Nrdc v. Usepa

FOR PUBLICATION

UNITED STATES COURT OF APPEALS
FOR THE NINTH CIRCUIT

NATURAL RESOURCES DEFENSE No. 20-70787
COUNCIL; PESTICIDE ACTION
NETWORK NORTH AMERICA, EPA No.
Petitioners, EPA-HQ-OPP-
2009-0361
v.

U.S. ENVIRONMENTAL PROTECTION
AGENCY,
Respondent,

NATIONAL ASSOCIATION OF WHEAT
GROWERS; NATIONAL COTTON
COUNCIL OF AMERICA; AMERICAN
FARM BUREAU FEDERATION;
NATIONAL CORN GROWERS
ASSOCIATION; AMERICAN SOYBEAN
ASSOCIATION; NATIONAL SORGHUM
PRODUCERS; AGRICULTURAL
RETAILERS ASSOCIATION; NATIONAL
ASSOCIATION OF LANDSCAPE
PROFESSIONALS; MONSANTO
COMPANY; GOLF COURSE
SUPERINTENDENTS ASSOCIATION OF
AMERICA; AMERICAN SUGARBEET
GROWERS ASSOCIATION,
Intervenors.
2 NRDC V. USEPA

RURAL COALITION; ORGANIZACION No. 20-70801
EN CALIFORNIA DE LÍDERES
CAMPESINAS; FARMWORKER EPA No.
ASSOCIATION OF FLORIDA; BEYOND EPA-HQ-OPP-
PESTICIDES; CENTER FOR FOOD 2009-0361
SAFETY,
Petitioners,
OPINION
v.

U.S. ENVIRONMENTAL PROTECTION
AGENCY; MICHAEL REGAN, in his
official capacity as Administrator,
Respondents,

NATIONAL ASSOCIATION OF WHEAT
GROWERS; NATIONAL CORN
GROWERS ASSOCIATION; NATIONAL
COTTON COUNCIL; AMERICAN
SOYBEAN ASSOCIATION; AMERICAN
SUGARBEET GROWERS
ASSOCIATION; NATIONAL SORGHUM
PRODUCERS; AMERICAN FARM
BUREAU FEDERATION;
AGRICULTURAL RETAILERS
ASSOCIATION; NATIONAL
ASSOCIATION OF LANDSCAPE
PROFESSIONALS; GOLF COURSE
SUPERINTENDENTS ASSOCIATION OF
AMERICA; MONSANTO COMPANY,
Intervenors.
 NRDC V. USEPA 3

On Petition for Review of an Order of the
Environmental Protection Agency

Argued and Submitted January 10, 2022
Pasadena, California

Filed June 17, 2022

Before: J. Clifford Wallace, Danny J. Boggs, * and
Michelle T. Friedland, Circuit Judges.

Opinion by Judge Friedland

*
The Honorable Danny J. Boggs, United States Circuit Judge for
the U.S. Court of Appeals for the Sixth Circuit, sitting by designation.
4 NRDC V. USEPA

SUMMARY **

Pesticides

The panel (1) granted in part and denied in part a petition
for review challenging the U.S Environmental Protection
Agency’s decision determining that glyphosate, the active
ingredient in the weedkiller Roundup, does not pose “any
unreasonable risk to man or the environment”; and (2)
remanded to the agency for further consideration.

The Federal Insecticide, Fungicide, and Rodenticide Act
(“FIFRA”) requires the U.S Environmental Protection
Agency (“EPA”) to regulate pesticides, which are defined to
include herbicides. A pesticide product may not be
distributed or sold in the United States until EPA has issued
a registration pursuant to FIFRA. A registration functions as
a license setting forth the conditions under which the
pesticide may be sold, distributed, and used. The EPA may
not issue a registration for a pesticide that causes
“unreasonable adverse effects on the environment.” In 2007,
Congress added a new process called “registration review”
to the FIFRA scheme governing pesticides, instructing EPA
to periodically review pesticide registrations every fifteen
years. For pesticides registered before 2007, such as
glyphosate, EPA must complete the first registration review
by October 1, 2022.

EPA began its registration review of glyphosate in 2009
and completed a preliminary ecological risk assessment of
the pesticide in 2015. That assessment concluded that

**
This summary constitutes no part of the opinion of the court. It
has been prepared by court staff for the convenience of the reader.
 NRDC V. USEPA 5

glyphosate may pose certain risks to mammals and birds and
may adversely affect terrestrial and aquatic plants, primarily
from spray drift. The EPA also released a draft human-
health risk assessment and a paper about glyphosate’s
carcinogenic potential, entitled the Revised Glyphosate
Issue Paper: Evaluation of Carcinogenic Potential (“Cancer
Paper”), which concluded that glyphosate posed no serious
human-health risks and should be classified as “not likely to
be carcinogenic to humans.”

In January 2020, EPA issued an Interim Registration
Review Decision for glyphosate (“Interim Decision”),
which: (1) announced that its earlier draft human-health and
ecological risk assessments were final; (2) contained a brief
cost-benefit analysis concluding that the benefits
outweighed the potential ecological risks when glyphosate is
used according to label directions; and (3) laid out various
mitigation measures, in the form of label changes for
glyphosate products, to reduce the potential ecological
risks. According to the Interim Decision, EPA still planned,
among other things, to complete an assessment of
glyphosate’s effect on endangered and threatened species,
pursuant to the Endangered Species Act (“ESA”).

Two groups of petitioners filed petitions for review of
the Interim Decision: one led by Rural Coalition and the
other led by Natural Resources Defense Council
(“NRDC”). Rural Coalition’s petition made two attacks on
the Interim Decision. It challenged EPA’s conclusions on
human health and insisted that EPA should have followed
the ESA’s procedural requirements before issuing the
Interim Decision. NRDC’s petition primarily challenges
EPA’s ecological risk assessment, cost-benefit analysis, and
risk-mitigation requirements.
6 NRDC V. USEPA

The panel first considered Rural Coalition’s challenge to
EPA’s conclusion that glyphosate poses “no risks to human
health.” That conclusion rested in important part on EPA’s
determination, explained in its Cancer Paper, that glyphosate
was not likely to be carcinogenic to humans. The panel held
that EPA’s conclusion was in tension with parts of the
agency’s own analysis and with the 2005 Guidelines for
Carcinogen Risk Assessment (“Cancer Guidelines”), which
EPA purported to follow. The panel noted that earlier in the
Cancer Paper, EPA had explained that a conclusion
regarding the association between glyphosate exposure and
risk of non-Hodgkin’s lymphoma (“NHL”) could not be
determined based on the available evidence. The panel
stated that EPA could not reasonably treat its inability to
reach a conclusion about NHL risk as consistent with a
conclusion that glyphosate is not likely to cause cancer
within the meaning of the Cancer Guidelines. Because
inconsistent reasoning cannot survive substantial-evidence
review, the panel concluded that EPA’s determination that
glyphosate was not likely to be carcinogenic was not
supported by substantial evidence. The panel therefore
vacated the human-health portion of the EPA’s Interim
Decision and remanded for further analysis and
explanation. Given that vacatur, the panel did not reach
Rural Coalition’s arguments of other errors pertaining to
human health or NRDC’s petition challenging the public-
comment process that informed the human health portion of
the Interim Decision.

The panel next addressed Rural Coalition’s claim
alleging that EPA impermissibly failed to follow the ESA
consultation procedures before issuing the Interim
Decision. The ESA protects endangered and threatened
species, in part, by requiring federal agencies to consult with
the U.S. Fish and Wildlife Service or the National Marine
 NRDC V. USEPA 7

Fisheries Service. The consultation procedures begin with
an agency reviewing its actions at the earliest possible time
to determine whether any action may affect listed species or
critical habitat, resulting in an effects determination.

The panel determined that Rural Coalition had standing
to bring an ESA claim. Rural Coalition’s members
submitted declarations stating that they regularly engaged in
educational and recreational activities involving a variety of
endangered species and that glyphosate was threatening their
interests by exposing those species to toxic runoff and
residues on vegetation. Members therefore had cognizable
interests for purposes of standing. Rural Coalition also
established causation by showing that EPA might have
required more mitigation efforts had the agency completed
the ESA’s procedures before issuing the Interim Decision
and redressability by showing that, at the time the petition
was filed, court-ordered relief was possible. The panel
rejected intervenor Monsanto’s argument that EPA’s recent
consultation efforts mooted the case.

Turning to the merits of the ESA claim, the panel held
that EPA’s registration review decision under FIFRA was an
“action” that triggered the ESA’s consultation requirement;
EPA actively exercised its regulatory power, completing an
assessment of glyphosate’s risks under FIFRA and
delineating what constituted acceptable glyphosate use
under the statute’s safety standard. EPA therefore had to
comply with the ESA by making an effects determination
before issuing the decision. It was undisputed that EPA did
not do so. Accordingly, EPA violated the
ESA. Nevertheless, the panel declined to order relief for the
ESA violation, noting that, according to the timeline
imposed by Congress, EPA must complete its final
registration review decision—including formal
8 NRDC V. USEPA

consultation—by October 2022. Given that the FIFRA
deadline was fast approaching, shortening EPA’s time to
consult would be only moderately beneficial to Rural
Coalition but potentially very disruptive to the agency. The
panel declined to vacate the Interim Decision, other than to
the extent specified regarding the human-health portion,
because it was not clear that vacatur would be beneficial; the
Interim Decision included certain mitigation efforts
designed to limit the ecological impact of glyphosate use,
and vacatur would eliminate those mitigation requirements.

The remaining issue involved petitioners’ challenges to
the Interim Decision’s ecological risk assessment,
determination of glyphosate’s costs, cost-benefit analysis,
and mitigation requirements (collectively, the “ecological
portion”), and EPA’s responsive motion for remand. The
panel granted EPA’s motion to remand without vacatur as to
the ecological portion of the decision but required EPA to
issue a new ecological portion by the October 2022 FIFRA
deadline. Because the panel granted EPA’s motion, it did
not reach the parts of NRDC’s and Rural Coalition’s
petitions that challenged the remanded portion of the Interim
Decision.
 NRDC V. USEPA 9

COUNSEL

Amy van Saun (argued), George A. Kimbrell, and Ryan D.
Talbott, Center for Food Safety, Portland, Oregon, for
Petitioners Rural Coalition, Organización en California de
Líderes Campesinas, Farmworker Association of Florida;
Beyond Pesticides, and Center for Food Safety.

Lucas Rhoads (argued), Aaron Colangelo, and Tom
Zimpleman, Natural Resources Defense Council,
Washington, D.C., for Petitioners Natural Resources
Defense Council and Pesticide Action Network North
America.

Philip R. Dupre (argued), Attorney; Robert Williams, Senior
Trial Attorney; Benjamin Carlisle, Senior Attorney; Bruce
S. Gelber, Deputy Assistant Attorney General; Jean E.
Williams, Acting Assistant Attorney General; Environment
and Natural Resources Division, United States Department
of Justice, Washington, D.C.; Devi Chandrasekaran and
Forrest Pittman, Attorney Advisors, Office of General
Counsel, United States Environmental Protection Agency,
Washington, D.C.; for Respondent.

Richard P. Bress (argued), Philip J. Perry, Stacey L.
VanBelleghem, and Andrew D. Prins, Latham & Watkins
LLP, Washington, D.C., for Intervenors.

Shannen W. Coffin, Sara Beth Watson, and Mark C.
Savignac, Steptoe & Johnson LLP, Washington, D.C., for
Amicus Curiae CropLife America.
10 NRDC V. USEPA

OPINION

FRIEDLAND, Circuit Judge:

Glyphosate, the active ingredient in Roundup, is the
nation’s most heavily used weedkiller. The Environmental
Protection Agency (“EPA”) recently assessed whether
glyphosate poses “any unreasonable risk to man or the
environment” and answered, for the most part, “no.” A
group of petitioners challenged EPA’s decision, arguing,
among other things, that EPA did not adequately consider
whether glyphosate causes cancer and shirked its duties
under the Endangered Species Act (“ESA”). We agree and
remand to the agency for further consideration.

I.

A.

The Federal Insecticide, Fungicide, and Rodenticide Act
(“FIFRA”) requires EPA to regulate pesticides, which are
defined to include herbicides. 1 See 7 U.S.C. §§ 136 et seq.
FIFRA’s primary regulatory mechanism is called
“registration.” Id. § 136a(a). A pesticide product may not
be distributed or sold in the United States until EPA has
issued a registration, which functions as a license setting
forth the conditions under which the pesticide may be sold,
distributed, and used. See id. § 136a. Those conditions

1
Under FIFRA, a “pesticide” includes both “any substance or
mixture of substances intended for preventing, destroying, repelling, or
mitigating any pest” as well as “any substance or mixture of substances
intended for use as a plant regulator, defoliant, or desiccant.” 7 U.S.C.
§ 136(u).
 NRDC V. USEPA 11

include labeling requirements with directions for proper use.
Id. § 136a(c); see 40 C.F.R. § 156.10.

EPA may not issue a registration for a pesticide that
causes “unreasonable adverse effects on the environment.”
7 U.S.C. § 136a(c)(5)(C); see also 40 C.F.R. § 152.112(e).
“[U]nreasonable adverse effects on the environment”
include “any unreasonable risk to man or the environment,
taking into account the economic, social, and environmental
costs and benefits of the use of any pesticide.” 7 U.S.C.
§ 136(bb). This is commonly referred to as the FIFRA safety
standard.

In 2007, Congress added a new process called
“registration review” to the FIFRA scheme governing
pesticides, instructing EPA to “periodically review[]”
pesticide registrations every fifteen years. Id.
§ 136a(g)(1)(A). For pesticides registered before 2007, such
as glyphosate, EPA must complete the first registration
review by October 1, 2022. Id.

EPA has promulgated regulations delineating an
elaborate process for registration review. 40 C.F.R.
§§ 155.23–155.58. The regulations require EPA to assess
any new information regarding risks to human health and the
environment that has emerged since EPA last issued a
registration decision for a pesticide to verify that the
pesticide continues to satisfy the FIFRA safety standard.
See, e.g., id §§ 155.40, 155.53(a). The process concludes
with a registration review decision, which conveys “the
Agency’s determination whether a pesticide meets, or does
not meet,” the FIFRA safety standard. Id. § 155.57. The
regulations also permit, but do not require, EPA to issue an
“interim registration review decision” prior to the
registration review decision. Id. § 155.56. “[T]he interim
registration review decision may require new risk mitigation
12 NRDC V. USEPA

measures, impose interim risk mitigation measures, identify
data or information required to complete the review, and
include schedules for . . . completing the registration
review.” Id.

If EPA finds that a pesticide does not satisfy the FIFRA
safety standard, EPA may initiate cancellation proceedings
to rescind a pesticide’s registration, 7 U.S.C.
§§ 136a(g)(1)(A)(v), 136d(b); 40 C.F.R. § 155.40(a)(2), or
may require mitigation measures to reduce risk to acceptable
levels, see 40 C.F.R. § 155.58.

B.

Glyphosate is a chemical that kills a broad range of
plants by inhibiting an important enzyme. EPA registered
the first glyphosate product in 1974, when Monsanto, an
agrochemical and agricultural biotechnology company,
sought to sell the now-well-known weedkiller Roundup.
During its first two decades on the market, Roundup had
limited utility to farmers because it killed all vegetation in
an application area. But in the mid-1990s, Monsanto
developed a “Roundup Ready” crop system, selling
Roundup along with seeds genetically modified to tolerate
glyphosate. The system allowed farmers to apply glyphosate
over genetically modified crops, killing weeds but leaving
the crops unharmed. As a result, glyphosate use
skyrocketed. The nationwide acreage across which
glyphosate is currently used is roughly equivalent to three
times the size of California.

Glyphosate is generally applied by being sprayed from
planes, ground equipment, or handheld devices. Workers
and residential users are exposed to glyphosate when, for
example, they handle the chemical during application or
enter areas where it was recently sprayed. People are also
 NRDC V. USEPA 13

exposed to glyphosate when they eat food from crops treated
with it.

Whether these exposures create health risks has become
a hotly debated and litigated issue. Health concerns
proliferated when the International Agency for Research on
Cancer (“IARC”), a subdivision of the World Health
Organization, classified glyphosate as “probably
carcinogenic to humans” in 2015. IARC’s conclusion
stemmed in part from scientific studies that found an
association between glyphosate exposure and non-
Hodgkin’s lymphoma (“NHL”), a type of cancer that affects
white blood cells. The IARC classification spurred a wave
of lawsuits against Monsanto. Since 2015, tens of thousands
of individuals with NHL have sued Monsanto in state and
federal court, alleging that Roundup caused their illnesses.
See, e.g., In re Roundup Prods. Liab. Litig., 544 F. Supp. 3d.
950, 953 (N.D. Cal. 2021). Monsanto lost the first three
lawsuits to go to trial, and the plaintiffs were awarded tens
of millions of dollars. Id. at 955–57.

C.

EPA began its registration review of glyphosate in
2009. 2 In September 2015, the agency completed a
preliminary ecological risk assessment of the pesticide. The
assessment considered glyphosate’s effects on all “non-
target organisms”—that is, animals and plants not intended
to be killed by the pesticide. EPA concluded that glyphosate

2
For registration review, EPA may evaluate a “pesticide case . . .
composed of 1 or more active ingredients and the products associated
with the active ingredients” or may evaluate each pesticide product
registration individually. 7 U.S.C. § 136a(g)(1)(A)(iii). Here, EPA
decided to conduct registration review on glyphosate, an active
ingredient.
14 NRDC V. USEPA

may pose certain risks to mammals and birds. EPA also
determined that glyphosate may adversely affect terrestrial
and aquatic plants, primarily from spray drift.

Meanwhile, EPA was working on a human-health risk
assessment and, in particular, an analysis of glyphosate’s
carcinogenic potential. EPA’s pesticide unit made a
preliminary determination that glyphosate was not likely to
be carcinogenic and shared that determination with the
agency’s Office of Research and Development (“ORD”). In
December 2015, ORD offered comments in response,
including criticisms of the pesticide unit’s approach to
reviewing epidemiological studies—specifically, studies of
human populations investigating whether glyphosate
exposure causes cancer. ORD commented that the pesticide
unit seemed to “dichotomize” such studies as “either ‘causal’
or ‘not causal’” rather than recognize “gradations of
causality.” According to ORD, that approach contravened
the “[f]rameworks for data analysis and causal
determinations” employed by “the risk assessment
community” and “by EPA” in its 2005 Guidelines for
Carcinogen Risk Assessment (“Cancer Guidelines” or
“Guidelines”). The Cancer Guidelines are intended to guide
EPA in classifying chemicals according to their carcinogenic
potential. After stating its methodological concerns, ORD
expressed disagreement with the pesticide unit’s
determination that glyphosate was “not likely to be
carcinogenic.”

ORD’s criticisms did not change EPA’s overall “not
likely” determination, and, in September 2016, EPA
defended that determination in a draft paper entitled
Glyphosate Issue Paper: Evaluation of Carcinogenic
Potential. The agency requested feedback on that draft from
an EPA-commissioned Scientific Advisory Panel (“SAP”).
 NRDC V. USEPA 15

The SAP published a report in response. Many of the SAP’s
comments were similar to ORD’s, but the SAP focused on
EPA’s treatment of laboratory studies that examined
whether glyphosate causes tumors in rodents, rather than on
the epidemiological studies of human health that ORD had
emphasized. The SAP “concluded that the EPA evaluation
does not appear to follow the [Cancer Guidelines] in several
ways.” The SAP also criticized the criteria EPA used to
discount tumor results in rodent studies, opining that EPA’s
approach was not “a conservative approach for public health
protection” and was “not advisable” because it was “not
consistent with . . . standard ways in which . . . results are
typically interpreted.”

Ultimately, the SAP was divided as to whether EPA’s
“not likely” determination was appropriate. According to
the report, “[m]any Panel members believe[d] that the EPA
did not provide convincing evidence of a lack of
carcinogenic effects.” These panelists thought that the
rodent studies alone provided suggestive evidence of
carcinogenic potential. Some panelists, however, argued
that results from those studies “are consistent with what
would be expected by chance and not reflective of
[glyphosate]-induced effects,” emphasizing the “wealth of
[rodent] studies with insufficiently consistent findings” and
an inability to “definitively link[]” the “positive
[epidemiological] results . . . to glyphosate-exposure.”

One year after receiving the SAP’s feedback, EPA
released a draft human-health risk assessment for glyphosate
and an updated and final paper about glyphosate’s
carcinogenic potential, now entitled the Revised Glyphosate
Issue Paper: Evaluation of Carcinogenic Potential (“Cancer
Paper”). In the draft risk assessment, EPA concluded that
glyphosate poses no serious human-health risks, stating, for
16 NRDC V. USEPA

instance, that “[g]lyphosate exhibits low toxicity across
species, durations, life stages, and routes of exposure.” EPA
also concluded that “glyphosate should be classified as ‘not
likely to be carcinogenic to humans’” and explained that
conclusion in the Cancer Paper. Separately, EPA responded
to the SAP’s criticisms, revealing that those criticisms had
prompted very few changes between the earlier draft and the
finalized Cancer Paper.

In January 2020, EPA issued an Interim Registration
Review Decision for glyphosate (“Interim Decision”). The
Interim Decision had three main components. First, the
Interim Decision announced that the earlier draft human-
health and ecological risk assessments were now final—with
no changes from those drafts. In summarizing the human-
health risk assessment, the Interim Decision explained that
the agency “determined that there are no risks to human
health from the current registered uses of glyphosate and that
glyphosate is not likely to be carcinogenic to humans.” The
Interim Decision directed readers to the human-health risk
assessment and to the Cancer Paper for additional
information. According to EPA, there were “[n]o additional
human health data needs” for glyphosate’s registration
review. The Interim Decision then reaffirmed the ecological
risk assessment, confirming that “potential risks of concern
were identified for mammals and birds” as well as for
“terrestrial and aquatic plants.”

Second, the Interim Decision contained a brief cost-
benefit analysis. EPA reiterated that glyphosate poses
potential risks to mammals, birds, and plants. It also
summarized glyphosate’s various benefits, such as its ability
to provide a broad spectrum of weed control across
agricultural and non-agricultural sites and its low cost. EPA
concluded that “the benefits outweigh the potential
 NRDC V. USEPA 17

ecological risks when glyphosate is used according to label
directions.”

Third, the Interim Decision laid out various mitigation
measures, in the form of label changes for glyphosate
products, to reduce the potential ecological risks. One label
change involves application restrictions to reduce spray drift.
Another label change alerts users that glyphosate has the
potential to harm non-target organisms. A final label change
warns of the risk that glyphosate use can cause herbicide
resistance.

According to the Interim Decision, only three steps
remained before EPA would conclude registration review.
First, EPA planned to complete an assessment of
glyphosate’s effect on endangered and threatened species,
pursuant to the ESA. As necessary based on that assessment,
EPA would then consult with the U.S. Fish and Wildlife
Service or the National Marine Fisheries Service to mitigate
any adverse effects on those species. 3 Second, EPA planned

3
Since the Interim Decision issued, EPA began following the ESA’s
procedures. The first step is to determine whether an agency action “may
affect” an endangered or threatened species or critical habitat. In
November 2020, EPA completed a preliminary effects determination,
publishing a draft Biological Evaluation that assessed potential effects
from all registered uses of glyphosate on ESA-listed species. It found
that glyphosate “may affect” all listed species experiencing glyphosate
exposure—that is 1,795 endangered or threatened species. In November
2021, EPA issued a final Biological Evaluation with similar conclusions.
Under the ESA, a “may affect” determination triggers a requirement that
the agency consult with the relevant wildlife agencies to prevent adverse
effects. See Nat’l Fam. Farm Coal. v. EPA, 966 F.3d 893, 922 (9th Cir.
2020). Thus, EPA is now consulting with those agencies. To the extent
the draft Biological Evaluation and final Biological Evaluation are not
part of the record before us, we take judicial notice of them. See Dine
18 NRDC V. USEPA

to address a petition that had been filed under the Federal
Food, Drug, and Cosmetic Act (“FFDCA”), requesting that
EPA restrict glyphosate’s use on oats to reduce dietary
exposure to the herbicide. And third, EPA planned to
conduct an endocrine analysis of glyphosate pursuant to the
FFDCA. 4

D.

In March 2020, two groups of petitioners filed petitions
for review of the Interim Decision: one led by Rural
Coalition and the other led by Natural Resources Defense
Council (“NRDC”). Rural Coalition’s petition makes two
attacks on the Interim Decision. It challenges EPA’s
conclusions on human health and insists that EPA should
have followed the ESA’s procedural requirements before
issuing the Interim Decision. NRDC’s petition primarily
challenges EPA’s ecological risk assessment, cost-benefit
analysis, and risk-mitigation requirements, though NRDC
also asserts that EPA failed to address NRDC’s comments
on human-health risks made during the public-comment
period. We consolidated the petitions and granted a motion
to intervene by Monsanto and various agricultural and
landscaping groups (collectively, “Monsanto”).

In May 2021, EPA filed its answering brief, which
addresses only its human-health findings, along with a
motion for voluntary partial remand without vacatur. EPA
seeks partial remand of the portions of the Interim Decision

Citizens Against Ruining Our Env’t v. Bureau of Indian Affs., 932 F.3d
843, 848 n.1 (9th Cir. 2019).
4
An endocrine analysis strives to determine whether a substance is
an endocrine disruptor—for example, whether it has effects in humans
or wildlife similar to those of naturally occurring estrogen.
 NRDC V. USEPA 19

related to glyphosate’s ecological risks as well as the
agency’s cost-benefit analysis. EPA’s answering brief and
Monsanto’s brief do not substantively address those issues
but do offer defenses to Petitioners’ challenges to EPA’s
human-health analysis and the alleged failure to comply with
the ESA. NRDC agreed that the remand requested by EPA
would be appropriate, but Rural Coalition opposed any
remand.

We heard oral argument in January 2022.

II.

Under FIFRA, we review EPA’s Interim Decision for
“substantial evidence when considered on the record as a
whole.” NRDC v. EPA, 857 F.3d 1030, 1035–36 (9th Cir.
2017) (quoting 7 U.S.C. § 136n(b)). This standard requires
the administrative record to show “such relevant evidence as
a reasonable mind might accept as adequate to support a
conclusion even if it is possible to draw two inconsistent
conclusions from the evidence.” Id. at 1036 (quoting NRDC
v. EPA, 735 F.3d 873, 877 (9th Cir. 2013)). The agency’s
reasoning must also be coherent and internally consistent.
See NRDC v. EPA, 31 F.4th 1203, 1210 (9th Cir. 2022)
(relying on internal “inconsistencies” in holding that a
decision was not supported by substantial evidence); Lott v.
Colvin, 772 F.3d 546, 551 (8th Cir. 2014) (holding that a
decision marked by “internal inconsistencies” was not
supported by substantial evidence); Linear Tech. Corp. v.
Int’l Trade Comm’n, 566 F.3d 1049, 1065 (Fed. Cir. 2009)
(holding that a decision based on “[i]nconsistent[]” rulings
and “contradictory statement[s]” was not supported by
substantial evidence).

The Administrative Procedure Act (“APA”) governs
judicial review of administrative decisions involving the
20 NRDC V. USEPA

ESA. See Nat’l Fam. Farm Coal. v. EPA, 966 F.3d 893, 923
(9th Cir. 2020). Under the APA, courts “shall . . . hold
unlawful and set aside agency action, findings, and
conclusions” that are “arbitrary, capricious, an abuse of
discretion, or otherwise not in accordance with law” or
“without observance of procedure required by law.” 5
U.S.C. § 706; see NRDC v. Houston, 146 F.3d 1118, 1125
(9th Cir. 1998).

III.

We first consider Rural Coalition’s challenge to EPA’s
conclusion that glyphosate poses “no risks to human health.”
That conclusion rests in important part on EPA’s
determination, explained in its Cancer Paper, “that
glyphosate is not likely to be carcinogenic to humans.”
Rural Coalition contests the Cancer Paper’s reasoning,
primarily arguing that EPA contravened the Cancer
Guidelines it purported to follow. We agree.

EPA’s Cancer Guidelines lay out four steps for
conducting risk assessments of chemicals’ carcinogenic
potential. The first step—and the one most relevant here—
is hazard identification, which asks whether a chemical can
“present a carcinogenic hazard to humans and, if so, under
what circumstances.” The second step considers the “dose
response” to a chemical—in other words, the levels of
exposure at which adverse effects might occur. The third
step assesses “the conditions of human exposure.” The
fourth and final step evaluates “the character of the risk,”
including “[h]ow well . . . data support conclusions about the
nature and extent of the risk from various exposures.”

For the first step, hazard identification, the Guidelines
lay out strategies for reviewing and evaluating data from
human and animal studies. For example, the Guidelines
 NRDC V. USEPA 21

include criteria for identifying reliable epidemiological
studies as well as factors to consider when determining
whether observed effects in such studies are causal. The
Guidelines also provide methods for analyzing tumor data
from animal laboratory studies, including tests for
determining whether results are statistically significant. In
addition, the Guidelines identify potential observations, such
as cellular metastases or tumors detected in multiple species,
whose presence or absence should add to or detract from the
weight of studies’ findings.

The culmination of the hazard-identification step is a
“weight of evidence narrative.” According to the Cancer
Guidelines, that narrative should explain the available
evidence and summarize how the evidence supports a
conclusion about human carcinogenic potential. The
Guidelines lay out five standard hazard descriptors for
expressing such a conclusion, with criteria for when each
applies: “Carcinogenic to Humans,” “Likely to Be
Carcinogenic to Humans,” “Suggestive Evidence of
Carcinogenic Potential,” “Inadequate Information to Assess
Carcinogenic Potential,” and “Not Likely to Be
Carcinogenic to Humans.” Although the choice of a
descriptor is holistic and does not automatically compel an
ultimate FIFRA conclusion, EPA’s choice among those
descriptors generally has practical consequences. As the
parties explained at oral argument, hazard descriptors
indicating higher levels of risk usually prompt more
mitigation efforts—for example, labeling requirements
intended to protect human health.

EPA ties itself to the Cancer Guidelines in its Interim
Decision. The Interim Decision relies on the Cancer Paper
to explain its conclusions about human health. And, in that
Cancer Paper, EPA explicitly states that it is completing an
22 NRDC V. USEPA

“Evaluation of Cancer Classification per the 2005 EPA
Guidelines for Carcinogen Risk Assessment [i.e., the Cancer
Guidelines].” The Cancer Paper directly quotes the Cancer
Guidelines’ language about the five hazard descriptors.

In the Cancer Paper, the agency ultimately concludes
that “[t]he strongest support is for [the hazard descriptor]
‘not likely to be carcinogenic to humans.’” As the Cancer
Paper and the Cancer Guidelines explain, the “not likely”
descriptor “is appropriate when the available data are
considered robust for deciding that there is no basis for
human hazard concern.” According to EPA’s conclusion in
the Cancer Paper, glyphosate is “not likely” to be
carcinogenic to humans because animal-tumor and
genotoxicity studies showed no reason for concern. 5 But this
conclusion is in tension with parts of the agency’s own
analysis and with the guidelines it purports to follow.

A.

EPA’s choice of the “not likely” descriptor conflicts with
a determination EPA made earlier in the Cancer Paper.
Earlier, EPA explained that “a conclusion regarding the
association between glyphosate exposure and risk of NHL
cannot be determined based on the available evidence.”

In coming to that determination, the Cancer Paper
discussed human epidemiological studies showing what
could be considered suggestive evidence that glyphosate
exposure causes NHL. For example, the Cancer Paper stated
that “reported effect estimates across case-control studies
and the associated meta-analyses [were] greater than 1,”

5
The Cancer Paper explains that genotoxicity studies examine
whether chemicals damage genetic material on a cellular level.
 NRDC V. USEPA 23

meaning that most studies EPA examined indicated that
human exposure to glyphosate is associated with an at least
somewhat increased risk of developing NHL. 6 The Cancer
Paper also acknowledged that some epidemiological studies
provide evidence of an exposure-response relationship
between glyphosate and NHL. One study, for instance,
indicated that there was an increased risk of NHL for those
with more than ten years of glyphosate exposure. In
addition, that same study as well as another indicated that
those who are exposed to relatively more glyphosate in a
year face a higher risk of NHL.

But EPA discounted epidemiological studies showing
increased NHL risk by concluding that “chance and/or bias”
could be “an explanation for observed associations in the
database.” EPA stated that some of the study results
showing increased NHL risk were not statistically
significant, which raises the concern that those results were
due to chance. EPA raised the possibility that confounders,
such as exposure to other pesticides, animals, or diesel
fumes, were driving the NHL results. EPA also emphasized
what it deemed “contradictory results”—that a few studies
did not detect a positive association between glyphosate
exposure and NHL or an exposure-response relationship.
EPA opined that such inconsistencies and limitations

6
These meta-analyses—which aggregate and analyze the results
from individual studies—quantify the increased risk found across the
many case-control studies EPA considered. See Definition: meta-
analysis, National Cancer Institute, https://www.cancer.gov/
publications/dictionaries/cancer-terms/def/meta-analysis. The effect
estimates from the meta-analyses range from 1.3 to 1.5, indicating that
those exposed to glyphosate were 30 to 50 percent more likely to develop
NHL.
24 NRDC V. USEPA

precluded it from coming to any firm determination on
glyphosate’s potential to cause NHL.

As Rural Coalition correctly argues, EPA’s own
conclusion from that epidemiological evidence is
inconsistent with its ultimate selection of the “not likely”
hazard descriptor. According to EPA’s Cancer Guidelines,
that hazard descriptor is appropriate when the agency
determines that “available data are considered robust for
deciding that there is no basis for human hazard concern.”
EPA therefore cannot reasonably treat its inability to reach a
conclusion about NHL risk as consistent with a conclusion
that glyphosate is “not likely” to cause cancer within the
meaning of the Cancer Guidelines.

We made a similar point in Pollinator Stewardship
Council v. EPA, 806 F.3d 520 (9th Cir. 2015), in rejecting
an argument by EPA “that since the studies are inconclusive
as to the risks of sulfoxaflor for bees, the studies
affirmatively prove that sulfoxaflor does not cause
unreasonable adverse effects on bees.” Id. at 531. We
explained that “[n]either logic nor precedent can sustain this
position. We have previously held that an agency cannot
rely on ambiguous studies as evidence of a conclusion that
the studies do not support.” Id.; see also League of United
Latin Am. Citizens v. Regan, 996 F.3d 673, 701 (9th Cir.
2021) (“[Because] EPA represents that there are
‘uncertainties concerning the impact of chlorpyrifos on
children’ . . . EPA has not determined, and on this record
reasonably could not determine to a ‘reasonable certainty’
that aggregate chlorpyrifos exposures under the current
tolerances pose no risk of harm.”).
 NRDC V. USEPA 25

B.

The analysis underpinning EPA’s “not likely” descriptor
is also flawed in various other ways. EPA relies on two main
propositions to support its chosen hazard descriptor, but
neither withstands scrutiny under the agency’s own
framework.

1.

EPA’s first proposition is that “the agency did not
consider any of the tumors observed in the animal
carcinogenicity studies to be treatment-related” (i.e., caused
by glyphosate). According to EPA, “none of the tumors
evaluated in individual rat and mouse carcinogenicity
studies are treatment-related due to lack of pairwise
statistical significance, lack of a monotonic dose response,
absence of preneoplastic or related non-neoplastic lesions,
no evidence of tumor progression, and/or historical control
information (when available).” But EPA’s reliance on at
least two of these indicia to infer a lack of treatment-related
effects—historical-control data and pairwise statistical
significance—conflicts with the Guidelines that the agency
contends it is following. 7

EPA’s Cancer Paper uses historical-control data
selectively and in a manner that is inconsistent with the
Cancer Guidelines. Historical-control data show the natural
frequency of different types of tumors in an animal strain.
As the Cancer Paper acknowledges, the Cancer Guidelines

7
EPA’s flawed use of two of the indicia to infer a lack of treatment-
related effects is sufficient to undermine the agency’s assessment of the
rodent studies it examined. EPA relied upon these indicia so often
throughout the Cancer Paper that it is impossible to know what
conclusion EPA would have reached without them.
26 NRDC V. USEPA

instruct that historical-control data can “add to the analysis,
particularly by enabling identification of uncommon tumor
types or high spontaneous incidence of a tumor in a given
animal strain.” For example, according to the Guidelines,
historical-control data that show a particular type of tumor is
very rare in an animal strain could indicate that, when tumors
of that type do occur in a study using that strain, “the result
is in fact unlikely to be due to chance.” By contrast, the
Guidelines explain that historical-control data showing
“common tumors” in an animal strain could reduce the
importance of “results that are barely statistically significant
or in which incidence rates in concurrent controls are
unusually low in comparison with historical controls.”

Rather than using historical-control data both when the
data bolster and when the data undermine studies’ results, as
would be supported by the Cancer Guidelines, EPA uses this
type of data only to discount studies indicating that
glyphosate may cause tumors. According to the SAP, there
were numerous instances in which historical-control data
could add weight to tumor findings, but EPA never used the
data in that manner. As the SAP observed, “[t]o subjectively
choose to use historical control incidence data only in
situations where” it ultimately undermines tumor results “is
to potentially introduce biases.” Replying to the SAP’s
criticism, EPA states that it “recognize[d] the concerns
raised by the panel” and “[a]s a result, historical control data
have been presented for those studies where historical
control data are available from the performing laboratory for
the same species and strain for a study” in the revised Cancer
Paper. But although EPA presents more historical-control
data for various studies in the final Cancer Paper than in the
draft reviewed by the SAP, the agency did not change the
way in which it factored those data into its analysis: EPA
still uses the data only to undermine tumor results, even
 NRDC V. USEPA 27

ignoring the specific instances in which the SAP explained
that historical-control data should increase the import of
tumor results.

EPA’s reliance on a second indicium, “lack of pairwise
statistical significance,” is also inconsistent with the Cancer
Guidelines. Pairwise comparison and trend tests are two
different tests for assessing statistical significance, which is
an indication that a particular result is unlikely due to
chance. A pairwise comparison test asks whether tumor
incidence in a treatment group was higher than in the control
group, while a trend test asks whether tumor incidences in
all treatment groups increased as the glyphosate dose
increased. EPA acknowledges in its Cancer Paper that,
according to the Cancer Guidelines, “[s]ignificance in either
kind of test is sufficient to reject the hypothesis that chance
accounts for the result.” (emphasis added). But as both
ORD 8 and the SAP pointed out, in analyzing various rodent
studies, EPA discounts tumor incidences because those
incidences were not statistically significant in pairwise
comparison tests—when those same tumor incidences were
apparently statistically significant using trend tests.
Criticizing EPA’s approach, “the [SAP] noted that requiring
a significant pairwise comparison . . . in addition to a
significant trend is neither consistent with the [Cancer
Guidelines] nor a conservative approach for public health
protection.” (emphasis in original). 9

8
Rural Coalition moves to add another ORD document to the
administrative record. Because our resolution of these petitions does not
depend on that document, we DENY Rural Coalition’s motion.
9
If EPA had determined that these tumor results were not
statistically significant in trend tests either, one would expect EPA’s
summary of the animal studies to have pointed to a lack of statistical
28 NRDC V. USEPA

Responding to that criticism, EPA asserts that, although
its Cancer Guidelines correctly convey that satisfaction of
either a trend test or a pairwise test is sufficient for statistical
significance, they do “not imply that statistical significance
alone in an individual test is sufficient to determine that
observed tumors are treatment-related.” (emphasis added).
According to EPA, it “evaluated the animal carcinogenicity
data using [the Guideline’s] weight of evidence approach,
which included both the trend and pairwise analyses.” But
the Guidelines’ reliance on these tests for determining
statistical significance is precisely for the purpose of sorting
out whether a result is treatment-related rather than caused
by chance. The Guidelines clearly explain that “[t]rend tests
and pairwise comparison tests are the recommended tests for
determining whether chance, rather than a treatment-related
effect, is a plausible explanation for an apparent increase in
tumor incidence.” Thus, EPA’s bare assertion that a lack of
pairwise statistical significance suggests that tumor results
in rodent studies are not treatment-related fails to account
coherently for the evidence of statistical significance from
trend tests, which the Cancer Guidelines deem similarly
probative (though not necessarily conclusive).

2.

EPA’s second proposition in support of its selection of
the “not likely” descriptor is that concerning results only
occurred at high doses. In the Cancer Paper, EPA invokes
the Cancer Guidelines’ criterion that a chemical can be

significance in general rather than to have focused solely on “a lack of
pairwise statistical significance.” The agency’s response to the SAP’s
criticism also assumes that various trend tests indicated that tumor results
were statistically significant, but it asserts that this did not matter because
significance in both types of tests was required to support a conclusion
that the tumor results were treatment-related.
 NRDC V. USEPA 29

considered “not likely to be carcinogenic” when there is
“convincing evidence that carcinogenic effects are not likely
below a defined dose range.” EPA states that, even if the
tumors were treatment-related as “some believe,” those
tumors occurred only at very high glyphosate dosages.
According to EPA, increased tumor incidences generally
occurred at “the highest doses tested”—“approximately
equal to or greater than the limit dose (1000 mg/kg/day).”
EPA also mentions that positive results in genotoxicity
studies occurred only at “high doses.” Importantly, for both
the rodent studies and genotoxicity studies, “[t]hese high
doses [were] not considered relevant to human health risk
assessment based on the currently registered use pattern for
glyphosate” because “[m]aximum potential glyphosate
exposure [had] been estimated at . . . 7 mg/kg/day . . . which
[is] well-below the doses necessary to elicit the effects seen
in these animal carcinogenicity and genotoxicity studies.”

But EPA’s disregard of tumor results occurring at high
dosages conflicts with the guidelines EPA purports to
follow. The “not likely” descriptor can, in fact, be applied
when there is “convincing evidence that carcinogenic effects
are not likely below a defined dose range,” as EPA states.
But the Cancer Guidelines indicate that this use of the “not
likely” descriptor is appropriate “when the mode of action is
sufficiently understood to conclude that a key event in tumor
development would not occur below a certain dose range”—
to avoid discounting potentially concerning animal results
from laboratory settings unless there is a compelling reason
to believe that those results are irrelevant to humans.10

10
A mode of action, according to the Cancer Guidelines, is “a
sequence of key events and processes, starting with interaction of an
agent with a cell, proceeding through operational and anatomical
changes, and resulting in cancer formation.”
30 NRDC V. USEPA

Indeed, when that mode of action is sufficiently understood,
the Cancer Guidelines contemplate that the chemical would
merit “multiple descriptors”: it “could be described as likely
to be carcinogenic above a certain dose range but not likely
to be carcinogenic below that range.” Despite these
instructions, the Cancer Paper provides no explanation of
any “mode of action” indicating that tumors are not likely to
occur below a certain glyphosate dosage range, nor does it
offer two different hazard descriptors for dosages above and
below that range. Instead, EPA asserts that the potentially
concerning results observed at high doses in the studies it
reviewed are undeserving of significant consideration for
two related reasons: because the Cancer Guidelines
encourage considering human-exposure levels, which EPA
asserts are far below the high doses associated with those
results; and because those high doses exceeded a so-called
“limit dose” that EPA says is established by the agency’s
Health Effects Test Guidelines. 11

But the Cancer Guidelines do not support disregarding
results simply because they are based on exposures that
exceed typical human-exposure levels. More specifically,
no part of the hazard assessment Guidelines encourages
disregarding results occurring at high dosage ranges for any
reason other than when there is evidence of excessive
toxicity. Excessive toxicity occurs when a dose is so high
that the sheer amount of the chemical induces abnormal
responses in laboratory animals. Excessive toxicity is rare,

11
The Health Effects Test Guidelines provide directions to
researchers who are designing studies. They are published by EPA to
“minimize variations among the testing procedures that must be
performed to meet the data requirements” of EPA under FIFRA and
other laws. As explained below, those guidelines do not establish any
such limit dose.
 NRDC V. USEPA 31

so, according to the Cancer Guidelines, “[i]n general . . .
effects seen at the highest dose tested are assumed to be
appropriate for assessment.” That highest dose “is generally
selected to provide the maximum ability to detect treatment-
related carcinogenic effects while not compromising the
outcome of the study through excessive toxicity.” (emphasis
added). The Cancer Guidelines further provide that results
“may be regarded as not appropriate to include in assessment
of the potential for human carcinogenicity of the agent” only
“[i]f adequate data demonstrate that the effects are solely the
result of excessive toxicity rather than carcinogenicity of the
tested agent per se.” Despite EPA’s various invocations of
its Guidelines, the agency provides no evidence of excessive
toxicity yet nonetheless disregards high-dosage results.

Indeed, disregarding results occurring at high dosage
ranges seems contrary to the “purpose” of a hazard
assessment. According to the Cancer Guidelines, the
“purpose” of a hazard assessment “is to construct a total
analysis examining what the biological data reveal as a
whole about carcinogenic effects and mode of action of the
agent, and their implications for human hazard and dose-
response evaluation.” Consistent with this understanding,
subsequent steps in the Cancer Guidelines’ risk
assessment—ones that follow only once EPA has
determined that there is, in fact, a hazard—integrate human-
exposure patterns into the risk assessment through a variety
of complex methods. In fact, as explained above, choosing
a hazard descriptor is presented in the Guidelines as part of
the hazard-identification step, and a subsequent step focuses
32 NRDC V. USEPA

entirely on characterizing “the conditions of human
exposure.” 12

EPA suggests that other agency guidelines support
disregarding results above a so-called “limit dose” of 1,000
mg/kg/day, but that suggestion is also unsupported.
According to EPA, its Health Effects Test Guidelines
establish that there is a generally applicable “limit dose” of
1,000 mg/kg/day. Those guidelines, however, do not
establish any such limit dose. Instead, the guidelines simply
explain that, when researchers design an experiment, the
important consideration for choosing the highest dose is—
once again—excessive toxicity. According to the
guidelines, researchers should carefully choose the highest
dose to be administered with the goal that “[t]he highest-
dose level should elicit signs of toxicity without substantially
altering the normal life span due to effects other than
tumors.” The guidelines state that “[t]he highest dose tested

12
Quoting a fragment from the Cancer Guidelines in the Cancer
Paper’s “not likely” explanation, EPA seems to suggest that part of the
Guidelines supports using human-exposure patterns to exclude results.
The full passage from the Cancer Guidelines states the following, with
the underlined part reflecting the fragment EPA quotes in its Cancer
Paper: “Weighing of the evidence includes addressing not only the
likelihood of human carcinogenic effects of the agent but also the
conditions under which such effects may be expressed, to the extent that
these are revealed in the toxicological and other biologically important
features of the agent.” Read in context, this sentence in the Cancer
Guidelines is not encouraging the agency to integrate human-exposure
patterns into its hazard assessment. The sentence appears in an
introductory section with instructions for completing a “Weight of
Evidence Narrative” that will explain the hazard-assessment reasoning.
Rather than conveying a method for evaluating study results in a hazard
assessment, the sentence simply encourages the agency to describe the
biological pathways through which tumors develop—insofar as the
agency has such information available—as part of producing a
comprehensive narrative for the public.
 NRDC V. USEPA 33

need not exceed 1,000 mg/kg/day.” (emphasis added).
“Need not” indicates that the highest dose ultimately
selected does not have to exceed 1,000 mg/kg/day, but it can.
Also, because the highest dose tested in a completed study
was presumably selected to avoid excessive toxicity, these
guidelines do not support EPA’s disregard of tumor results
occurring at and above that 1,000 mg/kg/day dosage.

The SAP expressly stated the concern that EPA
improperly discounted study results involving dosages at or
above 1,000 mg/kg/day: “Disregarding responses at any
dose above a pre-selected ‘limit dose,’ even though the dose
did not exceed the maximum tolerated dose”—that is, the
maximum dose before excessive toxicity kicks in—“is not
in keeping with the way rodent bioassays are normally
interpreted . . . . Thus selecting 1,000 mg/kg/day a priori as
the limit dose appears to be an ad hoc decision that is not
well-justified, and is not justified on the basis of the [Cancer
Guidelines].” Despite the SAP’s criticism, EPA declined to
change its approach or to meaningfully respond.

C.

For these reasons, EPA’s choice of a hazard descriptor is
not supported by substantial evidence. Despite EPA’s
repeated invocation of its Cancer Guidelines, the Interim
Decision fails to abide by those Guidelines. Inconsistent
reasoning “is, absent explanation, ‘the hallmark of arbitrary
action.’” Nat’l Parks Conservation Ass’n v. EPA, 788 F.3d
1134, 1145 (9th Cir. 2015) (quoting Sierra Club v. EPA, 719
F.2d 436, 459 (D.C. Cir. 1983)). It cannot survive
substantial-evidence review. NRDC v. EPA, 31 F.4th 1203,
1210 (9th Cir. 2022) (considering internal “inconsistencies”
and “EPA’s decision to abandon its own guidance . . .
without a discernable rationale” in holding that a decision
was not supported by substantial evidence).
34 NRDC V. USEPA

Vacatur is the traditional remedy for erroneous
administrative decisions. See Pollinator Stewardship
Council, 806 F.3d at 532. To determine whether vacatur is
appropriate, we consider at least three factors. First, “we
weigh the seriousness of the agency’s errors against the
disruptive consequences of an interim change that may itself
be changed.” Nat’l Fam. Farm Coal. v. EPA, 960 F.3d 1120,
1144 (9th Cir. 2020) (quoting Pollinator Stewardship
Council, 806 F.3d at 532). Second, we consider “the extent
to which either vacating or leaving the decision in place
would risk environmental harm.” Id. at 1144–45. Third, we
examine “whether the agency would likely be able to offer
better reasoning [and] . . . adopt the same rule on remand, or
whether such fundamental flaws in the agency’s decision
make it unlikely that the same rule would be adopted on
remand.” Id. at 1145 (quoting Pollinator Stewardship
Council, 806 F.3d at 532).

Based on these considerations, we vacate the human-
health portion of EPA’s Interim Decision and remand for
further analysis and explanation. The first factor clearly
weighs in favor of vacatur. EPA’s errors in assessing
human-health risk are serious. Moreover, no disruptive
consequences will result from vacating the human-health
portion of the Interim Decision because that portion simply
maintained the status quo—the Interim Decision imposed no
new mitigation measures associated with human health. For
similar reasons, vacating the human-health portion is
unlikely to risk environmental harm, and thus the second
factor also weighs in favor of vacatur. The last factor is more
uncertain. It is possible that EPA could come to the same
human-health conclusion on remand, but the agency’s
explanation would need to be so different that we cannot
make a confident prediction on this factor. With two factors
 NRDC V. USEPA 35

weighing in favor of vacatur, uncertainty with respect to this
last one does not tip the scale in the other direction. 13

In light of this holding, we need not reach Rural
Coalition’s arguments about other alleged errors pertaining
to human health. See Pollinator Stewardship Council, 806
F.3d at 532 (“The matter must be remanded to the agency.
We need not reach the other claims of error raised by
petitioners.”). Similarly, our vacatur makes it unnecessary
to address NRDC’s objection that EPA did not comply with
its procedural obligation to respond to NRDC’s comments
about human health—because further proceedings,
including a new public-comment process, will be needed on
remand. 14

IV.

We next address Rural Coalition’s Endangered Species
Act claim. Congress enacted the ESA “to provide a means
whereby the ecosystems upon which endangered species and
threatened species depend may be conserved” and “to
provide a program for the conservation of such endangered
species and threatened species” (collectively, “listed
species”). 16 U.S.C. § 1531(b). The ESA reflects “a
conscious decision by Congress to give endangered species

13
We decline to rule on any effect this vacatur might have on
glyphosate’s registration. Even assuming that we could order
deregistration outright, we would not do so here. Given the errors we
have pointed out—and our uncertainty regarding how correcting those
errors will alter EPA’s final conclusion—deregistration would be a
disproportionate and highly disruptive remedy.
14
As NRDC pointed out and EPA did not dispute at oral argument,
vacatur of the human-health portion will require the agency to conduct a
new public-comment process. 40 C.F.R. §§ 155.56, 155.58(a).
36 NRDC V. USEPA

priority over the ‘primary missions’ of federal agencies.”
Tenn. Valley Auth. v. Hill, 437 U.S. 153, 185 (1978).

The ESA protects those species, in part, by requiring
federal agencies to consult with the U.S. Fish and Wildlife
Service or the National Marine Fisheries Service (the
“Services”). Section 7(a)(2) of the ESA mandates that every
federal agency “shall, in consultation with and with the
assistance of [the Services], 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 or result in the
destruction or adverse modification” of designated critical
habitat. 16 U.S.C. § 1536(a)(2). The consultation process
unfolds as follows: “[A]t the earliest possible time,” the
agency proposing the action assesses whether a proposed
action “may affect” an endangered or threatened species or
its critical habitat, 50 C.F.R. § 402.14(a), making a so-called
“effects determination.” “May affect” is broadly
understood. It includes “[a]ny possible effect, whether
beneficial, benign, adverse or of an undetermined character.”
Karuk Tribe of Cal. v. U.S. Forest Serv., 681 F.3d 1006,
1027 (9th Cir. 2012) (quoting California ex rel. Lockyer v.
U.S. Dep’t of Agric., 575 F.3d 999, 1018 (9th Cir. 2009)). If
the agency determines that its action “may affect” a listed
species or critical habitat, “formal consultation” is generally
necessary. 50 C.F.R. § 402.14(a)–(b). Formal consultation
requires the Services to prepare a “biological opinion” on
whether the proposed action “is likely to jeopardize the
continued existence of listed species or result in the
destruction or adverse modification of critical habitat.” Id.
§ 402.14(g)(4), (h). If the Services conclude that “the
agency action would place the listed species in jeopardy or
adversely modify its critical habitat,” Nat’l Ass’n of Home
Builders v. Defs. of Wildlife, 551 U.S. 644, 652 (2007), they
 NRDC V. USEPA 37

must recommend “reasonable and prudent alternatives” to
the proposed action, id. (quoting 16 U.S.C. § 1536(b)(3)(A)
and 50 C.F.R. § 402.14(h)(2)).

Rural Coalition argues that EPA failed to comply with
these obligations before issuing the Interim Decision. At the
threshold, Intervenor Monsanto argues that we should not
decide the merits of this claim because Rural Coalition lacks
standing, the claim is moot, and/or the claim was not
adequately preserved. To help explain our analysis of these
arguments, we first describe the aspects of the Interim
Decision relevant to the ESA claim. We then answer all of
the threshold questions in Rural Coalition’s favor and, on the
merits, agree with Rural Coalition that EPA violated the
ESA.

A.

The Interim Decision contains the critical pieces of
EPA’s registration review of glyphosate. It “finalizes” the
human and ecological risk assessments and announces that
“[n]o additional data are required.” As to human health, it
“determine[s] that there are no risks to human health from
the current registered uses of glyphosate” and imposes no
health-related mitigation requirements. As to ecological
risk, it finds potential risks to animals and plants and
“require[s]” mitigation in light of those risks, laying out
specific language for glyphosate product labels. Crucially,
the Interim Decision “concludes that the benefits outweigh
the potential ecological risks” when glyphosate is used
according to the restrictions imposed by the Interim
Decision. That conclusion is the critical determination that
the pesticide complies with FIFRA’s safety standard. See
Nat’l Fam. Farm Coal. v. EPA, 960 F.3d 1120, 1133 (9th
Cir. 2020) (“FIFRA uses a cost-benefit analysis to ensure
that there is no unreasonable risk created for people or the
38 NRDC V. USEPA

environment from a pesticide.” (quoting Pollinator
Stewardship Council v. EPA, 806 F.3d 520, 522–23 (9th Cir.
2015))). 15

B.

1.

We turn to the first threshold question: whether Rural
Coalition lacks standing to bring an ESA claim. Article III
standing requires “(1) a concrete and particularized injury
that is ‘actual or imminent, not conjectural or hypothetical’;
(2) a causal connection between the injury and the
defendant’s challenged conduct; and (3) a likelihood that a
favorable decision will redress that injury.” Pyramid Lake
Paiute Tribe of Indians v. Nev. Dep’t of Wildlife, 724 F.3d
1181, 1187 (9th Cir. 2013) (quoting Lujan v. Defs. of
Wildlife, 504 U.S. 555, 560 (1992)). 16

15
The Interim Decision outlines only two more tasks, in addition to
ESA consultation, that EPA intends to complete before issuing a final
registration review decision: an endocrine analysis required by the
FFDCA and resolution of a petition filed under the FFDCA. But EPA
made clear that there is nothing left for the agency to do under FIFRA’s
mandate itself. The ESA analysis, FFDCA analysis, and response to the
FFDCA petition are discrete processes guided by other statutes. At oral
argument, EPA confirmed, “[T]he Interim Decision . . . basically
finalized everything except for the Endangered Species Act consultation,
which the agency committed to do before making a final decision. It
looked at human health and ecological risk, which are the factors under
the statute—under FIFRA.”

16
Rural Coalition asserts an injury on behalf of its members and thus
must also meet the requirements for associational standing. “[A]n
association has standing to bring suit on behalf of its members when:
(a) its members would otherwise have standing to sue in their own right;
(b) the interests it seeks to protect are germane to the organization’s
 NRDC V. USEPA 39

“To satisfy the injury-in-fact requirement of the Article
III inquiry, ‘a plaintiff asserting a procedural injury must
show that the procedures in question are designed to protect
some threatened concrete interest of his that is the ultimate
basis of his standing.’” Salmon Spawning & Recovery All.
v. Gutierrez, 545 F.3d 1220, 1225 (9th Cir. 2008) (quoting
Citizens for Better Forestry v. U.S. Dep’t of Agric., 341 F.3d
961, 969 (9th Cir. 2003)). Rural Coalition’s members have
submitted declarations stating that they regularly engage in
educational and recreational activities involving a variety of
endangered species, including the Indiana bat, whooping
crane, and least tern. These members also allege that
glyphosate is threatening their interests by exposing those
species to toxic runoff and residues on vegetation. The
interests identified in the Rural Coalition declarations are
“undeniably . . . cognizable interest[s] for [the] purpose of
standing.” See Lujan, 504 U.S. at 562–63. In addition, the
consultation procedures that Rural Coalition claims were
required but not completed are intended to protect these
types of interests. See Salmon Spawning, 545 F.3d at 1226
(“These procedures are designed to advance the ESA’s
overall goal of species preservation, and thus the [members’]
specific goals as to . . . preservation, by ensuring agency
compliance with the ESA’s substantive provisions.”).
Accordingly, the injury-in-fact requirement is met.

Rural Coalition has also established causation. We have
held that an alleged violation of the consultation requirement
constitutes a “procedural injury” for standing purposes.

purpose; and (c) neither the claim asserted nor the relief requested
requires the participation of individual members in the lawsuit.” Hunt v.
Wash. State Apple Advert. Comm’n, 432 U.S. 333, 343 (1977). Criteria
(b) and (c) are not contested, and we are satisfied that they are met here.
We therefore focus our analysis on (a).
40 NRDC V. USEPA

Citizens for Better Forestry, 341 F.3d at 971. When a
procedural injury is asserted, “[t]he causation requirement is
satisfied by showing a ‘reasonable probability of the
challenged action’s threat to [the petitioner’s] concrete
interest.’” Nat’l Fam. Farm Coal. v. EPA, 966 F.3d 893, 910
(9th Cir. 2020) (quoting Hall v. Norton, 266 F.3d 969, 977
(9th Cir. 2001)). Here, EPA reconsidered the conditions
under which the most heavily used herbicide in the nation
may be used, acknowledging in the Interim Decision that
glyphosate poses “potential risks of concern” for mammals,
birds, and plants. Yet the agency imposed few limitations
on glyphosate use. Thus, there is a reasonable probability
that the requested ESA procedures—that is, an effects
determination and consultation—would lead to greater
restrictions on glyphosate, thereby reducing the threat posed
to the species that are the focus of Rural Coalition members’
interests.

EPA’s recent analysis of glyphosate’s impact on listed
species—that is, its ESA effects determination—in
preparation for its final registration review decision adds
support for our conclusion. After issuing the Interim
Decision, EPA published a draft Biological Evaluation
(“BE”) followed by a final BE. According to EPA, the BE
is a “comprehensive, nationwide assessment of the effects of
glyphosate on ESA-listed species and critical habitats that
determines the need for consultation, and its scope.” Put
differently, the BE broadly assesses the effect of current
glyphosate use—the very use that is the subject of the
Interim Decision. 17 The BE found that glyphosate “may

17
Although EPA and Monsanto urge otherwise, we assume that
whatever consultation would have occurred if EPA had consulted with
the Services on the Interim Decision would have been equivalent to
EPA’s current consultation. From a practical standpoint, the only
 NRDC V. USEPA 41

affect” all ESA-listed species that experience glyphosate
exposure—that is, 1,795 species—and is likely to adversely
affect 93% of those species. EPA has begun formal
consultation about how to mitigate these adverse effects.

Monsanto argues unconvincingly that the causation
requirement is not satisfied. According to Monsanto, the
Interim Decision did not grant or extend glyphosate’s
registration, and EPA could not cancel that registration
through such a decision. Instead, if the agency intends to
cancel a pesticide’s registration, it must initiate cancellation
through an elaborate statutory process. See 7 U.S.C.
§ 136d(b). Accordingly, on Monsanto’s view, the Interim
Decision is not the reason glyphosate can continue to be
sold, and thus there is no causal link between Rural
Coalition’s injury and EPA’s action. But causation does not
require that the defendant’s unlawful conduct be the only
cause of the alleged injury. Had EPA observed the
consultation requirement prior to issuing the Interim
Decision, even if glyphosate’s registration had not been
cancelled outright, the Interim Decision might have imposed
more restrictions on glyphosate’s use than the Interim

differences between the Interim Decision and the anticipated final
registration review decision in terms of subject matter for potential
consultation are the outstanding analyses under the FFDCA. It is
difficult to believe that those analyses would make a difference in the
ESA consultation process, and no party has even suggested that they
would. Accordingly, we treat the hypothetical consultation process that
could have been done prior to issuance of the Interim Decision and the
current consultation process EPA has undertaken prior to its final
decision as essentially the same endeavor. Monsanto’s brief is in accord.
In arguing that Rural Coalition’s ESA claim is not redressable, Monsanto
contends that “[t]he very process that Petitioners want EPA to undertake
is already being undertaken,” effectively acknowledging that the scope
of EPA’s current consultation endeavor is equivalent to whatever might
have been required for the Interim Decision.
42 NRDC V. USEPA

Decision’s new label requirements actually did. That is
sufficient to satisfy the causation requirement given that the
alleged violation is procedural in nature. See Salmon
Spawning, 545 F.3d at 1229 (holding that causation is
satisfied for a procedural injury when “[t]he asserted injury
is not too tenuously connected to the agencies’ failure” to
take action).

When petitioners allege a procedural violation, the
redressability prong is satisfied by showing that the agency
decision “could be influenced” by the procedures at issue.
Hall, 266 F.3d at 977; see also Nat’l Fam. Farm Coal., 966
F.3d at 911 (explaining that the redressability requirement is
satisfied when relief “may influence the agency’s ultimate
decision of whether to take or refrain from taking a certain
action” (quoting Salmon Spawning, 545 F.3d at 1226–27)).
Broadly speaking, Rural Coalition requests that the agency
complete the consultation procedures found in the ESA. As
explained above, it is apparent that EPA might have required
more mitigation efforts had the agency completed an effects
determination and consulted before issuing the Interim
Decision.

Monsanto also argues that any alleged injury is not
redressable because EPA began following the ESA
consultation procedures before Rural Coalition filed its
petition for review of the Interim Decision. By the time
Rural Coalition filed its opening brief, the agency had
released a draft BE in preparation for its final registration
review decision. Since then, EPA has released a final BE
and has begun formally consulting. Monsanto argues that an
order to complete consultation procedures would not prompt
EPA to do anything more than it is already doing, and that
any injury is therefore not redressable. See Massachusetts v.
EPA, 549 U.S. 497, 518 (2007) (requiring “some possibility
 NRDC V. USEPA 43

that the requested relief will prompt the injury-causing party
to reconsider the decision that allegedly harmed the litigant”
(emphasis added)).

We disagree. “When evaluating whether [the standing]
elements are present, we must look at the facts as they
exist[ed] at the time the complaint was filed.” Slayman v.
FedEx Ground Package Sys., Inc., 765 F.3d 1033, 1047 (9th
Cir. 2014) (alteration in original) (quoting ACLU of Nev. v.
Lomax, 471 F.3d 1010, 1015 (9th Cir. 2006)). At the time
Rural Coalition filed its petition, court-ordered relief was
possible. EPA had not even completed a draft BE, and it
was unclear when the agency would do so. Thus, we could
have, for example, remanded the Interim Decision, directed
the agency to comply with the ESA, and even imposed a
deadline for completing the effects determination required
by the statute—that is, ordered EPA to finalize a BE and to
initiate any required consultation by a certain date. See, e.g.,
League of United Latin Am. Citizens v. Regan, 996 F.3d 673,
703 (9th Cir. 2021); see also Nat’l Wildlife Fed’n v. Nat’l
Marine Fisheries Serv., 524 F.3d 917, 936–38 (9th Cir.
2008); Alaska Ctr. For Env’t v. Browner, 20 F.3d 981, 986–
87 (9th Cir. 1994). An aggressive deadline—and any court
oversight that might accompany such a deadline—
presumably would have spurred EPA to act at least
somewhat faster than it otherwise would have, redressing
Rural Coalition’s injury. That is enough for Rural Coalition
to have standing.

2.

Monsanto alternatively argues that EPA’s recent
consultation efforts moot this case. “If an event occurs that
prevents the court from granting effective relief, the claim is
moot and must be dismissed.” Am. Rivers v. Nat’l Marine
Fisheries Serv., 126 F.3d 1118, 1123 (9th Cir. 1997); see
44 NRDC V. USEPA

Church of Scientology of Cal. v. United States, 506 U.S. 9,
12 (1992). On Monsanto’s view, even assuming there is
standing, we can no longer grant effective relief on the ESA
claim now that EPA has begun formally consulting with the
Services.

But Monsanto’s understanding of the relief available
here is too narrow. Broadly speaking, the allegedly unlawful
behavior targeted by the petition is a failure to complete
procedures required by the ESA before formally concluding
whether and how glyphosate may be used consistent with
FIFRA’s safety standard, and that behavior still has not been
rectified. The allegedly unlawful behavior could be
remedied by our imposition of an aggressive deadline for the
completion of consultation, an order for EPA to complete the
parts of consultation within its control with the utmost speed,
or an order for EPA to file status reports with our court on
consultation’s progress. 18 Such relief could meaningfully

18
A consultation deadline would implicate the obligations of the
Services in addition to obligations of EPA. That would seem permissible
under the principle that injunctive court orders may “bind[] the parties
defendant but also those identified with them in interest, in ‘privity’ with
them, represented by them or subject to their control.” Regal Knitwear
Co. v. NLRB, 324 U.S. 9, 14 (1945); accord Golden State Bottling Co. v.
NLRB, 414 U.S. 168, 179 (1973); see also Class Plaintiffs v. City of
Seattle, 955 F.2d 1268, 1277 (9th Cir. 1992) (“[A] judgment can bind
persons not parties to the litigation in question and not subject in
personam to the jurisdiction of the court if the persons are in privity with
parties to the litigation.”). In general, “[t]here is privity between officers
of the same government.” Sunshine Anthracite Coal Co. v. Adkins, 310
U.S. 381, 402–03 (1940); accord Fund for Animals, Inc. v. Lujan, 962
F.2d 1391, 1398 (9th Cir. 1992); see Ma Chuck Moon v. Dulles, 237 F.2d
241, 243 (9th Cir. 1956) (finding privity between the U.S. Secretary of
State and the Attorney General for res judicata purposes because the
actions at issue were “in effect suits against the United States”). Even if
we could not bind the Services, however, our ability to compel EPA to
act quickly is sufficient to avoid mootness.
 NRDC V. USEPA 45

hasten ESA compliance, especially because EPA appears
inclined to delay consultation given that it has already
pushed the entire consultation process until the final year of
registration review. Thus, Rural Coalition’s claim is not
moot. 19

3.

Monsanto additionally argues that Rural Coalition’s
ESA claim was not preserved during the public-comment
period. This argument is unconvincing. A Rural Coalition
petitioner, Center for Biological Diversity, discussed ESA
consultation in its comments on the proposed Interim
Decision. The Center argued that “EPA must consult with
the Services on its continuing and ongoing authority over
this pesticide to satisfy its duty to [en]sure that its use will
not jeopardize or adversely modify protected species or their
critical habitat well before it proposes a registration review
decision.” The Center also implored EPA to “[i]ncorporate
necessary factors into evaluation and any proposed decision”

19
Cases in which we have deemed consultation claims moot when
the agency had already begun consulting are not to the contrary. See All.
for the Wild Rockies v. U.S. Forest Serv., 907 F.3d 1105, 1121 (9th Cir.
2018); All. for the Wild Rockies v. U.S. Dep’t of Agric., 772 F.3d 592,
601 (9th Cir. 2014). The plaintiffs in those cases sought reinitiation of
consultation, which could not be ordered once it had already occurred.
Here, Rural Coalition seeks compliance with and completion of the
ESA’s consultation requirement, not just the initiation or reinitiation of
consultation. See Env’t Def. Ctr. v. Bureau of Ocean Energy Mgmt., —
F.4th —, 2022 WL 1816515, at *21 (9th Cir. 2022) (holding that,
because consultation with the Fish and Wildlife Service was still
ongoing, the court had jurisdiction over a claim that an agency failed to
consult before acting); see also All. for the Wild Rockies v. Savage, 897
F.3d 1025, 1031 (9th Cir. 2018) (holding that a reconsultation claim was
moot because the federal defendants completed reconsultation and the
plaintiff, therefore, “ha[d] obtained all that it sought with this claim”).
46 NRDC V. USEPA

including “effects on species listed as protected under the
ESA and their critical habitat.” Those comments sufficiently
raised the issue that EPA failed to comply with the ESA
consultation requirement before issuing its Interim Decision.
See Lands Council v. McNair, 629 F.3d 1070, 1076 (9th Cir.
2010) (“[A] claimant need not raise an issue using precise
legal formulations, as long as enough clarity is provided that
the decision maker understands the issue raised.”).

C.

We now turn to the merits of Rural Coalition’s ESA
argument. As explained above, the consultation procedures
begin with an agency “review[ing] its actions at the earliest
possible time to determine whether any action may affect
listed species or critical habitat,” 50 C.F.R. § 402.14,
resulting in an effects determination. It is undisputed that
EPA made no such effects determination before issuing the
Interim Decision. Thus, the key question is whether the
Interim Decision is an “action” that triggers the consultation
procedures found in the ESA. Rural Coalition argues that
the answer is “yes” because the Interim Decision was a
consequential milestone that essentially approved continued
glyphosate use across the United States. EPA and Monsanto
disagree. They argue that Rural Coalition’s grievance is
with EPA’s failure to require more mitigation measures, not
any affirmative action taken by the agency.

We have held that “agency action” under the ESA has
only two requirements. There is “agency action” whenever
an agency makes a decision that is (1) affirmative and
(2) discretionary about whether, or under what conditions, to
allow private activity to proceed. Karuk Tribe, 681 F.3d at
1011. That the second requirement is satisfied is not
disputed here—EPA clearly has the power to restrict
pesticide use by private parties through mitigation measures
 NRDC V. USEPA 47

to protect listed species. 40 C.F.R. §§ 155.40(a)(2), 155.56,
155.58(b)(2).

We agree with Rural Coalition that the first requirement
is also satisfied, because the Interim Decision is an
affirmative act. An agency must adhere to the consultation
requirement when it makes an “affirmative” act or
authorization. Cal. Sportfishing Prot. All. v. FERC, 472 F.3d
593, 595 (9th Cir. 2006). But “[w]here private activity is
proceeding pursuant to a vested right or to a previously
issued license, an agency has no duty . . . under Section 7 if
it takes no further affirmative action regarding the activity.”
Karuk Tribe, 681 F.3d at 1021; W. Watersheds Project v.
Matejko, 468 F.3d 1099, 1108 (9th Cir. 2006) (“‘[I]naction’
is not ‘action’ for section 7(a)(2) purposes.”). Here, EPA
actively exercised its regulatory power, completing an
assessment of glyphosate’s risks under FIFRA and
delineating what constituted acceptable glyphosate use
under the statute’s safety standard. See Karuk Tribe, 681
F.3d at 1024 (finding affirmative action when “the Forest
Service formulated precise criteria for the protection of coho
salmon, communicated those criteria to prospective miners,
and approved the miners’ activities under a [Notice of Intent]
only if they strictly conformed their mining to the specified
criteria”).

EPA and Monsanto’s primary argument is that Rural
Coalition is objecting to inaction, not action, when it
complains that EPA would have instituted more mitigation
efforts if EPA had engaged in ESA consultation. That
argument fails. EPA and Monsanto mainly rely upon
Western Watersheds, 468 F.3d 1099. That case involved a
Bureau of Land Management (“BLM”) decision not to
regulate private parties’ diversions of water—diversions that
were completed pursuant to those parties’ pre-existing
48 NRDC V. USEPA

rights-of-way. We assumed that BLM had the power to
regulate the diversions but held that BLM’s decision not to
exercise that power was not an affirmative action. Id. at
1107–09. We explained, “BLM did not fund the diversions,
it did not issue permits, it did not grant contracts, it did not
build dams, nor did it divert streams.” Id. at 1109; see also
Cal. Sportfishing, 472 F.3d at 598, 595 (holding that “the
agency[] ha[d] proposed no affirmative act that would
trigger the consultation requirement” for operations of a
hydroelectric plant that were authorized by an earlier permit,
even though the agency was empowered to “unilaterally
institute proceedings to amend the license if it so chose”).
The Interim Decision is unlike the agency inaction in
Western Watersheds. Here, EPA did not simply stand aside
as regulated parties continued to use glyphosate. Instead,
EPA exercised its regulatory power under FIFRA—
engaging in a re-assessment that was required by statute—
and delineated the manner in which glyphosate could be
used consistent with the FIFRA safety standard.

Because EPA’s registration review decision under
FIFRA is the “action” that triggers the consultation
requirement, it is irrelevant—despite EPA and Monsanto’s
suggestion to the contrary—that Rural Coalition takes the
view that the Interim Decision does very little to protect
listed species and should have contained more mitigation
measures. Indeed, the notion that challenging an absence of
mitigation efforts is merely an objection to inaction is
inconsistent with the purpose of the ESA. Arguing that
protection for endangered and threatened species is
insufficient is precisely the point of an ESA claim. When
aggrieved parties file a lawsuit asserting that the government
should not have authorized a certain activity, such as the
diversion of water from a river, the core grievance is often
that the activity threatens—or, put differently, does not
 NRDC V. USEPA 49

sufficiently protect—listed species. See, e.g., NRDC v.
Houston, 146 F.3d 1118 (9th Cir. 1998). Many clearly
cognizable ESA claims, then, could easily be framed as a
complaint about the failure to institute more mitigation
efforts. That framing does not mean that the claim merely
challenges inaction.

D.

Because the Interim Decision was an affirmative,
discretionary action, EPA had to comply with the ESA by
making an effects determination before issuing the decision.
It is undisputed that EPA did not do so. Accordingly, EPA
violated the ESA.

Nevertheless, because of the odd confluence of
circumstances here, we decline to order relief for the ESA
violation. Although relief such as a consultation deadline
could hasten EPA’s compliance with the ESA (and thus we
have held that the ESA claim is not moot), we believe that
the FIFRA deadline to complete glyphosate’s registration
review by October 2022 is a sufficient backstop. According
to the timeline imposed by Congress, EPA already must
complete its final registration review decision—including
formal consultation—by that October 2022 deadline. Given
that the FIFRA deadline is fast approaching, shortening
EPA’s time to consult would be only moderately beneficial
to Rural Coalition but potentially very disruptive to the
agency.

Rural Coalition urges vacatur of the Interim Decision for
failure to comply with the ESA. But it is not clear that
vacatur would be beneficial here. The Interim Decision
includes certain mitigation efforts that EPA designed to limit
the ecological impacts of glyphosate use. Although Rural
Coalition argues that those requirements are insufficient, if
50 NRDC V. USEPA

the requirements affect glyphosate use at all, they likely
reduce ecological risk. Because vacatur would eliminate
those mitigation requirements, we decline to vacate the
Interim Decision—other than to the extent specified in Part
III above regarding the human-health portion.

V.

The remaining issue involves Petitioners’ challenges to
the Interim Decision’s ecological risk assessment,
determination of glyphosate’s costs, cost-benefit analysis,
and mitigation requirements (collectively, the “ecological
portion”), and EPA’s responsive motion for remand. We
grant EPA’s motion to remand without vacatur as to the
ecological portion of the decision, but we impose a time limit
on the remand.

NRDC argues that EPA failed to consider major
environmental and economic costs of glyphosate use,
including the costs of creating glyphosate-resistant weeds,
harm to soil health caused by glyphosate, and the decimation
of milkweed in agricultural fields. In addition, NRDC
argues that EPA failed to provide any explanation as to how
it weighed the purported benefits and risks of glyphosate use,
pointing out that EPA simply concluded in a single sentence
that the purported benefits outweigh the risks without any
sort of reasoned analysis. Finally, NRDC argues that EPA’s
decision rests on unsubstantiated assumptions that the
mitigation measures will, in fact, reduce the acknowledged
ecological risks posed by glyphosate use—without any
evidence that the mitigation measures imposed will ensure
that glyphosate use satisfies FIFRA’s safety standard. Rural
Coalition echoes many of these concerns, adding that EPA
failed to consider the cost of glyphosate drift as well as the
costs to pollinators and to monarch butterflies.
 NRDC V. USEPA 51

EPA does not respond to the attack on the ecological
portion of the Interim Decision. Instead of answering these
parts of the petitions, EPA asks us to remand the ecological
portion for further consideration without vacatur and thereby
asks us not to reach the corresponding claims in NRDC’s
and Rural Coalition’s petitions.

Courts generally grant an agency’s request for voluntary
remand unless the request is frivolous or made in bad faith.
Cal. Cmtys. Against Toxics v. EPA, 688 F.3d 989, 992 (9th
Cir. 2012); see also Ethyl Corp. v. Browner, 989 F.2d 522,
524 (D.C. Cir. 1993) (“We commonly grant [agency
remand] motions, preferring to allow agencies to cure their
own mistakes rather than wasting the courts’ and the parties’
resources reviewing a record that both sides acknowledge to
be incorrect or incomplete.”). Normally, when remand is
requested and granted, “the agency intends to take further
action with respect to the original agency decision on
review.” Util. Solid Waste Activities Grp. v. EPA, 901 F.3d
414, 436 (D.C. Cir. 2018) (quoting Limnia, Inc. v. U.S. Dep’t
of Energy, 857 F.3d 379, 386 (D.C. Cir. 2017)).
“[I]ntervening events outside of the agency’s control, for
example, a new legal decision or the passage of new
legislation,” counsel in favor of granting such a remand
request. SKF USA Inc. v. United States, 254 F.3d 1022, 1028
(Fed. Cir. 2001). That said, we have “broad discretion” in
deciding whether to do so. Util. Solid Waste, 901 F.3d at
436.

Here, EPA has neither conceded error nor given any
clear indication of how it will proceed on remand. Instead,
EPA has vaguely asserted that “intervening decisions from
this Court,” “EPA’s publication of its draft biological
evaluation for glyphosate,” and the “change in
Administration” warrant a partial remand. EPA specifically
52 NRDC V. USEPA

requests that the partial remand be without vacatur to “allow
EPA flexibility” to implement the mitigation requirements
set forth in the Interim Decision. EPA suggests that, if it
decides to reassess aspects of the Interim Decision, it will
likely do so in its final registration decision rather than in a
new interim decision.

Rural Coalition opposes the remand motion, arguing that
EPA’s actions are a bad-faith attempt to avoid judicial
review. According to Rural Coalition, “EPA seeks remand
of part of its action at the eleventh hour to avoid an adverse
court ruling with little or no binding commitment to . . .
actually change its decision.” Rural Coalition further argues
that EPA does not have a properly cognizable rationale for
its request. Rural Coalition dismisses the BE because it was
within EPA’s control and dismisses the intervening court
decisions that EPA raises because they established no new
law and simply required compliance with FIFRA’s core
mandate.

NRDC, on the other hand, does not oppose EPA’s
motion but responds to it by urging a 90-day deadline for
completing the reconsideration of the ecological portion of
the Interim Decision on remand. According to NRDC, EPA
is unlikely to issue a final registration review decision by the
October 2022 statutory deadline because the agency only
recently initiated formal consultation under the ESA, which
NRDC predicts will take years. NRDC additionally details
how “EPA’s pesticide approval process has been beset by
consistent delays,” arguing that “[t]his history of delay
means that a deadline for remand is appropriate.” NRDC
thus asks that the Interim Decision be re-issued within 90
days.

Although Rural Coalition’s arguments have force, we
decide to grant EPA’s motion to remand largely for practical
 NRDC V. USEPA 53

reasons. As stated above, the challenges to the ecological
portion are properly raised in Petitioners’ briefs, but EPA
chose not to respond to those challenges in this action,
instead filing a motion to remand on the deadline for filing
its answering brief. Monsanto also did not substantively
respond to those challenges. If we were to evaluate the
merits of NRDC’s challenges, we first would want to request
responsive briefs and to have oral argument on the
ecological issues, raising the possibility that the FIFRA
October 2022 deadline would arrive before we could
complete our review. Thus, while we hesitate to reward
what some might consider sloth or indolence, we also
recognize that fully litigating the issues could result in an
outcome nobody wants: more, and probably unnecessary,
delay. Because of these unusual circumstances, we GRANT
EPA’s motion to remand.

We are sympathetic, however, to Petitioners’ concerns
about delay and gamesmanship. That said, we do not believe
that NRDC’s proposed 90-day deadline is warranted or that
EPA must issue another interim decision before its final
decision. Instead, we require EPA to issue a new ecological
portion by the October 2022 FIFRA deadline. 20 See Clean
20
EPA points out that the Supreme Court has cautioned that, “[a]t
least in the absence of substantial justification for doing otherwise, a
reviewing court may not, after determining that additional evidence is
requisite for adequate review, proceed by dictating to the agency the . . .
time dimension of the needed inquiry.” Vt. Yankee Nuclear Power Corp.
v. NRDC, 435 U.S. 519, 544–45 (1978) (quoting Fed. Power Comm’n v.
Transcont’l Gas Pipe Line Corp., 423 U.S. 326, 333 (1976) (per
curiam)). This concern is not triggered by the deadline we set because
Congress has already “dictat[ed] . . . [the] time dimension” for EPA
action here. Id. at 545. We avoid any potential issue with “propelling
the court into the domain which Congress has set aside exclusively for
the administrative agency,” id. at 545 (quoting Transcont’l Gas, 423 U.S.
at 333)—which was the Court’s concern in Vermont Yankee—by
54 NRDC V. USEPA

Wis. v. EPA, 964 F.3d 1145, 1176 (D.C. Cir. 2020) (“EPA
offers no reason, nor can we think of one, why it should be
permitted to evade the Clean Air Act’s statutory deadline
through a voluntary remand.”).

VI.

EPA’s motion to remand the ecological portion of the
Interim Decision without vacatur of that portion is
GRANTED subject to the deadline described above.
Because we grant EPA’s motion, we do not reach the parts
of NRDC’s and Rural Coalition’s petitions that challenge the
remanded portion of the Interim Decision.

The remainder of Rural Coalition’s petition for review is
GRANTED in part and DENIED in part. We VACATE
the human health portion of the Interim Decision and
REMAND for further consideration. Given that vacatur, we
do not reach the remainder of NRDC’s petition challenging
the public-comment process that informed the human health
portion of the Interim Decision. And although we agree with
Rural Coalition that an ESA violation has occurred, we
decline Rural Coalition’s request for relief for the reasons
stated above.

REMANDED.

imposing a deadline that is the same as Congress’s deadline for review
of pesticide registrations. See Clean Wis. v. EPA, 964 F.3d 1145, 1176
(D.C. Cir. 2020).