A total of 37 samples of white-bellied pangolins were collected from the Douala market network. A first sample set included 24 samples from the Ebo forest (population ‘Ebo’), c. 75 km north-east of Douala, as ascertained by circumscribing the supply area that receives bushmeat from Ebo together with interviews with bushmeat sellers (data not shown). A second sample set included 13 samples likely originating from diverse geographic sources, possibly including Ebo (‘altEbo’), according to the same interviews. Samples consisted of fresh and smoked tissues (tongue or muscle) collected from dead animals sold on the market. Sampling was done, after explaining the aim of our study, with the initial agreement of the seller and following the guidelines of the Comité d'Ethique Institutionnel de la recherche pour la Santé humaine de l'Université de Douala (CEI-UDo). All the samples were preserved in 90% ethanol.
Genomic DNA was extracted with the NucleoSpin® Tissue Kit (MACHEREY-NAGEL, Hoerdt, France), following the manufacturer’s recommendations. The elution step was repeated twice in 50 μL BE to maximize DNA yield and concentration. DNA concentration was quantified using the NanoDrop 1000 Spectrophotometer (ThermoFisher Scientific). All the DNA extracts were stored at -20 °C.
Microsatellite markers were developed at Ecogenics, Balgach, Switzerland (https://www.ecogenics.ch/home.html). An Illumina TruSeq nano library was built from a single DNA extract, which was enriched for simple sequence repeat content using magnetic streptavidin beads and biotin‐labeled CT and GT repeat oligonucleotides. The library was sequenced on an Illumina MiSeq sequencing platform using a nano 2 (500 cycles) sequencing chip. The resulting paired-end reads which passed Illumina’s chastity filter were subject to de-multiplexing and trimming of Illumina adaptor residuals. Subsequently, the quality of the surviving reads was checked with the software FastQC 0.117 (http://www.bioinformatics.babraham.ac.uk/projects/fastqc). In a next step, the paired-end reads were merged with the software USEARCH 10.0.240  to reform in silico the sequenced molecule. The resulting reads were screened with the software Tandem Repeats Finder 4.09 . After this process, 8,488 merged reads contained a microsatellite insert with a tetra- or a trinucleotide of at least six repeat units or a dinucleotide of at least 10 repeat units. Primer design was performed with Primer 3 4.1.0 . Suitable primer design was possible in 6,025 microsatellite candidates. Individual tests for amplification and polymorphism of the 48 best primer-designed loci were performed on 14 individuals (from the western central African and Gabon lineages; see ), after which 20 polymorphic markers were retained (Table 1). Overall, eight loci were di-nucleotide, three tri-nucleotide and nine tetra-nucleotide repeats.
We used MULTIPLEX MANAGER 1.2  to optimize a design into four PCR multiplexes (4-6 loci) using four different ABI fluorescent dyes (Table 1). PCR amplifications for each multiplex were carried out in 20 μL reaction mixture containing approximately 20 ng of genomic DNA, 1x Multiplex PCR Master Mix (QIAGEN Multiplex PCR Plus Kit; Qiagen, Courtaboeuf, France) and 0.2 μM of each primer pair. PCR thermoprofiles included an initial denaturation step (95 °C for 5 min), followed by 32 cycles of denaturation (95 °C for 30 sec) – annealing (60 °C for 90 sec) – elongation (72 °C for 30 sec), and a final elongation step (60 °C for 30 min).
The PCR products were run on an ABI 3730 DNA Analyser (Thermo Fisher Scientific) at GeT-PlaGe (Génotoul, Institut National de Recherche Agronomique, Castanet-Tolosan, France; https://get.genotoul.fr/). Allele scoring and final extraction of genotypes were performed in Geneious 9.0.5  with the Microsatellites plugin (https://www.geneious.com/features/microsatellite-genotyping/).
As a geographically coherent population, Ebo (N=24) was used for the validation of our 20 microsatellite loci. We ran the detection of potential scoring errors and null alleles in MICROCHECKER 2.2.3  Deviations from the Hardy-Weinberg equilibrium were tested for each locus with GenAlEx 6.503 . We used a permutation test under GENETIX 4.05.2  to estimate linkage disequilibrium (LD) between each pair of loci (1000 permutations). The Bonferroni correction was applied to each of those statistical procedures.
The number of alleles per loci (Na) and the observed (Ho) and expected heterozygosities (He) were estimated in GenAlEx. Allelic richness (AR) was calculated in FSTAT 126.96.36.199 .
We used the whole sample set (Ebo+altEbo; N=37) to conduct a Principal Coordinates Analysis (PCoA) in GenAlEx with pairwise population matrix unbiased genetic distances  in order to explore genetic variance among individuals sold in the Douala markets.
Values of unbiased probability of identity and probability of identity among siblings (uPI and PIsibs) were calculated in Gimlet 1.3.3 . We used the Multilocus tagging option in GenAlex to detect identical genotypes in our dataset (sub-option ‘Matches’).