FOR PUBLICATION
  UNITED STATES COURT OF APPEALS
       FOR THE NINTH CIRCUIT

TUCSON HERPETOLOGICAL SOCIETY;            
DEFENDERS OF WILDLIFE; CENTER
FOR BIOLOGICAL DIVERSITY;
HORNED LIZARD CONSERVATION
SOCIETY; SIERRA CLUB; WENDY
HODGES; FRANCIS ALLAN MUTH,
                                                 No. 07-16641
              Plaintiffs-Appellants,
                v.                                D.C. No.
                                               CV 04-0075 NVW
KEN SALAZAR,* in his official
                                                  OPINION
capacity as Secretary of the
Interior; ROWAN W. GOULD,** in
his official capacity as Acting
Director of the U.S. Fish and
Wildlife Service,
               Defendants-Appellees.
                                          
         Appeal from the United States District Court
                  for the District of Arizona
           Neil V. Wake, District Judge, Presiding

                    Argued and Submitted
         January 16, 2009—San Francisco, California

                       Filed May 18, 2009

Before: John T. Noonan, A. Wallace Tashima, and William
                A. Fletcher, Circuit Judges.

  *Ken Salazar is substituted for his predecessor Dirk Kempthorne, as
Secretary of the Interior, pursuant to Fed. R. App. P. 43(c)(2).
  **Rowan W. Gould is substituted for his predecessor H. Dale Hall, as
Acting Director of the U.S. Fish and Wildlife Service.

                                5941
5942     TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR
                Opinion by Judge Tashima;
 Partial Concurrence and Partial Dissent by Judge Noonan
5944      TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR




                        COUNSEL

Neil Levine, Denver, Colorado, for the plaintiffs-appellants.
            TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR                5945
Sambhav N. Sankar, Environment & Natural Resources Divi-
sion, U.S. Department of Justice, Washington, DC, for the
defendants-appellees.


                              OPINION

TASHIMA, Circuit Judge:

   Conservation organizations and individual biologists (col-
lectively “Plaintiffs”) contend that the Secretary of the Interi-
or’s (the “Secretary”) decision to withdraw a rule proposing
that the flat-tailed horned lizard (the “lizard”) be listed as a
threatened species is contrary to the requirements of the
Endangered Species Act (“ESA” or the “Act”), 16 U.S.C.
§ 1531 et seq., and the Administrative Procedure Act
(“APA”), 5 U.S.C. § 706. They appeal from the district
court’s order granting summary judgment in favor of the Sec-
retary.

   The district court had jurisdiction under 28 U.S.C. § 1331
and 16 U.S.C. § 1540(g)(1)(C) (authorizing “citizen suits” to
compel the Secretary to perform non-discretionary duties
required by the ESA). We have jurisdiction to review the dis-
trict court’s final judgment under 28 U.S.C. § 1291, and we
reverse in part and remand.

                     STATUTORY BACKGROUND

   The ESA provides “a means whereby the ecosystems upon
which endangered species and threatened species depend may
be conserved.” 16 U.S.C. § 1531(b). In service of this goal,
the Act directs the Secretary to maintain a list of threatened
and endangered species, and defines the phrase “endangered
species” as “any species which is in danger of extinction
throughout all or a significant portion of its range.” 

A threatened species is one “which is likely to
  1
   Upon listing a species, the Secretary must issue regulations to provide
for the species’ conservation. See 16 U.S.C. § 1533(d).
5946        TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR
become an endangered species within the foreseeable future
throughout all or a significant portion of its range.” 

“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.” 

The Secretary has delegated
his authority to administer the ESA to the Fish and Wildlife
Service (“FWS”). 50 C.F.R. § 402.01(b).

   The ESA requires the Secretary to consider five factors
when determining whether a species is threatened or endan-
gered: (1) the present or threatened destruction, modification,
or curtailment of the species’ range; (2) overutilization for
commercial, recreational, scientific, or educational purposes;
(3) disease or predation; (4) the inadequacy of existing regula-
tory mechanisms; and (5) other natural or man-made factors
affecting the species’ continued existence. 16 U.S.C.
§ 1533(a)(1); 50 C.F.R. § 424.11(c). The Secretary must reach
a listing determination “solely on the basis of the best scien-
tific and commercial data available to him.” 16 U.S.C.
§ 1533(b)(1)(A).

             FACTUAL AND PROCEDURAL BACKGROUND

  The lizard at issue in this case is “a small, cryptically col-
ored iguanid . . . that is restricted to flats and valleys of the
western Sonoran desert.”3 58 Fed. Reg. 62,624, 62,625 (Nov.
29, 1993). Its natural habitat stretches across parts of southern
California, southwestern Arizona, and northern Mexico. In
California, the lizard can be found in the Coachella Valley,
  2
     Whether a species is threatened with extinction in a “significant portion
of its range” and whether a “distinct population segment” of a species is
threatened with extinction are independent legal questions. See Nw.
Ecosystem Alliance v. U.S. Fish & Wildlife Serv., 

, 1143
(9th Cir. 2007). Plaintiffs do not contend that any portion of the lizard’s
population qualifies for protection as a threatened or endangered “distinct
population segment,” and thus we consider only the former question here.
   3
     The lizard’s formal name is Phrynosoma mcallii.
           TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR               5947
the west side of the Salton Sea and Imperial Valley, and the
east side of the Imperial Valley. In Arizona, the lizard can be
found in the Yuma Desert south of the Gila River and west
of the Gila and Butler Mountains. 68 Fed. Reg. 331, 332
(Jan. 3, 2003). Very little is known about the lizard’s range in
Mexico, but it has been observed south of the California bor-
der in Baja California, south of Arizona from the international
border to the Piñacate Region, and around Puerto Peñasco and
Bahía de San Jorge, Sonora.

   FWS estimates that man-made factors are responsible for
the destruction of 1,103,201 acres of the lizard’s estimated
4,875,624-acre historic range. 71 Fed. Reg. 36,745, 36,749-51
(Jun. 28, 2006). Agricultural and urban development have
resulted in fragmentation of the lizard’s remaining habitat. 68
Fed. Reg. at 332. Fragmentation creates isolated sub-
populations that, because of their reduced size, have an
increased probability of extinction. 58 Fed. Reg. 62,624,
62,626-27 (Nov. 29, 1993). Remaining lizard habitat in the
United States can be divided into four distinct geographical
sections: the Coachella Valley, west side of the Imperial
Valley/Salton Sea, and east side of the Imperial Valley in Cal-
ifornia, and the Yuma Valley in Arizona. 68 Fed. Reg. at 332.

1993 Listing

   The Secretary first proposed listing the lizard as threatened
in 1993, citing documented and anticipated population
declines.4 See 58 Fed. Reg. at 62,624. The proposed listing
identified urban expansion, off-highway vehicle (“OHV”)
use, energy development, and military activities as the pri-
mary threats to the lizard’s remaining habitat, and concluded
that three of the five statutory listing criteria had been satis-
fied. 

  4
    In Defenders of Wildlife v. Norton, 

, 1138-40 (9th Cir.
2001) (“Defenders”), the court recounted the administrative history pre-
ceding the Secretary’s 1993 proposed listing in detail. We do not repeat
it here.
5948       TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR
   Upon proposing a species for listing, the ESA requires the
Secretary to make a final determination (or formally extend
the time for making a final decision) on whether to list within
twelve months. See 16 U.S.C. § 1533(b)(6)(A). That time
period lapsed without the Secretary issuing a final decision,
and Defenders of Wildlife, along with several of the plaintiffs
in this action, sued the Secretary to compel a determination on
whether or not to list the lizard. See 

. In response to a court order requiring a final decision
within sixty days, the Secretary issued a rule withdrawing the
1993 proposed listing (the “1997 withdrawal”). 62 Fed. Reg.
37,852 (July 15, 1997).

   The 1997 withdrawal relied, in part, on the adoption of the
Flat-Tailed Horned Lizard Conservation Agreement (the
“Conservation Agreement”). 

In 1997, seven
state and federal agencies jointly entered into the Conserva-
tion Agreement, and agreed to implement a management
strategy to address threats to the lizard’s remaining habitat on
public lands.5 

to the Conservation Agreement
agreed to take steps aimed at “reducing threats to the species,
stabilizing the species’ populations, and maintaining its
ecosystem.” In addition to reliance on the Conservation
Agreement, the 1997 withdrawal concluded that population
trend data did not conclusively demonstrate significant
declines, and that many of the threats cited in the 1993 pro-
posed listing had either been reduced or eliminated. See 

The Secretary also found that known threats to liz-
ard habitat on private lands did not warrant listing, because
adequate habitat existed on public lands to ensure the species’
continued viability. See 

  5
   The management strategy revolves around the designation of lizard
management areas (“MAs”) on public lands under the jurisdiction of the
Conservation Agreement signatories. 62 Fed. Reg. at 37,854. The MAs
contain 437,000 acres of lizard habitat — approximately thirty-five per-
cent of the lizard’s remaining domestic range. 

SOCIETY v. SALAZAR                  5949
   Defenders challenged the 1997 withdrawal in the U.S. Dis-
trict Court for the Southern District of California. 

. In 1999, the district court entered summary
judgment for the Secretary, and Defenders appealed. 

and remanded, concluding that the Secretary had
failed to consider whether the lost and threatened portions of
the lizard’s range amounted to a “significant portion” of the
species’ overall range, as the ESA requires. 

2003 Withdrawal

   On remand in Defenders, the district court ordered the Sec-
retary to reinstate the 1993 proposed withdrawal within sixty
days, and to issue a final decision on listing within twelve
months of reinstatement. 66 Fed. Reg. 66,384 (Dec. 26,
2001). Following a 120-day public comment period, the Sec-
retary again issued a rule withdrawing the proposed listing
(the “2003 withdrawal”). 68 Fed Reg. 331 (Jan. 3, 2003).7
Plaintiffs then brought this action challenging the 2003 with-
drawal, arguing that the withdrawal failed to comply with the
remand order in Defenders.

   In 2005, the district court held that the 2003 withdrawal
failed to comply with Defenders because it “assumed without
explanation that large swaths of lost habitat were of no signif-
icance at all.” The district court rejected Plaintiffs’ other chal-
  6
      Defenders also noted that the 1997 withdrawal failed to analyze the
putative benefits of the Conservation Agreement and management strategy
in a site specific 

.
   7
     The 2003 withdrawal noted that FWS solicited the opinion of four
independent lizard experts. 68 Fed. Reg. at 340-41. Of the four, two rec-
ommended listing the species as threatened, one did not express a firm
opinion, and one concluded that listing was not warranted. 

the biologist that did not favor listing, stated that a “significant
portion of the [lizard’s] range” has indeed been lost, but concluded that
listing would likely direct resources away from efforts to protect the spe-
cies on public lands, and toward unproductive efforts to protect lizard hab-
itat on private lands.
5950      TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR
lenges to the 2003 withdrawal, concluding that the Secretary’s
assessment of threats to the lizard’s current range was reason-
able and supported by the evidence in the administrative
record. The Secretary did not appeal the remand order. Pursu-
ant to that order, the Secretary withdrew the 2003 withdrawal,
and restored the lizard to proposed listing status. 70 Fed. Reg.
72,776 (Dec. 7, 2005).

2006 Withdrawal

   In 2006, after another round of public comment, the Secre-
tary again issued a rule withdrawing the proposed listing (the
“2006 withdrawal”). 71 Fed. Reg. 36,745 (June 28, 2006).
The 2006 withdrawal quantified the lizard’s lost range,
explained why that range is not “significant” within the mean-
ing of the ESA, and incorporated the findings in the 2003
withdrawal by reference. 

Plaintiffs then filed
a supplemental complaint challenging the 2006 withdrawal.

   In its final 2007 order, the district court granted summary
judgment in favor of the Secretary. Plaintiffs appealed, chal-
lenging both the district court’s 2005 order (insofar as it
upheld the agency’s assessment of threats to the lizard’s cur-
rent range) and its 2007 order (upholding the agency’s assess-
ment of the significance of the lizard’s lost range).

                     STANDARD OF REVIEW

  We review the district court’s summary judgment ruling de
novo. Nw. Envtl. Advocates v. Nat’l Marine Fisheries Serv.,

, 1132 (9th Cir. 2006).

   Under the APA, this court must set aside final agency deci-
sions that are “arbitrary, capricious, an abuse of discretion, or
otherwise not in accordance with law.” 5 U.S.C. § 706(2)(A).
“Review under the arbitrary and capricious standard ‘is nar-
row, and [we do] not substitute [our] judgment for that of the
agency.’ ” Lands Council v. McNair, 

, 987 (9th
              TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR                 5951
Cir. 2008) (en banc) (quoting Earth Island Inst. v. U.S. Forest
Serv., 

, 1156 (9th Cir. 2006)) (alterations in
original). The Secretary has an obligation, however, to “state
a rational connection between the facts found and the decision
made.” Gifford Pinchot Task Force v. U.S. Fish & Wildlife
Serv., 

, 1065 (9th Cir. 2004).

                               DISCUSSION

I.       The Lizard’s Lost Historical Range

     A.     The Secretary’s Compliance with Defenders

   The district court determined that Defenders requires the
Secretary to engage in a two-step process. The Secretary was
directed first to quantify the lizard’s historical range in order
to establish a “temporal baseline,” and then to determine
whether the lost habitat, measured against that baseline,
amounts to a “significant portion” of the species’ overall
range. On appeal, the Secretary challenges the district court’s
entire approach.8

   [1] Defenders stated that a species can be extinct
“ ‘throughout . . . a significant portion of its range’ if there are
major geographical areas in which it is no longer viable but
once 

(ellipsis in original). Before the
Secretary can properly consider whether the areas where the
lizard is “no longer viable” are significant, it is of course nec-
essary to identify those areas. On appeal, the Secretary clings
to his argument that lost historical habitat is largely irrelevant
     8
    Plaintiffs contend that if the Secretary wished to challenge the district
court’s 2005 remand order, he ought timely to have appealed that order
immediately after it was issued. See Alesa Valley Alliance v. Dep’t of
Commerce, 

, 1184 (9th Cir. 2004) (explaining the general
rule that only an agency may appeal a court’s remand order). Regardless,
the issue on appeal is whether the 2006 withdrawal complied with the
requirements laid out in Defenders, and thus we cannot avoid interpreting
that case on this appeal.
5952        TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR
to the recovery of the species, and thus the ESA does not
require him to consider it.

   [2] The Secretary attempts to re-litigate Defenders rather
than accept the opinion’s basic conclusions. Tellingly, a mem-
orandum from the Office of the Solicitor, Department of Inte-
rior, to the Director of FWS, dated March 16, 2007,
acknowledges that Defenders “appears to” require consider-
ation of historic range, but maintains that this court’s interpre-
tation “is not supported by [the ESA].” District courts
applying Defenders have also concluded that the holding
requires consideration of historic range.9 See Defenders of
Wildlife v. Sec’y of Interior, 

, 1168-69
(D. Or. 2005) (interpreting Defenders to require FWS to ana-
lyze the gray wolf’s entire lost historical range before remov-
ing the species from the endangered species list); Defenders
of Wildlife v. Norton, 

, 20 (D.D.C. 2002)
(offering a similar interpretation of Defenders). We agree that
Defenders requires the Secretary to analyze lost historical
range.

   To comply with the district court’s 2005 limited remand
order, the Secretary quantified the lizard’s range of approxi-
mately one hundred years ago — the period before agricul-
tural and commercial development began to encroach on the
lizard’s habitat — and identified the overall habitat loss from
that baseline. 71 Fed. Reg. at 36,751. The Secretary then con-
cluded that, for reasons we discuss below, the lost portions of
  9
    A district court in the Tenth Circuit has disagreed with this court’s
analysis in Defenders. See Ctr. for Biological Diversity v. Norton, 411 F.
Supp. 2d 1271, 1279 (D.N.M. 2005) (holding that the statutory phrase
“significant portion of its range” refers to “biological significance,” and
crediting the Secretary’s argument — specifically rejected in Defenders —
that in order to be “significant,” the species as a whole must be threatened
with extinction). In rejecting Defenders’ gloss on the ESA, however, the
district court does not suggest or imply that the opinion does not require
consideration of historical range.
            TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR                   5953
the lizard’s range are not geographically, biologically, or oth-
erwise significant. 

that the Secretary maintains a “wide
degree of discretion in delineating ‘a significant portion of its
range,’ since the term is not defined in the 

. However, he “must at least explain [his] conclusion
that the area in which the species can no longer live is not a
‘significant portion of its range.’ ” Id.10 Thus, Defenders left
the appropriate criteria for testing “significance” undefined,
but made clear that the Secretary must develop some rational
explanation for why the lost and threatened portions of a spe-
cies’ range are insignificant before deciding not to designate
the species for protection.

   In the 2006 withdrawal, the Secretary offered a set of rea-
sons for discounting the significance of the lizard’s lost his-
torical range (approximately 23% of the species’ baseline
range). He explained that: (1) lizard populations persist across
most of the species’ current range despite habitat loss and
fragmentation; (2) much of the lizard’s lost habitat was con-
verted to agricultural, commercial, and residential develop-
ment long ago; (3) the lost portions of the lizard’s range do
not carry any special biological or genetic importance for the
species as a whole; and (4) lost historical habitat represents a
   10
      Defenders accorded the Secretary’s interpretation of the statutory
phrase “in danger of extinction throughout all or a significant portion of
its range” no 

n.11. The court explained that
the Secretary had entirely ignored an aspect of the phrase (the use of the
disjunctive “or”), and thus there was no interpretation to defer to. See 

owe the Secretary’s interpretation of the statute deference under
Chevron U.S.A. Inc. v. Natural Res. Def. Council, Inc., 

(1984), because he offers an interpretation of an ambiguous statutory
phrase, and offers that interpretation via formal notice and comment rule-
making. See United States v. Mead Corp., 

, 230-31 (2001)
(holding that, although notice and comment procedures are not required to
trigger Chevron deference, such procedures are “significant . . . in pointing
to Chevron authority”).
5954       TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR
relatively small portion of the lizard’s baseline range. See 71
Fed. Reg. at 36,751.

   Plaintiffs are correct that the Secretary’s stated reasons all
rely, to varying degrees, on the premise that lizard popula-
tions persist throughout most of the species’ remaining range.
For example, the 2006 withdrawal states that lost portions of
the lizard’s range are not necessary for the preservation of
gene flow because the isolated lizard populations outside of
the Coachella Valley “are large enough to be self-sustaining.”
71 Fed. Reg. at 36,748. Elsewhere, the Secretary finds that the
habitat lost early in the 20th Century (the majority of the liz-
ard’s overall lost range) is not significant because the species’
“continued persistence over a span of nearly 100 years is a
strong indication that the species will continue to persist into
the foreseeable future despite the loss of historical habitat.”

   It is insufficient, under Defenders, to point to one area or
class of areas where lizard populations persist to support a
finding that threats to the species elsewhere are not signifi-
cant; the ESA requires a more thorough explanation. 

(faulting the Secretary for concluding that the
availability of habitat on public lands alone renders lost habi-
tat on private lands insignificant). Plaintiffs argue that this is
precisely what the Secretary has done here.

   [3] The Secretary’s explanations, however, exhibit more
nuance than Plaintiffs acknowledge. Although he places con-
siderable weight on the lizard’s persistence throughout most
of its remaining range, reliance on persistence is not per se
inconsistent with Defenders. The 2006 withdrawal analyzes
the lizard’s lost habitat in a site-specific manner, and cites liz-
ard persistence to corroborate its conclusion that the lost por-
tions of the lizard’s range do not provide any unique or
critical function for the well-being of the species. Moreover,
the Secretary offers two supplemental reasons that do not
depend on persistence alone. He cites to a study of migration
            TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR                   5955
between isolated lizard populations and reasonably concludes
that the lizard’s lost range does not represent a critical path-
way for maintenance of genetic diversity. 71 Fed. Reg. at
36,750. He also notes that most of the lizard’s lost range was
converted to agricultural or commercial uses decades ago, is
generally not recoverable, and is thus of limited significance
to the lizard’s long-term survival. 

Taking these
reasons together, and according the Secretary due interpretive
deference, we conclude that his understanding of the phrase
“significant portion of its range” does not run afoul of Defend-
ers.11

  B.    Is the Lizard Persisting Throughout Most of its Range?

   The next question is whether the administrative record sup-
ports the Secretary’s finding that lizard populations outside of
the Coachella Valley are persisting. An action will be deemed
arbitrary and capricious where the agency offers an explana-
tion for an action “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 U.S. v. State Farm Mut. Auto Ins. Co.,

, 43 (1983). “While our deference to the agency
is significant, we may not defer to an agency decision that ‘is
without substantial basis in fact.’ ” Sierra Club v. U.S. EPA,

, 961 (9th Cir. 2003) (quoting FPC v. Fla. Power
& Light Co., 

, 463 (1972)).

   [4] The Secretary argues that FWS relied on population
studies to conclude that the lizard “is persisting in the vast
  11
    Plaintiffs also argue that the Secretary erred in failing to consider the
cumulative significance of both the lost historical range, and the currently-
threatened Coachella Valley range. The argument is unavailing. Both the
2003 and 2006 withdrawals offer explanations for the insignificance of
threatened and lost range, and those explanations clearly apply to the liz-
ard’s condition as a whole. Moreover, the 2006 withdrawal explicitly
incorporates the findings from the 2003 withdrawal, and thus its analysis
can be presumed to take the conclusions in that document into account.
5956      TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR
majority of its range.” Appellee’s Br. at 22. Yet, in the 2003
withdrawal, the Secretary reported that “[i]nformation con-
cerning population dynamics of flat-tailed horned lizard popu-
lations is limited and inconclusive.” 68 Fed. Reg. at 332; see
also Teresa Woods & Steve Morey, Uncertainty and the
Endangered Species Act, 83 Ind. L.J. 529, 531-32 (2008)
(explaining the difficulty in carrying out population studies
from which inferences about extinction risk can be made,
given budgetary constraints and short regulatory and court-
ordered time frames for decision-making). Apparently, the
Secretary infers from the uncertainty in the population studies
that lizard populations “remain[ ] viable throughout most of
[the lizard’s] current extant range.” 71 Fed. Reg. at 36,751.
This conclusion is conceptually distinct from and relied on
different evidence than the Secretary’s determination that the
species does not face significant threats in most of its current
range. The latter conclusion is premised on an estimate of
existing and potential threats to lizard habitat, existing man-
agement plans that mitigate those threats, and a variety of
other risk assessment factors. The persistence finding, how-
ever, relies solely on the conclusion that lizard populations are
in fact viable and stable throughout most of the species’ cur-
rent range. It is this conclusion that ultimately requires rever-
sal.

   [5] If the science on population size and trends is under-
developed and unclear, the Secretary cannot reasonably infer
that the absence of evidence of population decline equates to
evidence of persistence. The absence of conclusive evidence
of persistence, standing alone, without persuasive evidence of
widespread decline, may not be enough to establish that the
Secretary must list the lizard as threatened or endangered. See
Cook Inlet Beluga Whale v. Daley, 

, 21-22
(D.D.C. 2001) (holding that the ESA does not require listing
“simply because the agency is unable to rule out factors that
could contribute to a population decline”); cf. Balt. Gas &
Elec. Co. v. Natural Res. Def. Council, Inc., 

, 103
(1983) (holding that when examining decisions made under
            TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR                   5957
conditions of scientific uncertainty “a reviewing court must
generally be at its most deferential”). But this is a different
case. The Secretary affirmatively relies on ambiguous studies
as evidence of persistence (i.e., stable and viable populations),
and in turn argues that this “evidence” of persistence satisfies
Defenders’ mandate and proves that the lizard’s lost range is
insignificant for purposes of the ESA. This conclusion is
unreasonable. The studies do not lead to the conclusion that
the lizard persists in a substantial portion of its range, and
therefore cannot support the Secretary’s conclusion.

   [6] Both parties acknowledge that the formerly common
“scat count” method of estimating lizard population size has
been discredited; thus, that nearly all of the existing data on
lizard populations (including FWS’ 1998 Population Viability
Assessment — which itself concluded that population trends
could not be reliably determined) must be set aside as unhelp-
ful. See 68 Fed. Reg. at 333 (“The relationship between scat
counts and lizard abundance is unclear, or weak at best.”
(citations to scientific studies omitted)). Results from studies
utilizing an allegedly more accurate “capture-mark-recapture”
methodology are just now emerging. 68 Fed. Reg. at 333. In
the 2006 withdrawal, the Secretary cites one such study for
the proposition that, between 2003 and 2005, in two discrete
sections of the lizard’s current range (both within designated
lizard MAs), there is no evidence of a “large decline in popu-
lation” for the areas for which the researchers had more than
one year of data. 71 Fed. Reg. at 36,751. This single attenu-
ated finding represents the extent of the agency’s evidentiary
support for its sweeping conclusion that viable lizard popula-
tions persist throughout most of the species’ current range.12
  12
     The Secretary states that viable populations persist in most of the liz-
ard’s current range, and we can therefore infer that he believes that popu-
lations persist in Mexico (where an estimated 59% of remaining lizard
habitat exists). 68 Fed. Reg. at 332. Yet, the Secretary has not produced
any evidence on lizard population outside of the United States; indeed, the
agency acknowledges that “the distribution of the species in Mexico is
poorly understood . . . .” 

HERPETOLOGICAL SOCIETY v. SALAZAR
Contrary to the lesson the Secretary draws from the study
(i.e., that lizard populations in the study areas show no sign
of decline), the author’s primary conclusion is that the study’s
population estimates can serve as a “baseline for future moni-
toring.” Further, the study’s author warns that the population
estimates it reports “should be viewed with caution as they
were based on sparse data.” We thus conclude that the admin-
istrative record does not support the Secretary’s determination
that lizard populations persist throughout most of the species’
current range.

    The Secretary’s erroneous reliance on lizard persistence,
however, does not end our inquiry. See 5 U.S.C. § 706 (stat-
ing that courts reviewing agency decisions should take “due
account . . . of the role of prejudicial error”). “In circum-
stances where an agency errs, we may evaluate whether such
an error was harmless.” Gifford Pinchot Task 

; see also Nat’l Ass’n of Home Builders v. Defenders
of Wildlife, 

, 2530 (2007) (citing with approval
PDK Labs, Inc. v. U.S. DEA, 

, 799 (D.C. Cir.
2004) (“In administrative law, as in federal civil and criminal
litigation, there is a harmless error rule . . . .” )). We have held
that the harmless error doctrine “may be employed only
‘when a mistake of the administrative body is one that clearly
had no bearing on the procedure used or the substance of
decision reached.’ ” Gifford Pinchot Task 

(quoting Buschman v. Schweiker, 

, 358
(9th Cir. 1982)); see New York, New Haven & Hartford R.R.
First Mortgage 4% Bondholders Comm. v. United States, 

, 440 (S.D.N.Y. 1968) (Friendly, J.) (remanding
where an agency’s erroneous factual finding “went to the
heart of its analysis”) (“Bondholders Comm.”); cf. Carnegie
Natural Gas Co. v. FERC, 

, 1294 (D.C. Cir.
1992) (“We will . . . sustain an agency decision resting on
several independent grounds if any of those grounds validly
supports the result, unless there is reason to believe the com-
bined force of these otherwise independent grounds influ-
enced the outcome.”).
            TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR                   5959
   We thus must determine whether the Secretary’s stated rea-
sons — after setting aside the erroneous persistence finding
— would have persuaded him that the lizard’s lost historical
range is not significant. The Secretary’s conclusion that the
lizard’s lost range holds no critical genetic value for the spe-
cies finds some support in the record, as does his determina-
tion that much of the lizard’s lost historical range was
converted to other uses decades ago and is thus not recoverable.13
Neither reason is entirely dependent on lizard persistence.

   [7] Nonetheless, the 2006 withdrawal repeatedly refers to
lizard persistence as persuasive evidence that the species’ lost
historical range is not significant. The Secretary offers persis-
tence as both an independent, and indeed primary, basis for
discounting the importance of lost range, and as support for
several other key conclusions. Because a reliance on the liz-
ard’s persistence throughout most of its current range cuts to
“the heart of [the agency’s] analysis,” Bondholders 

, we cannot readily say that the erroneous
finding clearly had no bearing on the Secretary’s ultimate
decision to withdraw the proposed listing. On remand, the
Secretary may be persuaded that, absent reliable evidence of
population persistence, the lizard’s lost historical range is
indeed significant.

II.    Threats to the Lizard’s Current Range

   [8] The 2003 withdrawal reports that the Coachella Valley
in California is the only segment of the lizard’s current range
where the species is in immediate jeopardy. See 68 Fed. Reg.
at 345. The Secretary concluded that the likely extinction of
  13
     The conclusion that the overall lost range represents a relatively small
portion of the lizard’s overall range (approximately 23%) — although not
entirely dependent on lizard persistence — amounts to a conclusory and
unhelpful factor for determining the significance of the species’ lost range.
That is, the Secretary has not explained why the loss of nearly one-quarter
of the species overall range is “relatively small.”
5960      TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR
the Coachella Valley lizard population is not significant
because of the population’s relatively small size (approxi-
mately sixteen thousand acres, or, one percent if the lizard’s
remaining range, 68 Fed. Reg. at 334), isolation from other
remaining lizard populations, and lack of importance for
maintenance of genetic diversity. See 

In essence,
the Secretary determined that the Coachella Valley is home to
a very small (in comparison with the three other segments of
the lizard’s remaining domestic range) non-unique lizard pop-
ulation. As such, the demise of the Coachella Valley popula-
tion would not be significant within the meaning of the ESA.
This is a reasonable approach to assessing the significance of
threatened range, and one that complies with Defenders. 

(requiring only that the Secretary “explain
[his] conclusion that the area in which [the lizard] can no lon-
ger live is not a ‘significant portion of its range’ ”).

   Plaintiffs also challenge the Secretary’s assessment of
threats to the lizard’s remaining habitat outside of the Coa-
chella Valley.

   [9] First, they object to the Secretary’s assertion that the
Conservation Agreement and management strategy have
diminished threats to the lizard’s remaining habitat on public
lands, and emphasize the management strategy’s slow and
still incomplete implementation. The Secretary acknowledges
that the Conservation Agreement has not yet been fully imple-
mented, but points to specific conservation benefits that the
agreement has achieved since it came into being in 1997. See
68 Fed. Reg. at 347 (noting that pesticide use and OHV racing
have been limited in the lizard MAs). Moreover, the 2003
withdrawal states that its assessment of threats to the species’
current range is not “dependent on full implementation” of the
Conservation Agreement’s management strategy. 

one of the four independent biologists that reviewed
FWS’ assessment of the lizard’s viability, offers support for
the Secretary’s position. He opined that the Conservation
Agreement’s protective measures alone were “adequate” to
             TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR                    5961
protect the species’ remaining domestic range (outside of the
Coachella Valley). We conclude that the limited benefits that
the 2003 withdrawal points to are supported by the record,
and the Secretary did not err in taking the Conservation
Agreement into account.

   Next, Plaintiffs argue that OHV use in the lizard’s current
range presents a much greater threat to the species than the
Secretary acknowledges. Plaintiffs’ argument, however, relies
on inferences from indeterminate scientific evidence. Plain-
tiffs have shown that OHV use is on the rise throughout the
lizard’s remaining range.14 The available studies analyzing
OHV-related impacts on the lizard, however, do not conclu-
sively show that OHV use amounts to a significant threat to
the species’ viability.15 Both Plaintiffs and the Secretary point
to scientific studies supporting their respective views on the
effects of OHVs, but the merits of the conflicting studies is
not a proper subject for this court to resolve. See Nw.
Ecosystem Alliance v. U.S. Fish & Wildlife Serv., 

, 1150 (9th Cir. 2007) (“[W]e must defer to the agency’s
interpretation of complex scientific data.”); see also Marsh v.
Or. Natural Res. Council, 

, 378 (1989) (“When
specialists express conflicting views, an agency must have
discretion to rely on the reasonable opinions of its own quali-
fied experts even if, as an original matter, a court might find
contrary views more persuasive.”).
  14
      A study that the Secretary relies on reports that over one-quarter of the
lizard’s remaining habitat in California is open to legal OHV recreation,
and illegal OHV recreation in closed areas is “not uncommon.”
   15
      Although the notion that lizards and OHVs can co-exist in the same
immediate area is counter-intuitive, the Secretary directs us to a recent
study reporting that lizards can be found in areas heavily impacted by
OHVs, and another reporting frequent lizard observations in an area of
high Border Patrol traffic. On the other hand, Plaintiffs point to a study
suggesting that fewer lizards can be found in areas where OHVs regularly
operate.
5962      TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR
   [10] Finally, Plaintiffs challenge the Secretary’s treatment
of scattered threats posed by energy and mineral development
projects, increased Border Patrol activity, and the possible
construction of large-scale infrastructure projects in the liz-
ard’s current range. Plaintiffs’ arguments follow the same
course as their attack on the agency’s analysis of OHV use.
In short, they have not presented conclusive evidence to rebut
the Secretary’s determination that such threats, either alone or
in concert, are not likely to cause the “destruction, modifica-
tion, or curtailment of [the species’] habitat or range.” 16
U.S.C. § 1533(a)(1)(A).

                         CONCLUSION

   For the reasons set forth above, we reverse and remand the
judgment of the district court with instructions that the matter
be further remanded to the Secretary so that the Secretary can
again consider whether to withdraw the proposed listing of
the lizard.

REVERSED and REMANDED.



NOONAN, Circuit Judge, concurring and dissenting:

   This case began in 1993 when the Secretary of the Interior
first proposed listing the species. After careful and conscien-
tious consideration by this court, it is now in 2009 remanded
to continue to be litigated for an indefinite time. The pattern
of the litigation is scarcely unfamiliar in environmental cases.
Congress has enacted law designed to conserve species of
wildlife threatened with extinction. 16 U.S.C. § 1531 et seq.
A federal agency has been entrusted with enforcement. Using
its expertise, the agency has determined what protection
should be afforded a particular species. Its determination has
been challenged by a private nonprofit organizations con-
cerned with the existence of the species. The district court has
          TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR           5963
heard the resulting litigation more than once, and this court
has heard it more than once. The various decisionmakers and
participants — the agency, the nonprofits, the district court,
and the court of appeals — are not motivated by private pas-
sion or grudge, but seek to see the fair application of broad
legislation to highly particularized and often elusive data. The
legal system does not confide the definitive judgment to the
agency entrusted with enforcement of the law but subjects
that judgment first to the challenges of the nongovernmental
organizations and then to the supervision of judges who are
not expert in the scientific matters at stake and not familiar
with the species whose survival is at stake.

   As if this interplay of governmental and private groups did
not create room for tension, misunderstandings, and passion-
ate disagreement, the problems in this case have been exacer-
bated by the simple absence of information. How many flat-
tailed horned lizards are there?

   No one knows the answer to that question. Nor does any-
one know how many lizards disappeared when portions of
their range disappeared. It is supposed that a diminution in
range correlates with a diminution in lizards. This hypothesis
is plausible. It has not been shown to be probable. Yet the
case turns on what measures are necessary to keep this
unknown population in existence. The court concludes that
the Secretary erred in finding that the lizard has not lost a sig-
nificant portions of its range. The old method of counting liz-
ards is out. A new method has not been tried very much. It’s
anybody’s guess whether the lizards are multiplying or declin-
ing. In a guessing contest one might defer to the government
umpire. The court, however, finds the Secretary’s conclusion
impacted by over-reliance on fragmenting evidence of the liz-
ard’s persistence; so the court decides to give the Secretary
another crack at the problem.

  If the Secretary does not know what the lizard population
was to begin with, or what it was in 1993, or what it is now
5964      TUCSON HERPETOLOGICAL SOCIETY v. SALAZAR
in May 2009, how will he know if it is increasing, staying the
same, or declining?

   A style of judging, familiar to readers of the old English
reports, characterizes the judge as dubitante. That is probably
the most accurate term for me, which leads me to concur in
the majority opinion insofar as it rejects the contentions of the
Tucson Herpetological Society and to dissent from the
remand whose command to the Secretary of the Interior is,
Guess again.