Appellants, the State of Louisiana and the Concerned Shrimpers of Louisiana (“Concerned Shrimpers”), appealed the district court’s entry of summary judgment in favor of the Commerce Department, upholding in all respects the Secretary’s regulations requiring shrimp trawlers to install and use “turtle excluder devices,” also known as “TEDs,” in their nets or to limit their trawling to 90 minutes or less at a time. On July 11, 1988, six days after oral argument, we affirmed the judgment of the district court but postponed issuance of an opinion detailing the reasons for our ruling in order that we could immediately notify the parties that we were vacating the district court’s order staying execution of its judgment.1 We now offer our reasons for affirming summary judgment below.
*325I.
Five species of sea turtles — the Kemp’s ridley, loggerhead, leatherback, green, and hawksbill — frequent the Gulf of Mexico and the Atlantic Ocean, off the southeast coast of the United States. All of these species are listed as either “endangered” or “threatened” under the Endangered Species Act of 1973 (“ESA”), 16 U.S.C. § 1531 et seq.2 Upon inclusion of any species on the list as endangered, section 9 of the ESA prohibits any person from “taking”3 any such species within the United States, the territorial waters of the United States, or upon the high seas. 16 U.S.C. § 1538(a)(l)(B)-(C). In the case of sea turtles, it is equally forbidden to take threatened and endangered species. 50 C.F.R. §§ 222.21, 227.71. In addition to these prohibitions, the ESA permits the Secretaries of Commerce and the Interior to promulgate protective regulations. 16 U.S.C. § 1533(d).
On June 29, 1987, the Commerce Department, through its National Marine Fisheries Service (“NMFS”), promulgated final regulations requiring shrimp trawlers in the Gulf and South Atlantic to reduce the incidental catch and mortality of sea turtles in shrimp trawls. The regulations attempt to supplement ESA’s prohibitions against the “taking” of protected species, and were to become effective for Louisiana on March 1, 1988. Specifically, the regulations require shrimpers operating in offshore waters and in vessels 25 feet or longer to install and use certified “turtle excluder devices,” or “TEDs”, in each of their trawls. 50 C.F.R. § 227.72(e)(2), (6)(i) (1987). If the vessel is less than 25 feet or is trawling in inshore waters, the shrimper may limit each towing period to 90 minutes or less as an alternative to using a TED. Id. at § 227.72(e)(3), (6)(ii) (1987).
The reason for the regulations is simple: Researchers have found that during shrimping operations sea turtles are caught in the large nets, or trawls, pulled behind commercial shrimping vessels. The nets drag the turtles behind the boats and thereby prevent them from surfacing for air. According to one study, once a turtle is within the mouth of a shrimp trawl, the animal’s initial reaction is to attempt to outswim the device. Of course, this strenuous effort consumes oxygen but affords the turtle no opportunity to replenish the supply. Once trapped, if the exhausted turtle is not released quickly, it will drown. Research cited in the administrative record indicates that trawl times in excess of 90 minutes are highly likely to result in the death of a captured turtle.4
The TED requirement thus applies without exception to large shrimping vessels that operate offshore, as these vessels frequently pull their nets for long hours prior to bringing their catches aboard. All of the presently certified TEDs5 are coated mesh, rope, or rigid frame devices inserted *326into the cone-shaped shrimp nets at an angle, at the point where the trawl begins to narrow. When a captured turtle reaches the TED, the device deflects the turtle to an escape portal in the top or bottom of the net. The alternative measure of restricting tow-time to 90-minute intervals applies both to smaller vessels, which tend to pull fewer nets at a time and for shorter durations, and to inshore areas, where TED use by either large or small vessels may be impaired by often heavier concentrations of underwater debris.
In October 1987, the State of Louisiana filed a complaint in federal district court, contending that both the TED and tow limit requirements are invalid. The State’s complaint alleged that the regulations are arbitrary and capricious, unsupported by the record, and were promulgated in violation of the Administrative Procedure Act’s procedural requirements and Executive Order 12291’s requirement of a regulatory impact analysis. The State further argued that the regulations violate the Louisiana shrimpers’ due process and equal protection rights.
In December 1987 the Environmental Defense Fund and the Center for Environmental Education were permitted to intervene as defendants. The parties filed cross motions for summary judgment and orally argued their motions on February 10, 1988. Two days before oral argument, on February 8, Concerned Shrimpers of Louisiana was granted leave to intervene as plaintiff. Concerned Shrimpers neither filed, nor responded to, any summary judgment motion, nor participated in oral argument.
On February 29, 1988, the district court entered summary judgment for the defendants. Louisiana ex rel. Guste v. Verity, 681 F.Supp. 1178 (E.D.La.), aff'd, 850 F.2d 211 (5th Cir.1988). On April 12, 1988, a stay pending appeal was issued. In the present appeal, the appellants again argue that the regulations are, arbitrary and capricious.6 In addition, the Concerned Shrimpers’ appeal carries forward the equal protection challenge raised below.
II.
A.
The district court entered judgment for the Secretary on cross-motions for summary judgment. When reviewing a grant of summary judgment, a court of appeals applies the same legal standard as that which guided the district court. Barbetta v. S/S BERMUDA STAR, 848 F.2d 1364, 1368 (5th Cir.1988); Brooks, Tarlton, Gilbert, Douglas & Kressler v. United States Fire Ins. Co., 832 F.2d 1358, 1364 (5th Cir.1987). Review in the first instance of a challenge made to regulations promulgated under the Endangered Species Act, 16 U.S.C. § 1531 et seq., is governed by standards set forth in the Administrative Procedure Act (“APA”), 5 U.S.C. § 706, and it is thus to these same standards that we will look in conducting our review on appeal. Environmental Coalition of Broward County, Inc. v. Meyers, 831 F.2d 984, 987 (11th Cir.1987). Under the APA, the administrative record is reviewed to determine whether the challenged action was arbitrary and capricious, an abuse of discretion, or otherwise not in accordance with law; contrary to constitutional right; in excess of statutory jurisdiction, authority or limitation; or without procedure required by law. 5 U.S.C. § 706(2)(A)-(D).7
*327Appellants primarily charge that the regulations should be invalidated as arbitrary and capricious. Our scope of review under this standard, however, is very narrow.8 The court is not to weigh the evidence in the record pro and con. C.A. White Trucking Co. v. United States, 555 F.2d 1260, 1264 (5th Cir.1977). Rather, our role is to review the agency action to determine whether the decision “was based on a consideration of the relevant factors and whether there was a clear error of judgment.” Motor Vehicle Mfrs. Ass’n, 463 U.S. at 43, 103 S.Ct. at 2866-67; Nueces County Nav. Dist. No. 1 v. ICC, 674 F.2d 1055, 1062-63 (5th Cir.), cert. denied, 459 U.S. 1035, 103 S.Ct. 446, 74 L.Ed.2d 601 (1982).
Thus, if the agency considers the factors and articulates a rational relationship between the facts found and the choice made, its decision is not arbitrary or capricious. Watkins Motor Lines, Inc. v. ICC, 641 F.2d 1183, 1188 (5th Cir. Unit B Apr. 1981). Indeed, the agency’s decision need not be ideal, so long as it is not arbitrary or capricious, and so long as the agency gave at least minimal consideration to relevant facts contained in the record.9 With these principles of review in mind, we turn now to the challenges posed by the appellants.
B.
The core of appellants’ challenge on appeal concerns the sufficiency of the administrative record to support the TED and trawling-period regulations. In particular, they assert that the record insufficiently demonstrates the impact of shrimp trawling on sea turtle mortality, the efficacy of the regulations as applied to inshore Louisiana waters, and the impact of the regulations on the Louisiana economy. Appellants also challenge the regulations insofar as the administrative record supporting them fails to address serious causes of sea turtle mortality other than shrimping. Based upon the limited scope of our review, we find that the record amply supports the Secretary’s decision to issue the regulations in question.
1. The Impact of Shrimp Trawling on Sea Turtle Mortality.
The relationship of shrimping to sea turtle mortality is strongly demonstrated by data contained in the administrative record. Since 1973, on-board observers have documented the capture and drowning of sea turtles by shrimp trawlers. Using extrapolations based upon more than 27,-000 hours of shrimp trawl observation, experts have concluded that more than 47,000 endangered and threatened sea turtles are caught in shrimp trawls each year; 11,179 of these turtles drown in the shrimpers’ nets.10 Tag returns on the Kemp’s ridley *328also provide a fertile source of information: 84% of the Kemp’s ridley turtles tagged by scientists and later recovered were captured by shrimp trawlers.
The capture and mortality statistics for Louisiana waters were derived largely from the so-called Henwood-Stuntz study, a series of extrapolations based upon 16,785 hours of observer effort in the Gulf of Mexico. Of this total, 4,333 hours were spent on shrimp boats off the Louisiana shore. During the Louisiana observation period, 12 sea turtles were taken, 5 of which had died by the time the trawl was retrieved. This mortality rate of 42% is among the highest of any state, the Gulf-wide rate being 29%. More than one-third of the turtles that were observed to have died in Gulf shrimp trawls, died off Louisiana.
Although the observers spent substantial hours on the shrimp boats, their efforts recorded the results of only a small fraction of the annual shrimping effort. Each year, commercial shrimpers are estimated to spend 2,063,074 hours trawling off Louisiana. Using a simple ratio of 5/4,333 = X/2,063,074 and solving for X, a total of 2,381 endangered and threatened turtles would be estimated to be killed annually off Louisiana alone.11
“Stranding” data further supports the conclusion that shrimping is responsible for large numbers of sea turtle deaths. Beginning in 1980, volunteers established the Sea Turtle Stranding and Salvage Network (“Network”) to monitor the number and types of sea turtle carcasses stranded on beaches and in marshes and bayous. More than 8,300 dead sea turtles, including nearly 600 Kemp’s ridleys, were reported to NMFS by the Network. Although determining the precise cause of a stranded sea turtle’s death is difficult, a causal link to shrimping appears reasonable in light of the fact that strandings occur predominantly in areas adjacent to shrimping grounds, and that the number of sea turtle strand-ings increases dramatically with the advent of the shrimping season.
In addition, the administrative record established that, based upon tag returns between 1966 and 1984, 32% of the Kemp’s ridley turtles incidentally captured are caught in Louisiana waters, by far the highest rate of any state or country. Twenty-two percent of the Kemp’s ridley strandings in the Gulf occur in Louisiana. In a 1984 study, 12 out of 15 Louisiana shrimpers interviewed said they caught from 1 to 2 sea turtles each year.12
In challenging the administrative fact-finding that links shrimp trawling to sea turtle mortality, appellants assert that the Secretary failed to consider the best scientific data available before issuing the regulations. The Henwood-Stuntz extrapolations heavily relied upon by the agency are flawed, appellants contend, because one of the field samples on which they are based is unscientifically small. Specifically, appellants point out that researchers conducting the study recorded a mere two captur-ings of the Kemp’s ridley turtles during the entire time test trawls were pulled off Louisiana’s shores. The insufficiency of this sample is borne out, they believe, by the discrepancy between the Henwood-Stuntz extrapolations and observations made by the Louisiana Department of Wildlife and Fisheries. From 1967 until 1986, the Louisiana Department conducted a total of 36,-837 trawl samples, but in none of these was a single sea turtle ever reported to have been captured. At the very least, appellants conclude, the methodology of Hen-wood-Stuntz is prone to grossly over-estimating the killing of sea turtles in shrimp trawls.
*329Although we believe appellants’ challenge is not totally without merit, we are mindful that under the arbitrary-and-capricious standard, our deference to the agency is greatest when reviewing technical matters within its area of expertise, particularly its choice of scientific data and statistical methodology.13 In reviewing such technical choices, “[w]e must look at the decision not as the chemist, biologist or statistician that we are qualified neither by training nor experience to be, but as a reviewing court exercising our narrowly defined duty of holding agencies to certain minimal standards of rationality.” Avoyelles Sportsmen’s League, Inc. v. Marsh, 715 F.2d 897, 905 (5th Cir.1983) (quoting Ethyl Corp. v. EPA, 541 F.2d 1, 36 (D.C.Cir.) (en banc), cert. denied, 426 U.S. 941, 96 S.Ct. 2662, 49 L.Ed.2d 394 (1976)). Accordingly, where, as here, the agency presents scientifically respectable conclusions which appellants are able to dispute with rival evidence of presumably equal dignity, we will not displace the administrative choice.14 Nor will we remand the matter to the agency in order that the discrepant conclusions be reconciled.
From our admittedly lay perspective, the Henwood-Stuntz method of extrapolating the magnitude of sea turtle takings in shrimp trawls does not necessarily appear unreasonable. There are more than 18,000 domestic shrimp vessels operating in the Gulf and South Atlantic. Each of these vessels simultaneously pulls from 1 to 4 trawls, generally for 2 to 6 hours at a time. Shrimping occurs in the Gulf year-round, with most activity concentrated between June and December. Therefore, while the 16,785 hours of observer effort invested in the Henwood-Stuntz study represents the equivalent of less than one hour of fishing by the entire shrimping fleet, we recognize that the size of the industry realistically precludes statistical findings based totally upon actual observation rather than extrapolation.
We are also unpersuaded by appellants’ mischaracterization of the data base from which Henwood-Stuntz extrapolations were made. Appellants’ attack is accomplished by isolating a narrow range of data concerning the most endangered of the five protected species — the Kemp’s ridley — and concluding that the data concerning this single species in waters off a single state was insufficient to justify the regulations. As already indicated, however, the regulations are intended to prevent the illegal taking of all of the five endangered and threatened sea turtle species, whose habitats are not confined by state boundaries but may encompass thousands of square miles of sea. Because each local area, indeed each shrimper, is responsible for catching only a few sea turtles each year, it is only by aggregating this information that the relevant statistics can be approximated.
2. The Efficacy of the Regulations as Applied to Louisiana’s Inland Waters.
The administrative record amply demonstrates that sea turtles are found in inshore waters. Kemp’s ridleys, for example, are known to frequent the inshore waters of the Gulf, which hosts their favorite food species, the blue crab. Kemp’s ridleys are most abundant in Louisiana, particularly in the near-shore white shrimp grounds. At least one expert has observed that “Kemp’s ridley could logically be labelled the Louisiana turtle, because its greatest abundance is found there.... It is beyond doubt the commonest marine turtle in the state, concentrated in the shallow water from Marsh Island to the Mississippi Delta.” Of all tag returns of nesting Kemp’s ridley females, 59% are from Louisiana near-shore waters. Other species, including loggerheads and *330green turtles, also frequent inshore Louisiana waters. The Kemp’s ridley is also found in nearby Texas coastal waters, as are green turtles and loggerheads.
Sea turtles not only frequent inshore waters; the record is replete with evidence to show that they are captured there as well. Most recoveries of “headstarted” Kemp’s ridleys have occurred in inshore waters.15 Fully 12% of these recoveries have been off Louisiana, even though most of the releases have been off Texas. Nearly all of the Louisiana shrimpers who reported catching sea turtles (80% of those interviewed) caught them in shallow near-shore waters. In testimony before the National Oceanic and Atmospheric Administration, a Georgia marine fisheries expert reported that seven of the shrimpers interviewed had captured between 20 and 100 sea turtles each, in a single shrimping season, mostly in inshore waters.
By 1976, as evidence mounted that shrimp trawling is a substantial cause of sea turtle mortality, NMFS began research into equipment modifications that would allow turtles to escape shrimp trawls without causing a significant loss of shrimp.16 To ensure that the new gear would accomplish the program’s objectives, NMFS required its proposed TED design to be at least 97% effective at releasing sea turtles. After more than 15,000 hours of testing in the Atlantic and Gulf of Mexico, the agency concluded that the NMFS TED met this standard, without resulting in significant shrimp loss.17 52 Fed.Reg. 24245 (1987). Tests also showed that the new gear created certain savings for shrimpers. For example, the TED not only deflected turtles out of the net, but also other “by-catch” — marine animals that shrimpers normally must sort through and discard after retrieving the trawl from the water. And the reduced “by-catch” appeared to result in fuel efficiency as well, by reducing the unproductive (i.e., non-shrimp) drag on the vessel’s engine.
Data concerning TED effectiveness was based primarily upon tests done in offshore waters, so although the agency had sufficient data that turtles frequent inshore waters and that shrimpers catch them there, the agency was receptive to time limitations on trawling as an alternative measure for inshore areas. During hearings on the proposed regulations, commenters urged *331that TEDs will not be effective inshore because the TEDs will clog with debris that reportedly lines the bottom of these waters, particularly the inshore Gulf waters. Commenters also urged that TEDs were unnecessary in inshore waters, because the debris makes it impossible for shrimpers to trawl for more than 60 to 90 minutes at a time, less time than it generally takes to drown a captured turtle. As Louisiana’s Senator Breaux put it, “in all probability, shrimping in [inshore] waters will not result in any significant sea turtle mortality because it is practically impossible to tow a net for a sufficient length of time to drown a turtle.” Based upon these assurances, the agency determined that tow-time restrictions offer a reasonable alternative to TED use in inshore waters.
Appellants nevertheless argue that, if the agency had insufficient data to apply the more intrusive TED requirement to inshore shrimping, it automatically had insufficient data to support any other type of regulation. We disagree. Since the data was more than adequate to support the tow-time restriction in inland waters, it follows that the Secretary’s decision to give shrimpers the option to use TEDs as an alternative is no less supportable. Inshore shrimpers who find the TED onerous need not use it. On the other hand, should shrimpers who experiment with the TED in inshore waters find the device compatible with their needs, they will be free to use it instead of having to haul aboard their nets after every ninety minutes of fishing. The shrimpers can hardly complain that the agency has given them a choice.
3. The Impact of the Regulations on Louisiana’s Economy.
The proposed regulations, which were to be phased in over a two-year period, will require 17,200 shrimpers using certain size nets to install TEDs and use them when fishing in offshore waters during the shrimping seasons. Shrimpers will purchase and install certified TEDs at an expected cost of $200-400 per TED.18 The average annual cost to the entire industry was estimated at $5.9 million, which included the cost of expected shrimp loss during the start-up period, before gear adjustments and changes in trawling techniques overcome any initial inefficiencies. There is substantial evidence in the administrative record indicating that anticipated catch loss resulting from use of the TEDs will amount to no more than 5%.
Although we do not denigrate appellants’ concern with the expense and inconvenience the regulations will visit on Louisiana’s shrimping industry, Congress has decided that these losses cannot compare to the “incalculable” value of genetic heritage embodied in any protected living species.19 While we do not mean to imply that economic impact can never be considered in determining whether a particular regulation promulgated under the ESA is arbitrary or capricious, the protections afforded by the regulations before us have not been shown to be achievable through less costly means. Thus, the costs shouldered by the industry are not arbitrary, but reasonably related to Congress’s purpose.20
*3324. The Secretary’s Failure To Regulate Other Major Causes of Sea Turtle Mortality.
Appellants argue that the TED regulations are arbitrary and capricious because they do not address other serious causes of sea turtle mortality. As we understand this argument, however, appellants are in fact raising two separate points. First, they appear to be urging us to adopt a novel proposition that regulations failing to address all of the causes of a problem are, for that reason, arbitrary and capricious. In doing so, they ignore the well-established rule that regulations need not remedy all evils, or none. See Railway Express Agency, Inc. v. New York, 336 U.S. 106, 110, 69 S.Ct. 463, 465-66, 93 L.Ed. 533 (1949); Central Lumber Co. v. South Dakota, 226 U.S. 157, 160, 33 S.Ct. 66, 67, 57 L.Ed. 164 (1912). Nor must the government “choose between attacking every aspect of a problem or not attacking the problem at all.” Dandridge v. Williams, 397 U.S. 471, 486-87, 90 S.Ct. 1153, 1162, 22 L.Ed.2d 491 (1970).21 Thus, the agency’s decision to attack one of the major causes of sea turtle mortality through regulation is entirely within its discretion. That dredges, commercial fishermen from other nations, and pollution also contribute to sea turtle deaths does not undermine the validity of these restrictions.
Appellants’ second contention is based upon the proposition enunciated in Connor v. Andrus, 453 F.Supp. 1037, 1041 (W.D. Tex.1978), that regulations issued under the ESA must halt, or even reverse, the population depletion of an endangered species. In Connor, the administrative record was found not to support a finding that banning all duck hunting in designated portions of New Mexico, Texas, and Arizona would increase, or tend to increase, the population of the endangered Mexican duck. The record did demonstrate, however, that the Mexican duck was threatened by other causes not targeted by the regulation, namely, the destruction of its natural habitat and its hybridization with another species, the mallard.
Having thus found that the agency had failed to consider all of the statutorily relevant factors, the Connor court held the challenged regulation to be arbitrary and capricious. Id. Accordingly, appellants believe the administrative record in the instant case is similarly deficient: There is no finding that the regulations will ultimately save sea turtles from extinction; nor does the record show that sea turtles saved from drowning in shrimpers’ nets will ultimately survive the other causes of sea turtle mortality for such time as to replenish or increase their numbers.
An essential part of appellants’ argument assumes that the ESA authorizes the Secretary to issue protective regulations only if found actually to save an endangered species from extinction. We believe, however, that this assumption finds no support in the statutory grant of regulatory authority, 16 U.S.C. § 1533(d).22 To be *333sure, that statute mandates the Secretary to “issue such regulations as he deems necessary and advisable to provide for the conservation of [threatened] species.” Id. “Conservation” is elsewhere defined in the ESA as “all methods and procedures which are necessary to bring endangered species or threatened species to the point at which the measures provided pursuant to this chapter are no longer necessary.” 16 U.S.C. § 1532(3).
In addition to this mandatory duty, however, the ESA also provides the Secretary discretionary authority to prohibit by regulation the taking of any threatened species of fish and wildlife. 16 U.S.C. §§ 1533(d), 1538(a)(1). This regulatory authority supplements the statutory prohibition against the taking of endangered species, see 16 U.S.C. § 1538(a)(1), the enforcement of which is not conditioned upon any showing that the prohibition will itself operate to restore the species to a level considered unendangered. Rather, Congress simply presumes that prohibited takings will deplete the species. We must honor that legislative determination.
In sum, therefore, regulations aimed at preventing the taking of a protected species cannot be invalidated on the ground that the record fails to demonstrate that the regulatory effort will enhance the species’ chance of survival. Insofar as Connor v. Andrus, supra, requires such a showing, we disapprove its holding.23 Rather, the record need only show that such regulations do in fact prevent prohibited takings of protected species. Here, the record developed by the Secretary amply satisfies this burden.
III.
The Concerned Shrimpers also challenge the regulations on equal protection grounds. The basic concepts of equal protection as set forth in the fourteenth amendment apply to federal action though the due process clause of the fifth amendment. United States v. Hawes, 529 F.2d 472, 477 (5th Cir.1976). Where administrative classifications are made, absent a fundamental right or suspect classification such as race, equal protection requires only that the classification bear a rational relationship to the legislative purpose of the enabling statute. See Turner v. Weinberger, 728 F.2d 751, 757-58 (5th Cir.1984).
Concerned Shrimpers argue that the regulations cannot meet the “rational basis” test because (1) they do not constrain activity north of North Carolina, even though sea turtles are believed to frequent those waters; (2) they regulate Gulf shrimpers with the same intensity as they regulate Atlantic shrimpers, even though Atlantic shrimpers catch more sea turtles per capita; and (3) they regulate based upon boat size, rather than net size. With respect to the first of these challenges, the Shrimpers themselves provide the best answer: “It is no requirement of equal protection that all evils of the same genus be eradicated or none at all.” See Railway Express, supra. Nor, as we have already noted, must the agency “choose between attacking every aspect of a problem or not attacking the problem at all.” Dandridge, supra. Even assuming, as the Shrimpers assert, that Kemp’s ridleys do turn up off Long Island, there is no evidence that they are caught there by shrimp trawlers. Moreover, the government’s regulation of a known evil at one place does not become invalid by the discovery of a similar, as-yet-unregulated evil at another.
The Concerned Shrimpers’ next argument — that the regulations violate equal protection because they regulate Gulf and Atlantic shrimpers in the same way — is similarly without merit. The agency is under no obligation to treat Atlantic and Gulf shrimpers differently, so long as its rule is reasonably related to its purpose. Here, both Gulf and Atlantic shrimpers apparently cause large numbers of sea turtle deaths. It is the agency’s purpose to pro*334tect these turtles from harm; it has no interest in penalizing shrimpers for the number of turtles caught, only to prevent such captures, to the extent possible, in the future. The regulations create an evenhanded regulatory scheme, rationally related to their legitimate purpose. This is all that equal protection requires.
Finally, the Shrimpers challenge the use of boat size as the trigger for the TED requirement, because, they urge, it is the size of the net, not the boat, that is critical. As originally proposed, the regulations indeed used net size as the basis for the TED requirement, but public comments persuaded the agency that this criterion could not work: It was not only difficult to enforce, but would cause significant intrusions into shrimper operations.24
The compromise reached by the agency— to use boat size as a proxy for net size— was entirely reasonable. Boats below 25 feet in length generally do not pull the larger nets. But even if some 24-foot vessels do pull larger nets, as the Shrimpers say, the government need not achieve the “mathematical nicety” the Shrimpers demand in order to meet constitutional requirements. Lindsley v. Natural Carbonic Gas Co., 220 U.S. 61, 78, 31 S.Ct. 337, 340, 55 L.Ed. 369 (1911). Rather, as the Supreme Court has long recognized, “[t]he problems of government are practical ones and may justify, if they do not require rough accommodations.” Metropolis Theatre Co. v. City of Chicago, 228 U.S. 61, 69, 33 S.Ct. 441, 443, 57 L.Ed. 730 (1913).25 The “rough accommodation” attained by these regulations plainly passes constitutional muster.
IV.
For the foregoing reasons, the judgment of the district court is AFFIRMED.
. See Louisiana ex rel. Guste v. Verity, 850 F.2d 211 (5th Cir.1988). As stated in that order, the stay will remain in effect until 11:59 p.m., August 31, 1988. We also informed appellants that if they wish to apply to the Supreme Court for a further stay of judgment or for other relief, such application should he made during the penden-cy of the stay ordered by the district court, as we will grant no further stay.
. An “endangered” species is one that is in danger of extinction throughout all or a significant part of its range. 16 U.S.C. § 1532(6). A “threatened” species is one likely to become endangered within the foreseeable future throughout all or a significant part of its range. 16 U.S.C. § 1532(20).
Although the Kemp’s ridley is the most critically endangered sea turtle, the four other species are in serious danger of extinction as well. By 1978, the Secretary of Commerce had listed the Kemp’s ridley, leatherback, and hawskbill turtles as endangered species, and the loggerhead and green turtles as threatened. 50 C.F.R. § 222.23 (1987) (list of endangered sea turtles, effective 1974); 50 C.F.R. § 227.4 (1987) (list of threatened sea turtles, effective 1978). Under the Secretary’s authority to list endangered or threatened species by region, the green turtle was also listed as endangered in Florida waters. 50 C.F.R. § 222.23 (1987). The Pacific, or Olive ridley, not at issue in this case, is listed as threatened, except for the population off the Pacific coast of Mexico, which is listed as endangered. Id.
. To "take” is to “harass, harm, pursue, hunt, shoot, wound, kill, trap, capture, collect, or to attempt to engage in any such conduct.” 16 U.S.C. § 1532(19).
. In tests involving trawls lasting from 30 to 270 minutes, researchers found that 1.3% of the captured sea turtles died in the 30-minute trawls, 4.4% died in 90-minute trawls, and 26.4% died in 270-minute trawls. A maximum tow time of 90 minutes, therefore, results in a 95% reduction in sea turtle mortality.
. TED designs currently certified by the NMFS are diagrammatically described at 50 C.F.R. § 227.72(e)(4) (1987).
. The State ends its brief with an argument completely new to this litigation: that the TED regulations constitute a de facto designation of critical habitat subject to, but not in compliance with, certain procedural requirements of 16 U.S.C. § 1533(b)(2). The argument appears nowhere in the State’s (or the Concerned Shrimpers’) complaint, summary judgment motion or memorandum, its opposition to the Secretary’s summary judgment motion, or any of the pleadings filed in support of the stay entered by the district court. Since we discern no manifest injustice in declining to consider this argument so raised, we offer no opinion as to its merits. See Vaughner v. Pulito, 804 F.2d 873, 877 n. 2 (5th Cir.1986); Frank C. Bailey Enterprises, Inc. v. Cargill, Inc., 582 F.2d 333, 334 (5th Cir.1978) (appellate court, in reviewing summary judgment order, can consider only those matters presented to the district court).
. These regulations were promulgated as an informal rulemaking under the Endangered Species Act, which does not prescribe compliance with the formal procedures required by 5 U.S.C. § 556. Consequently, substantial evidence re*327view under 5 U.S.C. § 706(2)(E) is not appropriate. F.C.C. v. National Citizens Comm. for Broadcasting, 436 U.S. 775, 803-04, 98 S.Ct. 2096, 2116-17, 56 L.Ed.2d 697 (1978).
.Concerned Shrimpers urge this court to undertake de novo review of the agency decision, including in our review consideration of new, extra-record materials never submitted to the agency or, for that matter, to the district court. We note, however, that both the Concerned Shrimpers' proposed standard of review and their attempt to reopen the administrative record in this court are improper. Under well-established principles of administrative law, de novo review is not the standard. The courts are not empowered to substitute their judgment for that of the agency. Motor Vehicles Mfrs. Ass’n of the United States, Inc. v. State Farm Mut. Auto Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 2866-67, 77 L.Ed.2d 443 (1983). Nor are the courts permitted to consider evidence outside the administrative record. Camp v. Pitts, 411 U.S. 138, 142-43, 93 S.Ct. 1241, 36 L.Ed.2d 106 (1973); Avoyelles Sportsmen’s League, Inc. v. Marsh, 715 F.2d 897, 904-05 (5th Cir.1983). Agency action is to be upheld, if at all, on the basis of the record before the agency at the time it made its decision. Motor Vehicle Mfrs. Ass’n, 463 U.S. at 51, 103 S.Ct. at 2870-71.
. City of Houston v. FAA, 679 F.2d 1184, 1190 (5th Cir.1982); American Petroleum Institute v. EPA, 661 F.2d 340, 349 (5th Cir. Unit A Nov. 1981). See Hondros v. United States Civil Service Comm 'n, 720 F.2d 278, 295 (3d Cir.1983) (action will not be deemed arbitrary, capricious, or an abuse of discretion simply because one may happen to think it ill-considered or to represent a less appealing alternative).
. The observer data further established that 90% of the captured turtles are loggerheads, 6% are the critically-endangered Kemp’s ridley, 3% are green turtles, and 1% are leatherbacks and hawksbills.
. Actually, the Henwood-Stuntz study used a more sophisticated and conservative analysis than this. The study nevertheless concluded that 6,396 sea turtles would be caught off Louisiana and that 1,407 of these would be drowned.
. This result is all the more striking because interview data tends to under-report sea turtle captures, as a result of the shrimpers' understandable reluctance to admit catching protected species. One study reported that, according to interview data, the turtle mortality rate was 7.9%, while on-board observer data showed a 15% mortality rate.
. See Baltimore Gas & Elec. Council, Inc. v. National Resources Defense Council, Inc., 462 U.S. 87, 103, 103 S.Ct. 2246, 2255, 76 L.Ed.2d 437 (1983); National Ass'n of Metal Finishers v. EPA, 719 F.2d 624, 657 (3d Cir.1983), rev’d on other grounds, 470 U.S. 116, 105 S.Ct. 1102, 84 L.Ed.2d 90 (1985).
. See Suntex Dairy v. Block, 666 F.2d 158, 162 (5th Cir.), cert. denied, 459 U.S. 826, 103 S.Ct. 59, 74 L.Ed.2d 62 (1982); cf. Chrysler Corp. v. EPA, 631 F.2d 865, 890 (D.C.Cir.), cert. denied, 449 U.S. 1021, 101 S.Ct. 589, 66 L.Ed.2d 483 (1980) (applying substantial evidence test).
. The “headstarting” program involves nurturing turtles in a protected environment for one year after they are hatched. At one year of age, the young turtles are released into the sea.
. NMFS explored several alternatives to accomplish effective sea turtle conservation and restoration, some of which have been implemented with varying degrees of success. An extensive turtle headstarting program for the Kemp's ridley has released more than 10,000 young turtles into the sea. The success of this program is highly uncertain, however, so long as the headstarted turtles fail to return to the nesting beach. Thus far, none have returned. In addition, NMFS regulations now require anyone who incidentally takes a protected turtle during fishing operations to attempt to resuscitate the turtle if it is comatose or inactive. See 50 C.F.R. § 227.72(c)(6)(iv), (e)(l)(i)(B) (1987). According to the regulations, attempted resuscitation is accomplished by ‘‘[p]lacing the turtle on its back (carapace) and pumping its breastplate (plastron) with hand or foot, or ... [by] placing the turtle on its breast-plate (plastron) and elevating its hindquarter several inches for a period of one up to twenty-four hours." Id. at § 227.72(e)(l)(i)(B). Turtles that revive are to be released over the stern of the boat only when trawls are not in use, when the engine gears are in neutral position, and in areas where they are unlikely to be recaptured or injured by vessels. Similarly, sea turtles that fail to move within several hours (up to twenty-four if possible) must be returned to the water in the same manner. Id.
.NMFS conducted more than 15,000 hours of testing under a variety of conditions. The testing was done during the day and at night, on board chartered vessels operated by project personnel and on cooperative shrimp vessels conducting normal commercial fishing operations. The gear was tested extensively in the Cape Canaveral Channel, Florida, area, where there are dense concentrations of sea turtles. It was also .tested on commercial shrimp grounds throughout the United States South Atlantic and the Gulf of Mexico. Testing in inshore waters, however, was limited.
NMFS TEDs were tested by paired tows, in which one side of the vessel used a TED and the other did not. After 15,000 hours of testing, the TED was found effective at releasing turtles, without reducing shrimp catch. During the testing, NMFS was able to reduce the TED from a 97-pound rigid frame to a collapsible 37-pound frame.
. Some certified TEDs cost less than $75. In addition, many shrimpers have successfully constructed and installed qualified, homemade TEDs, at a much lower cost. For example, shrimpers report constructing the certified Morrison Soft TED for as little as $25 per TED.
. See TVA v. Hill, 437 U.S. 153, 179, 98 S.Ct. 2279, 2294, 57 L.Ed.2d 117 (1978) (quoting H.R. Rep. No. 93-412, 93d Cong., 1st Sess. 4-5 (1973), U.S.Code Cong. & Admin.News 1973, p. 2989). In Hill, suit was brought against the TVA to enjoin completion of the $100 million Tellico dam, situated on the Little Tennessee River, because the anticipated inundation caused by impoundment of the reservoir would have jeopardized the continued existence of the snail darter, an endangered species under the ESA. Impressed by the fact that none of the "hardship exemptions” provided in § 10 of the Act, 16 U.S.C. § 1539, would even remotely apply to the Tellico project, the Court concluded that it was not empowered to create additional exemptions by judicial fiat. 437 U.S. at 189, 98 S.Ct. at 2299.
.In this regard, we must be ever cognizant that "[w]e are judges, not legislators." Boyle v. United Technologies Corp., — U.S. -, -, 108 S.Ct. 2510, 2528, 101 L.Ed.2d 442 (1988) (Brennan, J., dissenting). Congress has made the policy determination that endangered species are to be protected, and an administrative agency has decided that sea turtles are to be *332protected despite substantial economic consequences for an important industry. We express no opinion on those choices. If the trade-off between marine life and economic success has been skewed in the wrong direction, it is for the legislative and executive branches, not the courts, to correct that imbalance.
. See also Sender v. Oregon State Bd. of Dental Examiners, 294 U.S. 608, 610, 55 S.Ct. 570, 571, 79 L.Ed. 1086 (1935) ("the [government is] ... not bound to deal alike with all ... classes, or to strike at all evils at the same time or in the same way.")
. 16 U.S.C. § 1533(d) provides in relevant part as follows:
Whenever any species is listed as a threatened species pursuant to subsection (c) of this section, the Secretary shall issue such regulations as he deems necessary and advisable to provide for the conservation of such species. The Secretary may by regulation prohibit with respect to any threatened species any act prohibited under section 1538(a)(1) of this title, in the case of fish or wildlife....
16 U.S.C. § 1538(a)(1) provides in relevant part as follows:
Except as provided in sections 1535(g)(2) and 1539 of this title, with respect to any endangered species of fish or wildlife listed pursuant to section 1533 of this title it is unlawful for any person subject to the jurisdiction of the United States to—
(A) ...;
(B) take any such species within the United States or the territorial sea of the United States;
*333(C) take any such species upon the high seas; ....
. We also note that in Connor, the court recognized the Secretary’s duty under the Act to increase the population of any endangered species. See 453 F.Supp. at 1041.
. Patrols would have to inspect the nets themselves, by itself a cumbersome task, and would have to order trawlers to lift their nets out of the water for these inspections. Boat length, by contrast, is more readily discernable. 50 C.F.R. § 217.12 (1987).
. See also Mathews v. De Castro, 429 U.S. 181, 185, 97 S.Ct. 431, 434, 50 L.Ed.2d 389 (1976) (imperfect classifications are constitutional unless “clearly wrong, a display of arbitrary power, not an exercise of judgment”).