Plant DNA Barcoding: rbcL and matK

When a leaf dies or a seed desiccates, its nuclear DNA begins to degrade within days to weeks. The double-stranded nuclear genome is present in just two copies per cell, and once those copies are damaged the information is gone. The chloroplast is different. A single photosynthetically active cell typically contains 50 to 100 chloroplasts, each holding multiple copies of its 120-160 kilobase circular genome. This means a single plant cell can hold a thousand or more intact chloroplast DNA templates, which is why PCR amplification succeeds from dried herbarium specimens decades old or from carbonised seeds recovered from fire scenes.

The chloroplast genome evolves slowly compared to the nuclear genome, which is a forensic double-edged sword. Slow evolution means sequences are conserved enough that universal PCR primers work across the whole plant kingdom without needing to design taxon-specific primers for every new species encountered in casework. But it also means closely related species can share identical or nearly identical rbcL sequences, because they have not yet diverged at that locus. This is the fundamental resolution ceiling.

CBOL resolved the resolution problem partly by requiring two loci together rather than one. rbcL provides reliable genus-level identification across almost all vascular plants and is easy to amplify. matK sits inside the trnK intron and evolves roughly three times faster than rbcL, providing the additional discrimination that lifts genus-level assignments to species-level for many groups. Used in combination, they cover around 72 percent of all land-plant species at species resolution in the global BOLD database, and over 90 percent at genus level.

A forensic botanist receiving a sample for barcoding follows a workflow that mirrors human DNA analysis in structure but differs in the specific reagents and primer sets used. Getting every step right matters: a contaminated extraction can yield a sequence from the analyst's own fingers (human chloroplast DNA does not exist, so this particular error does not apply, but environmental contamination from pollen or soil bacteria can produce misleading results).

1. Sample documentation and subsampling
   Photograph the item in situ, note its condition, and take a sub-sample under clean conditions. A sample as small as a square centimetre of dry leaf or a single seed can be sufficient. Record the sub-sampling as part of the chain of custody record.
2. DNA extraction
   CTAB (cetyltrimethylammonium bromide) extraction is the traditional method for plant material, effective at removing polysaccharides and polyphenols that inhibit PCR. Commercial silica-column kits (e.g. DNeasy Plant) are faster and better suited to casework throughput. For soil-rich samples, an additional bead-beating step and secondary purification may be needed.
3. PCR amplification
   Universal primer pairs for rbcL (e.g. rbcLa-F/rbcLa-R targeting the 550 bp amplicon used by CBOL) and matK (e.g. 3F_KIM/1R_KIM or matK_390F/matK_1326R for angiosperms) are used in separate reactions. Degraded samples may require shorter, internally designed amplicons covering 150-200 bp.
4. Sequencing and quality trimming
   Sanger sequencing of cleaned PCR products yields reads of 600-900 bp. Bidirectional sequencing (forward and reverse primer reads) is standard for casework to confirm ambiguous bases. Sequence-quality trimming removes low-confidence ends before database submission.
5. Database search and identification
   The query sequence is searched against BOLD using the Identification Engine and against GenBank using BLAST. A match threshold of 99-100% identity to a single species is generally treated as a species-level identification; 97-98% identity to a single genus is a genus-level identification. Ambiguous results (multiple species within threshold) require additional markers or morphological corroboration.

The usefulness of any DNA barcode depends entirely on the completeness and accuracy of the reference database. If the species in question has not been sequenced or its record contains errors, the best laboratory work will return a misleading or incomplete answer. Both BOLD and GenBank have known weaknesses that forensic practitioners must account for.

A recurring issue is geographic coverage gaps. Plant families with the highest forensic relevance, such as Cannabis, many Solanaceae, and tropical timber species implicated in illegal logging, have been intensively sequenced. But large numbers of regional landraces, cultivated varieties, and tropical species remain absent from both databases. When a sequence fails to match anything above 95% identity, the analyst cannot conclude the plant is exotic or novel; it may simply be unsequenced. Expert testimony must be explicit about this ceiling.

The quality of a database identification also depends on accurate taxonomy in the database itself. GenBank has known instances of misidentified reference sequences deposited under incorrect species names. Cross-referencing against BOLD, which requires a specimen voucher, and against published regional flora treatments gives the analyst a defensible basis for the identification in court.

Across the broad sweep of vascular plants, rbcL plus matK resolves about 72 percent of species queries to species level in the BOLD database, according to the foundational CBOL validation study. That number sounds good until you ask what happens with the remaining 28 percent. Some fall in groups where speciation was rapid and recent, leaving no time for the chloroplast markers to diverge between daughter species. Others are hybrids, which may carry the chloroplast of one parent species while morphologically resembling the other.

• Grasses (Poaceae): a large, economically important family with substantial forensic relevance (crime-scene grass, fodder trace from livestock operations). Resolution to species by rbcL and matK alone is poor for many genera; trnH-psbA or ITS is routinely added.
• Oaks (Quercus): extensive hybridisation means chloroplast haplotypes cross species boundaries. Identifying the particular oak species from wood or leaf may require nuclear markers or leaf morphology.
• Cannabis: despite its high forensic profile, Cannabis sativa, Cannabis indica, and hemp cultivars are not reliably separated by chloroplast barcodes; microsatellite or whole-genome sequencing is used for variety-level questions.
• Tropical timber trees: many flagship species under CITES protection (rosewood, ebony, teak) do resolve well at species level, but gaps in reference coverage persist for lesser-known species in the same genera.

Plant DNA barcoding is applied in several distinct forensic contexts. Understanding which context applies to a given case helps the analyst choose the right markers and frame the evidential question correctly from the start.

• Trace evidence and scene linkage: leaf fragments, seeds, or pollen on clothing, footwear, or vehicles can be assigned to a plant species and compared to vegetation at a suspect location. The barcode provides a class-level link; whether that link is meaningful depends on how common or rare the species is in the area.
• Controlled-substance identification: identifying plant material as Cannabis, Papaver somniferum (opium poppy), Erythroxylum coca, or Catha edulis (khat) at species level supports charges that turn on the specific controlled substance. Morphological examination is the first-line method but barcoding resolves ambiguous or heavily processed material.
• Illegal wildlife and timber trade: CITES-listed timber species (e.g. Brazilian rosewood, Dalbergia nigra) can be identified from wood chips, sawdust, or small cross-sections using rbcL alone when the species is well-represented in BOLD. The US Lacey Act and EU Timber Regulation both require proof of species identity in enforcement actions.
• Food fraud and poisoning cases: identifying plant species in food products, herbal medicines, or contaminated feed where morphological characters are destroyed. Substitution of Digitalis lanata for harmless herbs has caused fatalities; barcode-based identification provides the definitive species call.
• Burial context and time-of-death: roots growing through grave fill can be identified to species, contributing to burial-interval estimates alongside entomological evidence. Species identity matters because different plants have known root-growth rates.

In US federal courts and in states that follow Daubert v. Merrell Dow Pharmaceuticals (1993), expert scientific testimony must clear four criteria: the theory has been tested, the error rate is known, it has been peer-reviewed, and it is generally accepted in the relevant scientific community. Plant DNA barcoding using rbcL and matK satisfies all four. The method was published in the Proceedings of the National Academy of Sciences in 2009 by the CBOL Plant Working Group, has been independently replicated across many laboratories, has documented species resolution rates, and is used routinely by herbaria, biodiversity institutes, and customs enforcement agencies worldwide.

In English and Welsh courts, a botanical expert must comply with Criminal Procedure Rule 19, which requires the expert to help the court rather than advocate for the party that retained them, to set out the facts relied on, and to indicate where there is a range of opinion. An expert presenting barcode evidence should state the percentage identity to the best database match, the number of other records within the match threshold, and the completeness of database coverage for the relevant plant family.

India's courts treat plant identification evidence under the Indian Evidence Act 1872, Section 45, which allows expert opinion on science, art, foreign law, and handwriting. A botanical DNA expert qualifies as a science expert. Chain-of-custody documentation linking the sequenced sample to the scene exhibit is particularly important because Indian courts scrutinise exhibit handling carefully, and any break in the chain can lead to exclusion.