August 8, 2013
United States of America, Plaintiff-Appellee,
Jonathan Michael Thomas, Defendant-Appellant.
Argued and Submitted April 18, 2013 —San Francisco, California
Appeal from the United States District Court for the District of Arizona D.C. No. 4:10-cr-00628-CKJ-GEE-1 Cindy K. Jorgenson, District Judge, Presiding
Brian I. Rademacher, Assistant Federal Public Defender, District of Arizona, Tucson, Arizona, argued the cause and filed the briefs for the appellant. With him on the briefs were Jon M. Sands, Federal Public Defender, and Richard W. Raynor, Assistant Federal Public Defender, District of Arizona.
Bruce M. Ferg, Assistant United States Attorney, District of Arizona, Tucson, Arizona, argued the cause and filed a brief for the appellee. With him on the brief were John S. Leonardo, United States Attorney, and Christina M. Cabanillas, Appellate Chief, District of Arizona.
Before: Alfred T. Goodwin, Diarmuid F. O'Scannlain, and N. Randy Smith, Circuit Judges.
Reversing the denial of a suppression motion and vacating a conviction, the panel considered the demands of the Speedy Trial Act in the case of a superseding indictment, and explored the emerging parameters for the constitutional use of drug-detection dogs.
The panel held that charges in a superseding indictment not required to be joined with the original charges come with a new seventy-day clock under the Speedy Trial Act, and that the conspiracy count introduced in the superseding indictment in this case was not required to be joined with the count in the original indictment charging possession with intent to distribute marijuana.
Because in accord with then-binding precedent marijuana seized from a tool box attached to the bed of the defendant's truck was not subject to exclusion on the basis of an unconstitutional trespass or physical intrusion, the panel did not decide whether agents violated or implicated the Fourth Amendment by directing the drug-detection dog to jump up and put his paws and nose on the toolbox.
The panel reversed the denial of the suppression motion and vacated the conviction because the government's failure to turn over a full complement of dog-history discovery was error, and that error was not harmless.
Rejecting an alternative basis for invalidating the search, the panel explained that evidence from a trained and reliable handler about alert behavior recognized in his dog can, depending on the facts and circumstances, be the basis for probable cause.
This criminal appeal raises two issues of first impression. We are called upon to consider the demands of the Speedy Trial Act in the case of a superseding indictment as well as to explore the emerging parameters for the constitutional use of drug-detection dogs.
In the early afternoon on February 28, 2010, Jonathan Thomas approached a highway checkpoint in southern Arizona manned by the United States Border Patrol. He was driving a silver pick-up truck with a large black toolbox attached to the bed. Border Patrol Agent Christopher LeBlanc had a partner that day: "Beny-A, " his drug-detection dog, who was trained in the detection of concealed humans and controlled substances. LeBlanc was stationed about fifteen feet in front of a "primary inspection" area. As Thomas's truck passed, Beny-A started to demonstrate what LeBlanc described as "alert behavior." The dog's tail and ears went up, his posture and breathing pattern changed, and he started "air-scenting."
Based on those responses, agents directed Thomas to secondary inspection where he and his three young children exited the truck. Starting at the tailgate, LeBlanc walked Beny-A counterclockwise around the truck. As they encountered areas of interest, LeBlanc signaled Beny-A to go there. The dog was "in odor" throughout, meaning he was very animated and excited. Near the gas tank on the passenger side the dog exhibited more alert behavior. Beny-A is trained to perform what is known in the trade as an "indication" when he discovers contraband: he "rock[s] back into a sit."
When the team came upon the toolbox, LeBlanc cast his hand low-to-high. In response, Beny-A jumped up and placed his paws on the vehicle and pressed his nose against Thomas's toolbox. LeBlanc testified that the dog then tried to sit, but that he did not allow him to complete that trained indication. Next, LeBlanc returned Beny-A to his kennel, obtained Thomas's keys, and searched the locked toolbox. Inside was a blanket and, underneath, bundles of marijuana weighing about 150 pounds. Thomas was arrested, advised of his Miranda rights, and transported to the Tucson Border Patrol station. During interviews with the Border Patrol, Thomas said he had knowingly transported the marijuana but under duress.
Thomas was indicted on a single count of Possession with Intent to Distribute Marijuana, 21 U.S.C. § 841(a)(1), (b)(1)(C), on March 24, 2010. On May 18, 2011, a superseding indictment issued. In addition to renewing the possession offense, the superseding indictment added a charge: Conspiracy to Possess with Intent to Distribute Marijuana, 21 U.S.C. § 846.
Invoking the Speedy Trial Act, 18 U.S.C. §§ 3161, et seq., Thomas sought to have both counts in the superseding indictment dismissed. The court noted that 454 calendar days had elapsed since the original single-count indictment. Of that time, seventy-six days were non-excludable. Because that period exceeded the seventy-day "clock" under the Speedy Trial Act, 18 U.S.C. § 3161(c)(1), the district court entered a dismissal of the possession charge without prejudice. However, the court kept the conspiracy count alive, reasoning that under the Act the government could avail itself of a second seventy-day clock, triggered by the superseding indictment.
Before trial Thomas also filed a motion to suppress evidence of the marijuana obtained at the checkpoint. Thomas, Agent LeBlanc, and K9 Coordinator Paul Dubois testified about the circumstances surrounding the February 2010 search. Thomas pursued two arguments. He claimed that the search of the toolbox had violated the Fourth Amendment because the drug dog's failure to indicate meant probable cause had not been established. And during the suppression hearing, Thomas also objected to receiving heavily redacted training- and performance-evaluation records on Beny-A and his handler. After deciding that these limited disclosures satisfied the government's discovery obligation under Federal Rule of Criminal Procedure 16, the district judge ruled that the government had met its burden of establishing probable cause. Thomas's motion to suppress was denied.
The case proceeded to trial. Thomas took the stand in his defense. Having been unsuccessful in excluding the marijuana, Thomas argued the legal excuse of duress. He did not deny having knowingly possessed the drugs, nor did he contest entering into a conspiracy to distribute them. Instead, he argued that he had been an unwilling courier. Thomas testified that a Hispanic man with a gun had lured him into the desert under false pretenses and threatened to harm him and his family unless he transported the marijuana. After a two-day trial, the jury returned a guilty verdict as to conspiracy—the only count remaining in the superseding indictment. The court sentenced him to thirty months of incarceration, followed by thirty-six months of supervised release. Thomas timely appealed.
First, we consider Thomas's claim that the Speedy Trial Act required the district court to dismiss both counts of the superseding indictment—not simply the possession offense that traced to the original one-count indictment. What distinguishes a superseding indictment from a reindictment is that the former is issued without the original charge first being dismissed. See United States v. Hoslett, 998 F.2d 648, 657 n.11 (9th Cir. 1993). Although the issue Thomas raises is one of first impression, other circuits have confronted it, and there is some prior guidance in our case law as to when a subsequent indictment triggers a new speedy-trial period.
The Speedy Trial Act has three "clocks" that are relevant to this statutory interpretation issue. For the most part, the Act is designed to expedite the trial in that: (i) it "establish[es] a thirty-day limit for filing an indictment after an arrest, " United States v. Wilson, 690 F.2d 1267, 1276 (9th Cir. 1982) (citing 18 U.S.C. § 3161(b)), and—central to this case—(ii) it demands that a defendant have the opportunity to be tried within seventy days of indictment. 18 U.S.C. § 3161(c)(1). Conversely, to safeguard against the risk of undue haste, section 3161(c)(2) prevents the government from insisting that the trial begin "less than thirty days from the date on which the defendant first appears.
We begin by considering the subsection creating the seventy-day clock:
[i]n any case in which a plea of not guilty is entered, the trial of a defendant charged in an information or indictment with the commission of an offense shall commence within seventy days from the filing date (and making public) of the information or indictment, or from the date the defendant has appeared . . . whichever date last occurs.
18 U.S.C. § 3161(c)(1) (emphasis added). The most straightforward reading is that such provision applies to the superseding indictment here. Under that view, the issuance of the new indictment triggered a new seventy-day period in which to bring Thomas to trial. Since the jury was empaneled just over one month after the filing of the superseding indictment, there would be no violation of the Speedy Trial Act.
Matters are not quite so simple, however. We previously have acknowledged that the Act does not expressly provide for superseding indictments. See United States v. Karsseboom, 881 F.2d 604, 606 (9th Cir. 1989). Instead, the Act speaks directly as to reindictments only. On that score, section 3161(d)(1) provides that when the defendant secures a dismissal "both the 30-day trial preparation period and the 70-day speedy trial time period start over." Karsseboom, 881 F.2d at 606. We have also held that when the defendant is reindicted after a sua sponte dismissal by the court, the seventy-day clock is reset. United States v. Feldman, 788 F.2d 544, 549 (9th Cir. 1986).
Thomas argues this new-clock rule, which applies to these reindictment scenarios, has no application to indictments that are superseding. In Karsseboom, we held that some superseding indictments do not restart the original speedy-trial clock. 881 F.2d at 607 ("When a superseding indictment merely corrects technical errors but charges again the same offenses the 70-day clock continues and does not begin anew unless the original indictment in its entirety has been previously dismissed."). However, that case left open whether in a situation such as this, where the new indictment introduces different charges, the seventy-day clock starts running anew.
To resolve this problem (to which section 3161(c)(1) does not provide a plain answer), we think it useful to consider other aspects of the Speedy Trial Act. See Dada v. Mukasey, 554 U.S. 1, 16 (2008) ("In reading a statute we must not look merely to a particular clause but consider in connection with it the whole statute." (internal quotation marks omitted)).
Somewhat analogous to the government obtaining a superseding indictment is the government's choice to have an indictment dismissed and recharged. See United States v. Kelly, 45 F.3d 45, 48 (2d Cir. 1995). In that circumstance, although the initial seventy-day clock still applies, section 3161(h)(5) slows its march by providing that, as "for the same offense, or any offense required to be joined with that offense, any period of delay from the date the charge was dismissed to the date the time limitation would commence to run as to the subsequent charge" should be excluded from the applicable speedy-trial clock. See United States v. Magana-Olvera, 917 F.2d 401, 405 (9th Cir. 1990). This language suggests one of two possibilities for the case at hand. Either separate and distinct charges brought by superseding indictment—which are not "required to be joined"—are meant to be governed by the most demanding timing requirements (both ineligible for exclusion and subject to the initial clock) or, alternatively, they stand on their own and are subject to a new timing period under section 3161(c)(1).
The latter conception is the better interpretation. We believe the Fifth Circuit's discussion in United States v. Alford, 142 F.3d 825 (5th Cir. 1998), to be instructive. As that court recognized, "as to charges that the government is not required to join with the offenses charged in the original indictment, the government may obtain a fresh speedy trial clock by simply waiting until completion of the prosecution for the charges contained in the original indictment and beginning a new prosecution on the additional charges." Id. at 829. To interpret the Speedy Trial Act as Thomas urges, would often relegate the government to that protracted and inefficient procedure at no clear benefit to the defendant who would have to endure trials seriatim. Like the Fifth Circuit, we perceive "no logical basis for concluding that" is what the Act requires. Id. We thus hold that charges in a superseding indictment not required to be joined with the original charges come with a new seventy-day clock under the Speedy Trial Act. See United States v. Rocha-Leon, 187 F.3d 1157, 1159 n.3 (9th Cir. 1999) (explaining that the "interpretation that 'is consistent with the language of the statute and avoids absurd results' is [to be] preferred" (quoting United States v. Alfeche, 942 F.2d 697, 698-99 (9th Cir.1991)).
This conclusion is in line with various out-of-circuit authorities. See, e.g., United States v. Hale, 685 F.3d 522, 537 (5th Cir. 2012) ("[T]he original indictment started one speedy trial clock for the original two charges, and the superseding indictment started a second speedy trial clock for the four new charges."); Alford, 142 F.3d at 829–30 (same); United States v. Lattany, 982 F.2d 866, 872 n.7 (3d Cir. 1992) (concluding that when a superseding indictment "charges a new offense that did not have to be joined with the original charges, then the subsequent filing commences a new, independent speedy trial period"); United States v. Marshall, 935 F.2d 1298, 1302 n.7 (D.C. Cir. 1991) ("If, conversely, the charge in the second count was not one required to be joined with the charges in the first indictment, then the speedy trial clock for that count did not begin to run until [the date the superseding indictment was filed]."); cf. United States v. Roman, 822 F.2d 261, 265 (2d Cir. 1987) (implicitly accepting "that a new 70-day clock begins to run when the superseding indictment contains a completely new charge").
We now apply today's holding to Thomas's case. The government had a new seventy-day clock under the Act so long as the conspiracy count introduced in the superseding indictment was not "required to be joined."
Other courts of appeals have taken that concept as referring to the "joinder required by the Double Jeopardy Clause of the Constitution" and neither party urges a different understanding here. United States v. Novak, 715 F.2d 810, 817 (3d Cir. 1983), overruling on other grounds recognized in United States v. Felton, 811 F.2d 190, 195 (3d Cir. 1987); see also Alford, 142 F.3d at 829; Marshall, 935 F.2d at 1302. As the district court recognized, it is clear that double jeopardy does not require the government to prosecute Thomas for the conspiracy offense in the same trial as for possession with intent to distribute. See United States v. Felix, 503 U.S. 378, 391–92 (1992) (conspiracy to commit drug offense distinct from substantive offense).
We conclude that Thomas's prosecution for Conspiracy to Possess with Intent to Distribute Marijuana complied with the Speedy Trial Act because trial commenced within seventy days of the superseding indictment.
Because the district court was correct not to dismiss the conspiracy count, we turn to Thomas's other pretrial challenge: his argument that the evidence of marijuana should have been suppressed pursuant to the Fourth Amendment and the exclusionary rule. Contained within that argument are three separate assertions.
To begin with, for the first time on appeal, Thomas contends that agents invaded a constitutionally protected area by directing the drug detection dog to "jump up and put his paws and nose on the toobox in the truckbed, and stay like that, refusing to move." This argument is based on the watershed Supreme Court opinion in United States v. Jones, 132 S.Ct. 945, 950–51 & n.3 (2012), as well as on a very recent application of the principle announced in Jones to the use of a drug dog within the curtilage of a home. See Florida v. Jardines, 133 S.Ct. 1409, 1414 (2013).
Second, Thomas argues that even if the dog's touching did not violate the Fourth Amendment in a way compelling suppression, the government failed to disclose adequate evidence of Beny-A's and his handler's proficiency and experience to justify the search of his toolbox.
Third, Thomas claims that even if the complete records about the drug dog establish Beny-A's reliability, his failure to indicate by sitting means that the border patrol agents lacked probable cause to search.
The government claims that it is frivolous for Thomas to contend that the dog's contact with his truck was a Fourth Amendment search. After Jones and Jardines, his argument cannot be so easily dismissed.
Before those cases, relying on a concurrence by Justice Harlan from Katz v. United States, 389 U.S. 347 (1967), courts typically "said that a violation occurs when government officers violate a person's 'reasonable expectation of privacy.'" Jones, 132 S.Ct. at 950 (citation omitted). Jones changed the jurisprudential landscape by holding that this was not the exclusive rubric. Id. at 950, 953. Rather, we were to understand "the Katz reasonable-expectation-of-privacy test [as having] been added to, not substituted for, the common-law trespassory test." Id. at 952; see also id. at 954–55 (Sotomayor, J., concurring).
Case law now directs that if "'the Government obtains information by physically intruding' on persons, houses, papers, or effects, 'a "search" within the original meaning of the Fourth Amendment' has 'undoubtedly occurred.'" Jardines, 133 S.Ct. at 1414 (quoting Jones, 132 S.Ct. at 950 n.3). Applying this principle, the Jones Court held that a de minimis physical intrusion with respect to the exterior of a car (installing a GPS-tracking device) was a search subject to the Fourth Amendment's requirement of reasonableness. 132 S.Ct. at 949, 954. And this past term, the Supreme Court held that the use of "a trained police dog to explore the area around the home in hopes of discovering incriminating evidence" implicated the Jones principle. Jardines, 133 S.Ct. at 1416. It did so because using the dog in that fashion exceeded the terms of the property-law license to "approach the home by the front path, knock promptly, wait briefly to be received, and then (absent invitation to linger longer) leave." Id. at 1415.
"It is beyond dispute that a vehicle is an 'effect' as that term is used in the [Fourth] Amendment." Jones, 132 S.Ct. at 949 (citing United States v. Chadwick, 433 U.S. 1, 12 (1977)). Thus, it is conceivable that by directing the drug dog to touch the truck and toolbox in order to gather sensory information about what was inside, the border patrol agent committed an unconstitutional trespass or physical intrusion.
Yet we need not decide whether it violated or even implicated the Fourth Amendment when agents directed the dog to touch Thomas's vehicle because, as the government correctly observes, not every constitutional violation leads to application of the exclusionary rule—a "judicially created remedy" the only purpose of which "is to deter future Fourth Amendment violations." Davis v. United States, 131 S.Ct. 2419, 2426–27 (2011) ("Exclusion is not a personal constitutional right, nor is it designed to redress the injury occasioned by an unconstitutional search." (internal quotation marks omitted)). As a result of the exclusionary rule's limited aims, the Supreme Court has instructed us not to apply it to "suppress the truth and set the criminal loose in the community without punishment" if the evidence was "obtained during a search conducted in reasonable reliance on binding precedent." Id. at 2427, 2429. This maxim aptly has been dubbed "the 'faith-in-caselaw' exception to the exclusionary rule." Caleb Mason, New Police Surveillance Technologies and the Good-Faith Exception: Warrantless GPS Tracker Evidence After United States v. Jones, 13 Nev. L.J. 60, 66 (2012).
Applying this exception in United States v. Pineda-Moreno, 688 F.3d 1087 (9th Cir. 2012), we determined that the defendant could not rely on Jones to obtain suppression of evidence derived from a GPS tracker on his car because that precedent had not been on the books when the tracker was installed, and because agents had acted "in objectively reasonable reliance on then-binding precedent." Id. at 1090. So too here. In reaching such conclusion, we do not answer the broad question of whether officers were on notice before Jones that physical intrusions or trespassory behavior counted as searches under the Fourth Amendment. That is because authoritative guidance from the Supreme Court allowed the particular actions of Beny-A and LeBlanc. In Illinois v. Caballes, 543 U.S. 405 (2005), the Court held that "the use of a well-trained narcotics-detection dog" on a motor vehicle did "not rise to the level of a constitutionally cognizable infringement." Id. at 409; see also id. at 410 ("A dog sniff conducted during a . . . lawful traffic stop that reveals no information other than the location of a substance that no individual has any right to possess does not violate the Fourth Amendment."). In City of Indianapolis v. Edmond, 531 U.S. 32 (2000), it was explained that "an exterior sniff of an automobile" is permissible, in part, because it "does not require entry into the car." Id. at 40. Holding that "the canine sniff is sui generis, " the Court earlier had held that utilizing a drug-detection dog to examine luggage is not "a search requiring probable cause." United State v. Place, 462 U.S. 696, 706–07 (1983).
Thomas asserts that it would have been objectively unreasonable for Agent LeBlanc to perceive these drug-dog cases from the Supreme Court as blessing his use of Beny-A to find the marijuana stowed in the toolbox of the pick-up. Caballes and Edmond are insufficient in his view, because they approved of officers walking "the drug dog around the car in public space, not to intrude in trespass on the vehicle." At oral argument, though, counsel for Thomas admitted that those cases "didn't say that"—in the sense that neither authority so much as hinted that officers were to avoid contact between the dog's nose and paws and the vehicle's exterior.
This is significant for two reasons. First, it closely parallels what the Supreme Court confronted in Davis when it withheld exclusion on the basis of faith in precedent. There, the Court addressed its prior holding that an officer who had lawfully arrested "the occupant of an automobile . . . may, as a contemporaneous incident of that arrest, search the passenger compartment of that automobile." New York v. Belton, 453 U.S. 454, 460 (1981). As the Supreme Court would later acknowledge, Belton's recitation of the facts did mention that the car's occupant had not been handcuffed or otherwise secured at the time of the search. See Arizona v. Gant, 129 S.Ct. 1710, 1717 (2009). Nonetheless, the Court strongly suggested that it was reasonable to understand Belton as authorizing a search even after the occupant "had been subdued by police." Davis, 131 S.Ct. at 2424. Even farther afield, the key precedent on using a drug-detection dog on a vehicle did not even recount whether the dog had made physical contact during the search it validated. See Caballes, 543 U.S. at 406–10.
Second, the absence of a previously expressed limit along the lines of Jones/Jardines matters because there can be no exclusion "when binding appellate precedent specifically authorizes a particular police practice." Davis, 131 S.Ct. at 2429. It is beyond dispute that as of February 2010, when Agent LeBlanc acted, Supreme Court precedent specifically authorized law enforcement to use a drug-detection dog to seek out illegal narcotics on a validly detained vehicle. See Caballes, 543 U.S. at 409. When affairs are such, "well-trained officers will and should use that tool to fulfill their crime-detection and public-safety responsibilities." Davis, 131 S.Ct. at 2429. "Police practices trigger the harsh sanction of exclusion only when they are deliberate enough to yield meaningful deterrence, and culpable enough to be worth the price paid by the justice system." Id. at 2428 (alterations and internal quotation marks omitted).
Because LeBlanc acted in accord with then-binding precedent, the marijuana seized is not subject to exclusion on the basis of an unconstitutional trespass or physical intrusion.
Separate from the dog's touching, Thomas's second and third arguments for reversing the denial of the motion to suppress identify the agents' opening and investigating of the toolbox as the unconstitutional search.
At fixed Border Patrol checkpoints within the nation's interior, the government can send a motorist for a brief secondary inspection upon "a minimal showing of suspicion, " but then (as is the case with an ordinary traffic stop) it needs probable cause or consent in order to search the vehicle. See United States v. Taylor, 934 F.2d 218, 220–21 (9th Cir. 1991). While he acknowledges that an alert by a reliable drug-detection dog establishes probable cause, Thomas challenges Beny-A's reliability and questions whether the behavior the dog exhibited was a probable-cause-generating "alert."
With respect to reliability, Thomas claims it was improper for the government to have supplied heavily redacted records concerning Beny-A and LeBlanc's training and experience in narcotics detection.
Such records show that LeBlanc and Beny-A had attended yearly certification programs from the Border Patrol and were up-to-date at the time of the search. Biweekly logs, called green sheets, were also produced. The team's performance during eight-hour-controlled evaluations was scored on a scale of one to six—the higher the score, the worse the performance. At least one record analyzed at the suppression hearing revealed marginal performance in "search skills." The team received a 3.50. Had it been one-tenth of a point higher it would have been "a failing score." The redactions obscure comments on nearly every page of the records. As to what is beneath the blacked-out paragraphs, the defendant, district judge, this court, and even the Border Patrol's custodian of records are entirely in the dark.
In United States v. Cedano-Arellano, 332 F.3d 568 (9th Cir. 2003), we held that when a defendant requests dog-history discovery to pursue a motion to suppress, Federal Rule of Criminal Procedure 16 compels the government to disclose the "handler's log, " as well as "training records and score sheets, certification records, and training standards and manuals" pertaining to the dog. Id. at 570–71. These materials were held to be "crucial to [the defendant's] ability to assess the dog's reliability, a very important issue in his defense, and to conduct an effective cross-examination of the dog's handler" at the suppression hearing. Id. at 571. These disclosures are "mandatory" when the government seeks to rely on a dog alert as the evidentiary basis for its search. See United States v. Cortez-Rocha, 394 F.3d 1115, 1118 n.1 (9th Cir. 2005).
A unanimous Supreme Court echoed this in another recent dog-sniff case. See Florida v. Harris, 133 S.Ct. 1050, 1057 (2013). Harris explained that a defendant must be afforded the opportunity to challenge "evidence of a dog's reliability, whether by cross-examining the testifying officer or by introducing his own fact or expert witnesses." Id. "The defendant, for example, may contest the adequacy of a certification or training program, perhaps asserting that its standards are too lax or its methods faulty. So too, the defendant may examine how the dog (or handler) performed in the assessments made in those settings." Id.
In flat contravention of the principle at the heart of Harris and Cedano-Arellano, Thomas's counsel was hamstrung by the fact that the certification records had been so redacted.The K9 Coordinator for the Border Patrol conceded during the suppression hearing that the redactions contained, in part, information about "the training methodology and the techniques used to train the K9s and evaluate the K9s." He also said that if the redactions were lifted, he would expect to see critiques of the team's competence as well as discussions about areas for improvement.
The government makes several arguments why Cedano-Arellano does not apply here. None is availing. First, it argues that because Thomas did not contest that he transported the marijuana—but rather raised the affirmative defense of duress at trial—the discovery was not "material" to his particular "defense." Fed. R. Crim. P. 16(a)(1)(E)(i).
In Cedano-Arellano, however, the defendant did not contest his guilt at all; he entered a conditional guilty plea that reserved only his suppression challenge. 332 F.3d at 570. What is more, Cedano-Arellano explicitly considered subsection (a)(1)(E) when announcing its rule that dog-history discovery ought to be allowed when requested. Id. at 571. As a three-judge panel, we are not free to interpret this rule of criminal procedure afresh. See United States v. Lee, 704 F.3d 785, 790 n.2 (9th Cir. 2012).
Second, the government cites law-enforcement privilege. But, neither the magistrate, nor the district court, relied on that doctrine as a justification for withholding the records. Furthermore, our caselaw calls for an in camera evaluation by the trial court of the government's claim to privilege if the defendant goes beyond "mere suspicion" that the undisclosed evidence will be helpful in his criminal case. United States v. Henderson, 241 F.3d 638, 645 (9th Cir. 2001). As we have nothing but a bare assertion of sensitivity and because there has been no in camera review, we cannot approve of the limited disclosure on this ground.
As a final refuge, the government appeals to the harmless error doctrine. We are persuaded that admission of the marijuana evidence affected Thomas's substantial rights at trial so as not to be harmless beyond a reasonable doubt. See, e.g., United States v. Alvarez, 358 F.3d 1194, 1204 n.3. (9th Cir. 2004); United States v. Stoddard, 111 F.3d 1450, 1458 n.11 (9th Cir. 1997). Thus, the only question is whether the redacted passages would have undermined the reliability of Beny-A so as to defeat probable cause. Unlike in Cedano-Arellano, because we have not been afforded access to the complete records, we cannot say that there is "nothing in those documents that would have changed the ultimate determination that the agents had [probable cause] to support their search." 332 F.3d at 574. The error was thus not harmless under Federal Rule of Criminal Procedure 52(a).
Because it is squarely presented for our review and it will arise again upon remand, we also take up Thomas's claim that whether or not the disclosure was adequate, Beny-A's behavior could not give rise to probable cause. His contention is that even assuming Beny-A and LeBlanc were a reliable duo, because the dog never completed his trained indication—the sitting discussed earlier—his behavior was an insufficient basis for searching the toolbox. Thomas says that the "alert" behavior described by Agent LeBlanc (both at his pre-primary post and during the secondary inspection) consists of "untrained" responses that a dog might exhibit at any time, which fall short of probable cause as a matter of law.
Faced with a similar argument, our sister circuit declined to adopt a rule "which would require the dog to give a final indication before probable cause is established." United States v. Parada, 577 F.3d 1275, 1281–82 (10th Cir. 2009); see also id. at 1275 (upholding as sufficient, the dog's "rapid deep breathing, body stiffening, and upbreaking from the search pattern . . . around a vehicle"). Its rationale was on the mark: probable cause is measured in reasonable expectations, not certainties. Id.; see also United States v. Cervantes, 703 F.3d 1135, 1139 (9th Cir. 2012) ("An officer will have probable cause to search if there is a fair probability that contraband or evidence of a crime will be found in a particular place, based on the totality of the circumstances." (internal quotation marks omitted)). Evidence from a trained and reliable handler about alert behavior he recognized in his dog can be the basis for probable cause. Whether a particular dog displays enough signaling behavior will depend on the facts and circumstances of each case.
The Supreme Court's Harris decision confirms the correctness of this view. 133 S.Ct. at 1055 ("We have rejected rigid rules, bright-line tests, and mechanistic inquiries in favor of a more flexible, all-things-considered approach."). In rejecting Florida's evidentiary checklist for a drug dog's reliability, the Court explained that "[a] gap as to any one matter . . . should not sink the State's case, " because that is "the antithesis of a totality-of-the-circumstances analysis." Id. at 1056.
Thomas's alternative basis for invalidating the search therefore fails.
Because the government's failure to turn over a full complement of dog-history discovery was an error that was not harmless, we reverse the district court's denial of the motion to suppress and vacate Thomas's conviction. On remand, the district court must decide whether the government can establish the dog's reliability in light of an adequate record,  or alternatively, whether it can establish that Thomas freely consented to Agent LeBlanc's search. If either probable cause or consent is shown, then the court may reinstate the conviction. See, e.g., United States v. Bernal-Obeso, 989 F.2d 331, 337 (9th Cir. 1993). If not, a new trial without evidence of the marijuana will be appropriate.
VACATED, REVERSED and REMANDED.