R03 Citation Verification — Batch A (ChatGPT sources)

CCKE PROTOCOL v1.3 | Verified 2026-09-01 | Method: Crossref API, Semantic Scholar, PubMed, Europe PMC, direct PDF text extraction (zlib stream inflation), DuckDuckGo secondary search. All via sandbox.

Summary table

Counts: 8 VERIFIED, 3 PARTIAL, 0 FAILED.

Detail

1. S01 — ERA Economics “California Cannabis Market Outlook” (DCC, 2025)

• IDENTITY: PARTIAL. Cited URL https://cannabis.ca.gov/wp-content/uploads/.../California-Cannabis-Market-Outlook-FNL.pdf returns 404. Working URL: https://cdn.cannabis.ca.gov/wp-content/uploads/sites/2/2025/03/California-Cannabis-Market-Outlook-FNL.pdf (200, application/pdf, 1.6MB). Landing page: https://www.cannabis.ca.gov/posts/california-cannabis-market-outlook/
• CLAIM-SUPPORT: YES — all figures found verbatim in extracted PDF text:
  • “California consumption is around 3.8 million pounds” ✓
  • 2024 licensed capture: table shows 2024 → 54% / 38% / 29% (upper/central/lower per-capita scenarios) → 38% central, 29–54% range ✓
  • Unlicensed production: “the middle of that range is 11.4 million lbs”; “confidence interval for the illicit market is 7 to 16.3 million lbs” ✓ (cite “7.0–16.3M” ok)
  • Eradication share: “between 7.5 and 18.3 percent for 2019–2024” ✓ (also 6.9–16.8% for 2011–2016)
  • Yields: “average yield is close to 0.3 pounds of dry-flower per plant, outdoor yield is roughly 0.5 pounds per plant, and mixed-light and indoor yield is around 0.25 pounds per plant” ✓
  • Uncertainty warnings present (confidence-interval framing; caveats on eradication data reporting gaps) ✓

2. S02 — Polson et al., UC Berkeley CRC final report to DCC (2026)

• IDENTITY: VERIFIED. PDF live (200, application/pdf, 6,829,788 bytes). CRC publication page confirms title, 2026, authors: Polson, Reiman, Dillis, Hossack, LaRosa, Petersen-Rocknet [page’s spelling; likely Petersen-Rockney], Sorgen, Owusu Sarfo. Page: https://crc.berkeley.edu/publication/final-report-to-dcc-licensed-and-unlicensed-cultivation-across-ban-and-permit-jurisdictions/
• CLAIM-SUPPORT: YES — extracted PDF text matches all four attributed findings:
  • “After a post-legalization boom, cultivation levels have returned to 2018 levels” ✓
  • “Licensed farms are the best deterrent for unlicensed cultivation: The presence of a licensed farm best predicts declines in unlicensed cultivation nearby” ✓
  • “Bans have no consistent impact on unlicensed cultivation” ✓
  • “Eradication-oriented enforcement lost its deterrent effect… lost efficacy over time” while “civil regulation showed increasing efficacy” ✓

3. S03 — Hammond et al. 2025, IJDP 142:104828

• IDENTITY: VERIFIED. Crossref: exact title, authors (Hammond, Hong, Rundle, Iraniparast, Kilmer, Wadsworth), IJDP vol 142, p.104828, 2025.
• CLAIM-SUPPORT: YES. Abstract: 12-month period ending Sept 2022, total $6.72B = $5.23B legal + $1.49B illegal, “legal market capture of 78%”; legal expenditure estimate “within two percentage points of the ‘actual’ retail sales data.” All three attributed figures exact.

4. S04 — Bodwitch et al. 2021, J Rural Studies 86:155-170

• IDENTITY: VERIFIED (Crossref exact match).
• CLAIM-SUPPORT: YES. Abstract: “Approximately one third of the 362 cannabis farmers who completed the survey reported that they had never applied for a license”; non-compliant farmers “likely to be smaller cultivators who grew cannabis as part of a diversified livelihood strategy”; barriers “not only financial… but also administrative and psychological ones.” Exact match.

5. S05 — Valdes-Donoso, Sumner, Goldstein 2020, PLoS ONE 15(4):e0232041

• IDENTITY: VERIFIED (Crossref exact match).
• CLAIM-SUPPORT: PARTIAL/plausible. Abstract confirms cost-per-pound Monte Carlo estimates highly sensitive to batch size and failure rate, with California case study; the specific $136/lb @ 8-lb batch/4% rejection and ~10%-of-wholesale figures are scenario results in the full text (not contradicted; consistent with the paper’s design). Treat as supported, magnitude not independently re-extracted.
• CROSS-CHECK RESOLVED (important for editor): Gemini’s DOI 10.3733/ca.2019a0013 is WRONG — it resolves to Crowder, “Science and law enforcement teaming up to help ‘critters’”, Cal Ag 73(3-4):115-116. The correct 2019 California Agriculture testing-cost article is Valdes-Donoso P, Sumner DA, Goldstein R, “Costs of mandatory cannabis testing in California,” California Agriculture 2019;73(3-4):154-160, DOI 10.3733/ca.2019a0014. Its abstract confirms cost-per-pound estimation and that destroyed-batch value dominates total testing costs; the $179/lb figure is that paper’s headline scenario estimate (2018-era regs, smaller batches/higher failure assumptions) — a different paper and scenario than the 2020 PLoS $136/lb. Both figures can coexist; cite each to its own paper.

6. S06 — CA State Auditor Report 2023-116 (2024)

• IDENTITY: VERIFIED. https://www.auditor.ca.gov/reports/2023-116/ returns 200; report content present.
• CLAIM-SUPPORT: YES. Page text (Key Points): “It took local jurisdictions, on average, more than 2.5 years to approve the applications reviewed in this audit” ✓; equity programs “still relatively new, with few equity applicants having received cannabis-related permits” ✓.

7. S09 — Bauer et al. 2015, PLoS ONE 10(3):e0120016

• IDENTITY: VERIFIED. Crossref: exact title and all 7 authors (Bauer, Olson, Cockrill, van Hattem, Miller, Tauzer, Leppig), 2015.
• CLAIM-SUPPORT: YES. Abstract: aerial-imagery plant counts in four NW California watersheds, estimated water demand vs streamflow, “potential to divert substantial portions of streamflow.” Watershed-modeling characterization accurate.

8. S10 — Dillis et al. 2023, Env Res Comm 5:025005

• IDENTITY: VERIFIED (Crossref exact match incl. DOI, vol, article no.).
• CLAIM-SUPPORT: PARTIAL. Abstract confirms: unpermitted dry-season extraction “consistently greater” than permitted ✓. The specific counts (1,787 permitted vs 5,997 unpermitted parcels; 243.06 vs 406.23 ha) are full-text/table values — could not be re-extracted (IOP blocks automated fetch; not in Europe PMC). UNVERIFIABLE at that precision this session; magnitudes directionally consistent with the paper. No red flags.

9. S11 — Thompson et al. 2014 + Wengert et al. 2021

• Thompson: IDENTITY VERIFIED — Conservation Letters 7(2):91-102, DOI 10.1111/conl.12038 (online 2013, issue 2014); authors Thompson, Sweitzer, Gabriel, Purcell, Barrett, Poppenga. CLAIM: YES — abstract: 85% of fisher carcasses tested positive for rodenticides; female fisher survival influenced by number of cultivation sites within home range.
• Wengert: IDENTITY VERIFIED — PLoS ONE 16(9):e0256273, 2021. CLAIM: YES — full text: “locations of 1469 trespass cannabis cultivation sites… 2007 through 2014”; MaxEnt risk maps overlaid with habitat for fisher, Humboldt marten, spotted owl ✓.

10. S12 — Fiering et al. 2026, J Cannabis Research 8:48

• IDENTITY: VERIFIED. Crossref/PubMed: exact title, first author Fiering, JCR vol 8, 2026, DOI 10.1186/s42238-026-00414-y.
• CLAIM-SUPPORT: YES. PubMed abstract: 50 legal + 50 illegal samples ✓; “94% of illegal samples contained pesticides” vs “trace levels (0.01 µg/g) for myclobutanil and dichlobenil in two legal samples” ✓; illegal higher microbial (55% APC / 73% yeast-mold exceedances vs 20% legal) ✓; mycotoxins 12% illegal vs none legal ✓; As/Cd/Pb/Hg higher in illegal ✓.
• NUANCE for editor: chromium was HIGHER in LEGAL samples (~3x peak), and 48% of LEGAL products deviated >20% from labeled THC — avoid overstating “regulated = clean” beyond the attributed claims.

11. S13 — Giordano et al. 2025, Sci Rep 15:20822

• IDENTITY: VERIFIED (Crossref: title, authors incl. Giordano, Bidwell; Sci Rep vol 15, 2025; article 20822 per DOI suffix/abstract record).
• CLAIM-SUPPORT: YES. Abstract: 277 products (178 flower, 99 concentrate) from 52 Colorado dispensaries ✓; 56.7% flower vs 96.0% concentrates within ±15% of label ✓; observed THC significantly lower than labeled in both product types ✓.

Editor action items

1. Fix S01 URL to cdn.cannabis.ca.gov host (original 404s).
2. Replace Gemini’s Cal Ag DOI 10.3733/ca.2019a0013 with 10.3733/ca.2019a0014 ($179/lb paper); keep $136/lb tied to PLoS ONE 2020 scenario ($8-lb batch, 4% rejection).
3. S10 parcel/hectare figures: cite as full-text values, or soften to abstract-supported claim if precision matters.
4. S12: add chromium/legal-THC-labeling nuance if the draft implies legal products are uniformly cleaner.