Border and Cartel Trafficking, 2014-2024

KJKey Judgments

Eight judgments carry this brief. Each is an analytic conclusion, distinct from the reported facts beneath it, and each carries an explicit confidence level per §1.2. A judgment can be wrong even where every underlying fact is correct.

1How to read this brief

1.1 Source grading: the NATO/Admiralty scale

Every source in Annex A carries a two-character Admiralty grade: a letter for the reliability of the source itself, a number for the credibility of the specific information cited. The axes are independent - a highly reliable source can carry a claim that is only possibly true, and a source of unproven reliability can carry information later confirmed elsewhere.

This formalizes the four-tier system used in earlier passes (PRIMARY/OFFICIAL, NAMED INVESTIGATIVE, PEER-REVIEWED, SECONDARY/UNVERIFIED). Rough correspondence: PRIMARY/OFFICIAL typically grades A1-A2; NAMED INVESTIGATIVE and PEER-REVIEWED typically B1-B2; SECONDARY/UNVERIFIED typically B3-C4 or lower.

1.2 Analytic confidence and the probability lexicon

This brief distinguishes reported information (a fact directly stated by a source) from analytic judgment (this document's own conclusion about what the reported information means, which requires interpretation and could be wrong even if every underlying fact is correct). Judgments are labeled Assessment and carry a confidence level, plus a calibrated probability term where expressed as a likelihood.

• High - high-quality information, ideally corroborated by multiple independent sources using different methods; alternatives assessed and found significantly less likely.
• Moderate - credibly sourced but not independently corroborated, or corroborated by sources sharing a common origin, or sound analysis on sourcing too thin for High.
• Low - a single source, fragmentary information, or reasoning this document could not independently verify. Not a weak claim dressed up; an honest statement that the evidence supports no more than this.

Calibrated probability lexicon, where the words themselves carry the meaning rather than contextual tone: almost certainly 95-99%; highly likely / very likely 80-95%; likely / probably 55-80%; roughly even chance 45-55%; unlikely 20-45%; highly unlikely / remote below 20%.

A RECTUMmendation - this project's standing term for an unverified assertion presented as settled fact - is not the same failure as a properly labeled Low-confidence Assessment. The former hides its uncertainty; the latter states it. This brief aims never to do the former.

1.3 What stays a plain fact

Not every claim needs a confidence label. A directly quoted federal statistic, a named prosecution's outcome, or an on-the-record statement is reported information, presented as such, with its Admiralty grade doing the epistemic work. Assessment labels are reserved for interpretive judgments, concentrated in the Key Judgments, §8 (Signals), §9 (KAC), and §10 (ACH).

1.4 Verification history

This brief was built and checked across four separate passes, deliberately using different underlying models as a check against any one model's blind spots:

• Original research and drafting - Claude Sonnet.
• Adversarial verification - a dispatched Fable-model research subagent, given the completed document and instructed to find gaps, verify claims against primary sources, and flag anything thin, wrong, or outdated. That pass produced both new corroboration and several factual corrections to claims already in the document.
• Independent cross-check - OpenAI's ChatGPT, given the same general brief with no access to this document, produced its own reference from scratch. Preserved separately rather than merged, since it was not built to this grading standard. Where it surfaced a claim this document lacked, that claim was independently verified against a primary source before being added.
• Gap-closure pass (September 2026) - three parallel research agents tasked against the outstanding intelligence gaps, with all returned figures independently recomputed locally from the primary datasets rather than accepted on report. Findings marked New or Corrected throughout.

Every correction and addition from every pass was re-verified against its primary source before being incorporated, rather than taken on any model's word.

1.5 Abbreviations used in this brief

Spelled out here rather than on first use, because a reference document is read by jumping to a section rather than straight through. Organizations are given their own name in their own language where that is what the source uses.

2The decade in context: migrant encounters, FY2015-2025

Definitional caution, load-bearing. The left column reports Border Patrol apprehensions of migrants crossing between official ports of entry. It excludes the separate and large count of migrants processed at ports by CBP's Office of Field Operations, including CBP One appointment holders. Conflating the two is the single most common error made with this dataset. Note what the newly obtained OFO column shows: in FY2024 the at-ports flow was 604,482, up from 429,831 the prior year - rising while the between-ports figure fell sharply. A narrative built on the between-ports column alone reads FY2024 as a 25 percent decline; the total encounter figure declined only 14 percent.

The step-change pattern. The four FY2021-2024 fiscal years are the four highest between-ports totals in this fifteen-year series, roughly four to seven times the FY2017 low. The increase does not arrive gradually. It arrives as three step-changes, each landing inside a narrow window around a named, dated policy action:

• FY2020→FY2021, 400,651 to 1,659,206 (>4×), against MPP's enrollment stop. Corrected The operative date is 21 January 2021, not the 1 June 2021 formal termination memo. DHS's own contemporaneous statement is explicit: “Effective tomorrow, January 21, the Department will cease adding individuals into the program.” The June memo formalized a program that had already stopped enrolling four months earlier.
• FY2022 peak and FY2023-24 decline, against Title 42's termination on 11 May 2023 and the CBP One pathway scaling in the same window.
• FY2024→FY2025, 1,530,523 to 237,538 (~85 percent drop), against the administration change and accompanying policy reversal.

Two Title 42 termination events exist in the public record; do not conflate them. The CDC issued an order on 1 April 2022 terminating Title 42 effective 23 May 2022. That termination was blocked by federal court injunction (Louisiana v. CDC) and Title 42 remained in force. It finally ended 11 May 2023, tied to the separate expiration of the COVID-19 national public health emergency.

3Synthetic opioid overdose deaths, 2014-2023

*Category includes fentanyl, fentanyl analogs, and tramadol, per CDC's own definition.

Synthetic opioid deaths rose approximately 1,213 percent from 2014 to 2023. The steepest acceleration was 2015-2021, meaning the sharpest rise in the death toll predates the FY2021-24 encounter surge and was already well underway (2018: 31,335; 2019: 36,359) before 2021. The single highest year (2022) and the first-ever decline (2023) both fall inside the focal period. See KJ-8 and §8 Signal 5: this does not map cleanly onto a “border volume causes overdose deaths” narrative in either direction, and no causal link is asserted here.

4The Sinaloa Cartel: structure, territory, and the 2024 fracture

4.1 Organizational structure

The DEA's 2024 National Drug Threat Assessment describes the Sinaloa Cartel as operating on an umbrella model, not a single centralized chain of command. Four semi-independent factions shared the umbrella as of the assessment:

• Los Chapitos - the four sons of Joaquín “El Chapo” Guzmán Loera (Iván, Alfredo, Joaquín, Ovidio)
• Ismael “El Mayo” Zambada García's faction - Zambada having co-led the organization for over 30 years
• Aurelio “El Guano” Guzmán Loera - El Chapo's brother
• Rafael Caro Quintero - a Guadalajara Cartel founder, recaptured by Mexican authorities in 2022

The model theoretically allows factions to “share resources… without sharing profits or having to answer to a main chain of command.” That theoretical looseness is precisely what fractured in 2024.

Ríodoce, the Culiacán-based investigative outlet, independently reviewed the same assessment and flagged that the four-faction breakdown was the DEA's first formal public acknowledgment that the cartel had fragmented into separate organizations rather than remaining one command structure with internal factions. Ríodoce also noted the assessment's finding that Los Chapitos' publicly claimed 2023 fentanyl-production ban was not reflected in seizure or price data, calling it “a public relations trick.” This is a Mexican outlet independently reading the same primary DEA document - it corroborates the DEA's characterization rather than constituting a separate data source.

4.2 Geographic control relevant to the Arizona/Sonora corridor

The DEA assessment states the Sinaloa Cartel holds “near-total control” over the border region immediately south of Arizona, providing access to the San Luis Río Colorado and Nogales ports of entry, and stages shipments near these crossings for onward US distribution. Los Chapitos were, as of the assessment, actively contesting Caro Quintero's “Caborca Cartel” for dominance of the broader Sonora Desert corridor. A separate large presence operates across the California border, controlling access to San Ysidro and Otay Mesa - San Ysidro being the busiest land border crossing in the Western Hemisphere.

4.3 The 2016-2023 leadership transition

El Chapo was arrested January 2016 and extradited to the United States January 2017. Per DOJ's April 2023 indictment of the Chapitos (S.D. Cal.), they “assumed their father's former role as leaders of the Sinaloa Cartel” following the arrest and extradition, consolidating power partly through violence against rival factions. That indictment reaches back roughly 15 years and, in Attorney General Merrick Garland's on-the-record characterization, describes the resulting organization as “the largest, most violent, and most prolific fentanyl trafficking operation in the world.”

The clearest prior instance of the cartel absorbing a direct state hit and holding together is the Battle of Culiacán, 17 October 2019: Mexican authorities captured Ovidio Guzmán, the cartel mounted an armed response across the city, and authorities released him. Territorial control was not broken.

Ovidio Guzmán López was recaptured 5 January 2023 in a second, more violent confrontation. He was extradited to the United States 15 September 2023 and arraigned 18 September 2023 in Chicago before US District Judge Sharon Johnson Coleman; both dates are confirmed directly against Garland's own statement and ICE/HSI Nogales' own release. Unlike 2019, the 2023 recapture held - a second data point that the cartel's ability to absorb direct state action against its leadership weakened between 2019 and 2023-24.

4.4 The 2024 internal war

• July 2024: Ismael “El Mayo” Zambada and Joaquín Guzmán López (a Chapitos leader) were both taken into US custody. El Mayo's account is that Guzmán López lured him into a meeting that was in fact a setup - a claim that, whether or not fully verified at the time, created a durable rupture. See §10.1 for the ACH on this question and the material that has since resolved it.
• 9 September 2024: large-scale violence erupted in Culiacán in open warfare between Los Chapitos and La Mayiza (El Mayo's faction, reportedly led by his son “Mayito Flaco”) for control of the same corridor described in §4.2.

Sinaloa, homicidio doloso victims, all sex/age/modality rows summed. SESNSP Incidencia Delictiva de Víctimas, Fuero Común, Nueva Metodología. Feminicide tracked separately: 31 (2023), 28 (2024), 69 (2025).

The war signature is unmistakable and lands exactly where it should. August 2024: 44. September 2024: 143. The conflict began 9 September. September-December 2024 totals 656 victims against 163 in the same months of 2023 - a 302 percent increase. Annual: 994 against 531, up 87 percent. Note also that 2025 escalated far beyond 2024 (1,663), which is outside this brief's scope but material to any timeline extending past 2024.

Independent confirmation from a second government body

INEGI, Mexico's national statistics institute, publishes homicide data from death certificates - a wholly different collection method than SESNSP's count of victims in prosecutorial case files. Its Defunciones por Homicidio bulletin (1 August 2025) records:

The state prosecutor's own tally, and disappearances

• 9 Sept - 9 Dec 2024 (Fiscalía General del Estado de Sinaloa, via Infobae): 555 homicides, 695 disappearances (241 subsequently located), 7 femicides - an average of more than six killings daily across ninety days. Material caveat, from the FGE itself: these figures exclude deaths occurring during armed forces operations. The real toll is higher than the state number. This directly confirms the undercount direction assumed in KAC-1.
• 9 Sept 2024 - 9 Feb 2025 (FGE, via La Jornada): 823 homicides, 952 disappearances, 2,100 vehicle thefts.
• Disappearances, independent federal registry: the Comisión Nacional de Búsqueda (RNPDNO), via Ríodoce, records 1,358 people disappeared in Sinaloa during 2024, roughly 50 percent of them during the Chapitos/Mayiza conflict - approximately 679 in the 9 Sept-31 Dec window, closely corroborating the FGE's 695 through 9 December from an entirely separate federal registry. CNB further gives 914 disappearances from 9 Sept 2024 to 13 March 2025 (816 men, 98 women; 67 percent still missing; 66 found dead; 69 minors disappeared, 39 still missing).
• Economic impact (Coparmex business association, via OCCRP/El Universal, 13 Dec 2024): losses exceeding 18 billion Mexican pesos (~890 million USD) and roughly 25,000 formal jobs lost in Culiacán, alongside on-the-ground detail - schools converted to military barracks, an informal 7 p.m. curfew, residents disappeared and unaccounted for.

4.5 Contrast with the 2019 precedent

2019: a direct state action against cartel leadership; the cartel responded with organized force; the state backed down; structure held.
2024: a direct state action against cartel leadership, but enabled or precipitated by internal betrayal rather than external law enforcement action alone - and the structure did not hold. This is a genuine break in the pattern and should be treated as a discrete event, not a repeat of 2019.

4.6 Corroborating primary evidence: US State Department security alerts

Four US Embassy/Consulate alerts, fetched directly from mx.usembassy.gov, independently corroborate the corridor and conflict with real-time government documentation:

• 23 June 2020 - Consulate General Nogales, for Caborca and surrounding areas, citing “credible reports of ongoing cartel violence…including heavily armed groups patrolling in that area, with active fighting and attacks among the cartel factions,” and prohibiting US government personnel from Highway 2 between Santa Ana and Sonoyta. This predates the 2024 fracture by several years and documents the corridor as already contested and violent. Read as evidence of a persistent contested corridor, not of the 2024-specific conflict.
• 3 June 2024 - Consulate General Nogales, for San Luis Río Colorado, reporting “the targeted assassination of a high-ranking law enforcement official” and warning of “potential for further confrontations.” Independently corroborated by KAWC public radio, which named the victim as Police Commander Gerardo Camacho, killed alongside a bodyguard outside a veterinary clinic, with two suspects subsequently detained. A genuine correlation: an official alert and a named regional outlet, different institutions and methods, independently confirming the same dated event.
• 12 September 2024 - Consulate Hermosillo, for Culiacán, three days after the war's outbreak: car thefts, gunfire, security-forces operations, roadblocks, burning vehicles, closed roadways, schools closed across four municipalities, Independence Day events cancelled.
• 31 October 2024 - Consulate Hermosillo follow-up, seven weeks later: violence had not subsided, with armed carjackings and robbery specifically targeting US citizens on the Culiacán-Mazatlán Federal Highway 15, and some Mazatlán universities moving evening classes online.

These are official situational reporting, not analysis or casualty tallies, but they confirm from a government source with no stake in any competing framing that the corridor and conflict were serious enough to trigger formal US travel and safety warnings. The September and October alerts are treated as one continuing thread on the same event, not two independent sources.

5Narcotics: supply chain, seizures, and money

5.1 Supply chain

DOJ indicted eight China-based chemical companies and eight named individuals - Guangzhou Tengyue Chemical, Hubei Shanglin Trading, Jiangsu Jiyi Chemical, Tianjin Furuntongda Tech, Wuhan Jinshang Import & Export, Wuhan Mingyue Information Technology, Henan Oumeng Trade, Shanghai Senria New Materials - for trafficking in protonitazene and fentanyl precursor chemicals (piperidone, 1-BOC-4-AP), the ingredients needed to synthesize fentanyl in Mexico after China banned direct fentanyl production in 2019.

Attorney General Garland, on the record: “The global fentanyl supply chain, which ends with deaths of Americans, often starts with chemical companies based in China.” Precursors move from China, directly or through US intermediaries, to Mexico, where the Sinaloa Cartel and CJNG synthesize finished fentanyl in clandestine labs for US distribution.

5.2 Seizure data

CBP nationwide, both components (OFO + Border Patrol), all areas of responsibility.

This is the corridor's most analytically useful contrast. Tucson Sector - the highest-apprehension Border Patrol sector on the entire border in FY2024, at 460,000-plus apprehensions - seized 2,231 pounds of all drugs combined. The Arizona ports of entry in the same geographic footprint seized 11,995 pounds of fentanyl alone, roughly five times as much of a single drug. Whatever else is true about the corridor, drug volume moves through the ports, not between them.

Other 2024 detail retained: a single Lukeville, Arizona port seizure of approximately 4 million fentanyl pills, described by CBP as historic in scale; Nogales described by Miller as the location seeing “the most fentanyl in the nation,” with 50 percent of cargo screened by non-intrusive inspection technology; and roughly 50,000 pounds of fentanyl seized border-wide over the two years preceding October 2024.

Independent federal corroboration of the enforcement-volume trend, from two agencies outside CBP. The US Sentencing Commission's own data shows fentanyl trafficking cases sentenced in federal court rising from 1,023 (FY2020) to 1,533, 2,366, 3,085 (FY2023), and 3,639 (FY2024) - a 255.7 percent increase since FY2020. Separately, GAO-25-107667 found DHS components conducted approximately 15,000 fentanyl-related seizures FY2021-FY2024, totaling about 99,500 pounds of fentanyl and 61,100 pounds of precursor chemicals - while also finding that DHS “has not established a program to collect data and develop measures to assess the effectiveness of efforts to combat fentanyl trafficking into the U.S., as required by law.” GAO's finding is a direct, sourced caution against over-reading any single agency's seizure totals, CBP's included, as a measure of enforcement effectiveness.

5.3 Who is actually smuggling it

Per CBP's own data as cited by Acting Commissioner Miller, the substantial majority of fentanyl smuggling occurs through official ports of entry, not in the spaces between them, and is conducted primarily by US citizens, not by migrants crossing outside official crossings. This complicates the popular framing of fentanyl smuggling as a function of illegal border crossings: the two phenomena - migrant encounters between ports, and fentanyl smuggling through ports - are substantially different in mechanism and in who is involved.

5.4 Methamphetamine, cocaine, and Nogales-specific detail

Per the DEA's 2024 NDTA: methamphetamine and other synthetic stimulants are attributed to roughly 30 percent of drug-related deaths, and nearly all methamphetamine sold in the United States today is manufactured in Mexico, purer and more potent than in prior years, with the Sinaloa and Jalisco cartels named as controlling much of the production, transportation, and smuggling infrastructure generally. Cocaine for the US market is predominantly a South American (Colombian) supply chain that Mexican cartels transport rather than manufacture - and with the CBP dataset now in hand, the FY2024 volume is no longer an open question: 68,215 pounds seized nationwide, down 16 percent from 81,085 in FY2023. Note that cocaine outweighed fentanyl seizures by more than three to one in FY2024, a proportion easily lost in fentanyl-dominated coverage.

At the Nogales ports specifically, two named consecutive-day October 2024 seizures document multiple drug types moving through the same corridor by different concealment methods:

• 13 October 2024: over 650,000 fentanyl pills and approximately 180 pounds of methamphetamine, concealed in a hidden floor compartment of a bus.
• 14 October 2024: over 158,000 fentanyl pills, 1 pound of heroin, and 20 pounds of methamphetamine, concealed in car doors, seats, and speakers.

A commercial bus and a private vehicle, on consecutive days, at the same ports - consistent with the DEA's description of Nogales as a high-volume, multi-substance staging point.

5.5 Money laundering: the Chinese underground banking connection

Per a DEA release of 18 June 2024, a federal indictment (“Operation Fortune Runner”) charges Los Angeles-based Sinaloa Cartel associates with laundering more than 50 million dollars in drug proceeds through Chinese underground money-laundering networks, October 2019 through October 2023.

The alleged mechanism is elegant and worth understanding precisely. Wealthy Chinese nationals seeking to move money out of China face a strict capital control - a 50,000-dollar annual limit - which creates demand inside China for US dollars already sitting in the United States. The cartel network supplied that demand directly: US drug proceeds, in cash, were delivered to Chinese money-exchange operators in the Los Angeles area for a fee, using trade-based laundering, structured deposits kept under federal reporting thresholds, cryptocurrency, and the purchase of luxury goods and property shipped to China as alternate value transfer. In exchange, the Chinese operators deposited equivalent value in renminbi into mainland accounts on behalf of their own clients, while the cartel received clean US dollars domestically. Neither side's money ever crossed the Pacific.

Lead defendant Edgar Joel Martinez-Reyes, of East Los Angeles, allegedly coordinated the scheme after meeting Sinaloa Cartel members in Mexico in January 2021; he and 23 co-defendants face mandatory minimum 10-year sentences if convicted. DEA Administrator Anne Milgram, on the record: “Laundering drug money gives the Sinaloa Cartel the means to produce and import their deadly poison into the United States.”

This connects directly back to §5.1: the same Chinese financial and chemical-supply networks that source fentanyl precursors to the cartel are, per this separate indictment, also the mechanism by which the proceeds those precursors generate are laundered. The same relationship appears on both the supply side and the money side, in two separately prosecuted matters. See §8 Signal 4.

6Human trafficking and the unaccompanied-children pipeline

6.1 Who actually runs it

This is the most important complicating finding in the brief. The popular assumption - that large cartels run human trafficking and smuggling top to bottom the way they run drug trafficking - is not what the evidence shows.

InSight Crime's dedicated field investigation found:

• Small, clan-based criminal groups, often family organizations, run most human trafficking and smuggling networks along the border, frequently by directly paying off or collaborating with corrupt local officials.
• Where large cartels are involved, their role is usually one of two things: taxing smuggling traffic through their territory via extortion, or supplying victims into a pipeline run by someone else. Direct, total operational control by a cartel is, in the investigation's own words, rare.
• The specific dynamic varies by corridor and region; there is no single border-wide model.
• The investigation itself flags a “serious dearth of data” limiting corridor-by-corridor mapping - meaning specific claims about exactly which group controls trafficking through Nogales/Sonora were not available at this brief's sourcing standard.

A second, independently produced finding points the same direction. RAND Corporation (2019), verified against its own primary publication rather than secondary coverage, states plainly: “Little evidence exists that drug-trafficking TCOs engage directly in human smuggling.” RAND found a spectrum of operators - independent individuals, ad hoc groups, looser or more formal networks - most of which are not cartels or cartel-affiliated. Where cartels profit, RAND's finding is specific: they impose a piso, a transit tax, on migrants and smugglers moving through their territory, and sometimes coordinate crossings to divert law enforcement attention or coerce migrants into carrying drugs - rather than running the operation. RAND estimated total 2017 smuggling revenue from Guatemala, Honduras, and El Salvador combined at 200 million to 2.3 billion dollars, with the piso specifically at 30 to 180 million dollars - a real but minority share under RAND's own range.

Two institutions, two entirely different methods (field investigation and source reporting versus quantitative revenue modeling and case-data analysis), no citation relationship, same conclusion. See §8 Signal 1.

6.2 A concrete case between “cartel” and “lone smuggler”

The “La Mesa” network, based in Yuma, Arizona, operated on behalf of Los Rusos, a Mexico-based transnational criminal organization - not one of the major cartels. Eighteen members were sentenced in 2024. Per DOJ's own release:

• The network used stash houses, drivers, and scouts to smuggle hundreds of undocumented migrants across the border.
• Some drivers used were juveniles.
• From 2022-2024, members used kidnapping, armed robbery, and extortion against the people they were smuggling - including a documented January 2023 incident in which five members kidnapped and filmed the assault of a man, and a mid-2023 incident in which a woman was kidnapped and threatened with torture or death for failing to deliver smuggled individuals as arranged.
• Sentences ranged from time served to 90 months; leader Carlos Moreno-Serrano received the longest.

This sits at exactly the scale InSight Crime describes: a named TCO below cartel scale, with real organizational structure and real violence, distinct from both a lone smuggler and a cartel-run pipeline.

6.3 Unaccompanied minors: the full pipeline

This subsection follows one population - unaccompanied children, “UC” in federal usage - through five stages: scale and demographics, screening at the border, custody and sponsor placement, post-release monitoring, and what happens after release. At each stage, multiple independent sources are cited where available, and convergence or contradiction is stated directly.

6.3.1 Scale and demographics

• FY2024: 99,419 unaccompanied children released to sponsors. FY2023: 113,602 - meaning FY2024 is a real decline, consistent with the broader encounter decline in §2, though ORR release counts and CBP encounter counts are not the same measurement and must not be treated as interchangeable.
• Time in HHS custody before release: 30 days average in FY2024, up from 27 in FY2023. Historically this has ranged 30-69 days over the past decade, meaning FY2024 sits at the fast end of the historical range.
• FY2024 demographics: 61 percent male, 38 percent female. By age: 33 percent 17 or older, 31 percent 15-16, 12 percent 13-14, 24 percent 12 or younger. 87.95 percent Spanish speakers.
• Post-release services referral rate rose sharply: 59.5 percent (67,620 of 113,602) in FY2023 to 82.1 percent (81,588 of 99,419) in FY2024. A genuine, documented improvement in the same year overall volume declined - worth stating as its own fact rather than folding into the volume story.

ORR referrals (not CBP apprehensions), by fiscal year. FY2020 omitted for space (15,381 total, COVID-suppressed). Cross-validates against CRS Table 1 referral totals exactly for FY2019-FY2022.

Mexico rose from 1.4 percent of referrals in FY2021 to 20.3 percent in FY2024 - a fourteenfold share increase, and in absolute terms from 1,690 children to 19,971. Guatemala fell from 47.3 to 32.1 percent; “other” countries nearly tripled from 6.6 to 19.4 percent. CRS states the framing directly: “Children from the Northern Triangle countries - El Salvador, Guatemala, and Honduras - and increasingly from other countries dominate what was previously an almost entirely Mexican migrant flow.” The data through FY2024 shows that trend partially reversing on the Mexican side within the focal period. For a document grounded in the Sonora corridor specifically, this is the most locally relevant demographic fact in the brief.

6.3.2 Screening at the border itself

Ninety-seven recently deported Mexican unaccompanied minors were interviewed in Nogales and Matamoros shelters (Coulter et al., SAGE Open, 2020; University of Arizona, UTEP, NAU). Fieldwork was conducted 2016-2017. This remains the most specific, methodologically documented dataset located for the Nogales corridor, and should be read as documenting a structural pattern in CBP processing rather than a 2024 data point.

Demographic profile: 87 percent male, average age 15.6, 78 percent had family already in the United States.

• 31 percent reported having engaged in border-specific labor - acting as guides, smugglers, or drug transporters - before apprehension.
• 57 percent reported never being screened for trafficking victimization, a screening required under the Trafficking Victims Protection Reauthorization Act.
• 51 percent reported that CBP never explained the forms they were made to sign; 48.7 percent were not screened for fear of return.
• On conduct: 57 percent reported disrespectful treatment, 15 percent verbal abuse, 7 percent physical assault, 7 percent being threatened with a weapon.
• On Flores Settlement conditions: 67 percent reported being kept cold, 23 percent insufficient food, 33 percent of those requesting medical care reported it denied.
• 11 percent of the sample were under 14 - the age below which CBP's own policy presumes a minor “generally incapable” of independent decision-making - and were nonetheless processed for deportation rather than transferred to appropriate custody.

6.3.3 Sponsor vetting, after transfer to HHS custody

HHS OIG's February 2024 audit examined case files from the early-2021 referral surge specifically - the same period identified in §2 as the start of the sharpest encounter increase - making it a direct, government-documented look at how the system performed under exactly the volume conditions described earlier. Findings:

• In 16 percent of case files, one or more required sponsor safety checks had no documentation the check was ever conducted.
• For 19 percent of children released to sponsors with pending FBI fingerprint or state child-abuse-registry checks, the file was never updated with results - meaning it is not established whether those checks ever cleared.
• In 35 percent of files, sponsor-submitted identification had legibility concerns significant enough to flag.
• ORR failed to conduct legally mandated home studies in two identified cases, with four more raising concerns about discretionary home-study guidance.
• In 22 percent of cases, ORR did not conduct timely Safety and Well-Being Follow-Up Calls after release; 18 percent lacked documentation of whether such a call happened at all.
• HHS's Administration for Children and Families concurred with all six recommendations.

6.3.4 Post-release monitoring, and the “failed to appear” problem

DHS OIG's 2024 Management Alert, ICE Cannot Monitor All Unaccompanied Migrant Children Released from DHS and HHS Custody (OIG-24-46):

• ICE transferred more than 448,000 unaccompanied children to HHS custody across FY2019-2023.
• Of those, more than 32,000 failed to appear for scheduled immigration court hearings in that window.
• The OIG's stated root cause is not abduction, trafficking capture, or intentional loss. It is that ICE relies on manual, non-automated processes to track court appearances and maintain address records, and as a direct result “could not monitor the location and status of all unaccompanied migrant children” nor “account for the location of all UCs who were released by HHS.”
• The report's own recommendations are administrative: build automated tracking, and create a formal process to flag and share information on children who miss hearings.

6.3.5 Labor trafficking after release

In February 2023, DOL and HHS jointly announced measures responding to reporting on migrant-child labor exploitation: an interagency task force for information-sharing between labor and immigration authorities; a strategic enforcement initiative targeting high-risk industries and supply chains; mandatory HHS follow-up calls for children reporting safety concerns through the ORR hotline; an expansion of post-release services explicitly targeting 100 percent coverage by 2025, up from roughly 60 percent as of FY2023 (§6.3.1's FY2024 figure of 82.1 percent shows real if incomplete progress); a four-week audit of HHS vetting for sponsors taking in more than one unrelated child; and a request to Congress to raise the maximum civil penalty for child labor violations above the then-current $15,138 per violation.

HHS Secretary Xavier Becerra, on the record: “Every child in this country, regardless of their circumstance, deserves protection.” Labor Secretary Marty Walsh, on the scale: “This is not a 19th century problem, this is a today problem.”

DOL's November 2024 Report to Congress puts a two-year figure on the resulting enforcement: $23,203,878 in civil money penalties assessed across FY2023-2024. Named cases illustrate the range: A1 Meat Solutions ($4.8M in back wages and damages, $221,919 civil penalty, $1M profit disgorgement); Exclusive Poultry ($3.5M in back wages and restitution plus $201,104); Fayette Janitorial Service ($649,000); Tuff Torq Corporation ($296,951 plus more than $900,000 disgorgement).

Cross-reference, peer-reviewed (Dank et al., Northeastern / NYU Marron / Loyola, NIJ-funded): a study of 71 prosecuted child labor trafficking cases involving 132 victims (average age 14) and 145 perpetrators found 58 percent of victims were foreign-born and 42 percent were US citizens - meaning child labor trafficking in the United States is not exclusive to migrant or unaccompanied populations, though migrant children are disproportionately represented relative to their share of the general child population. In 40 percent of cases the perpetrator was a family member of the victim - a detail directly relevant to the sponsor-vetting failures in §6.3.3, since a family-member sponsor is not automatically a safe placement by virtue of the family relationship. That finding and the 89-percent-to-family-member figure in §6.3.1 should be read together.

6.3.6 Synthesis: what the convergence actually shows

Five independent government and academic sources, produced by different institutions using different methods - a peer-reviewed field-interview study, an HHS internal audit, a DHS internal audit, a DOL policy response, and a second NIJ-funded academic study - converge on the same underlying shape. The system's points of failure are not concentrated at any single stage. Screening fails at the border (§6.3.2). Vetting fails at placement (§6.3.3). Tracking fails after release (§6.3.4). And the population most exposed to labor exploitation after release includes, but is not limited to, unaccompanied migrant children (§6.3.5).

This is stronger and more specific than any single report alone: it is not one broken checkpoint. It is a pipeline where each stage has a separately documented, separately caused gap, and the gaps are cumulative rather than redundant - a child missed at one stage is not reliably caught at the next.

7Regional focus: southeastern Arizona

7.1 Jurisdictional overview

Border Patrol's Tucson Sector covers the bulk of this geography, spanning roughly the Yuma Sector boundary west of Ajo through Cochise County in the east - a stretch including the Baboquivari/Sasabe corridor, the Nogales ports of entry, and the mountainous Cochise County borderlands (Huachuca, Mule, and Peloncillo Mountains) running toward New Mexico. Per §4.2, the Sinaloa Cartel holds near-total control of the border region immediately south of this entire stretch, staging through Nogales and San Luis Río Colorado. Everything in this section sits inside that corridor, at the US-side receiving end.

7.2 Cochise County

Tucson Sector FY2024: a 24 percent year-over-year increase in apprehensions, to more than 460,000, even as border-wide apprehensions fell - driven by a record volume of asylum seekers processed through the sector specifically. This is a sector-specific apprehension figure, not the statewide 364-case prosecution count; the two measure different things and must not be conflated. For the sector's seizure figures, now obtained, see §5.2.

Cochise County Sheriff Mark Dannels, on the record, on smuggling-related pursuits: “We deal with ‘got aways,’ these are people that fight and flight, do whatever they can to get away from us… that's where the risk comes in,” and “That's what we see pretty much daily, that kind of activity going on in Cochise County, and it's sad.” Douglas Mayor Donald Huish, on seasonal patterns: “Recently speaking with Border Patrol they anticipate that going up again like it does every fall and winter.”

7.3 A named case in the exact geography: the Huachuca Mountains

The single most geographically specific case located. Per ICE/HSI's release, corroborated by USAO-AZ's own separate release on the same case: Jesús Ernesto Dessens-Romero, a Mexican national, guided five undocumented migrants across the border in February 2021, moving them through the Huachuca Mountains near Fort Huachuca. When a 23-year-old woman in the group became critically ill from the crossing conditions, Dessens-Romero left her behind without calling for emergency help, instead notifying his own smuggling organization; he also pressured survivors to remove missing-person social media posts about her afterward. Her remains were found months later; cause of death was exposure. Cochise County Sheriff's deputies, Border Patrol, and military personnel from Fort Huachuca all participated in search efforts before recovery. He was sentenced to ten years in federal prison on 2 July 2024 on multiple smuggling charges including conspiracy and placing human life in jeopardy. Coordination for the operation was conducted over WhatsApp. The underlying incident is dated 13 February 2021; remains were discovered 20 November 2021.

This case demonstrates, with a named defendant, named location, named victim outcome, and federal sentencing date, that the general patterns described earlier - organized smuggling networks below cartel scale, abandonment of migrants in medical distress, coordination via ordinary consumer technology, and use of remote mountain terrain to avoid law enforcement - are not abstractions in this geography. They happened in the Huachuca Mountains themselves. Two separately issued federal releases, from the investigating agency and the prosecuting office, independently confirm the same case, defendant, and outcome.

7.4 Pima County: prosecutions, victim services, and migrant mortality

DOJ's February 2022 National Strategy to Combat Human Trafficking directed federal grant funding into this geography: Our Family Services, Inc. (Tucson) received $600,000 for housing and services to trafficking victims ages 18-24 in Pima County; the International Rescue Committee (Tucson) received $349,585 for trauma-informed services across Pima, Cochise, and Santa Cruz Counties specifically - the same three-county area this section covers. US Attorney Gary Restaino, on the record: the strategy “provides our office with an opportunity to bolster our commitment to prosecuting human traffickers.”

On mortality, Pima County Medical Examiner Greg Hess is the named authority. Per the Pima County Office of the Medical Examiner's own 2024 Medical Examiner Report: 155 undocumented border crosser (UBC) recoveries were made in CY2024, of which 52 percent (80 of 155) were identified, leaving 75 sets of remains unidentified as of the report. This is more than double the mid-year figure of 64, consistent with the report's own caution that recoveries accumulate through the second half of the year and that a given year's recovery count can include remains of people who died in an earlier year. “Recoveries” in a calendar year is not the same measurement as “deaths” in that year, and this brief uses the ME's own terminology.

More than 4,000 sets of remains have been registered in the county since 2000, with over half identified and returned to families. DHS's own cumulative figure is 9,520 deaths recorded since 1998 across the full southern border. Cochise County separately recorded three bodies discovered within a several-week span in spring/summer 2024 (two in June, one each in April and May). A volunteer search-and-rescue source (James Holeman, Battalion Search and Rescue) anticipated a significant year-over-year increase in bodies located in the Growler Valley specifically, from 30 the prior year to 50-plus.

7.5 Historical material excluded from load-bearing use

A 2019 CBP Tucson Sector release describing operational specifics - 108 rescues, a 134-person group apprehension near Sasabe, an ultralight-aircraft narcotics seizure, a partially built cross-border tunnel discovered near Nogales, and then-Chief Patrol Agent Roy Villareal's description of large-group crossings as a deliberate tactic “to create a vulnerability in operations” - was located and reviewed. It predates the focal period by two to five years and is not used as 2024 data anywhere. Retained only as background on smuggling methodology and infrastructure (tunnels, ultralight aircraft, engineered large-group diversions) with a documented history in this exact sector.

7.6 Santa Cruz County (Nogales)

Nogales is a direct setting, so this subsection brings it to the level of named detail the Cochise and Pima material carries. Load-bearing trafficking findings are separated from civic texture and marked accordingly.

Named federal smuggling cases venued at or investigated through Nogales

• 20 December 2022 - three human smuggling organizers, Ramón Villalobos-López, Javier García-Mondragón, and Daniel López-Guerra, sentenced to a combined 100 months, following an HSI-Nogales surveillance operation on smuggling-linked vehicles beginning April 2021.
• 31 August 2022 - Ian Esteban Serrano and Marco Antonio Romero, both of Nogales, Arizona, sentenced to 45 months each, for coordinating smuggling by “recruiting drivers and sending them to various locations to pick up or drop off undocumented individuals in the Nogales and Rio Rico communities.”
• 14 December 2023 - leadership of the “Malas Mañas” transnational criminal organization indicted on an 11-count federal indictment (alien transport conspiracy, money laundering, drug distribution, firearm brandishing). The investigation involved HSI-Nogales specifically, though the indictment itself was returned in Tucson - a distinction noted rather than blurred.

Operation Plaza Spike (2024)

CBP announced in July 2024 the expansion of a Nogales-based operation (launched April 2024) targeting a specific “plaza” - a cartel-controlled chokepoint directly across from a Nogales port of entry used for northbound drug and southbound weapons trafficking - with 27 similar plaza sites identified across Mexico for the same treatment. A 2024-dated, Nogales-specific operational detail sitting directly inside the corridor-control finding in §4.2.

The DeConcini asylum waiting-list system (2023-2024)

Reporting from AZ Luminaria documents an informal, extralegal waiting-list system operating at the DeConcini port of entry, run at the time by a named aide to the mayor of Nogales, Sonora (Ciria Campoy), who reorganized the list in December 2023 and stated on the record: “Nobody is in charge but me. I'm the first, second, and third.” The same reporting states a predecessor in that informal role “charged up to 15,000 pesos, or nearly $900 dollars, to put a family at the head of the line” - a specific, named allegation of extortion operating parallel to, and separate from, the cartel-run trafficking and smuggling structures described elsewhere in this brief. Tucson Sector Chief Patrol Agent John Modlin, on the bottleneck dynamic: “You may move the bottleneck further down, but eventually it's going to bottleneck again somewhere else.”

Civic and geographic texture

General background, not load-bearing trafficking claims. Nogales, Arizona (population approximately 20,000, the Santa Cruz County seat) and Nogales, Sonora (population over 300,000 as of 2024) form “Ambos Nogales,” Arizona's largest transborder city pairing, with an estimated 30,000 Mexican shoppers crossing north daily. The current permanent wall traces to a barrier built after the 1918 Battle of Ambos Nogales. The city has an extensive cross-border tunnel history: a CBP tunnel-detection specialist stated in 2017 that 111 of the 194 border tunnels found nationwide had been found in Nogales specifically, with the method shifting over time from the 1990s-era open storm-drain/wash system, before security gates sealed it, to small hand-dug tunnels requiring belly-crawling afterward. The three ports serve distinct functions: Mariposa is the primary commercial port, handling roughly 37 percent of all US produce imports from Mexico; DeConcini is the primary pedestrian/passenger-vehicle and tourism crossing, and the site of the informal waiting-list system above; Morley Gate is a pedestrian-only crossing immediately east of DeConcini.

8Signals: genuine cross-source correlation

This section applies a specific standard. A correlated signal requires two or more independently produced sources, from different institutions, using different methods, arriving at the same underlying finding without one citing or reprinting the other. Multiple outlets covering the same event, press release, or study is not correlation - it is one data point reported several times, and it is not counted as two. Where research turned up that kind of coverage, it is named below specifically to mark what it is not.

Signal 1 - Human trafficking is not primarily cartel-run High

Strong, fully verified. Almost certainly true. InSight Crime's field investigation and RAND's quantitative research on Central American smuggling revenue reach the same conclusion through entirely different methods (field investigation and source reporting versus revenue modeling and case-data analysis) and different institutions, with no indication either drew on the other: small clan-based groups and independent networks, not cartels, run most human trafficking and smuggling. Cartel involvement is specific and secondary. RAND puts a number on it - a piso worth an estimated 30 to 180 million dollars in 2017, against RAND's own estimated 200 million to 2.3 billion dollar total. The strongest and most fully verified correlated signal in this brief.

Signal 2 - Fentanyl smuggling is a port-of-entry phenomenon Moderate

Likely true, with an upgrade on one half. The attribution claim (US citizens rather than migrants) rests on Acting CBP Commissioner Miller's on-the-record statement, with Cato's independently published analysis agreeing but excluded from load-bearing use per the advocacy filter. This stays Moderate rather than High because both draw on the same underlying CBP enforcement data rather than two independently collected datasets: what correlates is two institutions independently reading the same public data, not two bodies of evidence.

Upgraded this pass The mechanism claim (volume moves through ports, not between them) is now independently confirmed from CBP's own seizure dataset, computed directly: Tucson Sector's 2,231 pounds of all drugs against 11,995 pounds of fentanyl alone at the Arizona ports. That is a separate body of evidence from the Miller interview, and it settles the mechanism half. USSC sentencing volume and GAO seizure totals do not upgrade this signal - they measure enforcement activity, not who was doing the smuggling. That distinction is worth preserving rather than collapsing.

Signal 3 - The UC pipeline fails at multiple independently documented stages High

Strong, qualitative rather than numeric. Almost certainly true as a pattern, though no single number is claimed. Five sources, five institutions, five methods, no citation relationship: a peer-reviewed field-interview study (screening failure at the border), an HHS internal audit (vetting failure at placement), a DHS internal audit (tracking failure after release), a federal interagency policy response (labor-exploitation risk after release), and a second peer-reviewed study (labor trafficking outcomes). None set out to measure the same number. What correlates is the qualitative shape: every independent look at a different stage of the same pipeline found a documented gap at that stage. The broadest correlation in this brief, spanning the most independent sources of any signal here.

Signal 4 - The cartel-China relationship appears on both the supply and money sides Moderate

Likely true. The DEA's October 2024 precursor-chemical indictment and its separate June 2024 money-laundering indictment are two federal prosecutions filed four months apart, built on different evidence, covering different conduct on opposite ends of the same trade - yet both document the same underlying relationship. Moderate rather than strong because both come from the same investigating agency; but they are separately built prosecutions targeting different conduct, not one case reported twice, so this clears the bar for a real signal.

Signal 5 - Border volume and overdose deaths: tested and found weak Low

Low confidence that a driving relationship exists; unlikely that encounter volume is a primary driver. The two series share a peak year (2022) and a first-decline year (2023), which could read as correlation. But they diverge sharply beforehand: synthetic opioid deaths rose steeply 2015-2020 (5,544 to 56,516) while encounter volume stayed comparatively low and did not begin its sharp rise until 2021. A pattern that shares a peak year but diverges for the six preceding years does not support a “border volume drives overdose deaths” reading, and this brief does not make that claim. Naming a tested-but-weak correlation explicitly is part of following the facts: the honest answer belongs in the record as much as a strong signal would.

Signal 6 - State Department alerts corroborate civilian-impact reporting Moderate

Likely true. InSight Crime's investigative reporting on the Chapitos/Mayiza war and the US Consulate Hermosillo's own security alerts of 12 September and 31 October 2024 independently confirm the same conflict was severe enough to close schools across four municipalities, cancel public events, and prompt an armed-carjacking warning on the Culiacán-Mazatlán highway. Two institutions, two methods (field/source journalism versus consular situational reporting), no citation relationship. Moderate rather than strong because the two alerts are not independent of each other - one institution reporting the same event twice - so only one side counts as a separate institutional confirmation.

Signal 7 - The 2024 Sinaloa casualty escalation High Upgraded from Moderate/Low

This signal changed materially in the gap-closure pass, and the change is KJ-2.

Previously: Infobae, AFP, and OCCRP/El Universal each reported the Sinaloa State Prosecutor's tally at three points across September-December 2024. All three traced to the same single government source, so they confirmed that the number was real, current, and consistently reported - but not its accuracy. Rated a narrow, genuine correlation: Moderate that the tally was real and escalating, Low confidence in the underlying figure's absolute accuracy.

Now: two methodologically independent Mexican federal bodies confirm the escalation directly. SESNSP counts victims in prosecutorial case files; INEGI counts registered deaths from death certificates. These are genuinely different collection instruments run by different institutions, and both show the same sharp 2024 Sinaloa spike against a flat national baseline (SESNSP: 656 Sept-Dec 2024 against 163 in 2023; INEGI: rate 21→33 while national moved 24.9→25.6). Separately, the Comisión Nacional de Búsqueda's disappearance registry (~679 in the conflict window) independently corroborates the state prosecutor's 695 through 9 December from a wholly separate federal registry.

That is three independent government instruments - prosecutorial case files, death certificates, and a missing-persons registry - converging on the same finding. The signal now meets the same bar as Signals 1 and 3 and is upgraded to High. The residual caution is on absolute figures, not on the escalation: the FGE states its own count excludes deaths during armed forces operations, and SESNSP and INEGI differ by 38 for the same year by design.

Named specifically as NOT correlation

To keep the distinction concrete: the following rest on one source each and are not corroborated by an independent second source. They are not weaker facts for that - a well-documented federal case or a named official's on-the-record statement is solid evidence - but they must not be read as correlated signals.

• Pima County's CY2024 UBC-recovery count of 155 - the Medical Examiner's own report, a single primary document. One institution's own count. Hess's own statement that no one has a reliable border-wide count is itself a caution against treating any single mortality figure, his own county's included, as fully corroborated.
• The Nogales October 2024 seizure figures - CBP's own local media release could not be fetched; figures reach this brief via regional news reporting CBP's announcement.
• Operation Plaza Spike details and the DeConcini waiting-list reporting - one named regional outlet each, with named on-the-record sources but no second institution.
• The January 2023 Culiacán casualty figure - SEDENA's own tally, but reaching this brief only through relay, with the aggressor sub-count actively contested (19 official versus 21 documented).

Removed from this list in the gap-closure pass: the Dessens-Romero case, now corroborated by two separately issued federal releases; and the Cochise County 364-case figure, which was not corroborated but corrected - it is a statewide figure, and is regraded rather than retained as a county-level claim.

9Key Assumptions Check

A Key Assumptions Check identifies the load-bearing assumptions this analysis rests on that were not, and in most cases could not be, independently verified. Naming them lets a reader see where an unstated assumption, if wrong, would change a conclusion. None are flagged because they are doubted; they are flagged because they are unverified, which is a different thing.

1. Mexican casualty figures are more likely undercounts than overcounts. Standard conflict-analysis practice: bodies not recovered, disappearances not immediately classified as killings, rural or informal burials outside any official count. Now partially confirmed The FGE's own December 2024 figures explicitly exclude deaths occurring during armed forces operations, which confirms the undercount direction from the source itself rather than from analytic assumption. The assumption survives, better supported than before.
2. Federal agency statistics are not systematically distorted by the political administration in office at time of release, and are graded A1-A2 on that basis. This is the standard assumption underlying PRIMARY/OFFICIAL sourcing; it was not separately audited. A reader who does not share it should mentally downgrade every A-graded federal source to B (usually, not completely, reliable) rather than discard it.
3. Named on-the-record statements quoted by news outlets accurately reflect what the official said, rather than being misquoted, cut, or reframed. Not checked against a recording or transcript in any instance. Note the live example: Miller's FY2023 Arizona fentanyl figure and its stated direction both proved wrong against CBP's own dataset (§5.2). Whether that is a misquote, a misstatement, or a different construction of “Arizona” is unresolved - but it is a concrete instance of this assumption failing.
4. Where multiple outlets independently reach the same conclusion, the convergence reflects genuinely separate reporting rather than one outlet's framing propagating through the others via shared wire copy, stringers, or press-pool access. Checked only loosely, by confirming institutional distinctness (different owners, different countries in several cases); not checked at the level of each outlet's actual sourcing chain.
5. The DEA's four-faction “umbrella model” reflects the cartel's actual command structure at the time of assessment, rather than a simplified model adopted for law-enforcement or prosecutorial clarity. Ríodoce's independent read of the same assessment supports treating this as a genuine finding rather than prosecutorial convenience, but does not fully resolve it.
6. A US consular security alert reflects a genuine, credible threat assessment at time of issue, not a precautionary or liability-driven default. State Department alerts do carry some incentive toward over-caution; treated here as reliable evidence that a threat was assessed as credible, not as proof the threat fully materialized as described.
7. New CBP's published seizure datasets accurately reflect its own enforcement records, and the recomputation performed on them is faithful. The figures were recomputed locally from the raw CSV rather than accepted from any summary, and cross-checked as byte-identical across two independently published vintages of the same series - but the underlying agency data collection itself was not audited, and GAO has separately found DHS lacks a program to assess the effectiveness of its own fentanyl-trafficking efforts (§5.2).

10Analysis of Competing Hypotheses

ACH is a structured technique for evaluating a set of explanations against the same evidence side by side, specifically to guard against anchoring on the first plausible explanation and then collecting only confirming evidence. For each question below, competing hypotheses are listed and the available evidence weighed against each - not only for the leading one.

10.1 What precipitated El Mayo Zambada's July 2024 capture, and what was the US role?

Within the 2014-2024 scope, the only available account was El Mayo's own claim that Guzmán López lured him into a meeting that was a setup. That came from a single interested party - El Mayo himself, facing prosecution, with an obvious incentive to characterize himself as a kidnapping victim rather than a captured cartel leader.

• H1: Guzmán López deliberately and independently orchestrated the abduction as an intra-cartel power move, without US solicitation or advance coordination beyond receiving him once delivered.
• H2: US agencies solicited, induced, or had advance operational knowledge beyond simply receiving Zambada once he crossed into US custody.
• H3: El Mayo's account is substantially self-serving and the actual circumstances were materially different from either H1 or H2.

Material outside this brief's scope, directly bearing on the question: on 1 December 2025, Joaquín Guzmán López pleaded guilty in US federal court and, as part of that plea, confessed under oath to orchestrating the abduction on 25 July 2024 at a ranch outside Culiacán - luring Zambada under false pretenses, having armed men hood and handcuff him, sedating him, and flying him to a private airport in southern New Mexico where DEA agents were waiting. This is a sworn admission by the person with the most direct knowledge and, post-plea-deal, the strongest legal incentive to tell the truth rather than shade it. Strong diagnostic evidence against H3 and for H1's basic mechanics.

The plea agreement's own language states the US government “did not request, induce, sanction, approve, or condone” the kidnapping - direct evidence against H2 in its strongest form. But InSight Crime's analysis of the same plea raises a specific, evidence-based question the plea language does not resolve: DEA agents were positioned at the New Mexico landing site in a way suggesting advance knowledge of the operation's timing and destination, plausibly obtained through separate negotiations with Guzmán López's brother, already in US custody. That is evidence of advance US knowledge - short of H2's stronger claim of solicitation, but not fully consistent with H1's cleanest form either.

10.2 Which faction controls the Sonora corridor adjacent to Arizona?

§4.2 states that per the DEA's 2024 assessment, Los Chapitos held near-total control of the border region immediately south of Arizona and were actively contesting Caro Quintero's Caborca Cartel for the broader Sonora Desert corridor.

• H1: Los Chapitos retained effective control of the Sonora/Arizona corridor specifically through 2024, even as La Mayiza gained ground elsewhere in Sinaloa.
• H2: La Mayiza's post-September 2024 gains extended into this corridor along with the rest of former Chapitos territory.
• H3: Neither faction holds settled control; the corridor remains contested, or control is fragmented among smaller local groups, independent of the conflict's outcome elsewhere.

Material checked specifically for its bearing on this corridor: ACLED (7 May 2025) maps Los Chapitos as controlling Culiacán's city center and Pacific coastal areas while maintaining presence in Sonora and Baja California, and maps La Mayiza (under “Mayito Flaco”) as dominating Durango and rural Sinaloa with additional strongholds also in Baja California, Sonora, and Zacatecas - meaning ACLED's own account has both factions claiming a presence in Sonora rather than one holding it outright. ACLED explicitly names the contested axis as the Culiacán-Mazatlán corridor (Federal Highways 15 and 15D), not the Sonora/Arizona border corridor. Separately, an October 2025 report states La Mayiza controls “up to 90 percent” of formerly Chapitos-held territory - but its own stated geographic focus is central Sinaloa and Culiacán, and it contains no specific claim about the Sonora/Arizona corridor at all.

11Sources rejected, and why

Advocacy and mission-driven organizations were excluded from load-bearing claims regardless of the direction of their conclusions. Where an advocacy source was the only place a claim appeared, the claim was either dropped or re-sourced to a primary document. None ship here on advocacy sourcing alone.

The Cato entry is the cleanest example of the standard working. Two sources point the same direction; the one used is the primary/official one, not the think tank. The claim is not lost - only the weaker citation is.

Two prior rejections have since been reversed, and both reversals are recorded rather than quietly absorbed. The RAND smuggling-revenue finding was once flagged as secondary/unverified, reached only through CNBC's coverage; RAND's own brief and full report were subsequently located and fetched directly, and it is now a confirmed A1 source carrying Signal 1. The congressional “300,000 or more” figure was once excluded as unverifiable; it is now cited directly to OIG-24-46 at approximately 291,000, with the caution attached to what it measures rather than whether it exists (§6.3.4).

12Intelligence gaps

What this brief does not establish, stated plainly. Gaps closed in the September 2026 pass are listed first, so the record shows what moved.

Closed this pass

Still open

1. Port-level trafficking statistics for Nogales. A DeConcini/Mariposa/Morley Gate breakdown of human trafficking as distinct from smuggling does not appear to be published by CBP's Office of Field Operations. Likely requires FOIA rather than further search.
2. A Santa Cruz County-specific prosecution count. USAO-AZ publishes District of Arizona-wide totals only, with no county breakout.
3. A 2024-dated study of TVPRA screening at the border. Systematically searched and not found; Coulter et al. (2020, 2016-17 fieldwork) remains the only fieldwork-based treatment. This is a gap in the public evidence base, not in this search - and it is itself a finding worth carrying: no one has studied border-intake trafficking screening with fieldwork in the last eight years.
4. The ORR Category 2A/2B sponsor disaggregation. Used operationally in ORR policy guidance but not published as a national distribution. ORR's published Category 1/2/3 table covers FY2026 only, outside this brief's window.
5. A government-hosted transcript of the SEDENA January 2023 casualty statement. Every gob.mx and presidential-archive host returned 403 or a proof-of-work bot challenge; no Wayback snapshot exists. Requires a browser session that can clear the challenge, not a fetch.
6. Direct confirmation of the Nogales October 2024 seizure figures from CBP's own local media release rather than regional news reporting of it.
7. Which faction holds the Sonora/Arizona corridor. See §10.2 - no source located, in scope or out, makes a specific current claim about this exact corridor. The highest-value open question for a document set grounded in this geography.
8. 2025 escalation, out of scope but material. A large multi-agency operation reportedly producing over 600 arrests, a September 2025 narcoterrorism charge against senior Sinaloa figures, and SESNSP data showing Sinaloa homicides rising to 1,663 in 2025 against 994 in 2024. If the timeline extends past 2024, this needs its own verification pass rather than folding in as an afterthought.

13Note on the eventual World-Building Document

AAnnex A: Sources cited

Every source referenced above, with its Admiralty grade. Sources added or regraded in the September 2026 gap-closure pass are marked New.