. RECLAMATION’S LIABILITY UNDER THE ESA.

All Plaintiffs move for summary judgment that Reclamation violated the ESA by adopting and implementing the BiOp. Following the issuance of a biological opinion, the ESA regulations require the action agency, here, Reclamation, to “determine whether and in what manner to proceed with the action in light of its section 7 obligations and the Service’s biological opinion.” 50 C.F.R. § 402.15 (a). In making that determination, a federal action agency “may not rely solely on a[ ] biological opinion to establish conclusively its compliance with its substantive obligations under section 7(a)(2).” Pyramid Lake Paiute Tribe of Indians v. U.S. Dept. of Navy, 898 F.2d 1410, 1415 (9th Cir. 1990). In City of Tacoma v. Fed. Energy Regulatory Comm’n, 460 F.3d 53, 76 (D.C.Cir.2006), the D.C. Circuit summarized the caselaw culminating in Pyramid Lake:

City of Tacoma, 460 F.3d at 75-76 . The D.C. Circuit rejected the City of Tacoma’s claim that the consultant agency in that case, FERC, was liable under the ESA because the City had not “presented FERC with new information that was unavailable to [NMFS] or [FWS] and that would give FERC a basis for doubting the expert conclusions in the BiOps those agencies prepared.” Id. at 76 .

Reclamation clearly disagreed with NMFS’s approach to many important elements of the BiOp’s analysis. See Doc. 431 at 119 (Plaintiffs’ opening brief citing pages in the record containing Reclamation’s critiques of the BiOp). This is not alone the litmus test for Reclamation’s liability. In the context of ESA consultation, Reclamation is the regulated party and will not necessarily agree with every aspect of NMFS’s opinion on the impacts of Reclamation’s project on Listed Species. Under City of Tacoma, Plaintiffs must demonstrate that, at the time it adopted the BiOp’s RPA, Reclamation was in possession of any “new information” not considered by NMFS that provided Reclamation a basis for questioning the BiOp’s expert conclusions. They have not. Absent such a showing, even though the BiOp is flawed in many ways, Reclamation could rely upon it without incurring ESA liability.

All Plaintiffs motions for summary judgment that Reclamation violated the ESA and/or the APA are DENIED; Federal Defendant and Defendant-Intervenors’ cross motions are GRANTED.

VIII. CONCLUSION

For all the reasons set forth above:

(A) Plaintiffs’ and DWR’s motions for summary judgment that the BiOp violates the ESA and the APA are GRANTED IN PART AND DENIED IN PART; and Federal Defendants’ and DefendanNIntervenors’ cross motions are GRANTED IN PART AND DENIED IN PART based on the following findings:

(B) Stanislaus River Plaintiffs’ motion for summary judgment that the BiOp violates the ESA and the APA is GRANTED IN PART AND DENIED IN PART; and Federal Defendants’ and Defendant-Intervenors’ cross motions are GRANTED IN PART AND DENIED IN PART based on the following findings:

(C) All Plaintiffs’ motions for summary judgment that Reclamation violated the ESA and/or the APA are DENIED; Federal Defendant and Defendant-Intervenors’ cross motions are GRANTED.

It is undisputed that the law entitles the winter-run and spring-run Chinook, CV steelhead, Southern DPS of green sturgeon, and Southern Resident killer whales to ESA protection. Plaintiffs have succeeded on some of their challenges to the BiOp’s justifications and analyses of Delta and Stanislaus River operations. The BiOp discusses and prescribes RPAs to address many other sources of harm, including adverse temperature conditions and blockages caused by dams on the Sacramento River. The BiOp’s jeopardy conclusion is lawful. Project operations negatively impact the Listed Species and adversely modify their critical habitat in various ways that remain incompletely described and quantified.

Some of NMFS’s analyses rely upon equivocal or bad science to impose RPA Actions without clearly explaining or otherwise demonstrating why the specific measures imposed are essential to avoid jeopardy and/or adverse modification. Given the potential serious impacts of these measures, the agency must do more to comply with the law.

The 2009 Salmonid BiOp and its RPA are ARBITRARY, CAPRICIOUS, and UNLAWFUL, and are REMANDED to NMFS for further consideration in accordance with this decision and the requirements of law.

Plaintiffs shall submit a form of order consistent with this memorandum decision within five (5) days of electronic service.

Within five (5) days of service of this decision, Federal Defendants shall provide a proposed date by which they shall file the new BiOp and any RPA.

SO ORDERED

. Export Plaintiffs previously prevailed on their claims that the Bureau of Reclamation violated the National Environmental Policy Act ("NEPA”) by failing to perform any NEPA analysis prior to provisionally adopting and implementing the BiOp and its RPA. Doc. 288.

. More recent population figures were presented during hearings on motions for injunc *814 tive relief; only data available at the time the BiOp was issued has been considered.

. The term "species'' includes "any subspecies of fish or wildlife or plants, and any distinct population segment of any species of vertebrate fish or wildlife which interbreeds when mature." 16 U.S.C. § 1532 . NMFS has issued guidance on how to apply the ESA’s DPS concept, see Modesto Irr. Dist. v. Gutierrez, 619 F.3d 1024, 1028 (9th Cir.2010) (citing 56 Fed.Reg. 58,612 (Nov. 20, 1991)), and has "struggled for two decades over how to apply the term DPS to steelhead,” id.

. Fall-run Chinook salmon are not listed as threatened or endangered under the ESA.

. Generally, where the listed species in question is marine or anadromous, consultation must involve NMFS. For terrestrial and freshwater species, the United States Fish and Wildlife Service ("FWS”) must be consulted.

. Export Plaintiffs repeatedly rely on Tucson to argue that NMFS erred by relying on "ambiguous studies” as affirmative proof of scientific fact. Federal Defendants suggest that Export Plaintiffs' reading of this holding is incorrect, and emphasize that the Ninth Circuit re-affirmed the general rule that "when examining decisions made under conditions of scientific uncertainty 'a reviewing court must be at its most deferential.' ” Tucson, 566 F.3d at 879 . Federal Defendants suggest that the holding in Tucson resulted from the special circumstances in that case, where FWS relied on a single study to affirmatively conclude that a species persisted in a significant portion of its range, even though that one study only addressed two discrete sections of the species’ current range. Doc. 484 at 34 (citing Tucson, 566 F.3d at 882 ). This is a distinction without a difference. Tucson stands generally for the proposition that, while a court must be deferential in areas where there is scientific uncertainty, such deference is not unlimited. More specifically, an agency may not rely on an ambiguous study for affirmative proof of something the study does not establish.

. The State and Federal pumping facilities use louvers to divert salmonids entrained by *825 the pumping process into collection tanks where operators attempt to "salvage” them by returning them to other areas of the Delta, BiOp at 341, 345.

. ESA procedural errors are subject to a harmless error analysis. See Idaho Farm Bureau Fed'n v. Babbitt, 58 F.3d 1392 , 1405 (9th Cir.1995).

. The term "undipped'' refers to wild fish with intact adipose fins, while "dipped” fish have had their adipose fins dipped before release from a hatchery. BiOp at 337

. Plaintiffs also point to several documents in the administrative record where scientific experts recommended that NMFS use a quantitative life cycle model. See, e.g., AR 00108170 (CALFED Science Panel commissioned by NMFS to review a draft version of the BiOp noted that the “lack of quantitative modeling ... and lack of integrative life-cycle modeling for salmon” was "[o]ne of the most important limitations to the analyses used in the draft [BiOp]”). However, these recommendations are not dispositive of whether the models discussed were "available.”

. All references to the NMFS administrative record are noted as "AR” references. References to the separate Reclamation administrative record will be noted as “USBR AR.”

. NMFS also claims it did not have the expertise to make use of such a model. Plaintiffs rejoin that any lack of modeling resources was manufactured by NMFS, "which could easily have made modeling resources available to the BiOp team,” pointing to various individuals within NMFS who could have *841 done the work. Doc. 487 at 17-18. This debate raises difficult issues. On the one hand, an agency cannot be permitted to ignore the best available science simply because it refuses to assign to the task personnel with the expertise to understand and apply that science. On the other hand, NMFS has limited resources, and a court cannot instruct an agency how to allocate those assets. While it may be reasonable to demand that NMFS assign to the BiOp team individuals who can correctly apply readily available statistical methodologies, at some point Plaintiffs’ demands that NMFS assign specific experts on its staff to fine tune the application of OBAN to the purposes of the BiOp becomes a demand that NMFS develop new science. Where the line between these two extremes should be drawn is not clear. It is relevant that no outside expert (government, academic, or consultant) had applied OBAN in the manner Plaintiffs demand prior to the issuance of the BiOp. Plaintiffs were free to submit their own studies and results for NMFS's consideration. They did not.

. In their opening brief, Plaintiffs also argued that the conceptual model applied in the BiOp was biologically implausible because it relied on a particular study, Naiman & Turner (2000), for the proposition that it is possible to drive population to extinction by only slight changes in survival at each life history stage. Doc. 431 at 27-31. Plaintiffs complained that NMFS was misusing a "thought experiment” from that study as though it were scientific fact. Id. Federal Defendants respond that “NMFS did not rely upon the Naiman and Turner conceptual model to conclude that any one project effect causing small reductions in one life stage would jeopardize the species. The Naiman and Turner model was used in evaluating how slight incremental changes in life history stages affect already-diminished populations, as such changes are difficult to quantify and may take years to resolve.” Doc. 477 at 28. Plaintiffs appear to have abandoned this argument.

. The BiOp explains why the PDO is not necessarily the ideal measurement of ocean conditions off the California coast. While general ecosystem indicators, like the PDO, can be used to provide an understanding of ocean conditions, and their relative impact on marine survival of juvenile salmon, throughout the broader, northern portion of the California Current... they may not provide an accurate assessment of the conditions observed on a more local scale off the California coast.” BiOp at 151. Instead, the BiOp examined available data using the Wells index, which does provide specific information about conditions off the California coast. Id.

. The AR contains a portion of the FCRPS BiOp. The complete BiOp is available at http://www.nwr. noaa.gov/Salmon-Hydropower/ Columbia-Snake-Basin/final-BOs.cfm.

. Census size N can be used if direct estimates of effective size Ne are not available, assuming Ng/N 0.2.

. Decline within last two generations to annual run size £ 500 spawners, or run size > 500 but declining at > 10% per year. Historically small but stable population not included.

. Run size has declined to < 500, but now stable.

. Catastrophes occurring within the last 10 years.

. Decline < 90% but biologically significant.

. See Figure 1 for assessing hatchery impacts.

. Plaintiffs invoke Lindley (2009) to argue that the impacts to the species in the freshwater phase during recent years were inconsequential in comparison to the impacts resulting from poor ocean conditions. This argument fails for the reasons discussed above. Lindley recognized that the period of deteriorated ocean conditions, which were a major short-term cause of population decline, acted in conjunction with a long-term steady degradation of the freshwater environment leaving Chinook vulnerable to other stressors. See BiOp at 149; AR 00123517.

. Federal Defendants attempt to defend this analysis by asserting that the winter-run “population trend has been consistently negative for several decades.” Doc. 477-1 at 50. Plaintiffs point out that Federal Defendants rely on a comparison of 2008 figures to 1969 figures to reach this conclusion. Doc. 487 at 47. Lindley (2007) states that "[p]opulation growth (or decline) [] is estimated from the slope of the natural logarithm of spawners versus time for the most recent 10 years of spawner count data.” AR 00123481. In fact, when Lindley applied this standard to the most recent 10 years of data available at the time of publication, the population showed growth not decline. AR 00123486. Defendants do not explain this inconsistency.

. NMFS’s and FWS’s joint Consultation Handbook "provides internal guidance and *868 establishes national policy for conducting consultation and conferences pursuant to Section 7 of the Endangered Species Act of 1973, as amended.” AR 00217635.

. This finding applies with equal force to the analysis of the causes of indirect mortality discussed in the Consolidated Delta Smelt Cases' December 14, 2010 Summary Judgment Decision, namely the negative influence of Project operations on delta smelt food supply and the exacerbation of the impacts of pollution and contaminants by Project operations. Although the reasonably certain to occur standard was applied in that case, the link between Project operations and these purported sources of indirect mortality were not clearly articulated in the BiOp or justified by record evidence, so the application of the incorrect standard did not make a material difference.

. Plaintiffs also argue that NMFS must affirmatively acknowledge its own regulatory standard in the BiOp, presumably by making direct reference to the relevant regulatory language. Doc. 431 at 67. Although a court "cannot infer an agency's reasoning from mere silence,” PCFFA v. U.S. Bureau of Reclamation, 426 F.3d 1082, 1091 (9th Cir.2005), so long as the record, as evidenced by the agency's reasoning in the BiOp, supports a finding that Project operations reasonably would be expected to cause indirect mortality, the ESA does not require NMFS to use “magic words” in a biological opinion. An agency's rationale must be upheld if it can "reasonably be discerned.” See Modesto Irr. Dist. v. Gutierrez, 619 F.3d 1024, 1035 (9th Cir. 2010).

. PCEs are those elements of a critical habitat designation deemed essential for the conservation of the listed species and are described as the sites and habitat components that support one or more life stages or requirements of the species. PCEs are made up of essential features, which are needed to support that specific life-stage requirement. An example is the PCE of spawning habitat, which includes such essential features as clean spawning gravel, clean water, and appropriate water temperatures. See BiOp at 56.

. The pages in this document appear to be out of order in the AR. What appears to be page 3, AR 00005208, is before what appears to be page 2, AR 00005209.

. It is unclear whether Dr. Cummins is correct to assert that validation is standard practice in the field. Garwin Yip opines that his review of tagging studies in the Pacific Northwest reveals that the surrogate validation process is not typically used due to increased time and funding required to complete the validation process. Third Yip Decl., Doc. 518 at ¶ 16.

. VAMP is a multi-agency collaborative effort that is part of the San Joaquin River Agreement (“SJRA”). "SJRA is a negotiated settlement agreement between SJR water suppliers, water purveyors, and both State and Federal Fishery Agencies that calls for specific spring South Delta (e.g. SJR at Vernalis) river flows and Delta export pumping rates. The San Joaquin River Group Authority provides the flows necessary to attain the Vernal-is flow objectives. State and Federal agencies ensure that Delta exports rates are met. [VAMP] is a scientific study that evaluates the effects of Delta inflow, and outflow, upon fall-run Chinook salmon smolt survival.” AR 00212419.

. Contrary to Export Plaintiffs’ assertion that ocean conditions are the primary driver of salmonid abundance, the CDFG report concludes ocean conditions are far less influential than spring flows:

AR 00212424-25 (footnotes omitted).

. Plaintiffs repeatedly assert that NMFS was required to articulate in the BiOp how and provide supporting evidence demonstrating that each RPA action "will avoid jeopardy to the continued existence of a listed species.” Doc. 431 at 92, 101. This suggests a requirement that each individual RPA action must be designed to avoid jeopardy. The requirement is more subtle. The Handbook requires each aspect be an "essential” component of an overall RPA designed to avoid jeopardy and adverse modification. Although each element need not achieve avoidance on its own, Federal Defendants incorrectly assert that "the Court’s task here is not to dissect and reanalyze this RPA bit by bit, but analyze the overall management scheme proposed by the RPA *898 and determine whether NMFS acted arbitrarily and capriciously in concluding that the RPA, in its entirety, was necessary to avoid the likelihood of jeopardizing the continued existence of the multiple species at issue.” Doc. 477-1 at 71. In fact, the Handbook requires an action-by-action analysis. NMFS must thoroughly explain how "each” component of the RPA is "essential.” While the Handbook is not deserving of Chevron deference, N. Cal. River Watch v. Wilcox, 633 F.3d 766, 778-79 (9th Cir.2011), as its purpose is to provide "information and guidance,” its text is routinely cited as NMFS's and FWS’s interpretation of the ESA, entitled to at least Skidmore deference, Az. Cattle Growers' Ass’n v. Salazar, 606 F.3d 1160 , 1165 (9th Cir. 2010). In at least one case, the Secretary of the Interior argued that tire Handbook was not binding on the consulting agencies. Nat'l Wildlife Fed’n v. Babbitt, 128 F.Supp.2d 1274, 1292 (E.D.Cal.2000). But no such suggestion has been made here, nor is the agency's alternative interpretation that it may omit specific justification of each RPA action reasonable. "Although interpretations contained in agency manuals and comments are not entitled to the highest level of deference, a court may nevertheless defer to an agency's interpretation of its own regulation, depending upon 'the thoroughness evident in its consideration, the validity of its reasoning, its consistency with earlier and later pronouncements, and all those factors which give it power to persuade, if lacking power to control.' ” Medina County Envt’l Action Ass’n v. Surface Transp. Bd., 602 F.3d 687, 700-701 (5th Cir.2010) (quoting United States v. Mead Corp., 533 U.S. 218, 227-30 , 121 S.Ct. 2164 , 150 L.Ed.2d 292 (2001)). Applying this standard, the Fifth Circuit concluded that an interpretation contained in the Handbook was entitled to deference. Id. at 701.

. NMFS is entitled to deference in its interpretation and application of the body of relevant science, most of which is equivocal on the issue, and had some basis to reach the conclusion that some form of flow:export ratio limitation should be imposed. It is as apparent that the record contains no strong evidence drat a flowiexport ratio limitation will improve salmonid survival.

. Federal Defendants cite Kandra v. United States, 145 F.Supp.2d 1192, 1207 (D.Or. 2001), which briefly discusses consistency with the project purposes, for the proposition that so long as wildlife protection is a legitimate purpose of a project, an RPA designed to protect a species is consistent with the purposes of that. In Kandra , water user plaintiffs sought to enjoin Reclamation from implementing a 2001 Annual Operations Plan for the Klamath Reclamation Project, which included RPAs that would modify flows to support listed species, resulting in complete curtailment of water deliveries to the majority of land within the Klamath Project. Id. at 1195-96. Plaintiffs argued that the purpose of the Klamath Project, pursuant to the Reclamation Act, is irrigation, and that the RPAs adopted by Reclamation benefit fish to the detriment of irrigation was inconsistent with the Project's purpose. Id. at 1207 . The district court found this argument unpersuasive:

Id. at 1207. This non-binding decision is decidedly unpersuasive, as it ignores the competing, Congressionally mandated irrigation purpose. Even if the logic of Kandra is ac *916 cepted, arguendo, the Agency has a duty to closely examine the adverse effects to prevent emasculation of the co-equal purpose of irrigation.

. Export Plaintiffs also challenge the BiOp's related conclusion that the 330,000 AF of export curtailments "can be offset by application of (b)(2) water resources, water conservation, groundwater use, water recycling and other processes currently underway.” BiOp at 580. This is not an essential element of the BiOp's reasoning, as NMFS later concedes that "NMFS could not be reasonably certain b(2) water would be available” and indicates that the BiOp's analysis of the RPA actions does not depend on the availability of (b)(2) water. Id. at 722.

. In light of D-1641's requirement that DWR and the Bureau comply with the ESA, Export Plaintiffs have not pointed to any substantive statute or jurisdictional limitation that precludes the Reclamation or DWR from implementing a lawful ESA RPA. The SR Plaintiffs have made such an argument. That argument is addressed below.

. DWR points out that, in response to DWR’s motion to admit expert testimony, Federal Defendants made a judicial admission that they would not rely on Ricker’s 1952 study or the statement that when survival is below 25%, a reduction of predator numbers to below 50% can double survival. See Doc. No. 464; Draft Tr. 7/19/10 at 61-67. Relying on that admission, the Court concluded expert testimony was not needed to explain application of the Ricker study. Doc. 464 at 1-2. DWR now asserts that it is prejudiced by Federal Defendants reliance on a passage that discusses Ricker because DWR "does not have an expert to explain DWR (2008) and the Ricker equation. However, DWR successfully explains the Ricker study and NMFS’s use of it. See Doc. 495 at 4-5. Expert clarification is unnecessary. Nonetheless, Federal Defendants are precluded from using the 1952 Ricker study after they said they would not.

. While Action IV.4.2(2) provides that NMFS may agree to an extended timeline, the Action provides no basis for determination of whether an extension should be given.

. SR Plaintiffs’ motion suffers from a lack of internal organization, repeatedly shifting back and forth between challenges to the effects analysis and challenges to the RPA, making evaluation of the merits of their arguments unnecessarily time consuming and difficult.

. SR Plaintiffs mention further legal constraints on the Bureau's use of water set forth in the September 30, 2009 Federal Circuit Ruling, Stockton E. Water Dist. v. United States, 583 F.3d 1344 (Fed.Cir.2009), which post-dates the June 4, 2009 issuance of the BiOp by several months. See Doc. 492 at 6 (discussing holding that Reclamation does not have discretion to breach SEWD CVP contract to comply with the ESA). Those subsequent constraints and their future effects do not apply to the reasonableness of the BiOp when issued.

. Defendant-Intervenors and Federal Defendants suggest that SR Plaintiffs’ opening brief failed to challenge NMFS’s determination that the actin would adversely modify CV steelhead critical habitat. Doc. 484 at 88 & Doc. 515 at 53. However, although the critical habitat analysis was not a direct subject of *935 discussion in SR Plaintiffs' opening brief, that brief did directly challenge the effects analysis regarding the Stanislaus River, which includes both effects on the species and critical habitat. Defendant-Intervenors’ discussion of critical habitat as an alternative justification for the RPAs, Doc. 484 at 82-85, invites SR Plaintiffs’ discussing critical habitat in reply-

. Both SRJN Exhibit 7 and 8 are public records subject to judicial notice for their content. San Luis Unit Food Producers v. United States, 772 F.Supp.2d 1210 , 1216 n. 1 (E.D.Cal.2011).

. SR Plaintiffs cite a May 2009 Memo authored by NMFS’s Rhonda Reed, which discussed the purported 1,500 non flood flow limit:

AR 105885. SR Plaintiffs erroneously assert that this paragraph indicates that an NMFS scientist, Roger Guiñee, offered a "legal opinion” that the injunction no longer applied. In fact, the paragraph states that the legal opinion came from Jim Monroe, a federal government attorney.

. Federal Defendants later clarify that Reclamation must support an invocation of the exception *948 strates varying allocations and delivery schedules do not let them meet the required temperatures.” BiOp at 621. But, this does not place a limit on the number of times the exception may be invoked, nor does it demonstrate the extent to which repeated invocation of the exception will undermine the purpose of the Action.

. New Melones Dam operates in conjunction with Tulloch Reservoir and Goodwin Dam on the Stanislaus River to form the East Side Division. See BiOp at 197.

. SR Plaintiffs' alternatively argue that the RPA definition impliedly incorporates the "impermissible major changes” prohibition contained in the RPM definition. SR Plaintiffs offer no support for this argument, which is contradicted by the general rule that the plain language of a statute governs, absent ióme indication of [] regulatory intent that overcomes plain language ... referenced in the published notices that accompanied the rulemaking process.” See Webb v. Smart Document Solutions, 499 F.3d 1078, 1084 (9th Cir.2007).

. Action III.2.2 specifically requires:

BiOp at 627.