Pangolin Trafficking Geographic Assignment (West & Central Africa)

This dataset, published on Zenodo (record #19286247) and derived from targeted DNA sequencing, documents detection signals from 30,704 pangolin trafficking samples across 8 measured dimensions — a rare window into the forensic genomics of one of the world's most illegally traded mammals. The core data captures predicted spatial coordinates (xpred, ypred) and their kernel-density counterparts (kd_x, kd_y), which together map DNA sequence reads to a 2D reference space; the near-perfect correlation between xpred and kd_x (r²=0.98) confirms the two estimation methods are in tight agreement, lending credibility to the detection pipeline. Most striking is the dist_kd column, which measures each sample's distance from a reference distribution: while the median sits at 46, the range blows out to nearly 2,000 — meaning a meaningful subset of samples look genomically anomalous, potentially representing mislabeled species, contamination, or novel trafficking routes mixing specimens of different origins. With over 1,600 statistical outliers flagged in the position predictions alone, the data suggests a significant fraction of seized pangolins do not fit the expected population profile — a finding with direct implications for conservation genetics and law enforcement efforts to trace trafficking supply chains back to source populations.

• 304 pangolin samples (mostly Phataginus tricuspis) were genetically assigned to geographic origins across 14 West and Central African countries
• Median prediction error is ~40 km, with 54% of all predictions under 50 km — tight enough to identify source forests
• Cameroon dominates with 248 of 304 samples (82%), nearly all sourced from local retail markets
• Samples from Democratic Republic of Congo and Central African Republic show the highest prediction errors (137–160 km), suggesting genetic diversity or sparse reference data
• Trade seizure samples show 47% higher mean prediction error than market samples (87 vs 59 km), possibly reflecting longer-distance trafficking routes

• 304 pangolin samples (99% Phataginus tricuspis) were genetically assigned to geographic origins across 14 African countries using 101 bootstrap iterations each
• Median prediction error is 45.9 km, but ranges from 0.3 km to 1,539 km — some samples cannot be reliably placed
• Samples flagged as seizures (n=11) have significantly worse geographic assignment than market-sourced samples, consistent with long-distance trafficking
• Predicted origins cluster tightly between 8°W–15°E longitude and 1°–7°N latitude, centering on the Gulf of Guinea rainforest belt

The 11 seized samples average 129 km prediction error vs. 58 km for 274 market samples. High error in seized specimens suggests these animals were transported across genetic population boundaries, making their origin harder to pinpoint — a hallmark of cross-border trafficking networks.