The sequencing of the crustacean collection of the MNHN, Paris, constitutes a promising yet very challenging barcoding project. For the collection’s crustacean specimens preserved in ethanol, some of which were collected up to 40 years ago, the conventional COI barcoding procedure of amplification with Folmer primers failed for more than half of the specimens (58%, n = 1920). We hypothesised that this failure may have been due to incompatible mismatches between the crustaceans targeted and the Folmer primer sequences and/or the amount of degradation of the DNA extracted from museum specimens. The comparison of the Folmer primers against the COI sequences from GenBank complete decapod mitochondrial genomes revealed that the annealing regions were, in fact, rather conserved, suggesting that the amplification failures were due more likely to the low quality of the DNA isolated. Using an alignment of all available decapod sequences we designed two internal primers in the middle of the barcoding COI region and also selected two additional external primers to be used as alternative to the standard Folmer primers. Using a two-overlapping-fragments amplification strategy and different primer combinations, our new protocol significantly increased the amplification success rate of the collection material from 42% with the Folmer primers to 84%, recovering an additional 364 complete barcodes and 443 minibarcodes (i.e. fragments of less than 400 base pairs), and expanding the species coverage from 254 to 397 barcoded crustaceans.

Introduction	Go To >> Top Introduction Materials and methods Results Discussion Acknowledgements References

The Muséum national d’Histoire naturelle, Paris (MNHN), has a long tradition of carcinological studies. Leading zoological figures of the 19th century such as P.A. Latreille and H. Milne-Edwards established the MNHN crustacean collection, which has grown into one of the more important international collections in existence today. Forty years of marine expeditions as part of the ongoing Tropical Deep-Sea Benthos program, led by the MNHN in collaboration with the Institut de Recherche pour le Développement, have assured a continuous flow of new material from the Pacific and Indian Oceans into the collection (Bouchet et al. 2008). According to conservative estimates the MNHN crustacean collection contains over 120 000 lots representing at least 20 000 species and includes ~8000 types.

We took advantage of the rich and comparatively recent material obtained during the Tropical Deep-Sea Benthos campaigns to initiate a DNA barcoding project at the MNHN, as part of the international Marine Barcode of Life project started in 2007 (www.marinebarcoding.org) (Bucklin et al. 2011). Two main taxonomic groups, the marine molluscs and the decapod crustaceans, were targeted as part of the MNHN contribution to the MarBOL project. The decapods were specifically chosen as they represent one of the major components of the MNHN crustacean collection. Thanks to the efforts of several international taxonomists, the decapod collection at the MNHN is very well studied. Almost all specimens are identified to species level and numerous new species have been described using this material (among the most recently published papers are, for example, Castro 2010; Komai 2011; Chan in press). This situation offers excellent opportunities to develop a Barcoding reference database, relying on numerous specimens identified to species that are preserved in 75% ethanol. Given its richness in type specimens, it will allow us to test the species-delimitation hypotheses established on morphological grounds (Puillandre et al. 2011).

Despite the potential of the MNHN decapod collection to be an excellent reference source for developing a DNA barcoding database, it soon became clear from our preliminary analyses that this material presented specific challenges in recovering COI sequences. More than half (58%) of the specimens chosen for initial barcoding analysis failed to amplify using the universal invertebrate COI Folmer primers (Folmer et al. 1994). Given that cross-species amplification success is dependent upon the number and position of primer-template mismatches (Housley et al. 2006) and that DNA degradation in museum specimens is a well known factor affecting the chance of recovering amplicons using standard PCR protocols (e.g. Dean and Ballard 2001; Mandrioli et al. 2006; Zimmermann et al. 2008), we formulated three hypotheses that might account for the amplification failures: (1) the Folmer primers are not compatible with decapod crustaceans as a whole; (2) the Folmer primers were suitable only for some groups but not all (a taxonomic bias within the decapods); and (3) despite the relatively young age of the specimens and the preservation conditions the DNA was degraded to the point that most DNA template molecules were of shorter size than the COI barcode fragment. Here we describe the procedure used to test these three hypotheses and to optimise a decapod-specific amplification protocol for collection material using internal and specific primers.

Materials and methods	Go To >> Top Introduction Materials and methods Results Discussion Acknowledgements References

For a preliminary barcoding analysis we selected 1920 specimens representing 449 decapod species obtained mostly in the Indo-Pacific Ocean over the last 40 years and preserved in ethanol in the MNHN collections. The taxa selection is representative of the benthonic decapod diversity in the Indo-Pacific Ocean. We followed the MarBOL workflow methodology established at the MNHN and detailed in Puillandre et al. (2012), as well as the standard protocol of DNA sequencing in Barcoding projects for animals (Hebert et al. 2004). The DNA was extracted from a part of one pleiopod using the NucPrep (AB Applied Biosystem) or the NucleoSpin 96 Tissue (Macherey–Nagel) extraction kits, following the manufacturer’s protocols. A preliminary screening with the Folmer primers (LCO1490 and HCO2198) resulted in only 805 barcodes (42%) from 254 species.

To test the hypothesis that the Folmer primers were not optimal for our target group as a whole, we aligned all COI sequences obtained from the complete decapod mitochondrial genomes available in GenBank. Then we compared the Folmer primer sequences with the COI alignment to verify the degree and pattern of mismatches.

To test a potential taxonomic bias in the primer mismatch, we verified whether any taxon-specific bias was associated with the amplification failure using a Chi-square test. As a null hypothesis we assumed that the proportion of barcodes obtained from each decapod infraorder (sensu De Grave et al. 2009) did not differ significantly from the mean success rate for all decapods.

With a Chi-square test we also verified whether the specimen age affects the barcode recovery rate using the Folmer primers, comparing the proportions of barcodes obtained from two age classes (0–20 years versus older).

Following these tests, all 805 barcodes obtained with the Folmer primers were aligned and the alignment was inspected by eye for conserved regions that were suitable for designing internal primers. The primers were checked for compatibility (primer dimer or hairpin formation and Tm) using the online application NetPrimer (www.premierbiosoft.com/netprimer/index.html). Then we tested the effectiveness of the new primer sets on the decapod specimens that failed to amplify with the Folmer primers.

The PCRs were conducted in 20-μL reaction volume, containing 1–5 ng of DNA and to a final concentration of 1× reaction buffer, 2.5 mm MgCl₂, 0.26 mm dNTP, 0.3 μm of each primer, 5% DMSO and 1.5 units of Qiagen Taq polymerase. For all primer pair combinations the amplification profile was: 5 min initial denaturation at 94°C, 35–40 cycles of 40 s at 94°C, 40 s at 48°C and 60 s at 72°C, followed by a final extension of 5 min at 72°C. PCR products were visualised on a 1.5% agarose gel stained with ethidium bromide and the positive PCRs were sequenced in both directions using the Sanger method.

As a case-study we illustrate the significant input provided by our specific primers in recovering a barcode library for the genus Plesionika Spence Bate, 1888 (Family Pandalidae). Pandalidae is the second most commercially important family of prawns after the Penaeidae. While the fishery for Penaeidae is mostly confined to the tropics and subtropics, that for Pandalidae is located in colder waters of both the Northern and Southern Hemispheres. All barcodes for Plesionika were aligned with Bioedit 7.0.9.0 (Hall 1999). The Neighbour-Joining tree was generated with PAUP* (Swofford 2003) using the Kimura 2-parameter nucleotide-substitution model. The tree was rooted using three Pandalus species obtained from GenBank (accession numbers indicated in Fig. 1). The full specimen details and BOLD/GenBank accession numbers are supplied in Appendix 1.

Fig. 1.  Schematic representation of the primer positions on the COI gene. The amplicon lengths generated with the different combinations of internal primers are indicated below. The regions of the complete mitochondrial genomes alignment corresponding to the primer annealing sites are shown above, with the corresponding primer sequence at the bottom. For the three reverse primers the reverse complement sequence is shown. The two-letter code before the generic names indicates the major decapod subdivisions: Achelata, Anomura, Astacidea, Brachyura, Caridea, Dendrobranchiata. For improving the clarity some congeneric sequences have been omitted.

Results	Go To >> Top Introduction Materials and methods Results Discussion Acknowledgements References

The alignment of COI sequences from the 28 complete mitochondrial genomes available in GenBank revealed that this gene is structurally conserved in decapods. It has a total length of 1534–1544 bp without any indels. The Folmer primers align well with the COI sequences, having, on average, only 4.1 and 3.3 mismatching bases, in the forward and reverse primers, respectively (Fig. 1, primers LCO1490 and HCO2198).

We did not detect any taxonomic bias in recovering the barcodes with the Folmer primers. While amplification success ranged between 26 and 66% across the different decapod infraorders (Fig. 2), the differences were not statistically significant (χ² = 13.44, d.f. = 7, P > 0.05). Conversely, sample age has a significant effect: we recovered barcodes with the Folmer primers from 45% of the more recent samples (<20 years) but from only 35% of the specimens collected >20 years ago (χ² = 12.00, d.f. = 1, P < 0.01).

Fig. 2.  Proportion of complete barcodes and minibarcodes recovered for each decapod subgroup. The sample size is indicated in parentheses. The heading ‘Others’ includes the few specimens of Stenopoidea (n = 3) and Thalassinidea (n = 14). Light grey: barcodes obtained with the Folmer primers; dark grey: complete barcodes amplified with the specific primers; black: minibarcodes; white: missing barcodes, either full or mini.

We designed two new internal primers specific for decapods (Table 1 and Fig. 1). One forward (COI-IntF1) and one reverse (COI-IntR1) primer are located near the centre of the barcoding region and they are intended to pair with the opposite Folmer primer. The barcode fragment of the COI gene is thus split into two overlapping fragments of ~350–400 bp each.

Table 1.  Primer sequences used in this study with their annealing temperatures (Tm, in °C)

We observed that although the Folmer primers appear to be conserved within decapods as a whole, occasional mismatches near the 3′ end of the primer might lower the annealing strength and significantly reduce the amplification success. Therefore we designed a new forward primer (LCOI-V1) annealing 39 bp downstream of LCO1490 which can be used as an alternative to the latter. As an alternative to HCO2198 we selected the primer Fish-R2 (Ward et al. 2005), which aligns well with decapod sequences.

To maximise the effectiveness of the new primers we took care that they have similar melting temperatures and that they are all compatible, assuring that the same amplification conditions (Table 1) were effective for every possible pairing. All three new primers have two or three degenerate bases to compensate for variability of the 3rd codon position in the 3′ half of the primer annealing site; the 1st and 2nd codon positions, though, appear highly conserved in the species analysed.

First we screened the specimens that failed to amplify with the Folmer primers using the primer pairs LCO1490/COI-IntR1 and COI-IntF1/HCO2198 (i.e. we employed one of the Folmer primers in combination with the respective internal primer specifically designed for the decapods). We obtained sequences from ~36% of samples from both primer combinations. Then with those samples that still failed to amplify we used the primer pairs LCO-V1/COI-IntR1 and COI-IntF1/Fish-R2, obtaining 26% and 23% new sequences, respectively. The success rate was rather homogeneous across the different decapod suborders (Table 2).

Table 2.  Amplification success rates with the different primer combinations for the major decapod groups The first primer combinations for both fragments (Columns 3 and 5) were tested on those samples that failed to amplify with the Folmer primers (Column 2), and in turn the second primer combination for fragments (Columns 4 and 6) were used with those that failed with the previous attempts. Folmer, LCO1490/HCO2198; Folmer-intR, LCO1490/COI-IntR1; V1-intR, LCO-V1/COI-IntR1; intF-Folmer, COI-IntF1/HCO2198; intF-Fish, COI-IntF1/Fish-R2

Using the new primers we sequenced 364 complete new barcodes and added 98 species to the barcode library obtained with the Folmer primers, while for an additional 443 specimens only partial barcodes were obtained, i.e. only one of the two fragments. Overall, we doubled the sequenced samples, from 42% to 84%, and increased the species coverage by 58%, from 254 to 397 barcoded taxa.

The example offered by the Plesionika shrimps illustrates clearly the effectiveness of our approach. In addition to the 25 individuals sequenced with the Folmer primers, our primer combinations generated 43 barcodes and account for 9 of the 20 barcoded species. Of the four barcodes recovered from type specimens, only one was sequenced with the Folmer primers (Fig. 3).

Fig. 3.  The Plesionika shrimps illustrate how the optimised protocol adds more barcodes, more species and more barcode clusters to the barcoding library. The grey branches denote the topology generated by the barcodes recovered with the Folmer primers. The additional barcodes obtained with the internal primers are highlighted with open diamonds (full barcodes) or full diamonds (minibarcodes) and the resulting topology is in black. Name-bearing specimens are indicated with ‘Type’ after the MNHN accession numbers. The tree was edited in MrEnt 2.2. (Zuccon and Zuccon 2010).

Discussion	Go To >> Top Introduction Materials and methods Results Discussion Acknowledgements References

In over 200 years of activity, natural history museums have amassed an astonishing repository of the world’s biological diversity. In terms of both taxonomic diversity and geographical coverage, museum collections offer unique opportunities for adding a molecular perspective to the traditional morphological approach in biodiversity investigations. At the same time, collections not established for molecular studies pose specific challenges for modern-day use (e.g. Martínková and Searle 2006; Lee and Prys-Jones 2007).

The age of the biological material and the preservation conditions are largely responsible for the degree of DNA damage (Dean and Ballard 2001; Mandrioli et al. 2006). Today the crustacean wet collection at the MNHN is stored in 75% ethanol in glass jars at 14°C in a temperature-stable room. Unfortunately, little is known of the post-collection fate of the individual specimens used in our study and this uncertainty is a recurrent problem in museum collection-based studies. Some material, especially from the first Tropical Deep-Sea Benthos campaigns, might have been briefly fixed in formalin before definitive preservation in ethanol. The ethanol concentration might have varied over time and the removal of specimens from the ethanol for study might have further affected the tissue preservation. Some of the material might also have experienced temperature fluctuation. All these factors are known to affect the DNA preservation and the chance of recovering amplicons (Dean and Ballard 2001; Mandrioli et al. 2006), but with little and mostly anecdotal information available it is impossible to specifically correlate the amplification success that we observed with the individual specimen preservation histories.

Despite all these possible shortcomings, the MNHN decapod collection remains an effective source of genetic material and it proved useful for establishing a barcoding library for this taxon. For fresh material or for tissue samples preserved under appropriate conditions, the Folmer primers are likely to be highly useful for the amplification of the COI gene in decapods. Indeed, they have been used successfully in several studies for barcoding freshly collected decapods e.g. Radulovici et al. (2009), Filipová et al. (2011) and Matzen da Silva et al. (2011). The exploitation of decapod specimens not originally intended for genetic studies is still feasible, but may need ad hoc protocols like the one proposed here.

Because we invalidated the hypothesis that failure of amplification of the COI gene with the Folmer primers was linked to a mismatch of the primers with the decapod COI, our working-hypothesis was that, in several cases, DNA was degraded to fragments shorter than the 709-bp amplicon generated with the Folmer primers. The amplification of shorter, overlapping fragments circumvents the DNA shearing and is a widely followed strategy for recovering the full barcoding region from archival material (e.g. Hajibabaei et al. 2006). We favour a two-fragment approach, as it is a reasonable compromise between the additional laboratory effort and cost required, and the proportion of additional barcodes that can be retrieved. We were also interested in designing internal primers that would work across a wide set of taxa because it would enhance their utility in multiple contexts. Indeed, the two internal primers that we designed are compatible with all decapod groups, making them suitable even for barcoding samples with limited prior taxonomic information, such as processed food or larval stages.

Targeting fragments 350–400 bp long increases significantly the recovery success from archival specimens. With the two-fragment amplification strategy we were able to double the number of barcoding sequences and increase by 58% the number of species in the barcoding library. A quarter (443 specimens) of the barcodes obtained are actually minibarcodes, because only one of the two fragments was recovered from those specimens. Sequences of only 320–359 bp are clearly less optimal, but it must be remembered that when designing the internal primers we decided to limit their number to two to maximise the taxonomic inclusiveness of our protocol. While 45 species (11%) are represented only by minibarcodes, in all other cases these minibarcodes are additional sequences and at least one full-length barcode was obtained from another conspecific individual. Moreover, simulation studies showed that even 250-bp minibarcodes might provide enough information for a correct species identification (Hajibabaei et al. 2006; Meusnier et al. 2008).

The strategy of using alternative primer pairs contributed significantly to the amplification success, accounting for 33 full barcodes and 126 minibarcodes. The Folmer primers are located in rather conserved regions in decapods, as in many other animal groups (Folmer et al. 1994). However, any large barcode study will inevitably encounter species that fail to amplify due to a mutation near the 3′ end of the annealing region and the effects of primer mismatches are even more important in the case of damaged DNA such as those recovered from archival specimens (Van Houdt et al. 2010). The use of alternative primers should always be taken into account in the optimisation of barcoding protocols.

The specimens in the MNHN collection represent 68% of all species in the genus Plesionika. They have been critical for a taxonomic revision of the group and some new species have been described designating MNHN specimens as types (e.g. Crosnier 1986; Chan and Crosnier 1997; Chan 2004). The barcodes derived from the Folmer primers covered only part of the sampled Plesionika taxa. Our two-fragment amplification protocol significantly increases the genetic dataset (43 additional samples and nine species) but also provided a clearer picture of the cryptic genetic diversity within Plesionika (Fig. 3). Our barcoding approach uncovered 28 distinct sequence clusters, which may represent cryptic species, amongst the 20 morphologically defined species-level taxa. Of these, 15 were recovered only using our specific primers. The sequencing of some of the type specimens provides an objective criterion to anchor available names to the correct barcoding cluster.

Acknowledgements	Go To >> Top Introduction Materials and methods Results Discussion Acknowledgements References

We are grateful to Bertrand Richer de Forges and Philippe Bouchet, cruise leaders of several deep-sea cruises of the Tropical Deep-Sea Benthos program on board R/V Alis, that generated the deep-sea samples used in this study. All material has been collected under appropriate collection permits and approved ethics guidelines. The MarBOL project in the MNHN, Paris, is a joint effort with funding from (1) the Alfred P. Sloan Foundation; (2) the Consortium National de Recherche en Génomique and the Service de Systématique Moléculaire (UMS 2700 CNRS-MNHN), part of the agreement 2005/67 between the Genoscope and the Muséum National d’Histoire Naturelle on the project Macrophylogeny of Life; (3) the ATM ‘Taxonomie moléculaire: DNA Barcode et gestion durable des collections’ (MNHN); (4) the Fondation pour la Recherche sur la Biodiversité; and (5) the EDIT program. Sequencing was done as part of the project @SPEED-ID ‘Accurate SPEciEs Delimitation and IDentification of eukaryotic biodiversity using DNA markers’ proposed by F-BoL, the French Barcode of life initiative.

References	Go To >>\ Top Introduction Materials and methods Results Discussion Acknowledgements References

Appendix 1.  Specimen data for the genus Plesionika included in Fig. 1, with the primer combinations used for each specimen Barcode: F, full barcode; M, minibarcode (<400 bp)