SELYA, Circuit Judge.

The "Big Dig" is a massive highway project, built largely with federal funds, which has transformed vehicular travel in the city of Boston. Defendant-appellee Aggregate Industries-Northeast Region, Inc. (Aggregate) supplied concrete needed to construct the project. On various occasions, Aggregate surreptitiously substituted substandard material for the concrete required by its contract specifications.

Certain Aggregate employees, including plaintiff-appellant Joseph Harrington, learned of this chicanery and brought a sealed qui tam action against Aggregate pursuant to the False Claims Act (FCA), 31 U.S.C. §§ 3729-3733. Eventually, the federal government intervened, see id. § 3730(b)(2), and settled the case for several million dollars. The appellant received a percentage of the settlement proceeds. See id. § 3730(d).

A few days after the appellant had signed the settlement agreement in the qui tam action, Aggregate cashiered him. It premised the dismissal on the appellant's refusal to take a drug test. The appellant sued, asserting that Aggregate had stacked the deck, that the stated reason for discharging him was pretextual, and that his ouster was in retaliation for his whistleblowing activities. The district court granted summary judgment in favor of Aggregate, and the appellant now seeks review of that ruling.

As a matter of first impression, we apply a burden-shifting analysis to this FCA retaliation claim. Then, after careful consideration of a scumbled record, we conclude that the circumstances of the appellant's firing are open to legitimate question and that the record, viewed as a whole and in the light most favorable to the appellant, does not warrant the entry of summary judgment. Accordingly, we vacate the order appealed from and remand for further proceedings.

I. BACKGROUND

We sketch the facts and travel of the case, reserving more exegetic detail for our analysis of the issues on appeal. Aggregate supplied large amounts of concrete needed for the construction of the Big Dig. The concrete was supposed to meet certain specifications, and Aggregate pledged that all of its product did. But the appellant (whom Aggregate employed as a truck driver) and others came to doubt this pledge; they insisted that Aggregate frequently cut corners and provided an inferior product.

In June of 2005, the appellant, along with a fellow driver (Donald Finney), filed a qui tam action in the United States District Court for the District of Massachusetts.1 Their complaint named Aggregate Industries, Inc. and several of its subsidiaries as respondents and alleged multiple FCA violations.

The qui tam action proceeded under seal, see id. § 3730(b)(2), and both the appellant and Finney continued working for Aggregate. The seal eventually leaked and, in March of 2007, Aggregate management became aware of the identities of the relators.

As a result of a slowdown in construction activity, the appellant did not work during early 2007. He briefly returned in the spring but underwent a major dental procedure that resulted in more time away from his job. By July, he was physically fit and ready to resume his duties.

Aggregate scheduled the appellant to drive on July 20. According to the appellant, a company representative said "that [he] needed a `return to work' physical and drug test." The appellant disputed this precondition, citing his union contract. Aggregate did not press the matter. It nonetheless continued to insist upon a drug test, this time saying that the appellant was required to undergo such a procedure because he had tested positive for cocaine in 2005.

The appellant again objected, arguing that he had completed his probationary period and was no longer subject to follow-up drug testing on account of the 2005 incident. After the union's business agent interceded, Aggregate backed down and told the appellant to report for work as scheduled.

The appellant arrived at the yard on July 20. A supervisor, John Arsenault, informed him that his name had appeared on a list, generated by a third-party testing firm, for random drug testing and that he had to provide a urine sample. The appellant produced a four-ounce sample for examination by the testing firm. He then proceeded with his normal duties.

When the appellant came to the yard on July 25, he was confronted by two members of management (Frank Bradley and Steve Mikolop). Bradley related that the appellant's urine sample had yielded an inconclusive result ("negative dilute") and that he needed to submit to a second drug test.

Standard practice called for urine specimens to be "split" into two parts so that one split could be tested and the other held in reserve. Rather than acquiescing to the new test, the appellant asked to have the unused "split" tested. Bradley told him that the split was lost and that he could either take the follow-up test or leave the property. The appellant declined to submit to a retest and was not allowed to drive.

At around this same time, the government, negotiating without having formally intervened, reached a settlement of the claims asserted in the qui tam action. As part of the paperwork needed to wrap up the settlement, the appellant—two days after he had refused to take the new drug test—signed a settlement agreement that compensated him for his role as a relator and released any existing claims. On that same day, Aggregate sent the appellant a letter stating that his refusal was deemed the equivalent of a positive drug test and setting out the necessary steps for returning to duty. On July 30, Aggregate discharged the appellant for refusing to submit to the follow-up drug test.

The appellant was not alone in parting ways with Aggregate. His co-relator, Finney, experienced significant harassment beginning in mid-August 2007. Finney's employment was terminated for a scheduling violation in late September.2

In due season, the appellant and Finney sued Aggregate. Each of them alleged that he had been retaliated against for his role as a whistleblower. The pleaded causes of action were whittled down to the FCA's anti-retaliation provision.

Before the completion of pretrial discovery, Aggregate moved for summary judgment. It argued that neither the appellant nor Finney could make out a claim of retaliation. Despite the differences in their situations, the appellant and Finney submitted a consolidated opposition. They resisted summary disposition on the merits and asserted, in the alternative, that the motion should be denied without prejudice under Federal Rule of Civil Procedure 56(d).3

Ruling from the bench, the court denied the motion as to Finney. Thereafter, in a brief written order, it granted summary judgment against the appellant. With respect to the latter ruling, the court stated tersely that the appellant had failed to present evidence of a causal connection between his role as a relator and Aggregate's decision to terminate his employment.

Finney's case proceeded, and further discovery yielded deposition testimony, discussed infra note 7, that the appellant thought was a game changer. The appellant moved for reconsideration of the summary judgment order in light of this new evidence, but the district court demurred. When Finney's suit settled, the district court entered a separate final judgment against the appellant. This timely appeal followed.

II. ANALYSIS

"We review orders granting or denying summary judgment de novo, considering the record and all reasonable inferences therefrom in the light most favorable to the non-moving part[y]." Estate of Hevia v. Portrio Corp., 602 F.3d 34, 40 (1st Cir. 2010). "We will affirm only if the record reveals `that there is no genuine dispute as to any material fact and the movant is entitled to judgment as a matter of law.'" Avery v. Hughes, 661 F.3d 690, 693 (1st Cir.2011) (quoting Fed.R.Civ.P. 56(a)). We are not wedded to the district court's reasoning but, rather, may affirm the entry of summary judgment on any ground made manifest by the record. Houlton Citizens' Coal. v. Town of Houlton, 175 F.3d 178, 184 (1st Cir.1999).

Against this backdrop, we turn to the legal framework underlying the plaintiff's claim. The FCA is a statutory scheme created to forestall fraud against the federal government, Robertson v. Bell Helicopter Textron, Inc., 32 F.3d 948, 951 (5th Cir.1994), by proscribing the presentment of any "false or fraudulent claim for payment or approval" to the United States, 31 U.S.C. § 3729(a)(1)(A). One enforcement mechanism permits a private party to bring a civil action—a so-called qui tam action—in the name of the United States. Id. § 3730(b)(1). The government has the right to intervene and assume control of the action, but it need not do so. See id. § 3730(b)(2). If the qui tam action is ultimately successful (and regardless of who prosecutes it), the private party, known as a relator, gets a percentage of the funds recovered. Id. § 3730(d).

In an effort to prevent companies from discouraging potential relators from coming forward, Congress amended the FCA to include an anti-retaliation provision. The current version of the statute reads:

Id. § 3730(h)(1). The cause of action asserted by the appellant rests on this provision.

The parties to this case operate under the assumption that because the appellant showed no direct evidence of retaliation, the familiar McDonnell Douglas burden-shifting framework, see McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802-05, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973), applies to this retaliation claim. Several district courts have agreed with this construct and have used the McDonnell Douglas framework for retaliation suits brought under the FCA. See, e.g., United States ex rel. Parato v. Unadilla Health Care Ctr., Inc., 787 F.Supp.2d 1329, 1341 (M.D.Ga.2011); United States ex rel. Scott v. Metro. Health Corp., 375 F.Supp.2d 626, 643-44 (W.D.Mich.2005); Mann v. Olsten Certified Healthcare Corp., 49 F.Supp.2d 1307, 1316-17 (M.D.Ala.1999). There are, however, no published decisions on this point at the federal appellate level.4 We fill that void today.

The McDonnell Douglas approach fits comfortably with the test that courts generally apply to retaliation claims under section 3730(h)(1). This test requires an employee to show that (i) he was engaged in conduct protected under the FCA; (ii) the employer had knowledge of this conduct; and (iii) the employer retaliated against the employee because of this conduct. See Mann v. Heckler & Koch Def., Inc., 630 F.3d 338, 343 (4th Cir.2010); Maturi v. McLaughlin Research Corp., 413 F.3d 166, 172 (1st Cir.2005).

In a case such as this, the McDonnell Douglas framework provides a principled mode for analyzing retaliatory intent. See Furnco Constr. Corp. v. Waters, 438 U.S. 567, 577, 98 S.Ct. 2943, 57 L.Ed.2d 957 (1978). We hold, therefore, that the FCA's anti-retaliation provision is amenable to the use of the McDonnell Douglas framework. Cf. Mesnick v. Gen. Electric Co., 950 F.2d 816, 827 (1st Cir.1991) (noting in the ADEA context that "[a]bsent direct evidence, the McDonnell Douglas burden-shifting framework remains the option of choice in retaliation cases").

Adapting McDonnell Douglas to the FCA's anti-retaliation provision, a plaintiff first must set forth a prima facie case of retaliation. Id. Once this is accomplished, the burden then shifts to the defendant to articulate a legitimate, nonretaliatory reason for the adverse employment action. Id. This imposes merely a burden of production, not one of proof. See Collazo v. Bristol-Myers Squibb Mfg., Inc., 617 F.3d 39, 46 (1st Cir.2010). Thus, if the employer produces evidence of a legitimate nonretaliatory reason, the plaintiff must assume the further burden of showing that the proffered reason is a pretext calculated to mask retaliation. Mesnick, 950 F.2d at 827.

This burden-shifting framework is a useful screening device in the summary judgment milieu, but courts typically put it aside once the third step is reached. See, e.g., Collazo, 617 F.3d at 50; Ruiz v. Posadas de San Juan Assocs., 124 F.3d 243, 248-49 (1st Cir.1997). In such circumstances, an inquiring court looks to the record as a whole to determine whether there is sufficient evidence of "pretext and retaliatory animus" to make out a jury question. Mesnick, 950 F.2d at 827. This means that to succeed here the appellant must have adduced sufficient evidence to create a genuine issue as to whether retaliation was the real motive underlying his dismissal. See Collazo, 617 F.3d at 50. We proceed to apply these principles to the facts at hand.

The fact that some of Aggregate's high-level executives learned of the appellant's relator status in March 2007, along with the fact that Aggregate discharged the appellant 72 hours after he signed the settlement agreement, combine to evince a prima facie case of retaliation. Because Aggregate disputes every aspect of the prima facie case, we proceed step by step.

As to the knowledge element, Aggregate argues that the appellant failed to produce evidence indicating that any of its on-site managers were aware of his relator status. In our view, this quantum of proof was not necessary. To clear the low bar required to establish a prima facie case, the fact that high-level Aggregate executives learned of the appellant's whistleblowing several months before his firing suffices to show knowledge. See Gordon v. N.Y.C. Bd. of Educ., 232 F.3d 111, 116 (2d Cir.2000) (explaining, in the Title VII context, that general corporate knowledge is sufficient to satisfy the knowledge element of a prima facie case).

As to the protected activity element, Aggregate asserts that the appellant's execution of the settlement agreement was not protected activity because it was not conduct in furtherance of an action under the FCA. Indeed, Aggregate says, the appellant's execution of the settlement agreement had exactly the opposite effect: it ended the action.

Aggregate's "no protected activity" argument relies on language from our decision in United States ex rel. Karvelas v. Melrose-Wakefield Hospital, in which we stated that "conduct protected by the FCA is limited to activities that `reasonably could lead' to an FCA action." 360 F.3d 220, 237 (1st Cir.2004). This reliance is mislaid. In Karvelas, the court was discussing which actions short of formal qui tam litigation are sufficient to trigger whistleblower protection. See id. at 235-38. This is a matter of concern because, in pre-litigation cases, courts have struggled to place tangible limits on what sort of activity qualifies for protection under the FCA's anti-retaliation provision. See Mann, 630 F.3d at 343-44.

Here, however, the appellant was a relator against Aggregate when the qui tam action was resolved; and the appellant's execution of the settlement agreement was surely conduct in furtherance of that action. The limitations discussed in Karvelas are simply inapposite.

As to the causation element, Aggregate argues that the four-month gap between the time that the company learned of the appellant's relator status and the time that he was fired is too long an interval to support a finding of causation. This argument is belied by the realities of the situation. If Aggregate had dismissed the appellant while the qui tam action was unresolved, it might well have made settlement more difficult. Under these circumstances, we think that a reasonable factfinder could focus on the close temporal proximity between the appellant's signing of the settlement agreement and his ouster. Such a focus is consistent with the notion that, in the context of temporal proximity, courts typically look to the time between protected activity and retaliation. See, e.g., Collazo, 617 F.3d at 50; Calero-Cerezo v. U.S. Dep't of Justice, 355 F.3d 6, 25-26 (1st Cir.2004). So viewed, the temporal span here—72 hours—is sufficiently brief to satisfy the causation element of the appellant's prima facie case. See, e.g., Mariani-Colón v. Dep't of Homeland Sec. ex rel. Chertoff, 511 F.3d 216, 224 (1st Cir.2007) (finding prima facie case of retaliation where plaintiff alleged that approximately two months had transpired between protected conduct and adverse employment action).

Our verification of the existence of a prima facie case does not end our inquiry. Aggregate has proffered a facially legitimate, nonretaliatory reason for discharging the appellant: his refusal to submit to a repeat drug test. The juxtaposition of the prima facie case with this proffered reason makes further use of the burden-shifting framework unnecessary. Instead, we grapple directly with the question of whether a reasonable factfinder could infer that the appellant was fired because he was a relator in a qui tam action, rather than for refusing a drug test. See Collazo, 617 F.3d at 50.

At this stage, we must determine whether the appellant has produced "specific facts sufficient to deflect the swing of the summary judgment scythe." Ahern v. Shinseki, 629 F.3d 49, 54 (1st Cir.2010) (internal quotation marks omitted). In making this determination, "[a] properly supported summary judgment motion cannot be defeated by relying upon conclusory allegations, improbable inferences, acrimonious invective, or rank speculation." Id.

But if a "nonmoving party has produced more than that, trial courts should use restraint in granting summary judgment." Hodgens v. Gen. Dynamics Corp., 144 F.3d 151, 167 (1st Cir.1998) (internal quotation marks omitted). Courts should be especially cautious before granting summary judgment when pretext and retaliatory animus are at issue. See id. "[W]eaknesses, implausibilities, inconsistencies, incoherencies, or contradictions in the employer's proffer[]" can give rise to an inference of pretext. Morgan v. Hilti, Inc., 108 F.3d 1319, 1323 (10th Cir. 1997). So can deviations from standard procedures, the sequence of occurrences leading up to a challenged decision, and close temporal proximity between relevant events. See Hodgens, 144 F.3d at 168-70.

Viewing the record as a whole, we think that the facts underlying Aggregate's efforts to force a drug test on the appellant, along with the temporal proximity between the time that he signed the settlement agreement and the time of his dismissal, create a trialworthy issue about whether Aggregate's proffered reason for firing him was a sham. We explain briefly.

To begin, the ink was still wet on the settlement agreement when Aggregate dismissed the appellant. Such close temporal proximity strongly suggests retaliation. See Collazo, 617 F.3d at 50; Hodgens, 144 F.3d at 168.

There is more. The record casts substantial doubt on whether Aggregate, at several points, was following its own drug testing protocol. For one thing, when the appellant sought to return to work after several days off, Aggregate insisted that he take a drug test. The appellant objected, citing the union contract. Aggregate did not press the matter, and the record contains no explanation for this.

Following its initial botched attempt to get a drug test, Aggregate tried a new approach. This time, it told the appellant that he needed to take a follow-up drug test because of a positive test result in 2005. The appellant again challenged Aggregate's demand and brought the matter to the attention of the union's business agent. Aggregate relented, and there is no clear evidence indicating that it had a right to impose this requirement.5

The events that occurred on the day that the appellant returned to work also lend sustenance to an inference of pretext. On that day, the appellant was singled out for a drug test. Aggregate avers that he was selected at random by a third-party testing company. Coincidences happen, but this sequence of events raises eyebrows. This eyebrow-raising effect is heightened by the fact that Aggregate never produced an affidavit or other statement from its third-party contractor as to when or how the appellant was chosen. Moreover, the record contains a data sheet showing the test results.6 This data sheet did not describe the test as "random," but as a "follow-up" drug test.

The appellant provided a four-ounce urine sample to the third-party testing firm. Standard procedure required this sample to be split so that one specimen could be tested and the other held in reserve. The appellant's test yielded an inconclusive result.

When the appellant arrived at work a few days later, he was greeted by two supervisors. One of them (Frank Bradley) informed him that his drug test had been inconclusive and that he needed to take a repeat test. The appellant demurred, claiming a right to have the reserve split tested instead. When the appellant stuck to his guns, he was ordered to leave the yard.

Aggregate argues that everything about this encounter was on the up and up. Its human resources director, Rick Winter, signed a declaration stating that the company's standard policy was to require a follow-up test after an inconclusive result. But in his deposition, Winter contradicted his earlier declaration; he admitted that the appellant was entitled to have the split tested. When juggling these inconsistent accounts of Aggregate's protocol and construing the record in the light most favorable to the nonmoving party, a reasonable factfinder might well conclude that Aggregate was not following its own procedures and was demanding that the appellant take a new test despite his right to a test of the reserve split.

The whereabouts of the split contributes to the air of mystery. Bradley told the appellant that the split was lost, but the record does not show that Bradley had any firsthand knowledge. The mystery deepens because Winter (Aggregate's Rule 30(b)(6) representative, see Fed.R.Civ.P. 30(b)(6), and its human resources director) testified that he did not know whether the split was lost. He added that standard testing procedures made that eventuality highly unlikely. To compound the uncertainty, the record contains no evidence about the split from the third-party testing firm.

In retaliation cases, the whole is sometimes greater than the sum of the parts. Here, for example, the bits and pieces of evidence recounted above, taken collectively, have significant probative value. After all, irregularities in an employer's dealings with an employee who has fallen out of favor can support a reasonable inference of pretext. See, e.g., Collazo, 617 F.3d at 52; Hurlbert v. St. Mary's Health Care Sys., Inc., 439 F.3d 1286, 1299 (11th Cir.2006); Garrett v. Hewlett-Packard Co., 305 F.3d 1210, 1220 (10th Cir. 2002). Moreover, the fact that Aggregate repeatedly insisted on a drug test for spurious reasons contributes to the inference of pretext. Because the appellant had tested positive for cocaine in 2005, a jury could infer that Aggregate focused on this weak point in his employment history as a convenient way of getting rid of him.

Even so, this is a close case. When looking to the record as a whole, however, we deem summary judgment improvident. Aggregate's adamant insistence on subjecting the appellant to drug testing is pockmarked by irregularities. When this behavior is combined with the appellant's termination immediately following his signing of the settlement agreement, it creates a sufficient foundation for a reasonable inference that the appellant was terminated for retaliatory reasons.7

There is one loose end. Aggregate advances an alternative ground for summary judgment: the release contained in the settlement agreement.

There is no doubt but that the appellant released Aggregate from all claims that he could have asserted up to July 27, 2007 (the date on which the settlement agreement and release were executed). Although the appellant's firing came after that date, Aggregate argues that the conduct underlying the firing occurred beforehand and that, under Massachusetts law, the claim "ar[ose] at the time of the `underlying incident' giving rise to the claim." Eck v. Godbout, 444 Mass. 724, 831 N.E.2d 296, 302 (2005). To bolster its argument, Aggregate notes that the appellant admitted in his deposition that he knew he would be terminated once he refused to take the repeat drug test.

The rule in Eck does not bar the appellant's claim. In light of the release, the appellant's retaliation claim is limited to his termination, and his termination is the underlying incident. Although Aggregate contends that he was effectively terminated when he refused to take the drug test, the letter that Aggregate sent to the appellant on July 27 (the date on which he signed the settlement agreement) belies that characterization. At no point in the letter was there any indication that Aggregate was ending his employment. To the contrary, the letter specifically set out a series of steps that he needed to take in order to return to duty. None of those steps included reapplying for a job. The only indication in the record of when the appellant was terminated is Aggregate's statement that his official termination date was July 30—three days after the release was signed.

To sum up, the appellant was not terminated until after the release was signed. The termination itself is the incident out of which this retaliation action arises. On that basis, the release does not bar the action.

III. CONCLUSION

We need go no further. For the reasons elucidated above, we vacate the entry of summary judgment and remand for further proceedings.

Vacated and remanded. Costs are to be taxed in favor of the appellant.