United States Court of Appeals
                        For the First Circuit


Nos. 22-1795, 22-1796

  AMERICAN TRUCKING ASSOCIATIONS, INC; CUMBERLAND FARMS, INC.;
                  M&M TRANSPORT SERVICES, INC.,

                        Plaintiffs, Appellees,

                NEW ENGLAND MOTOR FREIGHT, INC.,

                             Plaintiff,

                                 v.

          RHODE ISLAND TURNPIKE AND BRIDGE AUTHORITY,

                        Defendant, Appellant,

 PETER ALVITI, JR., in his official capacity as Director of the
           Rhode Island Department of Transportation,

                        Defendant, Appellant.


         APPEALS FROM THE UNITED STATES DISTRICT COURT
                FOR THE DISTRICT OF RHODE ISLAND

          [Hon. William E. Smith, U.S. District Judge]


                               Before

                  Rikelman, Lipez, and Kayatta,
                         Circuit Judges.


     Ian Heath Gershengorn, with whom Adam G. Unikowsky, Michelle
S. Kallen, Elizabeth B. Deutsch, Maura E. Smyles, and Jenner &
Block LLP were on brief, for appellants.
     Peter F. Neronha, Attorney General of Rhode Island, Michael
W. Field, Assistant Attorney General of Rhode Island, and Keith
Hoffmann, Special Assistant Attorney General of Rhode Island, on
brief for appellant Peter Alviti, Jr.
     John A. Tarantino, R. Bart Totten, Nicole J. Benjamin, and
Adler Pollock & Sheehan PC on brief for appellant Rhode Island
Turnpike and Bridge Authority.
     David S. Coale and Lynn Pinker Hurst & Schwegmann LLP on brief
for International Bridge, Tunnel and Turnpike Association, amicus
curiae.
     Charles A. Rothfeld, with whom Evan M. Tager, Reginald R.
Goeke, Eric A. White, Mayer Brown LLP, Richard Pianka, and ATA
Litigation Center were on brief, for appellees.
     Tyler S. Badgley, Jonathan D. Urick, U.S. Chamber of Commerce
Litigation Center, Mark C. Fleming, Sharon K. Hogue, and Wilmer
Cutler Pickering Hale and Dorr LLP on brief for the U.S. Chamber
of Commerce, amicus curiae.
     Prasad Sharma and Scopelitis, Garvin, Light, Hanson & Feary,
P.C. on brief for the American Highway Users Alliance, Intermodal
Association   of   North   America,   NATSO,   Truckload   Carriers
Association, TRALA, Rhode Island Trucking Association, Inc., Maine
Motor   Transport   Association,   Inc.,   Maryland   Motor   Truck
Association, Inc., Trucking Association of Massachusetts, Motor
Transport Association of CT, Inc., New Hampshire Motor Transport
Association, Trucking Association of New York, Pennsylvania Motor
Truck Association, Vermont Truck and Bus Association, Alabama
Trucking Association, Inc., Georgia Motor Trucking Association,
Inc., Arkansas Trucking Association, Florida Trucking Association,
Inc., Louisiana Motor Transport Association, Inc., North Carolina
Trucking Assoc., Inc., South Carolina Trucking Assoc., Inc.,
Mississippi Trucking Association, Virginia Trucking Association,
Tennessee Trucking Association, Kentucky Trucking Association,
Inc., West Virginia Trucking Association, Inc., Idaho Trucking
Association, Nevada Trucking Associations, Inc., South Dakota
Trucking Association, Washington Trucking Associations, Arizona
Trucking Association, Hawaii Transportation Association, New
Mexico Trucking Association, North Dakota Motor Carriers Assoc.,
Inc., Colorado Motor Carriers Association, Montana Trucking
Association, Oregon Trucking Associations, Inc., Indiana Motor
Truck Association, Inc., California Trucking Association, Utah
Trucking Association, Alaska Trucking Association, Inc., Ohio
Trucking Association, Illinois Trucking Association, Inc.,
Missouri Trucking Association, Minnesota Trucking Association,
Texas Trucking Association, Michigan Trucking Association, Inc.,
Iowa Motor Truck Association, Inc., Nebraska Trucking Association,
Oklahoma   Trucking    Association,   Wisconsin    Motor   Carriers
Association, Wyoming Trucking Association, Inc., and New Jersey
Motor Truck Association, amici curiae.
December 6, 2024
          KAYATTA, Circuit Judge.            In 2016, Rhode Island passed

the   Rhode    Island     Bridge   Replacement,      Reconstruction,     and

Maintenance Fund Act ("RhodeWorks").           Under RhodeWorks, tractor-

trailers and larger trucks (collectively, "tractor-trailers") pay

a toll when they cross any one of thirteen bridges within Rhode

Island.   The state uses the toll revenue to replace, reconstruct,

operate, and maintain its bridges on the National Highway System.

The RhodeWorks tolls are subject to three statutory caps.            A truck

cannot pay a toll more than once in each direction, cannot pay

more than $40 per day, and cannot pay more than $20 for making a

single "through trip" from Connecticut to Massachusetts.

          In    this    lawsuit    brought    by   the   American   Trucking

Associations and several trucking companies (collectively, "ATA"),

the district court permanently enjoined the imposition of tolls

under RhodeWorks.       In so doing, it concluded that the collection

of tolls from only tractor-trailers and the capping of the tolls

each caused the tolls to run afoul of the dormant Commerce Clause.

For the following reasons, we agree that the caps render the tolls

unlawful, but hold that the statute's application to only tractor-

trailers does not.        We also hold that the unlawful caps are

severable from the rest of the statute.




                                    - 4 -
                                    I.

                                    A.

            In 2008, the Rhode Island Department of Transportation

("RIDOT") began to consider new sources of revenue for repairing

the state's transportation system.         Rhode Island has historically

underinvested   in   its   transportation     infrastructure,      with   one

estimate by a state blue-ribbon commission placing its annual

funding gap at $285 million.

            One option was to convert interstate highway bridges to

tolled   bridges.     Generally,    states    may   not   toll    interstate

highways.    

.     But Congress carved out an exception

in the Intermodal Surface Transportation Efficiency Act of 1991

("ISTEA"):   A state may "reconstruct[]" or "replace[]" a toll-free

bridge and then convert that bridge to a toll facility.                   

§ 129(a)(1)(E).     The state may use the resulting funds to maintain

and improve the bridge.       Id. § 129(a)(3)(A)(iii).           It may also

divert any excess funds to "any other purpose for which [f]ederal

funds may be obligated by a [s]tate" under Title 23 of the U.S.

Code.    Id. § 129(a)(3)(A)(v).

            By 2015, RIDOT had decided to use the ISTEA exception to

implement a truck-tolling program.           As initially proposed, the

tolling program applied to both what we are calling tractor-

trailers (vehicles in Classes 8 and above of the Federal Highway

Administration's ("FHWA's") vehicle-classification scheme) and the


                                   - 5 -
smaller vehicles in Classes 6–7 (which, for ease of reference, we

define as "single-unit" trucks).       The program did not seek to toll

smaller trucks, vans, pick-ups, buses, automobiles, and the like,

which fall under Classes 1–5 of the federal vehicle-classification

scheme.

           The Rhode Island General Assembly converted the truck-

tolling proposal into draft legislation.         In June of 2015, then-

Governor Gina Raimondo asked the legislature to revise the draft

legislation so that it        also exempted single-unit trucks.             A

collection of   so-called "equivalent single-axle load"              studies

compiled by RIDOT suggested that tractor-trailers were responsible

for between seventy-two percent and ninety-one percent of highway

damage.   RIDOT also pointed to a Government Accountability Office

("GAO") study from 1979, which concluded that a five-axle tractor-

trailer   weighing   80,000   pounds    "has   the   same   impact    on   an

interstate highway as at least 9,600 automobiles."           The governor

also requested an amendment that would place caps on the tolls

paid by frequent users of the tolled facilities.                In public

statements, then-Senate Majority Leader Dominick Ruggerio stated

that the proposed amendments reflected "the concerns of the local

trucking industry," while RIDOT Director Peter Alviti said they

"came as a result of us listening to the various stakeholders and

transportation industries in Rhode Island."




                                  - 6 -
            After   the   proposed    amendments   exempting    single-unit

trucks and adding the three statutory caps were incorporated into

the statute, RhodeWorks took effect in 2016.            Among other things,

the legislative findings credited RIDOT's "estimate[] that tractor

trailers cause in excess of seventy percent (70%) of the damage to

the state's transportation infrastructure, including Rhode Island

bridges, on an annual basis," while contributing less than twenty

percent of the state's annual transportation-maintenance revenues.

            In its final form, RhodeWorks covers thirteen bridges on

interstate highways in Rhode Island.         None of those bridges cross

the border into an adjoining state.1         The statute applies only to

tractor-trailers,     and   exempts    all   vehicles    in   lower   vehicle

classes, including single-unit trucks.         R.I. Gen. Laws § 42-13.1-

5 (2024).     Given these exemptions, ninety-seven percent of the

vehicles that cross RhodeWorks bridges do not pay tolls.              Out of

the tolled vehicles, nineteen percent are registered in Rhode

Island.




     1  Though ATA makes a glancing reference to Rhode Island's
choice "to impose tolls only on highway corridors that carry
substantial volumes of interstate traffic," it does not develop
any argument that this fact bears on the question of whether
RhodeWorks violates the dormant Commerce Clause. We thus find any
such argument waived. See United States v. Zannino, 

 (1st Cir. 1990) ("[I]ssues adverted to in a perfunctory manner,
unaccompanied by some effort at developed argumentation, are
deemed waived.").


                                     - 7 -
                                    B.

            In July 2018, ATA filed this lawsuit in the U.S. District

Court for the District of Rhode Island.        Alleging that RhodeWorks

violates the dormant Commerce Clause, ATA requested that the court

enjoin the statute's enforcement.          ATA argued that the tolling

system   contravenes     the   dormant    Commerce    Clause      because   it

(1) intentionally      discriminates     against     interstate       commerce,

(2) effectively discriminates against interstate commerce, and

(3) violates    the    "fair-approximation"    test     by     only     tolling

tractor-trailers, even though other vehicle classes also use Rhode

Island's bridges.

            Rhode Island moved to dismiss, arguing that the district

court lacked subject matter jurisdiction under the Tax Injunction

Act, which deprives federal courts of jurisdiction to hear certain

cases related to state taxes.      The district court agreed, and ATA

appealed.    This court reversed, concluding that the RhodeWorks

tolls were not "taxes" within the meaning of the Tax Injunction

Act.   Am. Trucking Ass'ns v. Alviti, 

, 46–47 (1st Cir.

2019) (ATA I).    The district court subsequently denied a motion

for a preliminary injunction.

            Discovery commenced.       ATA issued subpoenas requesting

documents and testimony from several Rhode Island officials, as

well as from an engineering consulting firm (CDM Smith).                    The

district court declined a motion to quash, and Rhode Island


                                  - 8 -
appealed.     This     court    refused    to    quash    the   subpoena    to   the

consultant, but we concluded that legislative privilege barred

discovery from the state officials. Am. Trucking Ass'ns v. Alviti,

, 80–81 (1st Cir. 2021) (ATA II).

            After a bench trial, the district court concluded that

RhodeWorks violates the dormant Commerce Clause, agreeing with ATA

on all three of its stated grounds.             Am. Trucking Ass'ns v. Alviti,

, 399–400 (D.R.I. 2022) (ATA).                     The district

court permanently enjoined Rhode Island from collecting tolls

under RhodeWorks.       

    This appeal followed.

                                         II.

            The    Commerce     Clause    empowers       Congress   to   "regulate

Commerce . . . among the several [s]tates."                 U.S. Const. art. I,

§ 8, cl. 3.       The Supreme Court has also read "a further, negative

command" into the Commerce Clause.              Okla. Tax Comm'n v. Jefferson

Lines, Inc., 

 (1995).               States may not pass laws

that   "discriminate[]         against    or     unduly    burden[]      interstate

commerce."    Gen. Motors Corp. v. Tracy, 

 (1997);

see also South Dakota v. Wayfair, 

 (2018).                  Absent

such a prohibition, states could erect barriers against interstate

commerce to protect local industries, triggering a spiral "into

economic isolation."       Jefferson Lines, 514 U.S. at 179–80.

            In furtherance of these principles, the Supreme Court

has crafted a three-part test to determine if a public-facility


                                     - 9 -
user fee comports with the dormant Commerce Clause.             See Nw.

Airlines, Inc. v. County of Kent, 

 (1994) (citing

Evansville-Vanderburgh Airport Auth. Dist.        v. Delta Airlines,

Inc., 

, 716–17 (1972)).        That Evansville/Northwest

Airlines test applies to highway toll programs. See Doran v. Mass.

Tpk. Auth., 

, 320–21 (1st Cir. 2003).      Under the test,

a tolling system survives dormant Commerce Clause review if it

(1) is based on "some fair approximation of use" of the tolled

facility, (2) "is not excessive in relation to the [governmental]

benefits   conferred,"   and   (3) "does   not   discriminate   against

interstate commerce."    Nw. Airlines, 

.

           The parties agree that, in this case, the second prong

of the Evansville/Northwest Airlines test (i.e., excessiveness)

has been statutorily displaced. As already discussed, ISTEA allows

states to reallocate excess toll revenues to "any other purpose

for which [f]ederal funds may be obligated by a [s]tate under

[Title 23 of the U.S. Code]."    

(a)(3)(A)(v).     Thus,

no toll for an ISTEA-authorized bridge can be stricken merely

because it exceeds the benefits conferred on the users, because

Congress "contemplated that tolls exceeding the amount needed to

fund a toll road would be collected and spent on non-toll road

projects."   See Owner Operator Indep. Drivers Ass'n v. Pa. Tpk.

Comm'n, 

 (3d Cir. 2019).




                                - 10 -
            So,   our    analysis      of     RhodeWorks      revolves       around   two

questions.        First,       does    the        statute    discriminate       against

interstate commerce?           And second, is the burden imposed by the

tolls based on "some fair approximation" of use of the Rhode Island

bridges?

                                            III.

                                             A.

            A state discriminates against interstate commerce when

it    enacts   "economic       protectionism"          by    imposing     "regulatory

measures    designed     to    benefit       in-state       economic    interests     by

burdening out-of-state competitors."                   New Energy Co. of Ind. v.

Limbach, 

 (1988).                 A court can identify "economic

protectionism" by looking to "either discriminatory purpose or

discriminatory effect."           Bacchus Imps., Ltd. v. Dias, 

 (1984) (citations omitted).                   "[B]oth inquiries present

questions of fact."           Waste Mgmt. Holdings, Inc. v. Gilmore, 

, 334 (4th Cir. 2001).

            The   Supreme      Court       has     nevertheless   warned       that   the

Commerce    Clause      primarily      "regulates        effects,      not    motives."

Comptroller of the Treasury of Md. v. Wynne, 

, 561 n.4

(2015).    Indeed, the Court's dormant Commerce Clause jurisprudence

has   consistently      focused       on    "whether    a    challenged      scheme   is

discriminatory in 'effect.'"               Associated Indus. of Mo. v. Lohman,

 (1994); see also Gregg Dyeing Co. v. Query, 286


                                           - 11 -
U.S. 472, 481 (1932) ("Discrimination, like interstate commerce

itself, is a practical conception.     We must deal in this matter,

as in others, with substantial distinctions and real injuries.");

Commonwealth Edison Co. v. Montana, 

 (1981)

(noting that review of challenges to state taxes under the Commerce

Clause focuses on the "practical effect of a challenged tax"

(citation omitted)). Moreover, in its most recent dormant Commerce

Clause case, the Court emphasized that in cases applying the

balancing test from Pike v. Bruce Church, Inc., 

(1970), "the presence or absence of discrimination in practice

[has] proved decisive."   Nat'l Pork Producers Council v. Ross, 

, 378 (2023) (emphasis added); see also id. (noting the

conceptual "congruity" between cases applying Pike and the Court's

"core dormant Commerce Clause precedents").

          Our own dormant Commerce Clause cases have made the same

point.   In deciding a prior appeal arising from this litigation,

we noted that it was "difficult to conceive of a case in which a

toll that does not discriminate in effect could be struck down

based on discriminatory purpose."      ATA II, 14 F.4th at 89.   An

even earlier case made the same point.   See All. of Auto. Mfrs. v.

Gwadosky, 

, 36 n.3 (1st Cir. 2005) (questioning whether

a sole showing of discriminatory intent would "invariably suffice"

to invalidate a statute under the Commerce Clause).    Were the law

otherwise, it would invalidate many more state statutes, given


                              - 12 -
that many legislators routinely claim to have crafted statutes to

accommodate local interests.           The dormant Commerce Clause exists

to eradicate economic protectionism among the states.                 A statute

that erects no barrier against interstate commerce -- despite the

state   legislature's     best   efforts     --   does   not   threaten      such

protectionism.     For that reason, "[w]e will not invalidate a state

statute    under    the   [Commerce]       Clause     merely      because    some

legislators [or officials] sought to obtain votes for the measure

on the basis of its beneficial side effects on state industry."

Minnesota v. Clover Leaf Creamery Co., 

, 463 n.7, 471

n.15 (1981) (first making this point in the context of the Equal

Protection Clause, and then reiterating it in the context of the

Commerce Clause).

            For the most part, the district court's analysis adhered

to the foregoing principles.           The court recognized the primacy of

discriminatory effects in the dormant Commerce Clause analysis of

facially neutral legislation, as well as the role of effects in

"smok[ing] out" discriminatory intent.              See Nat'l Pork Producers

Council, 598 U.S. at 379 (quoting Richard H. Fallon, Jr., The

Dynamic Constitution 311 (2d ed. 2013)); see also id. at 393

(Barrett, J., concurring in part) ("Where there's smoke, there's

fire.").    A discriminatory effect necessarily colors our analysis

because    it   "strengthens     the    inference    that   [a]    statute   was

discriminatory by design."         Fam. Winemakers of Cal. v. Jenkins,


                                   - 13 -


 (1st Cir. 2010).    According to the district court,

the "data" on RhodeWorks' "discriminatory effects" are some of the

"most important evidence" buttressing a finding of discriminatory

intent.   ATA, 630 F. Supp. 3d at 392.       In essence, the district

court's analysis, like ours, ultimately rises or falls with an

assessment of RhodeWorks' effects.2

          With these principles in mind, we turn to the facts of

this case.    Following the lead of the parties and the district

court, we divide our discrimination inquiry into two parts. First,

we consider whether RhodeWorks' exemption for single-unit trucks

discriminates against interstate commerce.        Second, we consider

whether the tolling caps discriminate against interstate commerce.

                                  B.

          ATA claims that RhodeWorks effectively discriminates

against interstate commerce by exempting single-unit trucks from

its tolls.    As we explain below, we disagree.

                                  1.

          The threshold question for courts considering this type

of discrimination claim under the dormant Commerce Clause is not

whether   a   statute   discriminates   at    all,   but   whether   it




     2  The district court's approach also aligned with our
guidance from ATA II, where we stressed that evidence about the
"discriminatory effects . . . of RhodeWorks toll collections is
more probative" of discriminatory intent than legislative
maneuvers by Rhode Island representatives. 14 F.4th at 90.


                               - 14 -
discriminates between "substantially similar entities . . . in a

single   market."   Tracy,   519   U.S.   at   298–300.      Indeed,    "the

principle that any notion of discrimination assumes a comparison

of substantially similar entities" is "a fundamental element of

dormant Commerce Clause jurisprudence."         Dep't of Revenue of Ky.

v. Davis, 

 (2008) (cleaned up).

           In Tracy, a purchaser of natural gas challenged Ohio's

grant of a tax exemption to local gas distributors but not out-

of-state   gas   distributors,     arguing     that   this   differential

treatment unlawfully favored in-state entities.           519 U.S. at 282–

83, 285.   Because the in-state and out-of-state gas distributors

sold different products to different consumer markets, however,

the Court found that they were not competitors.3          Id. at 310.    And

"in the absence of actual or prospective competition between the

supposedly favored and disfavored entities in a single market[,]

there can be no local preference, whether by express discrimination

against interstate commerce or undue burden upon it, to which the

dormant Commerce Clause may apply."       Id. at 300; see also Alaska

v. Arctic Maid, 

, 204–05 (1961) (holding that higher


     3  The Tracy Court did suggest that out-of-state gas sellers
may compete with in-state gas sellers in a more limited, noncaptive
market. See 

. But the record "reveal[ed] virtually
nothing about the details of that competitive market," and the
Court ultimately upheld the differential tax scheme due to its
"traditional recognition of the need to accommodate state health
and safety regulation in applying dormant Commerce Clause
principles." 

.


                                 - 15 -
taxes imposed by the state on freezer ships, which sold fish out

of state, compared to onshore storage facilities, which sold fish

in state, did not violate the dormant Commerce Clause because the

two entities did not compete).

          A   hypothetical       further    illustrates     the    difference.

Imagine that a state legislature proposes a toll on all motor

vehicles and bicyclists using a new roadway in the countryside.

Local voters who frequently bike on the roadway object to the

proposed legislation.      So, the legislature passes a revised bill

that applies only to motor vehicles, which are more likely than

bicyclists    to   come   from   outside    the    state.         Clearly,   the

legislature discriminated in some fashion:            It tolled motorists

and not bicyclists.       No one, though, would seriously argue that

motorists and bicyclists are "substantially similar entities . . .

in a single market."      Tracy, 519 U.S. at 298–300.         In the absence

of competition between similar entities, the amended statute would

impose "no local preference . . . to which the dormant Commerce

Clause [could] apply."      

.

          ATA charges RhodeWorks with discriminating against out-

of-state tractor-trailers in favor of smaller in-state single-unit

trucks.   So, the "threshold question" is whether out-of-state

tractor-trailers and      smaller in-state        single-unit trucks         "are

indeed similarly situated for constitutional purposes."                 

.   If they are not, then eliminating the RhodeWorks tolling


                                   - 16 -
disparity     "would    not   serve        the     dormant     Commerce    Clause's

fundamental    objective      of   preserving         a   national    market    for

competition undisturbed by preferential advantages conferred by a

[s]tate upon its residents or resident competitors."                   

            Were there at least one market in which out-of-state

tractor-trailers       competed    with      in-state        single-unit     trucks,

certainly Plaintiffs could easily prove it.                    ATA is a national

trade association of truck owners; Cumberland Farms is a business

founded in Rhode Island that, among other things, transports goods

throughout New England.       Nevertheless, as the district court found

in no uncertain terms, there is simply no "concrete evidence

demonstrating an increase in Rhode Island-based companies' use of

un-tolled trucks, changes in vehicle fleets, diversion, or any

other data demonstrating that [smaller] trucks compete in the same

market as [tractor-trailers]."4              ATA, 630 F. Supp. 3d at 398

(footnote omitted).       The district court also concluded that the

exemption for smaller trucks provided no "competitive advantage

[to   in-state    competitors]        at     the     expense     of   out-of-state

competitors that use [tractor-trailers]."                 Id. at 399.




      4 In a separate portion of its opinion, the district court
seemed to suggest that Classes 4–7 trucks and Class 8+ trucks
compete in an "undifferentiated market of businesses that
transport goods on interstate highways using trucks of varying
types." ATA, 630 F. Supp. 3d at 397. But the district court did
not identify any record evidence suggesting that this market is,
in fact, "undifferentiated." See id.


                                    - 17 -
            Pushing back on these findings, ATA argues that the

record     contains   evidence   that    out-of-state   tractor-trailers

compete with in-state single-unit trucks.           As support for this

contention, ATA points to testimony of one of its experts that ATA

says "demonstrates that, at least some of the time, either a

straight truck or a tractor-trailer may be used for deliveries,

and   in   these   circumstances   the    tolling   exclusion   gives   the

predominantly Rhode Island-owned straight trucks a competitive

advantage."    But as factfinder, the district court found the cited

testimony "mostly speculative."          Id. at 398.    We have squarely

held that "[c]onjecture . . . cannot take the place of proof" in

a dormant Commerce Clause analysis.         Cherry Hill Vineyard, LLC v.

Baldacci, 

 (1st Cir. 2007).          And we have rejected

dormant Commerce Clause challenges where "the district court found

no compelling evidence of discriminatory effect."         Wine & Spirits

Retailers, Inc. v. Rhode Island, 

 (1st Cir. 2007).

For all its conjecture, ATA simply offers no actual evidence that

tractor-trailers compete with single-unit trucks in Rhode Island,

let alone that out-of-state tractor-trailers compete with in-state

single-unit trucks in Rhode Island.           "The absence of any such

evidence is telling."     

            ATA next points to case law holding that two parties can

be similarly situated for dormant Commerce Clause purposes if they

indirectly compete.     See Bacchus, 

 (noting that two


                                   - 18 -
products can compete even if one does not pose a clear and present

"competitive threat" to the other); Trailer Marine Transport Corp.

v. Rivera Vazquez, 

 (1st Cir. 1992) (noting that

discrimination can occur between two "similarly situated" entities

that are not "direct business rivals").    True enough.   But here

there is no finding of even indirect competition between out-of-

state tractor-trailers and in-state single-unit trucks.

          Even were we to assume that a few Rhode Island single-

unit trucks compete in some manner with a few out-of-state tractor-

trailers, ATA's argument would still fall short.       The dormant

Commerce Clause is not an atomic fly swatter to be wielded against

any and all trivial effects on commerce.    A party challenging a

facially neutral statute under the dormant Commerce Clause must

prove that the statute has a substantial (i.e., beyond de minimis)

competitive effect on nonstate interests.      See Exxon Corp. v.

Governor of Md., 

 (1978) ("The fact that the

burden of a state regulation falls on some interstate companies

does not, by itself, establish a claim of discrimination against

interstate commerce."); Cherry Hill, 505 F.3d at 38–39 ("[A] de

minimis advantage to in-state [companies] . . . [is] insufficient

to establish a discriminatory effect." (quoting Brown & Williamson

Tobacco Corp. v. Pataki, 

 (2d Cir. 2003)) (second

alteration and omission in original)).




                              - 19 -
          In arguing otherwise, ATA points to no case in which a

facially neutral statute was struck down without a finding of more

than a de minimis impact on interstate commerce.         Instead, it

points only to cases involving facially discriminatory legislation5

or legislation from which a substantial discriminatory impact

could be easily inferred.6   Here, there is no claim that RhodeWorks

is facially discriminatory or that a substantial discriminatory

impact could be inferred from its exemption of single-unit trucks.

This is especially true given the district court's finding that

out-of-state tractor-trailers and in-state single-unit trucks do

not compete.

                                 2.

          The district court nevertheless concluded that, for two

reasons, "none of this matters."      ATA, 630 F. Supp. 3d at 399.




     5  See, e.g., Camps Newfound/Owatonna, Inc. v. Town of
Harrison, 

, 575–76 (1997) ("It is not necessary to
look beyond the text of this statute to determine that it
discriminates   against   interstate   commerce.");   Maryland   v.
Louisiana, 

 (1981) ("In this case, the Louisiana
First-Use Tax unquestionably discriminates against interstate
commerce in favor of local interests as the necessary result of
various tax credits and exclusions.       No further hearings are
necessary to sustain this conclusion."). Once a court finds that
a statute is discriminatory, it need not inquire into the extent
of that discrimination to conclude that the statute is
unconstitutional.
     6  See, e.g., Trailer Marine, 

 (noting that,
despite the "absence of firm statistics," "the inference [of
substantial disparate impact] is so compelling that only the amount
of the discrimination, and not its fact, can be plausibly
contested").


                               - 20 -
                                     a.

            First, the district court cited Trailer Marine, 

, to conclude that, given the "overtly protectionist" effects

of exempting single-unit trucks, it did not need evidence "of a

specific market impact" to find discriminatory effect. 

 (citing

Trailer Marine, 

).          But as Trailer Marine implied,

and as Tracy subsequently made clear, the threshold question in

this type of dormant Commerce Clause case is whether the statute

discriminates between similarly situated competitors.              See Trailer

Marine, 

 ("Such an imbalance in favor of local

interests    (here   local   trailers)    over   similarly    situated     non-

resident interests (transitory trailers) is a proper concern of

the   [dormant]      Commerce   Clause    whether    or   not      the   market

participants are direct business rivals.")7; Tracy, 519 U.S. at

298–99 ("[A]ny notion of discrimination assumes a comparison of

substantially similar entities." (footnote omitted)).                And here,

unlike in Trailer Marine, the in-state and out-of-state entities

are   not   similarly   situated   given    "the    absence   of    actual   or

prospective competition . . . in a single market."                 Tracy, 

.




      7 We do not read the reference to "direct business rivals"
as exhausting the relevant universe of competition between
similarly situated tractor-trailers and thus do not read Trailer
Marine as contrary to Tracy.


                                   - 21 -
           Nor does Trailer Marine otherwise provide the support

claimed by ATA.       Trailer Marine's analysis began with the apt

observation    that    "whether      discrimination        exists     is     heavily

dependent upon the facts."         

.    The court then pointed

to the case's most salient fact -- the challenged fee was in

substance a flat fee imposed on "all classes of motor vehicles

including trailers."         

    The flat fee -- like the flat fee in

American Trucking Ass'ns v. Scheiner, 

 (1987) -- was

"clearly discriminatory in impact," imposing a per-accident cost

on nonresident trailers that was between five and six times the

per-accident   cost    on     "similarly     situated"      resident       trailers.

Trailer Marine, 977 F.2d at 10–11.

           Here, by contrast, but for the caps (which we discuss

below), there is no evidence that RhodeWorks discriminates between

similarly situated entities to begin with.                 RhodeWorks does not,

for example, impose a per-mile (or per-bridge) fee on out-of-state

tractor-trailers      that    exceeds    the    fee   charged       to     similarly

situated   in-state    tractor-trailers.           Instead,     the      challenged

differential here is the fee charged to all tractor-trailers as

compared to no fee charged to smaller trucks that do not compete

with tractor-trailers.

                                        b.

           Second,     in     discussing       discriminatory       intent,       the

district   court   noted     that    "[t]here    is   no    question       that   the


                                     - 22 -
RhodeWorks legislation excluded lower-classed trucks to reduce the

financial burden on in-state businesses."         ATA, 630 F. Supp. 3d at

399.

            In public statements during the drafting process, RIDOT

Director     Alviti   and    then-Senate     Majority    Leader   Ruggerio

acknowledged that the "local trucking industry" advocated for the

caps and the small-truck exemption.          But neither man claimed that

the exemption would privilege in-state truckers over out-of-state

truckers.     Cf. ATA II, 14 F.4th at 89 n.7 (noting that similar

public statements by other Rhode Island officials "[did] not admit

that the [tolling burden on out-of-staters was] disproportionate

to the relevant use of the bridges by out-of-staters").             And the

statements were certainly not as nakedly protectionist as the ones

on which this court has previously relied to find discriminatory

intent.     See Fam. Winemakers of Cal., 

 (pointing to

legislator     statements    that    a   Massachusetts    statute     would

"inherent[ly] advantage" in-state wineries).

            In any event, we need not strike a facially neutral state

tolling statute that exempts both local and out-of-state similarly

situated entities merely because the statute responded to local

businesses' concerns.8      Moreover, as the district court explained,


       8Indeed, in this instance, most of the exempted vehicles in
Classes 4–7 bear out-of-state plates, ATA, 630 F. Supp. 3d at 392,
and there is no reason to suspect that the out-of-state percentage
of single-unit trucks in Classes 6–7 differs.


                                    - 23 -
its   finding    of     an   intent   to   discriminate          against    interstate

commerce rested principally on its finding that the legislation

had discriminatory effects.              ATA, 630 F. Supp. 3d at 392.                A

finding of intent so inferred cannot survive absent discriminatory

effect.

           Nor     does       the   amendment         of   the    broader    proposed

legislation      move    the    dial.      Generally        speaking,       "statutory

interpretation cannot safely . . . rest upon inferences drawn from

intermediate legislative maneuvers."                   All. of Auto. Mfrs., 

.       This is especially true when a party relies on a

statutory amendment, because "there are countless reasons why the

state legislature may have altered its position."                           

   That

general rule applies here.

           Much state regulation contains exemptions for smaller

employers, and smaller employers are more likely to be local than

are larger employers.           In Rhode Island, for example, state anti-

discrimination laws apply only to employers that have four or more

employees.      R.I. Gen. Laws § 28-5-6(9)(i) (2024).                That is to say,

many exemptions in state legislation effectively and foreseeably

reduce the regulatory burden imposed on local companies as compared

to the burden imposed on out-of-state companies.                      So, if such a

disparate impact were sufficient to strike down a statute in the

absence   of    facial       discrimination      or    a   substantial      impact   on




                                        - 24 -
competition, the dormant Commerce Clause would assume a role that

exceeds our understanding of its purpose.

              In sum, the record provides insufficient support for

ATA's contention that exempting all single-unit trucks from the

RhodeWorks tolling structure transgresses the dormant Commerce

Clause.

                                       C.

              We now turn our discrimination inquiry to the tolling

caps.

              To reiterate, RhodeWorks imposes three statutory caps

that reset daily.          See id. § 42-13.1-4(b)-(d).         The first cap

prevents any truck from paying more than once in each direction at

a given gantry.      The second cap prevents a truck from paying more

than    $20   in   tolls   on   a   "through   trip"   from   Connecticut   to

Massachusetts (or vice versa).           And the third cap prevents any

truck from paying more than $40 per day.           The district court found

that the second cap is "irrelevant," because a truck making such

a "through trip" would -- under current prices -- only pay around

$18 in tolls anyway.        ATA, 630 F. Supp. 3d at 394 n.50.        Neither

party argues otherwise.         So, like the district court, we focus on

our analysis on the remaining two caps.

              There is no question          that in-state tractor-trailers

compete in overlapping markets with out-of-state tractor-trailers.

Not even the state argues otherwise.            So, we focus on whether the


                                     - 25 -
caps in their effect provide a competitive advantage to in-state

tractor-trailers as compared to out-of-state tractor-trailers.

           We begin with two points on which the law is clear.

First, a flat tax on the right to "mak[e] commercial entrances

into [a state's] territory" would run afoul of the dormant Commerce

Clause.   Scheiner, 

.      Such a tax would not correlate

with road usage, and it would have "plainly" discriminatory effects

because it would impose a higher per-mile cost on out-of-state

vehicles relative to in-state vehicles.        

.   Second, and

conversely, a toll that is "directly apportioned to . . . mileage

traveled" would not offend the dormant Commerce Clause, because it

would "maintain state boundaries as a neutral factor in economic

decisionmaking."   

.   Under a usage-based tolling system,

a driver is "simply pay[ing] for traveling a certain distance that

happens to be within [a given state]."        

           Here, though, we are dealing with a hybrid model: a

usage-based toll that is capped after a certain number of gantries

are passed, and then reset daily.          One can imagine contrasting

scenarios in which such a capped toll resembles either a flat tax

or a usage-based toll.      For instance, if all drivers will easily

reach the caps, then the toll is effectively a flat tax because

everyone will ultimately pay the capped amount for the privilege

of using a state's roads.    By contrast, if drivers will never reach

the cap -- as is presently the case with the RhodeWorks toll on


                                 - 26 -
"through trips" -- then the toll functions as a nondiscriminatory

usage-based toll.

            It is more difficult to categorize tolls like the ones

before us: tolls with caps that are only sometimes reachable, more

likely by in-state tractor-trailers than by out-of-state tractor-

trailers, and that reset daily. In such circumstances, we consider

practical burdens that the caps place on out-of-state vehicles but

not on similarly situated in-state vehicles.         If the RhodeWorks

caps were facially discriminatory, even a de minimis burden on

out-of-state vehicles might be enough to invalidate them.           See

Camps Newfound/Owatonna, Inc. v. Town of Harrison, 

,

581 & n.15 (1997).     But they are not.    Accordingly, we look to see

whether the caps' burden on interstate commerce is more than merely

de minimis.    See Cherry Hill, 

.

            The evidence demonstrated, and the district court found,

that local tractor-trailers disproportionately benefited from the

caps   as   compared   to    out-of-state   tractor-trailers,   thereby

reducing the per-mile tolls paid on average by in-state tractor-

trailers, and in nontrivial amounts.        For example, 39.9% of the

reductions in what the tolls would have been but for the caps went

to Rhode Island vehicles even though they accounted for only 18.6%

of the transactions.        ATA, 630 F. Supp. 3d at 395.    This means

that in-state tractor-trailers pay, on average, substantially less




                                  - 27 -
than out-of-state tractor-trailers pay for each pass through a

RhodeWorks gantry.

            The state argues that the out-of-state vehicles are more

likely to get at least some benefit from the caps, but does not

dispute the district court finding that in-state vehicles receive

a disproportionate share of the cap benefits.               As Scheiner made

clear, our primary concern is whether "[i]n the general average of

instances, the [challenged] privilege is [less] valuable to the

interstate [carrier than to the] intrastate carrier."               

 (quoting Capitol Greyhound Lines v. Brice, 

(1950)    (Frankfurter, J.,    dissenting)).        Even     if    out-of-state

trucks are more likely to receive at least some benefit from the

caps,     the   evidence      from      trial     shows     that      discounts

disproportionately flow to in-state trucks.             ATA, 630 F. Supp. 3d

at 395.     In other words, out-of-state tractor-trailers receive

substantially less of a discount per bridge crossing than do in-

state tractor-trailers.       Therefore, the privilege of toll capping

is considerably more valuable for intrastate carriers than it is

for interstate carriers.

            Given   this   disparate     impact    on     similarly    situated

tractor-trailers, the caps are discriminatory under the logic of

Scheiner and Trailer Marine.           Even though out-of-state and in-

state tractor-trailers can both benefit from the statutory caps,

the caps still create "a privilege that is several times more


                                     - 28 -
valuable to a local business than to its out-of-state competitors."

Scheiner, 

.              Because the disparity between the

discounts received by in-state and out-of-state tractor-trailers

is     so    pronounced   in    favor     of     locals,   "the   inference    [of

discrimination] is so compelling that only the amount of the

discrimination, and not its fact, can be plausibly contested."

Trailer Marine, 

.                And because even the state does

not contest that out-of-state tractor-trailers compete with in-

state       tractor-trailers,    no     basis     exists   for    treating    this

discrimination as permissible.

               In   resisting   the     conclusion     that   RhodeWorks'     caps

unlawfully interfere with interstate commerce, Rhode Island relies

heavily on our 2003 decision in Doran, which involved an option to

obtain discounted tolls by purchasing a transponder.                 

.        But in Doran, there was no evidence that by offering the

option to buy a transponder and get discounted rates, the state in

any way affected competition between in-state and out-of-state

interests in any market.          Rather, "the incremental burden of the

undiscounted toll for the infrequent traveler" who did not make

use of the transponder program was "de minim[i]s," and the mere

fact that nonparticipants paid higher tolls did not mean "that

interstate commerce [would] be burdened, much less that it [would]

suffer discrimination."         

.        Put simply, in Doran, there

was neither proof of competitive harm nor the type of disparate


                                        - 29 -
treatment that would make such a competitive impact obvious.               So,

eliminating the state's transponder program would not have served

the "fundamental objective" of the dormant Commerce Clause -- to

"preserv[e]   a    national   market    for    competition   undisturbed   by

preferential advantages conferred by a [s]tate upon its residents

or resident competitors."       Tracy, 

.           RhodeWorks'

caps, by contrast, directly bear on competition between in-state

and out-of-state tractor-trailers.         The record shows that the caps

disproportionately benefit in-state tractor-trailers over out-of-

state tractor-trailers, and to a nonnegligible extent.              As such,

the caps resemble more closely the flat taxes struck down in

Scheiner.

            This    resemblance        finds     support     in    Scheiner's

internal-consistency test for identifying problematic tolls.               To

apply that test, we ask what would happen if each state adopted an

identical toll regime.        See Scheiner, 

.           As in

Scheiner, the toll regime here would mean that tractor-trailers

staying within state borders would pay on average substantially

less per mile than those engaged in interstate travel.

            The state insists that the RhodeWorks caps are not "flat

fees"; rather, they are per-use fees with retroactive "frequency-

based discounts."     To understand this framing, consider a highly

simplified version of the $40-per-day cap.            Imagine that a truck

pays $10 per toll and hits the $40 cap after paying four tolls.


                                  - 30 -
Because the truck has hit the cap, the fifth toll is free.               At

this point, the truck's effective per-toll payment is no longer

$10 per toll -- it is $8 per toll.        Thus, on Rhode Island's view,

RhodeWorks   is   simply   a   garden-variety    per-use   toll   with   a

retroactive discounting mechanism.         And per-use tolling systems

satisfy Scheiner    because the fees depend on         use, not states

entered.   See 

 (explaining that a fee that is "directly

apportioned to the mileage traveled in" a given state does not

violate the internal-consistency test because the fee is simply

"for traveling a certain distance that happens to be within" that

state); see also Doran, 

 (making the same point).

           But that same logic could apply to a flat-fee system.

Imagine that Rhode Island required an upfront fee of $40 to use

its roads.    Under Rhode Island's view, the flat fee is still

correlated with road usage because trucks that drive more will pay

a lower effective per-mile fee.       The more miles a truck drives,

the greater its retroactive "frequency-based" discount.              Yet,

under Scheiner, that flat fee would clearly fail to pass muster.9

           Rhode Island protests that the Supreme Court "pared

back" Scheiner in American Trucking Ass'ns v. Michigan Public

Service Commission, 

 (2005).        But there is no language


     9  ATA contends that the caps fail the internal-consistency
test but develops no argument that RhodeWorks' exemption of single-
unit trucks would also be problematic under the internal-
consistency test. Thus, we consider that argument waived.


                                 - 31 -
in that case so stating, nor has the Supreme Court subsequently so

suggested.        Instead, that case, like Scheiner, rested on a finding

that "neither record evidence nor abstract logic" suggest that the

challenged fee penalizes any trucks.               Wynne, 

 n.7

(cleaned up) (describing Michigan Public Service Commission).

             Rhode Island also seeks to distinguish Scheiner based on

the fact that the flat fee in Scheiner was not applied equally to

all trucks; rather, in-state trucks effectively received a rebate

for the fee, paying nothing net.                See 483 U.S. at 274–75.          But

this argument overlooks Trailer Marine.               That case involved a flat

fee    of   $35    assessed     on   trailers    in   Puerto    Rico,   which    the

commonwealth used to fund an accident-compensation plan.                   

.      Unlike the fee in Scheiner, the Puerto Rico fee was even-

handed.     

     It applied to in-state trailers, as well as out-of-

state "transitory trailers" so long as they remained in Puerto

Rico for more than a month.             

 at 3–4, 10.        Applying Scheiner,

we    nevertheless       concluded    that   the   fee   violated    the   dormant

Commerce Clause.          

 at 10–12.         Transitory trailers spent far

less time on Puerto Rico's roads.                  They were therefore likely

responsible        for   only   a    small   fraction    of    accidents   in   the

territory.        

.    But because Puerto Rico assessed flat fees

depending on whether the trailer stayed in Puerto Rico for thirty

days   or    fewer,      even   an   out-of-state     transitory    trailer     that

qualified for the reduced $15 fee would "effectively pay[] five or


                                       - 32 -
six times as much per accident" on average compared to the in-

state trailer staying in Puerto Rico year-round.            

   The fee was

therefore invalid.      

.

            Rhode Island also notes that in Scheiner, the "disparate

impact of the flat tax on in-state versus out-of-state interests

was much more dramatic [than the impact in this case]."                  To be

sure, the fee in Scheiner resulted in a cost-per-mile on out-of-

state trucks that was around five times greater than the one

imposed on in-state trucks.          

.      But Scheiner did

not suggest that the Pennsylvania axle fee would have survived if

its disparate impact on out-of-state trucks were smaller but still

substantial.    And for good reason.          At least when, as here, the

impact is clearly substantial, any attempt to identify how much

disparate impact is "too much" would result in a purely arbitrary

rule.   That is why, in Trailer Marine, we refused to endorse a

"specific   figure"     for   the    "cumulative    disparity"    that   would

justify invalidating a fee as effectively discriminatory.                   

.    Instead, we simply noted that the challenged fee's

disparate impact was sufficiently substantial and onerous for out-

of-state    interests    that   it    could   not   "be   brushed   aside   as

incidental."    

     That logic applies here.          Even if RhodeWorks

does not result in a disparate impact as large as the ones in

Scheiner or Trailer Marine, we cannot "brush[] [it] aside as

incidental."    

                                     - 33 -
            Rhode   Island   next   stresses        that     no    record   evidence

clearly demonstrates that the toll caps deter interstate commerce.

But as discussed above, when a fee disproportionately burdens

similarly situated out-of-state competitors, a court may -- as in

Scheiner and Trailer Marine -- infer discriminatory effect from

the   non-incidental    burden      on        out-of-state        interests.    See

Scheiner, 483 U.S. at 286–87; Trailer Marine, 

.

Neither Scheiner nor Trailer Marine involved precise evidence

about the extent to which the challenged fee limited interstate

commerce.    But in each case, the fee was presumptively invalid

because it disparately burdened similarly situated out-of-state

entities so much that the fee was "clearly discriminatory in

impact," such that "only the amount of the discrimination, and not

its fact, [could] be plausibly contested."                 Trailer Marine, 

; see also Scheiner, 483 U.S. at 286–87.

            For the foregoing reasons, we agree with the district

court that the RhodeWorks caps effectively discriminate against

interstate commerce and are therefore unconstitutional.

                                         D.

                                         1.

            Having concluded that the small-truck exemption survives

the discrimination inquiry, but the caps do not, we turn to the

district court's conclusion that RhodeWorks' small-truck exemption




                                    - 34 -
violates the fair-approximation test.10              ATA, 630 F. Supp. 3d at

380.    This is the first prong of the Evansville/Northwest Airlines

analysis.

            Evansville's fair-approximation test poses "essentially

a question of allocation; we ask whether the government is charging

each individual entity a fee that is reasonably proportional to

the entity's use, and whether the government has reasonably drawn

a line between those it is charging and those it is not." Industria

y Distribucion de Alimentos v. Trailer Bridge, 

(1st Cir. 2015).     With one possible exception,11 this inquiry is

distinct from the tests used to assess the other two prongs under

Evansville    --   i.e.,    whether    a     tolling   program   discriminates

against     interstate     commerce     or     is   excessive.     

    The

discrimination prong is concerned with user fees that unfairly

advantage in-state entities over similarly situated out-of-state

competitors, while the excessiveness prong (statutorily displaced

in this case) asks whether the fee imposed on users of a public

facility is reasonable compared to the costs incurred by the state




        Because we have already concluded that the RhodeWorks caps
       10

discriminate against interstate commerce (and thereby violate the
third prong of the Evansville/Northwest Airlines test), we need
not subject them to (or even consider if they fall within) the
fair-approximation test.
     11 It is not clear whether absolute excessiveness can doom a
fee in the context of an ISTEA waiver, or whether such an inquiry
might bear on a fair-approximation analysis. Because this issue
was not developed on appeal, we do not address it.


                                      - 35 -
to improve or maintain that public facility.                     

.     By

contrast, the fair-approximation inquiry concerns whether the fee

imposed     on    an    entity     "reflect[s]       a   fair,    if   imperfect,

approximation of the use of facilities for whose benefit they are

imposed."       Evansville, 

.             The standard here is a

lenient one:       We will strike down a public facility fee as not

fairly approximated only if the legislature allocated it in a

"wholly unreasonable" manner.                

; see also N.H. Motor

Transp. Ass'n v. Flynn, 

 (1st Cir. 1984) ("[T]he

Constitution requires not 'precision' but 'rough approximation' in

matching fee and benefit." (cleaned up)); Selevan v. N.Y. Thruway

Auth.,    

   (2d    Cir.   2013)   (asking   whether   a

distinction between paying and nonpaying motorists on Grand Island

Bridge was "wholly unreasonable").

                                         2.

            At first blush, RhodeWorks (minus the caps) would seem

to pass the fair-approximation test quite easily.                   In ATA's own

words to this court, the RhodeWorks toll is "a paradigmatic toll."

Opening Brief for Plaintiffs-Appellants at 27, ATA I, 

(1st Cir. 2019) (No. 19-1316).               "It is paid only by the user of

tolled bridges, for each use of the bridges; it is paid for the

privilege of using those facilities," so that "there is a direct

correlation between the fee . . . and the use of the property."

 at 25–26. Rhode Island's legislature "estimate[d] that tractor


                                       - 36 -
trailers cause in excess of seventy percent (70%) of the damage to

the    state's      transportation      infrastructure . . .      on     an   annual

basis."     R.I. Gen. Laws § 42-13.1-2(8) (2024).               Having so found,

the legislature granted RIDOT the "[a]uthority to collect tolls on

large commercial trucks only," with the tolls to "be fixed after

conducting a cost-benefit analysis."             Id. § 42-13.1-4.        The amount

of    the   tolls    was   to    be   determined   based   on    "the     costs   of

replacement, reconstruction, maintenance, and operation of Rhode

Island's system of bridges."             Id. § 42-13.1-8.       RIDOT's estimate

that tractor-trailers cause over seventy percent of the damage to

Rhode Island's bridges was principally based on its review of five

studies     employing      "an    equivalent     single-axle      load    ('ESAL')

methodology, which considers pavement thickness to measure the

impact of vehicle load."              And no party disputes that the tolls

would have been allocated to fund roughly seventy percent of the

repair costs.        At trial, Rhode Island presented additional expert

testimony using a different methodology (the "fatigue analysis"),

which similarly concluded that tractor-trailers caused between

seventy and eighty percent of bridge damage.12




        After RhodeWorks was enacted, RIDOT performed its own ESAL
       12

analysis, which likewise found that tractor-trailers accounted for
around eighty percent of bridge damage. It is not clear to us
that RIDOT's analysis is methodologically different from the ESAL
studies on which the agency initially relied.


                                        - 37 -
            The district court nevertheless held that the tolls did

not represent a fair approximation of tractor-trailers' use of the

bridges.    ATA, 630 F. Supp. 3d at 389.         The district court seemed

to view the fee as excessive because a tractor-trailer's "use" of

a bridge is no different from a car's use of the bridge, which the

district court defined as "to cross" the bridge.                     Id. at 384.

Clearly, though, a state can charge users of a facility a fee that

covers all or some portion of the damage that use does to the

facility.     See Cont'l Baking Co. v. Woodring, 

(1932). For this reason, the Supreme Court has "sustained numerous

tolls based on a variety of measures of actual use, including . . .

[the]   manufacturer's     rated    capacity    and    weight   of    trailers."

Evansville, 

.

            The district court relied alternatively on a review of

expert testimony concerning how best to estimate the damage done

to bridges by the different types of vehicles that cross them.

ATA, 630 F. Supp. 3d at 385–87.          The state's own expert (Dr. Small)

stated that ESAL studies are generally used to measure pavement

(not bridge) damage, and that he would not "use [them] to look at

bridges."    Id. at 386.    The district court further credited expert

testimony    that    the   GAO   study    was   only   designed      to   analyze

overweight    and    oversized     vehicles'    impact    on    pavement,     not

bridges.     Id.    Moreover, the district court rejected the state's

fatigue analysis, noting that bridges on interstate highways "are


                                    - 38 -
designed to withstand the flow of heavy trucks."              Id.     Instead,

the court agreed with ATA that a fourth type of analysis -- called

"highway cost allocation studies," or HCASs -- was the superior

method for measuring bridge damage.             Id. at 381.       Under that

method, the court noted, tractor-trailers are responsible for

around twenty to forty percent of bridge damage.            Id. at 388.

            While the court found that the ESAL studies used by Rhode

Island are "flawed" measures, it never found that it was "wholly

unreasonable" for Rhode Island to rely on a vehicle's relative

contribution to pavement damage as a proxy for estimating relative

damage to the paved bridges.         Id. at 386–87.    Nor could it have so

found.     As noted earlier, the Supreme Court has blessed a broad

"variety     of    measures   of    actual   use,"    including     "gross-ton

mileage . . . and manufacturer's rated capacity and weight of

trailers."        Evansville, 

.        In Evansville itself,

the Court cited "aircraft weight" as an example of a permissible

measure of use.       

.    And in one case, the Court upheld a

Kansas statute that taxed heavy trucks to fund highway maintenance,

expressly holding that the legislature could allocate the tax to

those trucks whose "character of use" tore up the state's highways

and created the need for the maintenance tax.             See Cont'l Baking

Co., 

.

            Logically, if a bridge's pavement is impassable and

potholed, a driver cannot use the bridge safely even if the


                                    - 39 -
bridge's other components remain perfectly healthy.                       Pavement

health is, therefore, at least somewhat correlated with the safety

and    utility    of   entire   bridges,        the   maintenance    of   which   is

RhodeWorks' primary goal.            For instance, under federal regulations

for implementing the National Highway Performance Program, Rhode

Island must measure a bridge's structural integrity under "the

minimum of condition rating method."13                  

(b)

(2024).     Basically, a bridge's condition rating is the lowest of

the condition ratings assigned to its component parts.                     See 

In other words, a bridge is only as strong as its weakest link.

So,    if   a    bridge's     deck     (which    includes     the   pavement)     is

significantly degraded, federal law requires a state to assume

that the entire bridge is significantly degraded.                   See 

            The state's reliance on the ESAL analyses also appears

not wholly unreasonable given that its subsequent fatigue analysis

also    concluded      that   tractor-trailers        cause   around   seventy    to

eighty percent of bridge damage.             To be sure, the district court

found flaws with the state's fatigue analysis, concluding that

HCAS analyses are more reliable.            ATA, 630 F. Supp. 3d at 386–88.

For example, the district court noted that the fatigue analysis

ignored "other ways vehicles impact bridges."                  

.        The




        We take judicial notice of regulations published in the
       13

Federal Register. See 

.



                                        - 40 -
court also highlighted that most interstate highways "are designed

to withstand the flow of heavy trucks," pointing to a statement by

Rhode Island's fatigue-analysis expert (Dr. Nowak) that certain

bridge components have a theoretically "infinite fatigue life."14

            These may be perfectly fair criticisms.    But they do not

suggest that the fatigue analysis is so flawed that Rhode Island's

initial estimate was wholly unreasonable.     Consider the district

court's reference to Dr. Nowak's comment about "infinite fatigue

life." 

 Elsewhere, Dr. Nowak testified that, while some bridge

components are "over-designed," others "are designed exactly to

the code requirements," and those "are the components which would

wear out first."    Those "others" presumably include the pavement,

which no one claims does not need to be repaired and replaced from

time to time.     Moreover, Dr. Nowak noted that if a bridge is not

well-maintained (as many Rhode Island bridges are not), certain

components will fatigue even faster when subject to heavy vehicle

loads.     So, while some bridge components are indeed "designed to

withstand the flow of heavy trucks," others will fatigue much

faster under the burden of heavy truck loads.    

      14The district court noted several other potential flaws
with the fatigue analysis, but then decided that it "need not
interrogate them given its holdings." ATA, 630 F. Supp. 3d at 386
n.39. We therefore do not review them here.


                                - 41 -
             Separately, Dr. Small offered some criticisms of the

HCAS studies offered by ATA, noting that HCAS studies allocate

maintenance costs to different vehicle classes based on the initial

construction costs incurred to accommodate those vehicle classes.

In   other   words,   the   assumed   maintenance   costs   for   a   given

classification of vehicles are a function of the money spent to

make the bridge usable for that vehicle class.          They are not a

function of "what actually happens on the bridge over time[,] which

might be quite different."      The district court did not discredit

this testimony.

             We need not (and therefore do not) hold that the district

court's factual findings were clear error.      The question before us

is not whether the district court correctly concluded that an HCAS

analysis is more accurate than an ESAL or fatigue analysis for

measuring bridge damage.      Rather, the question is whether it was

"wholly unreasonable" for Rhode Island to rely on the ESAL (and

GAO) studies when concluding that the larger trucks as classified

by the FHWA cause the most damage to RhodeWorks bridges.                See

Evansville, 405 U.S. at 717–18.

             We cannot say that it was.    At the most basic level, it

does not strike us as wholly unreasonable to presume that bigger

trucks will cause more damage to bridges, and smaller trucks, less.

And Rhode Island's conclusion that tractor-trailers cause most of

the damage to the pavement is consistent with that common sense.


                                 - 42 -
We therefore do not substitute our own judgment for that of the

Rhode Island legislature.

                                 3.

          Finally, the district court held that even if Rhode

Island could equate bridge "use" with bridge "damage," and even if

Rhode Island could rationally show that tractor-trailers caused

most bridge damage, the state still could not impose tolls on

tractor-trailers to recoup that cost unless it also imposed tolls

on all "users having more than a 'negligible' impact on the tolled

facilities."    ATA, 630 F. Supp. 3d at 387.

          The court's main authority on this point was Bridgeport

& Port Jefferson Steamboat Co. v. Bridgeport Port Authority, 

 (2d Cir. 2009).   See ATA, 630 F. Supp. 3d at 387.   In that

case, the Second Circuit held that a municipal port authority could

not fund most of its operating budget -- which covered ferry and

nonferry programs -- via a flat tax on ferry passengers.     567 F.3d

at 82–83, 88.   Nonferry passengers also used the port authority's

other facilities.     Id. at 84.      Therefore, fair approximation

required that they have some skin in the game and contribute

something to the port authority's operating budget.     See id. at

88.

          In response, Rhode Island argues that Evansville allows

it to exempt a class of nonnegligible users from RhodeWorks




                               - 43 -
tolling.15 There, the Court upheld a flat fee on commercial airline

passengers to fund airport maintenance.       405 U.S. at 720–21.

Critically, that fee did not apply to noncommercial passengers,

even though those passengers also used the airport's runways,

navigational facilities, and aviation-related services.     Id. at

717–18.   The fee also did not apply to commercial passengers on

nonscheduled flights, commercial passengers on light aircraft,

military passengers, or nonpassenger airport users (e.g., people

dining at airport restaurants).        Id.   Taken together, these

exemptions covered most of the airport's users.   Id. at 717.




     15  Rhode Island cites several other cases for the proposition
that a fee can pass fair approximation even when it exempts a class
of nonnegligible users. But none of those other cases meaningfully
supports the state's argument. Most of those cases held that a
facility fee may target a specific class of payers when those
payers are also the facility's only users. See Nw. Airlines, 

 (upholding runway maintenance fee assessed on airlines
and not on concessionaries, because only the airlines used the
runways); Flynn, 751 F.2d at 49–50 (holding that New Hampshire
could assess a fee for funding hazardous-waste programs solely on
hazardous-waste truckers, because the supported programs existed
entirely to benefit those truckers).      A fourth case discussed
whether a fee could pass fair approximation if it was overinclusive
(i.e., incidentally levied on non-users of a service), rather than
underinclusive of nonnegligible users, as is the case here. See
Trailer Bridge, 797 F.3d at 145–46.     And a fifth case explored
whether a facility fee may be assessed on "indirect" users, whose
businesses rely on the existence of the public facility. Alamo
Rent-a-Car, Inc. v. Sarasota-Manatee Airport Auth., 

 (11th Cir. 1990) (holding that an airport could assess a
percentage-based fee on an off-site car-rental service, because
the service still indirectly used the access roads to the airport
and therefore benefitted from the facility's existence).


                              - 44 -
              We cannot square the district court's conclusion with

Evansville's holding that a public authority may assess a fee on

only the most significant group of facility users, even if other

nonnegligible users of the facility are exempt, at least as long

as its justification for doing so is not wholly unreasonable.16

    at    717–18.      Evansville    made    clear   that   assessing    the

maintenance fee only on commercial passengers made sense, because

commercial      travel    "require[d]    more    elaborate     navigation   and

terminal facilities, as well as longer and more costly runway

systems, than [did] flights by smaller private planes."                 Id. at

718.        Commercial    aviation   demanded    more   from    the   airport's

facilities, so commercial passengers needed "to bear a larger share

of the cost of facilities built primarily to meet [their] special

needs."      Id. at 718–19.

              Evansville further suggested that the state may exempt

nonnegligible users for administrability reasons.              See id. at 716.

Administrability is one of the reasons the fair-approximation

standard exists.         It would be onerous and expensive to require a




        ATA attempts to distinguish Evansville on the grounds that
       16

the exempted passenger classes in that case were "trivial."
Nothing in the text of the opinion suggested this.         On the
contrary, the fact that most airport users were exempt from the
fee cuts against ATA's characterization of the exempt passengers
as a negligible segment of overall users. See 

.
To be sure, many exempt users were nonpassengers. See 

 But
Evansville simply did not say that the exempt passenger users were
a "trivial" chunk of the overall user population.


                                     - 45 -
state to assess "every factor affecting appropriate compensation

for [facility] use" before constructing a tolling system.                  

(quoting Capitol Greyhound Lines, 

).               Facility fee

systems that reasonably exempt certain payer classes to minimize

the "administrative burdens of enforcement" can comport with the

fair-approximation test.         

    In Evansville, commercial airlines

were responsible for collecting the maintenance fee.               Id. at 709.

This was a more administrable approach than extracting a fee from

each   airport   user,    many    of    whom    would   have   lacked   billing

relationships with associated airlines, making collection much

harder.

           Evansville's logic applies to RhodeWorks.              Just as the

commercial passengers in Evansville were the most intensive users

of airport facilities, Rhode Island concluded with at least some

reason that tractor-trailers cause the most wear and tear to Rhode

Island's bridges.        Thus, like the airport in Evansville, Rhode

Island may collect a fee from the most intensive users without

having to also collect a fee from lesser users.                Moreover, Rhode

Island urges -- and ATA does not dispute -- that charging only the

largest trucks is more administrable than charging each of the

tens of thousands of smaller vehicles.           Furthermore, by relying on

the preexisting federal vehicle classification system and focusing

on a classification that corresponds to an observable physical




                                       - 46 -
characteristic (i.e., the "gap" between tractor and trailer),17

Rhode Island can reasonably point to a benefit from deciding to

apply its toll to only tractor-trailers rather than to all the

varied and much more numerous vehicles that cross its bridges.

            While     Evansville     is      on     point,     Bridgeport     is

distinguishable.      In that case, fees on ferry passengers covered

almost the entire port authority operating budget, which supported

ferry and nonferry services.        

.          So, the key fair-

approximation     problem   there    was     that    ferry    passengers    were

supporting port facilities that they did not use at all and often

could not even access.      See 

.          Moreover, even where some

port facilities did indirectly benefit ferry passengers, there was

"nothing in the record to indicate how the portion of . . . costs

borne by the ferry passengers compare[d] to the costs, if any,

borne by large vessels" that were the primary beneficiaries of

those services.      

.     In other words, the Port Authority

presented    no     discernable    rationale        behind   how   costs    were

apportioned between ferry passengers -- who benefitted minimally

from those port facilities but bore the entirety of the fee -- and

large vessels, which made extensive use of those same facilities

but paid none of the fee.     

    That is not what is happening here.




     17 RhodeWorks gantries assess tolls with lasers that detect
the "telltale gap between the tractor and trailer that marks a
vehicle in Class[es] 8–13."


                                    - 47 -
Tractor-trailers are not paying to maintain bridges that they do

not (or cannot) use.   Instead, they are paying to maintain bridges

that they use. And there is a plainly discernible rationale behind

how costs are apportioned under RhodeWorks:          The fee is levied

only on vehicles that Rhode Island regards as inflicting the most

damage to the bridges they use.       Nothing in Bridgeport suggests

that this allocation scheme is impermissible under the fair-

approximation test.

                                IV.

          We are not quite done.        Because we conclude that the

RhodeWorks caps violate the dormant Commerce Clause, but the small-

truck exemption does not, we must determine whether the caps are

severable from the rest of RhodeWorks.      We hold that they are.

          Severability is a matter of state law.        Leavitt v. Jane

L., 

 (1996) (per curiam).           In Rhode Island, a

court may sever an unconstitutional provision when it "is not

indispensable to the rest of the statute and can be severed without

destroying legislative purpose and intent."     Landrigan v. McElroy,

 (R.I. 1983).        Ultimately, "[t]he test for

determining" severability "is 'whether, at the time the statute

was enacted, the legislature would have passed it absent the

constitutionally   objectionable      provision.'"       

   (quoting

Scheinberg v. Smith, 

 (5th Cir. 1981)).              A

severability provision is "probative," but not dispositive, of


                               - 48 -
legislative intent.      R.I. Med. Soc'y v. Whitehouse, 

 (1st Cir. 2001) (citation omitted); see also Landrigan, 

.

            It is not difficult to discern the "purpose and intent"

behind RhodeWorks.       The legislature told us when it passed the

statute.     Specifically, the legislature found that there was

"insufficient revenue available from all existing sources to"

maintain Rhode Island's transportation infrastructure.                R.I. Gen.

Laws § 42-13.1-2(4) (2024).         It wanted to ameliorate this "funding

gap" by creating "recurring" revenue sources that would "fund

transportation infrastructure on a pay-as-you-go basis."                   Id.

§ 42-13.1-2(7).        The    legislature      also   included   an    express

severability provision, which states that if any part of RhodeWorks

is held unconstitutional, "all valid parts that are severable from

the . . . unconstitutional part [should] remain in effect."                Id.

§ 42-13.1-14.

            Given this language, it seems clear that severing the

RhodeWorks   caps     would   not   "destroy[]    legislative    purpose    and

intent."     Landrigan, 

.             Rather, invalidating

RhodeWorks based on nothing more than the unconstitutionality of

the caps would cut against the legislature's resolve to raise funds

for   its   bridges    and    its   stated    preference   for   --    wherever

possible -- only excising the statute's defective provisions.




                                     - 49 -
               ATA counters by pointing to legislative history.                        It

notes that the legislature -- at the request of then-Governor

Raimondo -- added the caps to assuage "vociferous local opposition

to   the    tolls."          Thus,    ATA    argues,    it    is   unlikely     that   the

legislature would have passed RhodeWorks without the caps, which

ensured that the local trucking industry fell in line behind the

bill.      We see several problems here.

               First,        ATA      points      to    no     evidence       reasonably

demonstrating that RhodeWorks would not have passed without the

caps.      At most, it shows that as between RhodeWorks without both

a small-truck exemption and the caps, and RhodeWorks with both the

exemption and the caps, the legislature preferred the latter.                          The

but-for       scenario       posed    by    our   severance    inquiry     is   markedly

different:       It asks whether the legislature would have foregone

RhodeWorks and its revenues altogether without the caps.                               ATA

points to nothing that would allow us confidently to discern an

answer to that question.                Put slightly differently, ATA has not

shown that the caps were "indispensable" to RhodeWorks' passage.

Id.;    cf.    All.     of    Auto.    Mfrs.,     

    ("[S]tatutory

interpretation cannot safely be made to rest upon inferences drawn

from intermediate legislative maneuvers.").

               Second, this case is unlike those in which Rhode Island

courts have refused to sever unconstitutional provisions.                              For

instance, ATA cites In re Advisory Opinion to the Governor, 856


                                            - 50 -
A.2d 320 (R.I. 2004).        But that case is clearly distinguishable.

There, the Rhode Island Supreme Court examined a statute governing

the establishment and operation of a casino in West Warwick, Rhode

Island. See id. at 323.          Among other things, that statute required

voter approval of the casino in a public referendum.                          Id.     The

court held this referendum unconstitutional.                     Id.     And because

"[t]he whole casino [was] dependent on voter approval [via] the

referendum," the rest of the statute had to fall as well.                        Id. at

333.   The referendum provision was the linchpin of the entire

statute   because   "[a]ll       the    provisions     of    the   [statute         were]

subsumed by the referendum question."                Id.    That is not the case

with RhodeWorks.     The statute's remaining provisions can function

perfectly well without the caps provision, meaning the caps are

not "indispensable to the rest of the statute."                    Landrigan, 

.

            The other major example here is Bouchard v. Price, 

 (R.I. 1997).       There, Rhode Island passed a statute saying

that when a felon tried to "commercial[ly] exploit[]" a crime

(e.g., by receiving royalties from a movie about the crime), the

money owed to the felon would instead flow into "a criminal

royalties   fund    from    which       victims   of       the   crime    may       claim

reimbursement for damages."            

.     The court held that this

structure    violated      the    First    Amendment        because      it    was    an

overinclusive restraint on free speech.                

.            Thus, the


                                       - 51 -
provision redirecting funds from "commercial exploitation" into a

"criminal royalties fund" was unconstitutional.     

.

The court held that this provision could not be severed because it

was "indispensable" to the act's purpose "of compensating victims

by utilizing the proceeds that a criminal has derived from the

criminal activity."   

 at 676–78.    Again, the caps provision in

RhodeWorks does not play the same kind of central role.

           We therefore conclude that although the RhodeWorks caps

are unconstitutional, they are severable from the rest of the

statute.   Thus, RhodeWorks may go into effect (absent the caps)

without offending the dormant Commerce Clause.

                                V.

           For the foregoing reasons, the judgment of the district

court is affirmed in part and reversed in part.     Each party shall

bear its own costs, and the case is remanded for the entry of

judgment in accord with this opinion.




                              - 52 -