ARIZONA CATTLE GROWERS'ASS'N v. Salazar

                   FOR PUBLICATION
  UNITED STATES COURT OF APPEALS
       FOR THE NINTH CIRCUIT

ARIZONA CATTLE GROWERS’                  
ASSOCIATION,
                  Plaintiff-Appellant,
                  v.
KEN SALAZAR, in his official                  No. 08-15810
capacity as Secretary of the
Interior; H. DALE HALL, in his                  D.C. No.
official capacity as Director of            2:06-CV-01744-
Fish & Wildlife Service; UNITED                   SRB
STATES DEPARTMENT OF THE                        OPINION
INTERIOR; PAUL K. CHARLTON; ERIC
H. HOLDER JR., Attorney General,
              Defendants-Appellees,
CENTER FOR BIOLOGICAL DIVERSITY,
     Defendant-intervenor-Appellee.
                                         
        Appeal from the United States District Court
                 for the District of Arizona
         Susan R. Bolton, District Judge, Presiding

                 Argued and Submitted
       November 4, 2009—San Francisco, California

                      Filed June 4, 2010

    Before: Betty B. Fletcher, William C. Canby, Jr., and
             Susan P. Graber, Circuit Judges.

                Opinion by Judge B. Fletcher



                              8025
             ARIZONA CATTLE GROWERS v. SALAZAR           8029
                         COUNSEL

Norman D. James, Fennemore Craig, Phoenix, Arizona, for
the plaintiff-appellant, Arizona Cattle Growers’ Association.

Andrew C. Mergen, Rebecca Riley, & Robert H. Oakley, U.S.
Department of Justice, Environment and Natural Resources
Division, Washington, D.C., for the defendant-appellee, U.S.
Fish and Wildlife Service.

Karen Budd-Falen, Budd-Falen Law Offices, LLC, Cheyenne,
Wyoming, for amicus curiae New Mexico Cattle Growers’
Association.

Marc D. Fink, Center for Biological Diversity, Duluth, Min-
nesota, and Matt Kenna, Western Environmental Law Center,
Durango, Colorado, for defendant-intervenor-appellee Center
for Biological Diversity.


                         OPINION

B. FLETCHER, Circuit Judge:

   Arizona Cattle Growers’ Association (“Arizona Cattle”)
appeals from the district court’s grant of summary judgment
rejecting its challenge to the United States Fish and Wildlife
Service’s (“FWS”) designation of critical habitat for the Mex-
ican Spotted Owl. Arizona Cattle argues that the FWS unlaw-
fully designated areas containing no owls as “occupied”
habitat and that the FWS calculated the economic impacts of
the designation by applying an impermissible “baseline”
approach. We have jurisdiction under 28 U.S.C. § 1291, and
we affirm.
8030           ARIZONA CATTLE GROWERS v. SALAZAR
I.    BACKGROUND

     A.   Litigation History

   In 1993 the Mexican Spotted Owl was listed as a threatened
species under the Endangered Species Act (“ESA”). The list-
ing decision prompted a series of lawsuits alternately seeking
to compel the FWS to designate critical habitat for the owl
and, following the FWS’s designation of habitat, attacking
that designation.

   The first such lawsuit was in 1995 to compel the FWS to
designate critical habitat and resulted in the FWS’s issuing a
final rule designating 4.6 million acres of critical owl habitat,
a designation that was quickly challenged in court and then
revoked in 1998. After another lawsuit was filed to compel
the FWS to designate habitat, the FWS proposed a rule in
2000 to designate 13.5 million acres of critical habitat and in
2001 the agency promulgated a final rule that again desig-
nated 4.6 million acres. That rule was later struck down and,
rather than propose a new rule, the FWS reopened the com-
ment period on the rule it proposed in 2000. In 2004 the FWS
designated approximately 8.6 million acres of critical habitat.
It is this designation, the 2004 Final Rule, that Arizona Cattle
challenges in the current action.

   Arizona Cattle moved for summary judgment to set aside
the 2004 Final Rule as invalid on several grounds, only two
of which are appealed. First, Arizona Cattle argues that the
FWS impermissibly treated areas in which no owls are found
as “occupied” under the ESA and, in doing so, bypassed the
statutory requirements for designating unoccupied areas. Sec-
ond, Arizona Cattle challenges the FWS’s determination of
the economic impacts of the designation, arguing primarily
that the FWS applied an impermissible “baseline” approach
that did not account for economic impacts of the critical habi-
tat designation that are also attributable to the listing decision.
               ARIZONA CATTLE GROWERS v. SALAZAR                  8031
The district court rejected Arizona Cattle’s arguments and
granted the Appellees’ cross-motions for summary judgment.

  B.    The 2004 Final Rule

   The FWS relied on three types of habitat management
areas, first outlined in a Recovery Plan created in 1995, as a
starting point for the 2004 Final Rule: protected areas,
restricted areas, and other forest and woodland types. Pro-
tected areas are those areas containing known owl sites,
termed Protected Activity Centers (“PACs”); “steep slope”
areas meeting certain forest conditions; and legally and
administratively reserved lands. “PACs include a minimum of
600 acres . . . that includes the best nesting and roosting (i.e.,
resting) habitat in the area . . . . and the most proximal and
highly used foraging areas.” However, PACs contain only
75% of necessary foraging areas for the owl. Restricted areas
include non-steep slope areas with appropriate forest condi-
tions that are “adjacent to or outside of protected areas.”
“Areas outside of PACs, including restricted areas, provide
additional habitat appropriate for foraging.” According to the
2004 Final Rule, restricted areas “also provide habitat for
nonterritorial birds[,] . . . support dispersing juveniles, and . . .
provide replacement nest/roost habitat on the landscape
through time.”

   The FWS used these categories to “develop[ ] alternatives
for critical habitat designation,” selecting protected and
restricted areas as a starting point for potential owl critical habi-
tat.1 In the 2004 Final Rule the FWS adopted an alternative
that excluded all tribal lands from designation, refined critical
habitat unit boundaries, and excluded certain areas that did
not contain PACs. The FWS also excluded “Wildland-Urban
Interface” areas identified as being at high risk of catastrophic
  1
   As discussed in greater detail below, the FWS also analyzed the areas
meeting these habitat characteristics for evidence of owl presence and
used this information to refine its eventual designation.
8032            ARIZONA CATTLE GROWERS v. SALAZAR
wildfire. The 2004 Final Rule concluded that all of the desig-
nated habitat was occupied by the owl.

II.    STANDARD OF REVIEW

   We review the grant of summary judgment de novo,
reviewing directly the agency’s action under the Administra-
tive Procedure Act’s arbitrary and capricious standard. Gif-
ford Pinchot Task Force v. U.S. Fish & Wildlife Serv., 378
F.3d 1059, 1065 (9th Cir. 2004).

       [A]n agency rule would be arbitrary and capricious
       if the agency has relied on factors which Congress
       has not intended it to consider, entirely failed to con-
       sider an important aspect of the problem, offered an
       explanation for its decision that runs counter to the
       evidence before the agency, or is so implausible that
       it could not be ascribed to a difference in view or the
       product of agency expertise.

Motor Vehicle Mfrs. Ass’n of the U.S., Inc. v. State Farm Mut.
Auto. Ins. Co., 463 U.S. 29, 43 (1983). In recognition of the
agency’s technical expertise the court usually defers to the
agency’s analysis, particularly within its area of competence.
See Earth Island Inst. v. Hogarth, 494 F.3d 757, 766 (9th Cir.
2007); Nat’l Ass’n of Home Builders v. Norton, 340 F.3d 835,
843-44 (9th Cir. 2003). However, the court need not defer to
the agency when the agency’s decision is without substantial
basis in fact, and there must be a rational connection between
the facts found and the determinations made. Earth Island,
494 F.3d at 766.

III.    THE FWS PROPERLY DESIGNATED ONLY
        OCCUPIED AREAS AS CRITICAL HABITAT

  We first consider whether the owl “occupied” the desig-
nated areas, as defined by the ESA. We conclude that the
FWS permissibly interpreted the word “occupied” in the ESA
             ARIZONA CATTLE GROWERS v. SALAZAR                  8033
to include areas where the owl was likely to be present and
that, applying this definition, the FWS designated only “occu-
pied” areas.

  A.   The ESA and the Definition of “Occupied”

  [1] The ESA defines a species’ critical habitat as

       (i) the specific areas within the geographical area
    occupied by the species, at the time it is listed . . . ,
    on which are found those physical or biological fea-
    tures (I) essential to the conservation of the species
    and (II) which may require special management con-
    siderations or protection; and

       (ii) specific areas outside the geographical area
    occupied by the species at the time it is listed . . . ,
    upon a determination by the Secretary that such
    areas are essential for the conservation of the spe-
    cies.

16 U.S.C. § 1532(5)(A). The statute thus differentiates
between “occupied” and “unoccupied” areas, imposing a
more onerous procedure on the designation of unoccupied
areas by requiring the Secretary to make a showing that unoc-
cupied areas are essential for the conservation of the species.
Although this appeal turns primarily on the factual question
of whether the FWS treated unoccupied areas as occupied to
avoid this more onerous process, we face the preliminary
issue of what it means for an area to be “occupied” under the
ESA.

   [2] It is useful to unpack this inquiry into two components:
uncertainty and frequency. Uncertainty is a factor when the
FWS has reason to believe that owls are present in a given
area, but lacks conclusive proof of their presence. Frequency
is a factor when owls are shown to have only an intermittent
presence in a given area. Occasionally, both factors will play
8034            ARIZONA CATTLE GROWERS v. SALAZAR
a part in determining whether an area is “occupied.” Because
the ESA permits only one of two possible outcomes for this
inquiry—occupied or unoccupied—when the result is best
characterized by a spectrum, we must determine the scope of
the FWS’s authority to categorize as “occupied” those areas
that may not fit neatly into either pigeonhole.

   We have ample guidance on the “uncertainty” issue. The
ESA provides that the agency must determine critical habitat
using the “best scientific data available.” 16 U.S.C.
§ 1533(b)(2); see also id. § 1533(b)(6)(C)(ii). This standard
does not require that the FWS act only when it can justify its
decision with absolute confidence. See, e.g., Pub. Citizen
Health Research Group v. U.S. DOL, 557 F.3d 165, 176 (3d
Cir. 2009); Greenpeace Action v. Franklin, 982 F.2d 1342,
1354-55 (9th Cir. 1992). Although the FWS cannot act on
pure speculation or contrary to the evidence, the ESA accepts
agency decisions in the face of uncertainty. Compare Ariz.
Cattle Growers’ Ass’n v. U.S. Fish & Wildlife, 273 F.3d 1229,
1244 (9th Cir. 2001), with Sw. Ctr. for Biological Diversity v.
Babbitt, 215 F.3d 58, 60-61 (D.C. Cir. 2000).

   Turning to the “frequency” component, Arizona Cattle
asserts that the word “occupied” is unambiguous and must be
interpreted narrowly to mean areas that the species “resides
in.” In the context of the owl, they argue that such areas con-
sist only of the 600-acre PACs. The FWS argues for a broader
interpretation. It suggests that where a geographic area is used
with such frequency that the owl is likely to be present, the
agency may permissibly designate it as occupied. FWS con-
tends that, at a minimum, this includes the owl’s “home range”2
and may include other areas used for intermittent activities.
  2
    Home range is the “area used by an animal during its normal activi-
ties”; in the case of the owl, estimates of the owl’s home range size have
varied substantially. Studies cited in the 1995 Recovery Plan, for example,
estimated home range sizes varying from a low of 645 acres to a high of
3,831 acres.
             ARIZONA CATTLE GROWERS v. SALAZAR              8035
   [3] We cannot agree that “occupied” has an unambiguous,
plain meaning as Arizona Cattle suggests. The word “occu-
pied,” standing alone, does not provide a clear standard for
how frequently a species must use an area before the agency
can designate it as critical habitat. Cf. Amoco Prod. Co. v.
Vill. of Gambell, 480 U.S. 531, 548 n.14 (1987) (explaining
that there is “clearly” no plain meaning to the phrase “public
lands which are actually occupied”). Merely replacing the
word “occupied” with the word “resides” does not resolve this
ambiguity. Rather, Arizona Cattle’s argument that “occupied”
is limited to areas where the species “resides” only under-
scores the flexibility of determining whether an area is “occu-
pied.” Viewed narrowly, an owl resides only in its nest;
viewed more broadly, an owl resides in a PAC; and viewed
more broadly still, an owl resides in its territory or home
range. Determining whether a species uses an area with suffi-
cient regularity that it is “occupied” is a highly contextual and
fact-dependent inquiry. Cf. Cape Hatteras Access Pres. Alli-
ance v. United States DOI, 344 F. Supp. 2d 108, 119-20
(D.D.C. 2004). Relevant factors may include how often the
area is used, how the species uses the area, the necessity of
the area for the species’ conservation, species characteristics
such as degree of mobility or migration, and any other factors
that may bear on the inquiry. Such factual questions are
within the purview of the agency’s unique expertise and are
entitled to the standard deference afforded such agency deter-
minations. See Earth Island, 494 F.3d at 766.

   [4] Having found the term “occupied” dependent on a
number of factors, we must look to whether the agency’s pro-
posed interpretation is permissible as applied to the owl’s
habits and habitat. Arizona Cattle argues that the FWS has
never previously defined “occupied critical habitat.” Thus, it
contends, the agency interpretation urged on appeal is merely
a self-serving construction found only in the FWS’s legal
briefs and is entitled to no deference. But the agency has
defined “occupied critical habitat” in a manner very similar to
the proposed interpretation. In its Endangered Species Con-
8036         ARIZONA CATTLE GROWERS v. SALAZAR
sultation Handbook, the agency defines “occupied critical
habitat” as

    critical habitat that contains individuals of the spe-
    cies at the time of the [Section 7] project analysis. A
    species does not have to occupy critical habitat
    throughout the year for the habitat to be considered
    occupied (e.g. migratory birds). Subsequent events
    affecting the species may result in this habitat
    becoming unoccupied.

U.S. Fish & Wildlife Serv. & Nat’l Marine Fisheries Serv.,
Endangered Species Consultation Handbook 4-34 (1998),
available     at    http://www.fws.gov/endangered/pdfs/Sec7/
handbook/ch4.pdf. This definition recognizes that a species
need not be present continuously for habitat to be considered
“occupied.” It also demonstrates that “occupied” habitat is not
limited to areas in which the species “resides,” as it includes
habitat that merely “contains individuals of the species.” At
the very least, this definition is entitled to deference “propor-
tional to its power to persuade” pursuant to the Supreme
Court’s holding in Skidmore v. Swift & Co., 323 U.S. 134
(1944), and subsequent cases. See United States v. Mead
Corp., 533 U.S. 218, 235 (2001) (giving Skidmore deference
to interpretations contained in agency manuals or enforcement
guidelines); Bamonte v. City of Mesa, 598 F.3d 1217, 1228
(9th Cir. 2010) (explaining that, under Skidmore, agency posi-
tions not afforded the force of law are entitled to deference
“proportional to [their] power to persuade” (internal quotation
marks omitted)). The definition in the handbook appears to be
the result of the agency’s considered judgment and, for the
reasons we express below, we are persuaded by the agency’s
position that “occupied” should not be interpreted in a restric-
tive fashion.

   [5] The FWS permissibly rejected Arizona Cattle’s “re-
sides in” interpretation as too narrow. Looking to the context
of the present appeal provides a solid justification for this
                ARIZONA CATTLE GROWERS v. SALAZAR                        8037
rejection. The record demonstrates, for example, that PACs
include only 75% of the owl’s foraging habitat. Even if we
assume that each owl “resides in” a PAC, we are not per-
suaded that Congress intended a definition of “occupied” that
would exclude areas likely to be regularly used by the species.
This is particularly true where those areas contain resources
necessary for species conservation. Arizona Cattle’s proposed
interpretation would also exclude habitat for nonterritorial
owls that may not be under constant or uniform use despite
frequent owl presence. Cf. Cape Hatteras Access Pres. Alli-
ance, 344 F. Supp. 2d at 119-20 (noting the agency’s exami-
nation of areas for “consistent use”).

   [6] The FWS has authority to designate as “occupied”
areas that the owl uses with sufficient regularity that it is
likely to be present during any reasonable span of time. This
interpretation is sensible when considered in light of the many
factors that may be relevant to the factual determination of
occupancy. For example, Arizona Cattle’s “reside in” inter-
pretation would make little sense as applied to nonterritorial,
mobile, or migratory animals—including the owl—for which
it may be impossible to fix a determinate area in which the
animal “resides.”3 Such a narrow interpretation also would
mesh poorly with the FWS’s authority to act in the face of uncer-
tainty.4

   [7] We are further persuaded by our decision in Gifford
  3
     It is easy to envision other contexts in which this interpretation is even
less helpful: consider salmon that swim upstream to spawn, periodical
cicadas that live underground for years before emerging as adults, migra-
tory birds, or other animals that require a diverse array of habitats.
   4
     By way of example, because PACs represent only known owl sites,
Arizona Cattle’s proposed interpretation that only PACs are “occupied”
because they reflect where the owl “resides” is likely substantially
underinclusive and places a burden on the agency to count every owl. See
Sw. Ctr. for Biological Diversity, 215 F.3d at 60-61 (explaining that the
ESA’s requirement to use the best scientific data available did not require
the agency to conduct independent studies to count every single animal).
8038         ARIZONA CATTLE GROWERS v. SALAZAR
Pinchot. In that case we invalidated an agency interpretation
of the ESA that effectively eliminated the independent signifi-
cance of critical habitat as a measure to protect endangered
species. See Gifford Pinchot, 378 F.3d at 1070; see also
Sierra Club v. U.S. Fish & Wildlife Serv., 245 F.3d 434, 441-
43 (5th Cir. 2001). The same logic leads us here to reject Ari-
zona Cattle’s attempt to shackle the FWS with an overly nar-
row definition of “occupied.” Critical habitat—including
“occupied critical habitat”—is defined in relation to areas
necessary for the conservation of the species, not merely to
ensure its survival. See 16 U.S.C. § 1532(5)(A); Gifford Pin-
chot, 378 F.3d at 1070; Sierra Club, 245 F.3d at 441-42. Lim-
iting the agency to designating habitat only where the owl
“resides” focuses too narrowly on owl survival and ignores
the broader purpose of the critical habitat designation.

   Our decision is also informed by Supreme Court precedent
that has treated the word “occupied” with considerable
breadth. In Amoco Production Co., the Court distinguished a
statute referring to land “in Alaska” from a statute referring
to “public lands which are actually occupied.” 480 U.S. at
547-48 n.14. The Court explained that while “in Alaska” had
a “precise geographical/political meaning[ ],” the phrase
“public lands which are actually occupied” did not and was
properly construed to include substantial areas of adjacent
waters. Id. (citing Hynes v. Grimes Packing Co., 337 U.S. 86,
110-16 (1949)). The Court’s interpretation of “actually occu-
pied” as including adjacent waters suggests that it is permissi-
ble for the FWS to interpret “occupied” more broadly than
merely the area where an individual or species “resides.”

   Finally, this interpretation is supported by the purpose of
the ESA “ ‘to prevent animal and plant species endangerment
and extinction caused by man’s influence on ecosystems, and
to return the species to the point where they are viable compo-
nents of their ecosystems.’ ” Trout Unlimited v. Lohn, 559
F.3d 946, 949 (9th Cir. 2009) (quoting H.R. Rep. No. 95-
                ARIZONA CATTLE GROWERS v. SALAZAR                      8039
1625, at 5 (1978), reprinted in 1978 U.S.C.C.A.N. 9453, 9455).5
Where data are inconclusive or where habitat is used on a
sporadic basis, allowing the FWS to designate as “occupied”
habitat where the species is likely to be found promotes the
ESA’s conservation goals and comports with the ESA’s pol-
icy of “institutionalized caution.” See 16 U.S.C. § 1531; cf.,
e.g., Defenders of Wildlife v. Flowers, 414 F.3d 1066, 1074
(9th Cir. 2005); Sierra Club v. Marsh, 816 F.2d 1376, 1386
(9th Cir. 1987) (“Congress clearly intended that [agencies]
give the highest of priorities and the benefit of the doubt to
preserving endangered species.” (internal quotation marks
omitted)).

   It is possible for the FWS to go too far. Most obvious is
that the agency may not determine that areas unused by owls
are occupied merely because those areas are suitable for
future occupancy. Such a position would ignore the ESA’s
distinction between occupied and unoccupied areas. See Ariz.
Cattle Growers’ Ass’n, 273 F.3d at 1244. We note as a caveat,
however, that determining whether an area is occupied or
merely will be occupied in the future may be complicated in
the context of migratory or mobile species. The fact that a
member of the species is not present in an area at a given
instant does not mean the area is suitable only for future occu-
pancy if the species regularly uses the area.6
  5
     Arizona Cattle cites to the legislative history surrounding the enact-
ment of the statutory definition of “critical habitat.” Although this history
suggests that Congress was concerned that agencies were giving equal sta-
tus to “lands needed for population expansion” as to those presently occu-
pied by the species and that Congress intended the agency to be
circumspect about designating unoccupied areas, the history sheds little
light on what frequency of species use renders an area “occupied,” the
issue of substance in this case. See S. Rep. No. 95-874, at 10 (1978).
   6
     Consider, for instance, habitat that a currently living owl has migrated
through—and used—in the recent past, and through which owls are likely
to migrate and use similarly in the future. While owl presence in this area
at a particular time may be limited, we are disinclined to hold that the
FWS must find this area “unoccupied” simply because, despite owl use in
the recent past and foreseeable future, no owl is using the area during a
particular temporal window.
8040         ARIZONA CATTLE GROWERS v. SALAZAR
   Having thus framed the inquiry, we turn to the primary
issue before the court: whether the FWS included unoccupied
areas in its critical habitat designation.

  B.   The FWS Did Not Designate Unoccupied Areas as
       Critical Habitat

   [8] After a thorough review of the record we find that the
FWS did not arbitrarily and capriciously treat unoccupied
areas as occupied. We reiterate that when an agency is acting
within its expertise to make a scientific determination “a
reviewing court must generally be at its most deferential.”
Balt. Gas & Elec. Co. v. NRDC, 462 U.S. 87, 103 (1983);
Lands Council v. McNair, 537 F.3d 981, 993 (9th Cir. 2008)
(en banc), abrogated in part on other grounds by Winter v.
NRDC, 129 S. Ct. 365 (2008).

   [9] The FWS took, as a starting point for its 2004 designa-
tion, the three types of habitat management areas that it devel-
oped in the 1995 Recovery Plan. Simply by virtue of the
definitions of these habitat management types, there is a direct
link between the designated territory and owl occupancy.
PACs are explicitly defined with reference to frequent owl
presence, and non-PAC protected areas and restricted areas
are “devised around” and “adjacent to” PACs. More to the
point, we note significant record support for owl occupancy
of these areas in the form of studies correlating the habitat
characteristics of protected and restricted areas with owl pres-
ence. Cf. Gifford Pinchot, 378 F.3d at 1066; cf also Envtl.
Prot. Info. Ctr. v. U.S. Forest Serv., 451 F.3d 1005, 1017 (9th
Cir. 2006) (rather than counting individual animals, an agency
may in appropriate cases use habitat as a proxy).

  [10] The agency did not stop there. It further refined its
designation by adding and removing areas on the basis of evi-
dence of owl occupancy or lack thereof. A striking example
of this is the FWS’s analysis of owl occupancy in the pro-
               ARIZONA CATTLE GROWERS v. SALAZAR                    8041
posed critical habitat in Arizona.7 This analysis proceeds, unit
by unit, through the addition of areas to the critical habitat
proposal on the basis of information about known owl loca-
tions. It also demonstrates that the FWS, even where it could
not identify known owl sites (PACs) with certainty, was con-
sidering other evidence of current owl use in designating hab-
itat, such as “owl calling routes.”

   A point of recurring significance to our analysis is that
PACs reflect only known owl sites. Although the 2004 Final
Rule identified 1,176 PACs, owl populations have been esti-
mated to be significantly greater than the maximum 2,352
owls reflected by this number of PACs. For example, the
2004 Environmental Analysis notes that more owls than this
may exist in a single recovery unit: “a pilot study (Ganey et
al., 2000) conducted in 1999 estimated the number of Mexi-
can spotted owls for the Upper Gila Mountains Recovery
Unit, exclusive of tribal lands, as 2,950.” Since the listing of
the species, the FWS has repeatedly revised upward its esti-
mates of owl populations and identified new PACs. Likewise,
the 2004 Final Rule recognizes that “[a]dditional surveys are
likely to document more owls.” Efforts by the FWS to iden-
tify other evidence of owl presence when it is unable to fix the
location of a PAC with certainty are, therefore, highly signifi-
cant.

   Even more significant is the fact that the FWS excluded
areas with evidence of few or no owls. The 2004 Final Rule
explains that the FWS “did not designate some areas that are
known to have widely scattered owl sites, low owl population
densities, and/or marginal habitat quality.” We find this state-
ment supported by record evidence explaining the FWS’s
decision to exclude several areas due to an absence of owls.
We likewise find that the record demonstrates that where the
FWS did include areas in which owl presence was uncertain
  7
    This document is particularly significant because Arizona contains the
largest areas of designated critical habitat.
8042          ARIZONA CATTLE GROWERS v. SALAZAR
—such as the North Kaibab Ranger District (“NKRD”), the
Peloncillo Mountains, the Prescott National Forest, and cer-
tain “sky islands”—it did so after thoughtful consideration of
owl occupancy. Finally, we note that a comparison of the
areas designated in the 2004 Final Rule to PAC locations con-
firms that the FWS excluded the vast majority of critical habi-
tat units that contained no PACs and refined the boundaries
of the critical habitat units to exclude large areas that are dis-
tant from PACs.

  [11] The FWS’s process for designating critical habitat
gives us a strong foundation for our conclusion that the
agency did not arbitrarily and capriciously treat areas in
which owls are not found as “occupied.” With this context in
mind, we turn to Arizona Cattle’s specific arguments that the
agency improperly designated unoccupied areas as critical
habitat and our reasons for disagreement.

    1.   The FWS Did Not Impermissibly Change Course
         in the 2004 Final Rule from Its Approach to Prior
         Designations

   Arizona Cattle argues that the 1995 Recovery Plan and the
agency’s prior proposed habitat designations demonstrate that
the FWS considered only PACs to be occupied and intended
non-PAC areas—specifically restricted habitat—to provide
future owl habitat. This, it contends, is proof that the 2004
Final Rule must have falsely labeled restricted areas as “occu-
pied.” Arizona Cattle singles out an e-mail that it contends is
a smoking gun demonstrating that the FWS made an abrupt
shift in its description of protected and restricted areas, decid-
ing to refer to these areas as “suitable habitat outside of
PACs” rather than “unoccupied habitat.”

   We have already suggested some of the reasons why it
would be inappropriate to read the 2004 Final Rule as treating
PACs as the only areas occupied by the owl. We pause here
to explain this conclusion further. First, the 2004 Final Rule
                ARIZONA CATTLE GROWERS v. SALAZAR                       8043
is explicit that PACs represent only the best habitat used by
the owl. The record also demonstrates that the FWS believed
that owls may use habitat within a one-mile buffer around
PACs and that PACs were intended to “minimize activities
occurring in close proximity to owl nests . . . and preserve the
best habitat close to known nesting and roosting sites.” As
already discussed, the record reflects that known, territorial
owls regularly use substantial areas outside of their PACs for
foraging. For these owls, we find that the agency’s suggestion
that the owl’s home range is an appropriate measure of the
territory occupied by the owl is well-supported by the record.

   Even that measure would not present a complete picture of
the territory occupied by the owl because PACs reflect only
known owl sites. It is implausible to believe the FWS intended
—or was statutorily required—to limit “occupied” habitat to
PACs, or to the home range of only known owls, when such
a decision would be significantly underinclusive. Similarly, it
is clear that PACs and the owl’s home range do not reflect
areas used by nonterritorial owls or areas used for certain
other intermittent owl activities, such as dispersal or migra-
tion. The agency points out in the 2004 Final Rule, for exam-
ple, evidence that “some [owls] migrate considerable
distances 12-31 miles . . . during the winter.”8 It does not
appear that the FWS intended to limit “occupied” habitat to
PACs, nor was this decision arbitrary and capricious.

   [12] As to the FWS’s purported shift in approach between
the 2004 Final Rule and earlier agency actions, we find that
it reflects merely a change in the agency’s perspective, a
movement away from an unnecessarily restrictive view of the
areas the owl “occupies.” In other words, the apparent differ-
ence between the 2004 Final Rule and the agency’s prior
  8
   The 2004 Final Rule also notes that “some [owls] remain in the general
area but exhibited shifts in habitat use patterns.” Such shifts in habitat use
could also explain some of the statements in the previous rule that discuss
replacement of nesting and roosting habitat over time.
8044         ARIZONA CATTLE GROWERS v. SALAZAR
approach did not arise because the agency suddenly decided
to treat substantial areas where owls were not present as “oc-
cupied.” The agency simply reassessed its previous approach,
which focused narrowly on “known nesting sites” as the areas
occupied by the owl, adopting the broader approach that we
have held reflects the proper definition of “occupied.” See
Nat’l Ass’n of Home Builders v. Defenders of Wildlife, 551
U.S. 644, 658-59 (2007) (explaining that agencies may
change their minds if proper procedures are followed and fed-
eral courts ordinarily review only an agency’s final action);
Motor Vehicle Mfrs. Ass’n, 463 U.S. at 44; PLMRS Narrow-
band Corp. v. FCC, 182 F.3d 995, 1001-02 (D.C. Cir. 1999).
Because we conclude that the 2004 Final Rule remained
within permissible bounds, we find no fault in the rule on this
front.

   The e-mail that Arizona Cattle contends demonstrates the
FWS’s shift in approach is explicit that this change was one
of agency perspective and was entirely consistent with the
evidence of the areas used by the owl. The e-mail states that,
in its previous discussion of “occupied habitat,” the agency
was “really referring to . . . known nesting sites” and that
“non-PAC protected areas, restricted areas, and other forest
and woodland types within one mile of a PAC . . . may be
used by owls at any given time.” It explains that, although a
narrow definition of “occupied habitat” that focused exclu-
sively on nesting might not include these areas, non-PAC pro-
tected areas and restricted areas “may potentially harbor an
owl at any given time.” The same e-mail states that the FWS
believed these areas were, in fact, presently used by owls—
even if they had been previously labeled “unoccupied” under
an unnecessarily restrictive approach to that term. Thus, state-
ments in prior agency actions as to the suitability of certain
areas for “future owl occupancy” or characterizing restricted
areas as “unoccupied” lose their force because they reflect
this change in approach rather than a contradiction as to owl
presence or use of these areas. We will “uphold a decision of
less than ideal clarity if the agency’s path may be reasonably
                ARIZONA CATTLE GROWERS v. SALAZAR                    8045
discerned.” Nat’l Ass’n of Home Builders, 551 U.S. at 658
(internal quotation marks omitted); see also Wetlands Action
Network v. U.S. Army Corps of Eng’rs, 222 F.3d 1105, 1122
n.8 (9th Cir. 2000).

   We find the e-mail’s explanation for the agency’s change
in perspective supported by the record. For example, the
record contains notes taken at a meeting that demonstrate the
FWS’s focus on assuring that the agency could identify a
“reasonable expectation” of occupancy in the areas it planned
to designate. The record likewise demonstrates the agency’s
view that restricted areas are “temporally occupied” even if
not used full-time for nesting purposes. As we have already
explained, there is sufficient record support for the agency’s
decision in the 2004 Final Rule not to treat PACs as the only
areas occupied by the owl. Another example is an e-mail in
which the agency explains that, although the exclusion of cer-
tain land resulted in the FWS’s proposing to designate an area
that contained no PACs, the agency considered the restricted
area occupied.

   The FWS was not attempting to designate areas devoid of
owls as “occupied” in the 2004 Final Rule. Although seeming
inconsistencies between the FWS’s decisions may shed light
on the agency’s process, and changes from past positions that
are unsupported by evidence are unlawful, ultimately it is the
2004 Final Rule that is before the court and our inquiry is
whether the FWS exceeded its authority or deviated from the
evidence.9 See Nat’l Ass’n of Home Builders, 551 U.S. at 658-
  9
    Arizona Cattle does point to a few statements in the record questioning
owl presence in areas proposed for designation in what eventually became
the 1995 Final Rule. These statements are dated and their applicability to
the 2004 Final Rule is questionable. The record reflects evolving agency
knowledge as to owl population and location in the decade between the
listing of the owl and the 2004 Final Rule, a point that is explicit in the
1995 Recovery Plan and that Arizona Cattle acknowledges in its reply
brief. These isolated statements of disagreement with the FWS’s approach
in a prior rule promulgated in the mid-1990s do not overcome the record
support for the agency’s decision in 2004.
8046            ARIZONA CATTLE GROWERS v. SALAZAR
59; Wetlands Action, 222 F.3d at 1122 n.8. The FWS’s analy-
sis solidly demonstrates the connection between the desig-
nated areas and owl occupancy, notwithstanding that the
agency previously adopted an unnecessarily restrictive view
of the areas the owl occupied as limited to known nesting
sites. The agency’s approach in the 2004 Final Rule was sup-
ported by the evidence and within permissible bounds.

       2.   The Amount of Land Designated Is Not
            Disproportionate to the Number of Owls

   [13] Arizona Cattle also argues that even using the owl’s
substantially larger home range as the appropriate measure for
the territory occupied by the owl, the FWS has designated a
grossly disproportionate amount of land compared to the
amount the owl occupies. It ties this argument to a seemingly
simple calculation: multiplying the 1,176 PACs by the maxi-
mum estimated home range size of the owl of 3,831 acres, the
resultant area is only approximately 4.5 million acres, in con-
trast to the 8.6 million acres designated. This calculation,
however, rests on a faulty assumption that the PACs represent
all extant owls. We have already explained that PACs reflect
only known owl sites and that there is record support for the
existence of substantially greater numbers of owls and undis-
covered sites. Nor does this calculation, tied as it is to the
number of PACs, reflect areas used by nonterritorial owls,
areas used for juvenile dispersal, or areas used for owl migra-
tion.10 Arizona Cattle’s argument does not overcome the
  10
     We note again here the complexity of determining whether an area is
occupied in the context of a mobile species. Areas used for juvenile dis-
persal, for example, may be necessary for owl survival but only used for
portions of the year. The record reflects that juvenile dispersal involves the
connection of owl groups into “metapopulations.” It does not mean that
habitat used for dispersal is not used by owls, but is rather under intermit-
tent use for routine owl movement from one area to another. Regardless,
we do not rest our holding on this point because, even absent this ration-
ale, the FWS has sufficiently justified the designation.
               ARIZONA CATTLE GROWERS v. SALAZAR                   8047
strong evidence that the FWS was focused on designating
areas occupied by owls.

       3.   The FWS’s Decision to Include the North Kaibab
            Ranger District in the Designation Was Not
            Arbitrary and Capricious

   Finally, we turn to the single specific location11 where Ari-
zona Cattle contends that the FWS has failed to demonstrate
owl occupancy: the NKRD. Citing a letter to the FWS detail-
ing certain studies conducted in the area without owl sight-
ings, Arizona Cattle maintains that the agency treated the
NKRD as occupied despite evidence that owls were in fact
absent from the District. Arizona Cattle, however, overlooks
a responsive memorandum by the agency in which the FWS
explains that it declined to rely on those studies because it
concluded that the studies were not reliable evidence that
owls were not present. This memorandum also explains the
reasons for the agency’s conclusion that owls are present,
including a history of owl sightings in the NKRD. This is pre-
cisely the sort of decision within the agency’s technical exper-
tise that we are not free to second-guess. This memorandum,
along with the FWS’s diligent review of the proposed desig-
nation for owl occupancy, detailed above, persuades us that
the FWS’s decision had a sound basis in fact.

   [14] We conclude that the agency designated only “occu-
pied” areas as critical habitat, even though it may not have
identified with certainty in all cases a known owl constantly
inhabiting that territory. The process that the FWS used to
select habitat for designation, the measures it took to exclude
  11
    Although Arizona Cattle points to areas where it contends that the
amount of land designated is disproportionate to the number of PACs con-
tained in the area, this argument is simply a variant on Arizona Cattle’s
mathematical argument we reject above. In addition, even as to known
owls the argument relies on the 600-acre PACs, rather than a more accu-
rate measure of occupied area.
8048         ARIZONA CATTLE GROWERS v. SALAZAR
areas where owls were absent or use by owls was infrequent,
and its careful work to confirm the presence of owls in the
designated areas demonstrate that the FWS designated areas
that correspond to habitat where the owl is likely to be found.
The agency action was neither based on speculation nor
counter to the evidence.

IV.    THE FWS’S ECONOMIC ANALYSIS WAS NOT
       ARBITRARY AND CAPRICIOUS

  Arizona Cattle challenges the FWS’s analysis of the eco-
nomic impacts of the critical habitat designation. For the rea-
sons expressed below, we find no fault with the agency’s
economic analysis.

  A.   The FWS Permissibly Used the Baseline Approach
       in Conducting the Economic Analysis

   The decision to list a species as endangered or threatened
is made without reference to the economic effects of that deci-
sion. See N.M. Cattle Growers Ass’n v. U.S. Fish & Wildlife
Serv., 248 F.3d 1277, 1282 (10th Cir. 2001). Listing alone
results in certain protections for the species, including a
requirement that federal agencies “insure that any action
authorized, funded, or carried out by such agency . . . is not
likely to jeopardize the continued existence of any endangered
species or threatened species.” 16 U.S.C. § 1536(a)(2); see
also, e.g, id. § 1538. These protections may impose economic
burdens.

   [15] In contrast to the listing decision, under the ESA the
agency may designate critical habitat only after considering
the economic impact of the designation on any particular area.
Id. § 1533(b)(2). The agency has discretion to exclude any
area from the designation if the agency determines “that the
benefits of such exclusion outweigh the benefits of specifying
such area as part of the critical habitat,” unless exclusion
would result in extinction of the species. Id. This can be a del-
                ARIZONA CATTLE GROWERS v. SALAZAR                      8049
icate balancing act. After critical habitat is designated, the
ESA requires that federal agencies “insure that any action
authorized, funded, or carried out by such agency . . . is not
likely to . . . result in the destruction or adverse modification”
of critical habitat. Id. § 1536(a)(2); see also Gifford Pinchot,
378 F.3d at 1069.

   [16] The crux of the parties’ dispute over the FWS’s eco-
nomic analysis is whether the FWS was required to attribute
to the critical habitat designation economic burdens that
would exist even in the absence of that designation. The par-
ties agree that the FWS applied the “baseline” approach to the
economic analysis. Under this approach, any economic
impacts of protecting the owl that will occur regardless of the
critical habitat designation—in particular, the burdens
imposed by listing the owl—are treated as part of the regula-
tory “baseline” and are not factored into the economic analy-
sis of the effects of the critical habitat designation.12 Arizona
Cattle, relying on the Tenth Circuit’s decision in New Mexico
Cattle Growers Association, argues that this was error and
that the FWS was required to apply a “co-extensive” approach
to the economic analysis. Under the co-extensive approach,
the agency must ignore the protection of a species that results
from the listing decision in considering whether to designate
an area as critical habitat. Any economic burden that designat-
ing an area would cause must be counted in the economic
analysis, even if the same burden is already imposed by list-
ing the species and, therefore, would exist even if the area
were not designated.
  12
    For example, suppose that the decision to list the owl as endangered
resulted in a ban on logging in a particular area, and that designating that
area as critical habitat would independently result in the same ban.
Because the listing decision would result in the logging ban even if the
agency did not designate critical habitat in that area, the baseline approach
would not treat the ban as a burden that was imposed by the critical habitat
designation.
8050            ARIZONA CATTLE GROWERS v. SALAZAR
   [17] In New Mexico Cattle Growers Association the Tenth
Circuit held that the baseline approach was impermissible
under the ESA. See 248 F.3d at 1285. It did so, however, rely-
ing on an FWS regulation that defined “destruction or adverse
modification” as effectively identical to the standard for deter-
mining whether an agency action places a species in “jeopar-
dy.” See id. at 1283-85; see also Gifford Pinchot, 378 F.3d at
1069-70.13 The Tenth Circuit held that this regulation ren-
dered an economic analysis relying on the baseline approach
“virtually meaningless” because it allowed the agency, in all
cases, to find no economic impact to the critical habitat desig-
nation. See N.M. Cattle Growers Ass’n, 248 F.3d at 1283-85.
Our court and others have since found the agency’s definition
of “adverse modification” too narrow. See Gifford Pinchot,
378 F.3d at 1070; see also Ctr. for Biological Diversity v.
BLM, 422 F. Supp. 2d 1115, 1151-53 (N.D. Cal. 2006); Cape
Hatteras Access Pres. Alliance, 344 F. Supp. 2d at 128-30.
We therefore reject the Tenth Circuit’s approach in New Mex-
ico Cattle Growers Association as relying on a faulty premise
and hold that the FWS may employ the baseline approach in
analyzing the critical habitat designation.

  The baseline approach is, if anything, more logical than the
co-extensive approach. The very notion of conducting a
cost/benefit analysis is undercut by incorporating in that anal-
ysis costs that will exist regardless of the decision made.14
Moreover, the practical relevance of the economic analysis
  13
      The Tenth Circuit declined to address whether the FWS’s definition
of “adverse modification” was invalid. See N.M. Cattle Growers, 248 F.3d
at 1283-85.
   14
      We note further the confusion engendered by the co-extensive
approach on the “benefit” side of the equation. If the FWS must consider
“burdens” imposed by the critical habitat designation as if there were no
protections imposed by the listing decision, must it also assume that in the
absence of the critical habitat designation the species is entirely unpro-
tected in considering the “benefits” of designating a particular area? The
co-extensive approach runs the risk of becoming a purely academic exer-
cise.
                ARIZONA CATTLE GROWERS v. SALAZAR                      8051
under the ESA is to determine the benefits of excluding or
including an area in the critical habitat designation: if there is
no net benefit (such as a reduction in economic impacts) to
excluding the area, the agency must designate it. See 16
U.S.C. § 1533(b)(2). The baseline approach, in contrast to the
co-extensive approach, reflects this purpose.

   Congress has directed the FWS to list species, and thus
impose a regulatory burden, without consideration of the costs
of doing so. See 16 U.S.C. § 1533(a); N.M. Cattle Growers,
248 F.3d at 1282. It would be strange to conclude that Con-
gress intended the FWS to consider costs at the critical habitat
phase that the agency was barred from considering at the list-
ing phase where, as a result, the analysis would bear little
relationship to reality.15 It would also be strange to conclude
that Congress intended to use the critical habitat designation
to require the agency to consider the previously irrelevant
costs of listing the species, particularly given that the decision
to exclude an area from critical habitat for economic reasons
is discretionary. See 16 U.S.C. § 1533(b)(2); Bennett v. Spear,
520 U.S. 154, 172 (1997). The simpler explanation is that the
economic analysis of the critical habitat designation is exactly
what it sounds like and is not intended to incorporate the bur-
dens imposed by listing the species.

   Arizona Cattle argues that if the FWS designated critical
habitat at the same time as it listed the species, see 16 U.S.C.
§ 1533(a)(3), there would be no baseline to which to compare
the critical habitat designation. Even if the FWS lists the spe-
cies concurrently with designating critical habitat, however,
listing the species is a necessary antecedent to designating
habitat. We see little inconsistency with the FWS’s consider-
ing the burdens imposed by the critical habitat designation
  15
     Although the Tenth Circuit is likely correct that inclusion of the costs
of listing in the critical habitat analysis does not affect the FWS’s listing
process, see N.M. Cattle Growers, 248 F.3d at 1285, it has clear potential
to distort the critical habitat analysis.
8052         ARIZONA CATTLE GROWERS v. SALAZAR
while taking into account those necessarily imposed by the
listing decision even in these circumstances.

   Finally, Arizona Cattle argues that the baseline approach
allows the FWS to treat the economic analysis as a mere pro-
cedural formality. We reject the argument that, as a matter of
course, the FWS will neglect its duty to perform a thorough
economic analysis. To hold otherwise would amount to a pre-
sumption that the FWS will act in an arbitrary and capricious
fashion, a presumption that is inconsistent with the deference
the court affords agencies. See, e.g., Smith v. U.S. Forest
Serv., 33 F.3d 1072, 1077 n.2 (9th Cir. 1994). Furthermore,
contrary to Arizona Cattle’s contention that the impact of des-
ignating critical habitat cannot be negligible, the costs of a
critical habitat designation could, in fact, be subsumed by the
burdens imposed by listing the species—any burden that is
entirely “co-extensive” with the listing decision will reflect
exactly such a case.

  We hold that the FWS permissibly applied the baseline
approach in conducting the economic analysis of the effects
of the designation.

  B.   Remaining Arguments

   We can easily dispose of Arizona Cattle’s two remaining
challenges to the FWS’s economic analysis. First, Arizona
Cattle suggests that the FWS’s analysis ignored the economic
effects of designating unoccupied habitat. Having already
rejected this argument’s premise—that the FWS’s critical
habitat designation included unoccupied habitat—we need
give no further consideration to this argument. Second, Ari-
zona Cattle asserts that the FWS’s economic analysis ignored
the difference between the jeopardy and adverse modification
standard in light of Gifford Pinchot, and should have consid-
ered the economic impacts of additional consultations and
project modifications that the adverse modification standard
imposes. Arizona Cattle did not make this argument at the
               ARIZONA CATTLE GROWERS v. SALAZAR                  8053
district court level and we therefore do not consider it now.16
See Solis v. Matheson, 563 F.3d 425, 437 (9th Cir. 2009);
Harger v. Dep’t of Labor, 569 F.3d 898, 904 n.9 (9th Cir.
2009).

V.     CONCLUSION

   We find no fault with the FWS’s designation of habitat for
the Mexican Spotted Owl. The FWS did not impermissibly
treat unoccupied areas as “occupied,” and we hold that it per-
missibly applied the baseline approach in analyzing the eco-
nomic impact of the critical habitat designation.

     The judgment of the district court is AFFIRMED.




   16
      Although we need not reach it, we note that this argument appears
fundamentally inconsistent with Arizona Cattle’s primary complaint
regarding the FWS’s application of the baseline approach. The baseline
approach counts precisely these economic impacts and Arizona Cattle’s
disagreement with the baseline approach is, in fact, that it counts only
impacts like these.