PER CURIAM:

After the deaths of some whooping cranes — an endangered species — The Aransas Project ("TAP") sued directors of the Texas Commission on Environmental Quality ("TCEQ") under the Endangered Species Act ("ESA" or the "Act"). TAP sought and was granted an injunction prohibiting TCEQ from issuing new permits to withdraw water from rivers that feed the estuary where the cranes make their winter home. The injunction also required TCEQ to seek an incidental-take permit ("ITP") from the U.S. Fish and Wildlife Service ("FWS"). A motions panel of this court stayed the injunction pending appeal. We conclude that the district court's opinion misapplies proximate cause analysis and further, even if proximate cause had been proven, the injunction is an abuse of discretion. The judgment is reversed.

I.

The whooping crane is a majestic bird that stands five feet tall and has a wingspan of more than eight feet. It once came close to extinction and, despite international recovery efforts, is still endangered. The world's only wild flock, called the Aransas-Wood Buffalo ("AWB") flock, consists of almost 300 birds and inhabits the Aransas National Wildlife Refuge ("the Refuge") in Texas during the winter and Wood Buffalo National Park in Canada in the summer. Adjacent to the Refuge is San Antonio Bay, also known as the Guadalupe Estuary, which provides a critical habitat for the flock and receives freshwater inflows primarily from the San Antonio and Guadalupe Rivers. The State of Texas owns the state's surface water, including the water in the San Antonio and Guadalupe River systems, and holds it in trust for the citizens of the state. Under Texas law, surface-water capture and use is regulated by TCEQ, a state agency that, through permitting processes and regulatory powers, can affect the availability of fresh water to users throughout the state.

According to reports issued by the Refuge's biologist, the AWB flock consisted of about 270 whooping cranes in 2008. During a severe drought in the winter of 2008-2009, four crane carcasses were recovered in the Refuge. Necropsies were performed on two of them, and in both instances, emaciation was listed among other factors as a cause of death. Using aerial surveys, the biologist concluded that nineteen other cranes died during that season. Thus, by the end of the 2008-2009 winter, the flock had purportedly declined to 247 cranes.

When reports of those crane mortalities became known, various environmentalists, local coastal business owners, bird enthusiasts, and others formed TAP, a non-profit corporation whose objective is to protect the habitat of the whooping crane; its members have expressed direct interests in the continued vitality of the AWB flock and the Refuge, ranging from personal enjoyment of the birds to various business interests. TAP sued on behalf of itself and its members, alleging that various TCEQ officials (the "state defendants") had violated the ESA, 16 U.S.C. § 1531 et seq. The crux of TAP's complaint was that the state defendants' actions and failures to act in managing water diversion in the San Antonio and Guadalupe River systems violated the ESA by harming and harassing cranes in the flock and causing the deaths of twenty-three cranes.

A.

The ESA applies to all "persons," including "any officer, employee, [or] agent,... of any State." 16 U.S.C. § 1532(13). The Act forbids "takes" of endangered species such as the whooping crane. Id. § 1538(a)(1)(B). "The term `take' means to harass, harm, ... wound, [or] kill" protected species. Id. § 1532(19). "Harm" includes "significant habitat modification or degradation where it actually kills or injures wildlife by significantly impairing essential behavioral patterns, including breeding, feeding or sheltering." 50 C.F.R. § 17.3(c). "Harass ... means an intentional or negligent act or omission which creates the likelihood of injury to wildlife by annoying it to such an extent as to significantly disrupt normal behavioral patterns which include, but are not limited to, breeding, feeding, or sheltering." Id. "Congress intended `take' to apply broadly to cover indirect as well as purposeful actions." Babbitt v. Sweet Home Chapter of Cmtys. for a Great Or., 515 U.S. 687, 704, 115 S.Ct. 2407, 2416, 132 L.Ed.2d 597 (1995).

In 1982, Congress amended the ESA to provide exceptions to the strict prohibition on "takes." See id. at 691, 115 S.Ct. at 2409-10. Under the revised 16 U.S.C. § 1539(a)(1)(B), the Secretary of the Interior may issue an ITP authorizing "takes" that are "incidental to, and not the purpose of, the carrying out of an otherwise lawful activity." An ITP is issued by the U.S. Fish and Wildlife Service ("FWS") after the development and approval of a Habitat Conservation Plan ("HCP").1 HCPs must include, among other things, information regarding the applicant's plan to "minimize and mitigate" the impacts likely to result from incidental takes. 16 U.S.C. § 1539(a)(2)(A)(ii).

B.

TAP sued the state defendants pursuant to 16 U.S.C. § 1540, which authorizes citizen suits seeking to enjoin the actions of any person, including any "governmental instrumentality or agency (to the extent permitted by the [E]leventh [A]mendment to the Constitution), who is alleged to be in violation of any provision" of the ESA. TAP asserted that the state defendants' water permitting and regulatory practices had led to private parties' withdrawing water from the San Antonio and Guadalupe rivers, in turn leading to a significant reduction in freshwater inflow into the San Antonio Bay ecosystem. That reduction in fresh-water inflow, coupled with a drought, led to increased salinity in the bay, which decreased the availability of drinkable water and caused a reduction in the abundance of blue crabs and wolfberries, two of the cranes' staple foods. According to TAP, that caused the cranes to become emaciated and to engage in stress behavior, such as denying food to juveniles and flying farther afield in search of food, leading to further emaciation and increased predation. Ultimately, this chain of events led to the deaths of twenty-three cranes during the winter of 2008-2009.

TAP thus alleged that the state defendants' water-permitting practices effected a taking of whooping cranes, in violation of the ESA, and that such takings would continue to occur absent intervention by the court. Accordingly, TAP sought declaratory and injunctive relief designed to ensure that the AWB flock had sufficient water resources to prevent future takings.

Before trial, the Guadalupe-Blanco River Authority ("GBRA"), Texas Chemical Council, and San Antonio River Authority ("SARA") (collectively, the "intervenor defendants") were granted leave to intervene. The district court conducted an eight-day bench trial that included nearly thirty witnesses. On March 11, 2013, the court issued an exhaustive 124-page opinion, which adopted verbatim TAP's proposed fact findings. The court declared that the state defendants had violated the ESA through their water-management practices and were continuing to do so. The court granted an injunction ordering (1) that

and (2) that

Two days later, the state defendants, GBRA, and SARA moved in the district court for a stay pending appeal. The court denied the motions but amended the first portion of its injunctive relief to provide that the

A motions panel of this court granted the state defendants' and GBRA's motions for a stay pending appeal after setting an expedited briefing schedule. The state defendants and intervenor defendants appeal the judgment.2

II.

In their statement of jurisdiction, the state defendants "note two issues relevant to Article III standing." They note that TAP did not suggest a threat of future injury sufficient for standing, and they question whether the remote causal connection between TCEQ permits and crane deaths demonstrates traceability.

For standing, a party must demonstrate the "triad of injury in fact, causation, and redressability." Steel Co. v. Citizens for a Better Env't, 523 U.S. 83, 103, 118 S.Ct. 1003, 1017, 140 L.Ed.2d 210 (1998). The injury in fact must be "a harm suffered by the plaintiff that is `concrete' and `actual or imminent.'" Id. (citing Whitmore v. Arkansas, 495 U.S. 149, 155, 110 S.Ct. 1717, 1723, 109 L.Ed.2d 135 (1990)). Causation requires a "traceable connection" between the plaintiff's injury and the defendant's conduct. Id. Redressability requires "a likelihood that the requested relief will redress the alleged injury." Id. To seek injunctive relief, the plaintiff must show a real and immediate threat of future or continuing injury apart from any past injury. In re Stewart, 647 F.3d 553, 557 (5th Cir.2011). "Past exposure to illegal conduct does not in itself show a present case or controversy regarding injunctive relief." Id. (citation omitted). Although past wrongs may help establish the threat of a future injury, they are insufficient alone. See O'Shea v. Littleton, 414 U.S. 488, 495-96, 94 S.Ct. 669, 676, 38 L.Ed.2d 674 (1974).

There is little doubt that TAP alleged sufficient facts concerning the components of standing to justify pursuing this litigation. TAP alleged injury (death to cranes and injury to those who enjoy them) and a theory of causation (TCEQ water use permits ultimately affected the cranes' habitat), and it alleged that future deaths could be attributed to "takes" in violation of the ESA without injunctive relief. The state defendants' concerns about Article III standing boil down to a post hoc argument based on the results of trial. We think it prudent to review the issues on the merits in the following discussion. While Lujan requires that standing be maintained throughout the course of litigation, Lujan v. Defenders of Wildlife, 504 U.S. 555, 561, 112 S.Ct. 2130, 2137, 119 L.Ed.2d 351 (1992), this does not require a court to dismiss for lack of standing when a plaintiff fails to prove its case on any of the three essential components.

III.

Although the intervenor defendants do not challenge TAP's standing to sue, they raise a procedural question concerning the district court's decision to adjudicate the case instead of invoking the Burford abstention doctrine.3 We review an abstention ruling for abuse of discretion, but "we review de novo whether the requirements of a particular abstention doctrine are satisfied." Romano v. Greenstein, 721 F.3d 373, 380 (5th Cir.2013) (internal citations and quotations omitted). "A court abuses its discretion when its ruling is based on an erroneous view of the law or a clearly erroneous assessment of the evidence." Kipps v. Caillier, 197 F.3d 765, 770 (5th Cir.1999). We find no abuse here.

The federal courts have a "virtually unflagging obligation ... to exercise the jurisdiction given them." Colorado River Water Conservation Dist. v. United States, 424 U.S. 800, 817, 96 S.Ct. 1236, 1246, 47 L.Ed.2d 483 (1976); New Orleans Pub. Serv., Inc. v. Council of City of New Orleans, 491 U.S. 350, 359, 109 S.Ct. 2506, 2513, 105 L.Ed.2d 298 (1989) ("NOPSI"). A court may abstain from exercising its equity jurisdiction, however, where doing so would "be prejudicial to the public interest." Burford v. Sun Oil Co., 319 U.S. 315, 318, 63 S.Ct. 1098, 1099, 87 L.Ed. 1424 (1943) (internal citations omitted). The Court in Burford delineated an area of abstention where the issues "so clearly involve basic problems of [State] policy" that the federal courts should avoid entanglement. Id. at 332, 63 S.Ct. at 1106.

In NOPSI, the Court further articulated the narrow bounds of Burford abstention:

NOPSI, 491 U.S. at 361, 109 S.Ct. at 2514 (citations omitted). Even where a federal court would have to upset a "complex state administrative process," abstention may not be proper. Id. at 362, 109 S.Ct. at 2515. Although Burford abstention thus continues to be "permissible," it is "the exception, not the rule." Id. at 359, 109 S.Ct. at 2513.

Five factors govern the decision whether to abstain:

Wilson v. Valley Elec. Membership Corp. 8 F.3d 311, 314 (5th Cir.1993) (internal citations and quotations omitted). We have applied those factors consistently in reviewing Burford abstention. See, e.g., Sierra Club, Inc. v. Sandy Creek Energy Assocs., L.P., 627 F.3d 134 (5th Cir.2010). We consider each in turn.

1.

The first prong — whether the cause of action arises under federal or state law — is straightforward. This cause of action arises under the federal ESA. The first factor thus weighs in favor of not abstaining but does not settle the issue.4

2.

Regarding the second prong, "Burford abstention does not so much turn on whether the plaintiff's cause of action is alleged under federal or state law, as it does on whether the plaintiff's claim may be in any way entangled in a skein of state law that must be untangled before the federal case can proceed." City of San Antonio, 112 F.3d at 795 (citation omitted). Of primary concern in Burford was the involvement of the federal courts in deciding issues of essentially state law and policy. Federal courts were interpreting and applying state law in oil-well disputes, which "created a constant task for the Texas Governor" and forced the Texas Railroad Commission to "adjust itself to the permutations of the law as seen by the federal courts." Burford, 319 U.S. at 329-30, 63 S.Ct. at 1105-06.

In Wilson, 8 F.3d at 315, we stated that this factor turns in part on whether the court will be forced to weigh competing local interests and mostly review an agency's decision in an area in which that agency is arguably an expert. Abstention would be proper if "applying the seemingly clear legal standard ... would involve the federal court in an open-ended `fairness' inquiry into predominantly local matters." Id. What would amount to review of state agency action in a state law framework would be grounds for abstention: A "claim that a state agency has misapplied its lawful authority or has failed to take into consideration or properly weigh relevant state-law factors" might disrupt the state's programs and would immerse the court in local law and facts. NOPSI, 491 U.S. at 362, 109 S.Ct. 2506. This court thus required abstention in City of San Antonio, 112 F.3d at 794, where an injunction under the ESA would have entangled the court in issues of state law in part by forcing the administrative bodies to violate other state laws.

On balance, this factor weighs against abstention. The state defendants do not argue, as did the defendant in City of San Antonio, that they would be forced to violate state law by complying with the injunction. Additionally, the district court, to render a decision, did not, engage complex issues of state law or weigh state policy decisions. Instead, the court decided that (1) the ESA prohibits "takes"; (2) TCEQ causes takes; and (3) the court enjoins the actions that cause takes unless they are "approved" by the FWS. On its face, the formula does not require, as in Burford, examining individual permits and rendering decisions in favor of individual permittees. One key difference between this case and City of San Antonio is that the injunction there required the state to distribute or not distribute water in a certain fashion, whereas here the injunction is primarily focused on the ITP process and future permitting actions. Abstention is not required "merely because resolution of a federal question may result in the overturning of a state policy." NOPSI, 491 U.S. at 363, 109 S.Ct. at 2515.

3.

As for the importance of the state interests, "Texas clearly has an interest in uniform decision-making regarding [its] finite amount of water." City of San Antonio, 112 F.3d at 795. States have a strong interest in managing their own natural resources, and courts have recognized a strong state interest in, among other areas, utilities, train service, and insurance regulation.5

In Burford, the state had a strong interest in creating a coherent system of oil regulations and managing natural resources. Even though Burford concerned a constitutional challenge, the Court in NOPSI, 491 U.S. at 360, 109 S.Ct. at 2514, explained that any federal interest there was dwarfed by the state interest — the "constitutional challenge was of minimal federal importance, involving solely the question whether the commission had properly applied Texas' complex oil and gas conservation regulations."

In Wilson, 8 F.3d at 315, we similarly applied a sort of balancing between state and federal interests, noting the importance of regulation of utilities as a core part of the police power and affirmed abstention. We noted in City of San Antonio, 112 F.3d at 794, that, where both the water source and the endangered species were "entirely intrastate," the "management of the aquifer [was] a matter of peculiar importance to the state."

In City of San Antonio, we went on to explain that Texas has a strong interest in water regulation, "especially in times, like today, of devastating drought." Id. (internal quotations omitted). The regulation of the Edwards Aquifer, at issue there, was "vital to the general economy and welfare of the State of Texas," because the Aquifer was "the primary source of water for residents of the south central part" of Texas. Id. Further, "the State has the responsibility under the Texas Constitution to preserve and conserve water resources for the benefit of all Texans." Id.

Water management is undoubtedly an important state interest. But what distinguishes this case somewhat from City of San Antonio and Burford is that there is also a strong federal interest. The whooping crane is an interstate, and indeed international, species. The ESA is designed to "grant federal courts subject matter jurisdiction over suits like the one presently before us" because of the federal interest in endangered species. Sandy Creek, 627 F.3d at 144 n. 15. Though the state interest is strong in terms of managing water use, so is the federal interest.

4.

For the fourth prong, states have a strong need for coherent policy in the regulation of finite natural resources. See Burford, 319 U.S. at 325, 63 S.Ct. at 1103. For example, in Burford, id. at 319, 63 S.Ct. at 1100, the Court explained that the state needed a coherent policy, because "one operator can ... drain oil from the most distant parts of the reservoir." Similarly, in City of San Antonio, 112 F.3d at 793-94, we explained that "allowing one party to take water necessarily affects other parties."

The Court in NOPSI, however, explained that the need for coherence is not alone a reason for abstention. Although Burford abstention "is concerned with protecting complex state administrative processes from undue federal interference, it does not require abstention whenever there exists such a process or even in all cases where there is a potential for conflict with state regulatory law or policy." NOPSI, 491 U.S. at 362, 109 S.Ct. at 2515 (internal quotations omitted).

The Texas Water Code is similar to the regulations at issue in Burford and City of San Antonio. It regulates a scarce resource that necessarily interconnects its users. It requires a state oversight agency, TCEQ, to implement the regulatory scheme. Finally, it regulates water by instructing TCEQ to consider scientific data, balance stakeholder interests, and maintain a permit system through an elaborate system of orders, schedules, and reports. Federal intervention could easily upset that delicate balancing. This factor weighs in favor of abstention.

5.

To justify abstention, there must be a forum that offers "[t]imely and adequate state-court review." NOPSI, 491 U.S. at 361, 109 S.Ct. at 2514 (internal citations omitted). Review typically includes the ability to appeal agency orders to a state trial court, with available state appellate review, and such review may include initial review by the agency. Neither a private cause of action nor specific enforcement provisions are required, but review should be more than a fact finding venture with only the remote possibility of enforcement. See City of San Antonio, 112 F.3d at 797. In Burford, 319 U.S. at 333-34, 63 S.Ct. at 1107, the Court found sufficient state review where the state provided "a unified method for the formation of policy and determination of cases by the Commission and by the state courts." The review in both the agency and the state courts was "expeditious and adequate." In Wilson, 8 F.3d at 316, we explained that, where there was an administrative agency with "broad power to address legal issues related to regulatory duties," there was sufficient state judicial review. Similarly, in City of San Antonio, 112 F.3d at 797, we concluded that sufficient review was provided where the Edwards Aquifer Authority could sue for injunction in state court, and a separate entity, the Texas Natural Resource Conservation Commission, could "file suit for an order of mandamus against the Authority to compel it to perform its duties". The statute in City of San Antonio explicitly addressed the preservation of endangered species and required the Authority to "protect aquatic and wildlife habitat" and to "protect species that are designated as threatened or endangered under applicable federal or state law." Id. at 794.

The scheme here, at first glance, seems to afford sufficient state-court review. Under Section 5.351 of the Texas Water Code, "[a] person affected by a ruling, order, decision, or other act of the [TCEQ] may file a petition to review, set aside, modify, or suspend the act of the commission," and a suit in state court follows the standard state appeals process, just as in Burford, Alabama, Wilson, and City of San Antonio. Additionally, individuals may petition TCEQ to provide more water for environmental uses. See 30 TEX. ADMIN. CODE § 20.15.

There are, however, signs of inadequate review. In the first place, Section 11.0235(d)(1) of the Texas Water Code expressly forbids granting water rights for environmental needs. As TAP points out, there is thus no petition option; TCEQ is not authorized, under state law, to grant flows based on environmental concerns. Second, although the Code requires TCEQ to consider the environmental impact of permitting, it also requires, as "an essential part" of that scheme, that all permitting related to environmental flows be suspended "during emergencies," which includes drought emergencies. TEX. WATER CODE § 11.0235(c).

The key question is whether TCEQ actually has authority to remedy the problem: that is, whether, given a drought (which constitutes an emergency), TCEQ can still provide water for the cranes. Under Section 11.0235(c), TCEQ appears not to have that power. That essentially leaves the state courts as the only avenue for redress, but the parties cite no authority showing how one would bring such an action to force TCEQ to provide greater freshwater flows. At oral argument, counsel for TAP repeatedly suggested that there was no cause of action under which TAP could sue TCEQ in the Texas courts; that analysis seems correct. That factor, on which the district court focused, weighs against abstention, because it is not evident that TCEQ or the state courts have authority to provide TAP the type of relief it seeks. See Tex. Comm'n on Envtl. Quality v. San Marcos River Found., 267 S.W.3d 356 (Tex.App.-Corpus Christi 2008).

6.

In summary regarding abstention, the instant case is similar in certain ways to City of San Antonio, in which we held that a water regulatory scheme demanded abstention even in the face of an ESA suit. There are key differences, however, including the intrastate focus in City of San Antonio, the more highly developed environmental protections there, and the broader grant of administrative and judicial authority by state law to remedy environmental grievances.

Burford abstention is disfavored as an abdication of federal jurisdiction. This case arises under federal law, and, treading carefully, the federal courts need not become entangled in state law to adjudicate the ESA claim here. The district court did not abuse its discretion by declining to abstain. We turn to its findings and conclusions.

IV.

A.

"The standard of review for a bench trial is well established: Findings of fact are reviewed for clear error and legal issues are reviewed de novo." Kona Tech. Corp. v. S. Pac. Transp. Co., 225 F.3d 595, 601 (5th Cir.2000). "A finding is clearly erroneous if it is without substantial evidence to support it, the court misinterpreted the effect of the evidence, or this court is convinced that the findings are against the preponderance of credible testimony." Petrohawk Props., L.P. v. Chesapeake La., L.P., 689 F.3d 380, 388 (5th Cir.2012) (quoting French v. Allstate Indem. Co., 637 F.3d 571, 577 (5th Cir.2011)).

The basis of many of the district court's conclusions and remedy is a finding that twenty-three whooping cranes perished during the winter of 2008-2009 and the death toll represented the "takes" committed or caused by the state defendants. The court reached its mortality finding based primarily on evidence and testimony from TAP's expert, Tom Stehn, a Refuge biologist formerly employed by the FWS.

Stehn worked at the Refuge for twenty-nine years, until 2011, and was responsible for conducting annual surveys of the whooping-crane population. He flew over the entire Refuge in parallel lines and conducted a visual count of adult and juvenile birds; each flight could cover the Refuge twice. Because the cranes are territorial, Stehn concluded that not seeing a bird in its usual location for two or more flights meant that it had died. When tallying mortality rates, Stehn counted only adults and juveniles, not sub-adult birds.6

Stehn's methods changed somewhat over time. In the early years, he routinely conducted as many as twenty-six flights during the winter season. The planes flew at low speeds and were often 20-50 feet above the ground. Also, a high percentage of birds had colored bands used for tracking and identification.

In more recent years, Stehn flew at 200 feet, and budgetary constraints reduced the number and length of flights from twenty-six per season to between eight and twelve, and from eight to six hours. The colored bands had faded and were no longer useful for identifying specific birds.

During the 2008-2009 winter season, Stehn conducted eleven flights, six of which he categorized as reliable for purposes of conducting a mortality count. He noted that cranes seemed to be moving more than in past years; he surmised that was related to the drought and food conditions. His counts found nineteen birds absent from their usual territory, and those were counted as dead. Additionally, four carcasses were found, a high number considering that only twenty had been recovered in the Refuge since 1938.

The intervenor defendants contend that the finding of twenty-three deaths is clearly erroneous and unsupported by the evidence. They assert that Stehn's surveys and mortality calculations were inaccurate and unreliable. Though they have abandoned their challenge to Stehn's qualifications as an expert, they claim his data are "unreliable as a matter of law." The district court, however, found Stehn's methods reliable. He had employed the same counting method for almost thirty years, and the Refuge, the FWS, and national and international organizations relied on his work. No one else had attempted counts or challenged the validity of his findings for three decades. Moreover, Stehn had a vested interest in making accurate counts of whooping cranes — that was one of his primary job responsibilities, and he made counts after the 2008-2009 winter using the same methods. Both sides sought to compel Stehn's testimony, and he testified only after the court subpoenaed him and without having been prepared by TAP.

It is true that Stehn's methods changed somewhat over the years and may have led to a less accurate count in 2008-2009 than might have been made in the 1990s. The lack of bands, higher-flying surveys, increased crane movement, and fewer flights may reasonably be concluded to have contributed to inaccuracies. Moreover, as Stehn admitted, colored bands and GPS tracking would have been more accurate. But these considerations alone are not enough to say that his methods are unreliable "as a matter of law."

Further, although it was not peer-reviewed in the sense that a journal article would be, and may not even be the "best" method of counting, Stehn's methodology could be considered by the district court for whatever weight it might bear. The only indications that Stehn's methodology was sub-optimal are in the 2011 FWS report discussed below and the testimony of a statistician, Dr. Conroy, who had never conducted surveys of the whooping cranes.7 Consequently, although there may be some doubt as to the 2008-2009 mortality numbers, that doubt hardly leaves us with a "firm conviction" that a mistake has been made. The finding that twenty-three cranes died that winter is not clearly erroneous.8

B.

The intervenor defendants moved to reopen the evidence after trial to introduce an FWS report, the 2011-2012 Abundance Survey, critical of Stehn's aerial survey methodology. The district court, after reviewing and considering the survey, denied the motion, giving a lengthy analysis of why it would be improper to admit the survey. The intervenor defendants challenge that exclusion.

We review evidentiary rulings under a deferential abuse-of-discretion standard. Stover v. Hattiesburg Pub. Sch. Dist., 549 F.3d 985, 992 (5th Cir.2008). A court "abuses its discretion when its ruling is based on an erroneous view of the law or a clearly erroneous assessment of the evidence." United States v. Yanez Sosa, 513 F.3d 194, 200 (5th Cir.2008). If there is error, it is reviewed for harmlessness. We reverse a judgment based on an erroneous evidentiary ruling only if that ruling "affected the substantial rights of the parties." Stover, 549 F.3d at 992. "When, as here, the district court has conducted, on the record, a carefully detailed analysis of the evidentiary issues and the court's own ruling, appellate courts are [wary] about finding an abuse of discretion." Kelly v. Boeing Petroleum Servs., Inc., 61 F.3d 350, 356 (5th Cir.1995).

In deciding whether to reopen evidence, a court should weigh "the importance and probative value of the evidence, the reason for the moving party's failure to introduce the evidence earlier, and the possibility of prejudice to the non-moving party." Chieftain Int'l (U.S.), Inc. v. Se. Offshore, Inc., 553 F.3d 817, 820 (5th Cir. 2008) (citation omitted). The district court focused primarily on importance and probative value. Its discussion did not touch on the fact that the evidence could not be submitted earlier, and it did not discuss the possible prejudice to TAP by the report's admission.

Those last two factors undoubtedly weigh in favor of reopening evidence. The intervenor defendants did not have access to the report during trial, because it had not been published, and TAP would not have been prejudiced by late admission of the survey.

Probative value is the main dispute. The survey makes generally broad claims about Stehn's methods, which it concludes were flawed because he relied on the assumptions that cranes do not leave their territories and that territory is therefore a "defensible surrogate" for counting birds. Drawing on data from the 2008-2009 winter, when birds were found moving farther afield in search of food and water, the survey concludes that Stehn's assumption "[are] unnecessary and untenable given recent data." The survey criticized past methods as "not based on a statistically defensible sampling design and therefore [unable to] provide meaningful measures of precision." The survey also noted that results were difficult to duplicate and were dependent on the observer's "experience and judgment."

The court concluded that the survey lacked value for several reasons. First, it focused on a population count, rather than a mortality count, whereas the evidence at trial was focused on mortality. Second, the survey conflicted with evidence adduced at trial, such as the cranes' territoriality. Third, the court was unconvinced by the survey's data and noted that the explanation for the total count was lacking. Fourth, the survey described itself as "preliminary." Fifth, the court found the survey's "error rate" unacceptable. For those reasons, the court found that the survey was not sufficiently important and lacked enough probative value to require reopening the evidence.

Exclusion of the survey was error under Chieftain International, 553 F.3d at 820. In the first place, the court did not consider the second two factors. More importantly, it improperly acted as a trier of fact, weighing and then excluding the evidence, rather than as a gatekeeper, so it imposed a higher bar than the law allows.

Evidence is relevant where it has the tendency to make a fact more or less probable. FED.R.EVID. 401(a). The survey is relevant to, and highly probative of, Stehn's calculations of crane mortality, a central issue. The report tends to cast doubt on Stehn's methods and his count. There is no independent basis in the Federal Rules of Evidence for exclusion. The court essentially acted as if the evidence had been admitted, then weighed it against the evidence presented. The court was thus not merely making a threshold probative — value determination — it was trying the evidence. That, combined with disregard of the latter two factors, means there was error.

That said, the error was harmless. Although the district court did not admit the survey, it did carefully consider it, and its ultimate factual findings regarding Stehn's methods and the mortality count were unaffected. The court found the survey unpersuasive in light of the other evidence. Thus, even if the court had admitted the survey into evidence, the outcome would not have changed. The trier of fact explicitly stated that it would not have come to a different conclusion had it considered that evidence, which it did in fact thoroughly review. The defendants' rights were therefore not affected.

C.

The principal liability issue thus becomes whether the actions of TCEQ in administering licenses to take water from the Guadalupe and San Antonio rivers for human, manufacturing and agricultural use foreseeably and proximately caused the deaths of whooping cranes in the winter of 2008-2009.9 The district court either misunderstood the relevant liability test or misapplied proximate cause when it held the state defendants responsible for remote, attenuated, and fortuitous events following their issuance of water permits.

Proximate cause and foreseeability are required to affix liability for ESA violations. In the course of holding that "harm" under the ESA validly includes "significant habitat modification or degradation that actually kills or injures wildlife," 50 C.F.R. § 17.3 (1994), the Supreme Court squarely rejected the dissenters' assertions that a form of strict liability, unlimited by causal connection, could be imposed. Sweet Home, 515 U.S. at 690-708, 115 S.Ct. at 2409-2418 (Stevens, J., majority), 714-735, 115 S.Ct. at 2421-2430 (Scalia, J., dissenting). The Court reasoned that the ESA prohibits "takes" so long as they are "foreseeable rather than merely accidental." Sweet Home, 515 U.S. at 700, 115 S.Ct. at 2414. Indeed, the statute should be read to incorporate ordinary requirements of proximate causation and foreseeability. Id. at 696 n. 9, 700 n. 13, 115 S.Ct. at 2412 n. 9, 2414 n. 13 ("Nothing in the regulation purports to weaken [ordinary requirements of foreseeability and proximate cause]."). Justice O'Connor's concurrence elaborates that proximate cause, while "not susceptible of precise definition," is a concept that "`normally eliminates the bizarre'" and has "`functionally equivalent' alternative characterizations in terms of foreseeability ... and duty. Proximate causation depends to a great extent on considerations of the fairness of imposing liability for remote consequences." Id. at 713, 115 S.Ct. at 2420 (citations omitted).

The Court was not asked to apply its proximate cause definition to the facts in Sweet Home, but acknowledged that "[i]n the elaboration and enforcement of the ESA, the Secretary and all persons who must comply with the law will confront difficult questions of proximity and degree." Id. at 708, 115 S.Ct. at 2418. Later, in Exxon Co., U.S.A. v. Sofec, Inc., the Supreme Court affirmed that "proximate causation principles are generally thought to be a necessary limitation on liability." Exxon Co., 517 U.S. 830, 838, 116 S.Ct. 1813, 1818, 135 L.Ed.2d 113 (1996). "`In a philosophical sense, the consequences of an act go forward to eternity, and the causes of an event go back to the dawn of human events, and beyond.'" Id. (quoting W. Keeton, et al., Prosser and Keeton on the Law of Torts 264 (5th ed.1984)) (hereinafter Keeton). Nevertheless, the Exxon Court continued:

Id. at 838-39, 116 S.Ct. at 1818 (quoting Petition of Kinsman Transit Co., 338 F.2d 708, 725 (2nd Cir.1964) (Friendly, J.), quoted in 1 T. Schoenbaum, Admiralty and Maritime Law § 5-3, at 164 (2d ed.1994)). Most recently, the Court reiterated that "[a] requirement of proximate cause thus serves, inter alia, to preclude liability in situations where the causal link between conduct and result is so attenuated that the consequence is more aptly described as mere fortuity." Paroline v. United States, ___ U.S. ___, 134 S.Ct. 1710, 1719, 188 L.Ed.2d 714 (2014) (citing Exxon Co., 517 U.S. at 838-39, 116 S.Ct. at 1818).

Applying a proximate cause limit to the ESA must therefore mean that liability may be based neither on the "butterfly effect"10 nor on remote actors in a vast and complex ecosystem. Justice O'Connor's concurrence in Sweet Home is instructive. It disavows foreseeability, and thus ESA liability, where a farmer tills his field, causes erosion that makes silt run into a nearby river, which depletes oxygen in the water, and thereby injures protected fish. Sweet Home, 515 U.S. at 713, 115 S.Ct. at 2420 (O'Connor, J., concurring).

A district court's finding of proximate cause is reviewed for clear error. Bertucci Contracting Corp. v. M/V ANTWERPEN, 465 F.3d 254, 259 (5th Cir. 2006). When, as here, a court's factual finding "rest[s] on an erroneous view of the law", its factual finding does not bind the appellate court. See Pullman-Standard v. Swint, 456 U.S. 273, 287, 102 S.Ct. 1781, 1789, 72 L.Ed.2d 66 (1982). Accordingly, when the record permits only one resolution of the factual issue after the correct law is applied, remand is unnecessary. Swint, 456 U.S. at 292, 102 S.Ct. at 1792.11

In resolving the factual issue, the trial court maintained an erroneous view of proximate cause. The trial court cited Sweet Home's proximate cause requirement exactly twice in a 124-page opinion. Aransas Project v. Shaw, 930 F.Supp.2d 716, 727, 786 (S.D.Tex.2013) (stating that ordinary requirements of proximate causation apply). The court concluded in the very next paragraph to one of these citations that "[p]roximate causation exists where a defendant government agency authorized the activity that caused the take." Id. at 786. This is an erroneous view of proximate cause standards. Taken at face value, the court's statement eliminates "proximate" from "proximate cause" whenever a governmental entity's licensing activity is involved in a "take." In addition to the foregoing explanations about proximate cause in general and under the ESA, the Supreme Court succinctly states that, "[t]he concepts of direct relationship and foreseeability are, of course, two of the `many shapes [proximate cause] took at common law.'" Hemi Group, LLC v. City of New York, 559 U.S. 1, 130 S.Ct. 983, 991, 175 L.Ed.2d 943 (2010)(citing Holmes v. Sec. Inv. Protection Corp., 503 U.S. 258, 268, 112 S.Ct. 1311, 117 L.Ed.2d 532 (1992). The district court's formulation and its ensuing opinion ignore both of those concepts, as it nowhere mentions remoteness, attenuation, or the natural and probable consequences of actions. Nowhere does the court explain why the remote connection between water licensing, decisions to draw river water by hundreds of users, whooping crane habitat, and crane deaths that occurred during a year of extraordinary drought compels ESA liability. Indeed, the court's rule is open to the State's criticism that issuing drivers' licenses will "cause the take" of endangered species run over by cars, and it implies that governmental licensing of power lines, wind turbines or cell towers, with which many endangered birds collide, could violate the ESA. The court's simplistic phrasing begged the questions of remoteness and foreseeability inherent in proximate cause and required by Sweet Home.

Moreover, the court's rule establishing proximate cause from "authorizing" any activity that "caused" a take creates liability far beyond the contours of current ESA case law. In Sweet Home, for instance, the Court explained that a landowner who knowingly drained a pond that housed endangered fish should not escape ESA liability for destroying the aquatic habitat. Sweet Home, 515 U.S. at 699-700, 115 S.Ct. at 2413-14. This is the limited, albeit not definitive, Sweet Home conception of an "indirect" taking.12

Cases decided in the wake of Sweet Home also do not create this almost per-se proximate cause rule applied by the district court. Other circuit courts have held certain regulatory acts resulted in ESA liability where a close connection existed between the liable actor's conduct and habitat destruction or killing of endangered species. In Sierra Club v. Yeutter, the Forest Service permitted excessive timber removal in Texas forests whose trees are home for red cockaded woodpeckers. Sierra Club, 926 F.2d 429, 432-33 (5th Cir. 1991). In Strahan v. Coxe, the state's licensing of fishermen to use gillnets and lobster traps in certain areas was done with an awareness that right whales could be caught in the devices, and over fifty percent of right whales showed scars from previous encounters with the devices. Strahan, 127 F.3d 155, 165 (1st Cir.1997). In Loggerhead Turtle v. Cnty. Council of Volusia Cnty., Fla., 148 F.3d 1231 (11th Cir.1998), the court accepted for standing purposes a theory that the county's authorization of nocturnal vehicular beach traffic and regulation of outdoor lighting could directly result in the killing of newly-hatched loggerhead turtles by misdirecting them away from the sea.13 See also Anim. Welfare Inst. v. Martin, 623 F.3d 19 (1st Cir.2010) (licensing of animal traps that caught endangered lynx). The regulations or licensing in each of these cases concerned actions that directly killed or injured species or eliminated their habitat.14

Not every government action has such direct consequences. Indeed, in Strahan, the court held that "a governmental third party pursuant to whose authority an actor directly exacts a taking of an endangered species may be deemed to have violated... the ESA." 127 F.3d at 163 (emphasis added). Rejecting an intervening actor defense to proximate cause, the court was even more pointed: "In this instance, the state has licensed commercial fishing operations to use gillnets and lobster pots in specifically the manner that is likely to result in a violation of federal law." 127 F.3d at 164. In sharp contrast to Strahan and these other cases, the district court's untethered and virtually per-se linking of governmental licensing with ESA takes elides proximate cause rather than applying it. The standard simply assumes proximate causation where licensing occurs. Therefore, the district court had an erroneous view of the law and this court need not be deferential to the proximate cause finding. Swint, 456 U.S. at 287, 292 102 S.Ct. at 1789, 1792.

Even if we accept the court's subsidiary factual findings, there can be only one resolution of the proximate cause issue based on the record: proximate cause was lacking as a matter of law. The issue here is whether TCEQ's issuance of water permits "proximately caused" the unusual crane deaths in the winter of 2008-09. Unlike the cases just cited, there is a long chain of causation here between the TCEQ's issuance of permits to take water from the rivers and cranes' mortality. Until recently, the permits had few limits on users' ability to take water.15 TAP offered evidence, which the court accepted, that the licensed withdrawals of water from the rivers resulted in a decline in freshwater inflows to the San Antonio Bay. Continuing with the court's findings, with less freshwater inflows, the bay's salinity increased in various gradients by a few parts per thousand. In turn, the increased salinity of the estuary and marsh water affected the conditions in which blue crabs and wolfberry plants grow. These are principal food sources of whooping cranes following their thousands-mile migration across North America to their winter habitat. There were then fewer blue crabs and wolfberries for the cranes to eat. The cranes succumbed to "food stress," causing them to search for "upland" sources of food and water. Necropsies of two cranes that died during the 2008-2009 winter showed signs of emaciation, and overall an estimated 23 cranes died. The crane population, nevertheless, has continued to increase both before and after the winter of 2008-2009.

Every link of this chain depends on modeling and estimation. At best, the court found but-for causation. Proximate cause, however, requires the causal factors and the result to be reasonably foreseeable. Sweet Home, 515 U.S. at 697 n. 9, 115 S.Ct. at 2412 n. 9 (providing that ESA "take" must be foreseeable). TAP acknowledges in its brief the importance of foreseeability. ("The issue is not the number of steps of `causation' ... but foreseeability."). But the district court's opinion does not establish that the state could have reasonably anticipated the synergy among the links on the chain in 2008-2009. The court's only discussion of foreseeability in its entire opinion occurs with respect to the effect of water-permitting on freshwater inflows. The court found that the state defendants could foresee this link in the causal chain because a 2007 United States Fish and Wildlife Service International Whooping Crane Recovery Plan noted that "[u]pstream reservoir construction and water diversions for agriculture and human use reduce freshwater flows." Aransas Project, 930 F.Supp.2d at 747. All this statement indicates is that upstream diversions in general reduce freshwater inflows to the bay. It does not indicate that the freshwater inflows into the San Antonio Bay were materially decreasing from levels in prior years. The report, moreover, noted that "[m]any existing water rights are currently only partially utilized, but greater utilization is expected over time." Id. While the report explains generally the possibility of drought and attendant risks to the cranes, this, too, fails to satisfy TAP's burden to prove that this drought or its severity was foreseeable.

The report includes numerous non-specific, conditional, predictive statements not quoted by the district court. The report states that:

The report also states that:

In fact, a few pages later, the report states:

According to the report, decreased freshwater inflows "at times" have been "insufficient," and in future decades the decline may affect the bay's ecosystem. These statements do not establish foreseeability that decreased freshwater inflows in 2008-2009 would result in abnormal crane deaths. After all, during nearly six preceding decades, the same FWS report observes that human population along the rivers had steadily increased, leading presumably to increasing river water use, and the state had suffered periodic, severe droughts, but the whooping crane population was concomitantly steadily increasing after near extinction. The state defendants had no reason to anticipate a significant die-off because of decreased freshwater inflows only one year after this report issued.16

The lack of foreseeability or direct connection between TCEQ permitting and crane deaths is also highlighted by the number of contingencies affecting the chain of causation from licensing to crane deaths. The contingencies are all outside the state's control and often outside human control. To begin, the state's control over water usage is at a macro, not a micro level. Surface water is the property of the state, subject to the vested property rights of landowners. Texas Water Rights Comm'n v. Wright, 464 S.W.2d 642, 647 (Tex.1971). Texas law generally forbids appropriating water from the state's rivers without a permit. Tex. Water Code § 11.121. While permits authorize usage, however, they do not compel it. Further, some users, such as domestic and livestock users, need not obtain permits. Id. § 11.142. The independent choices of water users are also affected by the availability of water from alternative sources like reservoirs and the Edwards aquifer. The aquifer, indeed, is a major water source for South Texas, including San Antonio. TCEQ accordingly cannot control the amount of water that will be diverted from the rivers.

Even more unpredictable and uncontrollable are the forces of nature. The weather, tides and temperature conditions dramatically affect salinity within and throughout the bay.17 As the district court found, a few rains in autumn 2009, for instance, restored salinity to desirable levels for crab and wolfberry production. Aransas Project, 930 F.Supp.2d at 746. That these natural conditions can change quickly is a truism, and that the seriousness or duration of a drought cannot be foreseen in advance is equally trite. Texas is prone to cyclical drought conditions, but the winter of 2008-2009 was an outlier among those.

For another link in its chain of causation, the district court found that "with lower salinities, the greater the chances for a Whooping Crane to find a blue crab," id. at 753, and that "decreases in freshwater inflows to the San Antonio bay/Guadalupe estuary results [sic] in a decrease in blue crabs as well as wolfberries on the critical habitat of the AWB cranes." Id. at 754. Even accepting these findings, the salinity levels that affect blue crab habitat choices and wolfberry production are also subject to varying and unpredictable contingencies of weather, tides and temperature changes. In addition, the blue crab population in this bay (as in many places) suffered a consistent decline since the 1980s because of overfishing.18 Yet inversely to the ongoing blue crab decline, which must have decreased the chances for a whooping crane to find a blue crab, the whooping crane population grew nearly every year.

Contingencies concerning permittees' and others' water use, the forces of nature, and the availability of particular foods to whooping cranes demonstrate that only a fortuitous confluence of adverse factors caused the unexpected 2008-2009 die-off found by the district court. This is the essence of unforeseeability.

Proximate cause eliminates liability for actors when the resulting harm is too attenuated from their negligence (and there is no suggestion that TCEQ's actions were even negligent). The chain of causation here, unlike any in the reported case law concerning the ESA, may have had an impact on the whooping crane deaths in 2008-2009. Finding proximate cause and imposing liability on the State defendants in the face of multiple, natural, independent, unpredictable and interrelated forces affecting the cranes' estuary environment goes too far. Had the court not applied an erroneous proximate cause rule and instead considered foreseeability carefully, it must have concluded that the unusual die-off of cranes was, in the nearly half century of their population recovery process, a fortuity from the standpoint of TCEQ's water regulation. The situation is similar to Judge Henry Friendly's hypothetical, noted by the Supreme Court in the Exxon case, supra, in which a vessel colliding with a bridge should not be held liable for the death of a patient whose doctor arrived late because of the bridge closing. For these reasons, proximate cause and foreseeability are lacking as a matter of law.

V.

We review a district court's grant or denial of injunctive relief for abuse of discretion. Abraham v. Alpha Chi Omega, 708 F.3d 614, 620 (5th Cir. 2013). A district court abuses its discretion if it "(1) relies on clearly erroneous factual findings when deciding to grant or deny the permanent injunction, (2) relies on erroneous conclusions of law when deciding to grant or deny the permanent injunction, or (3) misapplies the factual or legal conclusions when fashioning its injunctive relief." Peaches Entm't Corp. v. Entm't Repertoire Assocs., Inc., 62 F.3d 690, 693 (5th Cir.1995). As was earlier noted, a plaintiff seeking injunctive relief must show a real and immediate threat of future or continuing injury apart from any past injury. In re Stewart, 647 F.3d 553, 557 (5th Cir.2011). "Past exposure to illegal conduct does not in itself show a present case or controversy regarding injunctive relief." Id. (quoting O'Shea, 414 U.S. at 495, 94 S.Ct. at 676). Although past wrongs may help establish the threat of a future injury, they are alone insufficient. O'Shea, 414 U.S. at 495-96, 94 S.Ct. at 676. The district court erred in three ways in granting injunctive relief. First, the relief is based on its failure properly to apply proximate cause and foreseeability to the circumstances of this case. Our reversal of the state defendants' liability commands the vacating of injunctive relief. No further discussion of this error is required. But even if the state defendants' issuance of water use permits had proximately caused the crane deaths, the court erred in claiming a "relaxed" standard for granting injunctive relief, and it erred, under the proper standard, in finding a real and immediate threat of future injury to cranes.

The district court's assertion that there is a "relaxed" standard for granting injunctions under the ESA is true only insofar as the balance of equities will lean more heavily in favor of protecting wildlife than it would in the absence of the ESA. See Amoco Prod. Co. v. Vill. of Gambell, AK, 480 U.S. 531, 545, 107 S.Ct. 1396, 1404, 94 L.Ed.2d 542 (1987). That does not, however, support "the proposition that courts no longer must look at the likelihood of future harm before deciding whether to grant an injunction under the ESA. Federal courts are not obligated to grant an injunction for every violation of the law." Nat'l Wildlife Fed'n v. Burlington N.R.R., Inc., 23 F.3d 1508, 1511 (9th Cir.1994). The court's power to order injunctive relief depends, as in all other cases, on whether plaintiffs have established by a preponderance of the evidence, that there is "a reasonably certain threat of imminent harm to a protected species." Defenders of Wildlife v. Bernal, 204 F.3d 920, 925 (9th Cir.2000). The court's misstatement of the standard represents an obvious abuse of discretion.

An injunction may thus be issued only if future injury is "certainly impending." Babbitt v. United Farm Workers Nat'l Union, 442 U.S. 289, 298, 99 S.Ct. 2301, 2309, 60 L.Ed.2d 895 (1979) (citation omitted); see also Winter v. Natural Res. Def. Council, Inc., 555 U.S. 7, 22, 129 S.Ct. 365, 375-76, 172 L.Ed.2d 249 (2008); Amoco, 480 U.S. at 545, 107 S.Ct. at 1404. The court's only finding regarding future, imminent harm was "that [the plaintiff] has established by a preponderance of the evidence that there is a reasonably certain threat of imminent harm." The court made no subsidiary findings to buttress this statement. The totality of the court's opinion focused almost exclusively on the injury that occurred in 2008-2009 and did not explain how from year to year following that unusually dry winter season the cranes' habitat or the cranes themselves suffer immediate jeopardy. The evidence is to the contrary, showing steadily increasing flocks in the Refuge: peak sizes 237 (winter 2006-2007); 266 (winter 2007-2008); 270 (winter 2008-2009); 264 (winter 2009-2010); 283 (winter 2010-2011); 300 (winter 2011-2012). There is no evidence of unusual crane deaths following 2008-2009; no evidence of dangerously higher salinities or blue crab or wolfberry deficiencies; no evidence of lack of drinking water in the Refuge; no evidence of emaciated birds or extreme behavioral patterns.

To sustain the court's barren findings, TAP contends that the cranes continue to be an endangered species and TCEQ continues to issue some water use permits. These observations are insufficient to show likely, imminent future harm by a preponderance of the evidence. Although the cranes have been endangered for many decades, it is also clear that TCEQ has been issuing permits continuously up until 2010, yet TAP neither alleged nor proved "takes" in any year before or after 2008-2009. Injunctive relief for the indefinite future cannot be predicated on the unique events of one year without proof of their likely, imminent replication.

CONCLUSION

Because the deaths of the whooping cranes are too remote from TCEQ's permitting withdrawal of water from the San Antonio and Guadalupe Rivers, the state defendants cannot be held liable for a take or for causing a take under the ESA. Even if the state defendants should be held liable, the injunction was an abuse of discretion. The district court's judgment is REVERSED.