Area of Research: DNA Barcoding

Krill are shrimp-like crustaceans found throughout the world’s oceans. The Antarctic krill, Euphasia superba, is thought to be the most abundant species on the planet in terms of biomass (500 million metric tonnes corresponding to 5 x 10^14 individuals), is a primary food source for whales, seals, and oceanic birds, and functions as a major planetary carbon sink by excreting waste that sinks to ocean floor. What does this very abundant, central-to-food-web species eat? For many animals, observation of eating behavior is impractical, and analysis of stomach contents or feces may be the only way to determine diet. However, such material may be morphologically unrecognizable. 

In August 2006 Marine Biotech 8:686, researchers from University of Tasmania and Department of Environment, Tasmania, compare DNA sequencing and light microscopy in identifying prey in stomach contents of E. superba. Passmore et al isolated DNA from stomach contents of 6 ethanol preserved krill and, using diatom-specific primers, amplified a 103 bp portion of nuclear small subunit RNA (ssRNA). ssRNA was used because at present it has the best taxonomic representation in GenBank for krill prey species. The researchers sequenced at least 50 clones from each individual krill and found 14 OTUs (operational taxonomic units), with 86% to 100% match to GenBank sequences. These results were compared to microscopic identification of diatom silica skeleton fragments in stomach contents, which involved counting 1000-3000 fragments per individual. Results were similar, although DNA analysis and light microscopy each appeared more sensitive for certain species. This study might be a best case for light microscopy because silica-skeletoned diatoms are not easily digested. As the authors point out, krill also consume a range of protozoa and small zooplankton, and the importance of these sources may be underappreciated.The authors conclude “the application of DNA diet analysis to krill warrants further investigation, particularly for prey that are difficult to study using other methods“.

This work shows the essential need for a comprehensive reference library, so far lacking. A study underway is examining mitochondrial and nuclear genes as barcodes for phytoplankton. Looking ahead, a “massively parallel” pyrosequencing approach could enable rapid and representative analysis of mixed environmental samples, such as stomach contents, without biases resulting from amplification and cloning. 

In early access Dec 2006 Nucl Acid Res, researchers from 9 laboratories in France, Italy, Norway, and Denmark examine a candidate barcode gene for land plants, the group I intron in the chloroplast leucine transfer RNA gene (trnL intron). Prior research has already shown that a simultaneous or tiered multi-gene approach will be needed to distinguish among closely-related land plant species. A project coordinated by Royal Botanic Gardens, Kew aims to identify the best overall approach.

Rather than cracking the tough nut of an ideal plant barcode, Taberlet and co-authors look at a simple approach “emphasizing the point of view of scientists other than taxonomists“, and test this on food plants in archeological and industrial applications. The chloroplast trnL intron is not the most variable non-coding region in chloroplast DNA and does not differ enough to separate many closely-related plant species. On the plus side, there are robust primers which amplify the intron from diverse species. Like other group I introns, the trnL intron sequence has catalytic activity and a conserved secondary structure with alternating conserved and variable sequence domains. Taking advantage of this feature, the researchers designed primers to amplify one of the variable domains, the P6 loop. Binding sites for both the trnL primers, which amplify the entire intron, and the P6 loop primers are “highly conserved among land plants, from Angiosperms to Bryophytes“. Importantly, the P6 loop is only 10 to 143 bp and can be amplified from degraded DNA.

Using “simulated ePCR” with the large GenBank data set, trnL intron and P6 loop sequences identified to species level 67% and 19% of cases respectively. However, in many practical applications, the number of possible species that need to be distinguished is relatively small and they are taxonomically diverse. Following this reasoning, Taberlet et al tested the intron and its P6 loop on a set of 132 species found in the Arctic and 72 species representing the commonest food plants. With Arctic plants, trnL intron and P6 identified to species level 85% and 47%, respectively. With the food data set, the tiny P6 loop was sufficient to identify 78% to species level. The P6 loop was successfully amplified from a 20,000-year old permafrost sample, from human feces, and from various processed foods including detecting potato, leek, and onion DNA in dried soup mix!

This is an exciting study, and DNA barcoding will likely have multiple applications in food safety. Whether or not these exact gene regions are adopted, a standardized approach will enable widespread and inexpensive use.

Relying on morphologic species descriptions lacking DNA sequences is like diagnosing patients without laboratory tests. An experienced clinician can often make the correct diagnosis from examination alone, but laboratory tests are frequently needed to confirm or point to other causes. Many advances in medicine reflect incorporating laboratory testing into routine evaluation. Two papers on leeches suggest similar benefits to taxonomy from incorporating mtDNA analysis into routine practice.

In May 2005 Conservation Genetics 6:467 researchers at the American Museum of Natural History analyze morphology, mitochondrial COI, and nuclear ND-I sequences of genus Helobdella leeches from Austrialia, New Zealand, South Africa, Hawaii, and South America. The tale starts with a leech discovered in Germany in 1985, H. striata, re-named H. europaea in 1987. 

Authors Siddall and Budinoff found that H. europaea is morphologically and genetically indistinguishable from a leech “discovered” in Australia in 1998, H. papillornata. Including COI sequences in initial species descriptions would have prevented wasted taxonomic effort, and a species native to South America would probably not be given the unfortunate name europaea.

In December 2006 Evol Devel 8:491, scientists from University of Maryland and University of California, Berkeley, apply COI barcoding to another Helobdella leech, H. robusta, a model organism in developmental biology and one of the lucky species selected for genome sequencing. Researchers Bely and Weisblat obtained leech specimens from laboratory cultures and field sites. They found that isolates thought to be H. robusta actually represent four species, 2 of which co-exist at the same locality. The authors observe “the perils of misidentification and taxonomic confusion in the lab are numerous and costly” and conclude with a call for routine application of DNA barcoding and collection of voucher specimens to confirm identity of laboratory organisms. 

For enjoyment or use, a US postage stamp featuring All Birds Barcoding Initiative (ABBI) and the Consortium for the Barcode of Life (CBOL) is available at zazzle.com.

 

 

 

 

 

 

 

 

In early online J Zool Syst Evol Res researchers from Natural History Museum and Imperial College, London, scrutinize “recent advances in DNA taxonomy…that follow the dramatic increase in data generation“. Authors Vogler and Monaghan provide a scientific update to

what has been learned so far: “a key finding from recent studies in animals is that variation in mitochondrial DNA is partitioned as tight clusters of closely related genotypes, which group specimens largely according to traditionally recognized species limits, and which are congruent with nuclear markers”,

the durability of clustering: “it can be expected that denser geographic and taxonomic sampling may result in the discovery of new clusters, and perhaps reduce their divergence from each other, but they are unlikely to erode the clustering altogether”,

the significance of incongruence between DNA-based and morphology-based methods for delimiting species: “the high degree of congruence of mtDNA groups and traditionally defined taxa appears to contradict the reported mismatch of established species boundaries…even well-studied groups may be in need of taxonomic revision before accurate tests of incongruence can be conducted”,

what the future holds: “a standard DNA taxonomic analysis will include broad sampling..followed by large-scale sequencing, and algorithmic procedures for delineating species limits. The taxonomic system will be derived from the data rather than expert opinion“,

and what is needed to harness DNA taxonomy in general and DNA barcoding in particular to speed description of the estimated 80% of earth’s biodiversity that is at yet undescribed: “a feedback loop that [uses] discrepancies between DNA and other data to refine species descriptions..founded in existing theory of evolutionary biology and phylogenetics”

I close with a pictorial analogy. The Coulter counter uses electrical sensing to gain the same information as morphologic diagnosis of blood smears, with dramatic improvements in speed, cost, and necessary expertise. In some situations, DNA sequencing may provide similar improvements over morphologic diagnosis for species-level identification.

 

Why DNA barcoding works as well as it does is an unsolved scientific puzzle. It is long observed that mitochondrial DNA differences within animal species are generally much smaller than those among species and, in the landscape of phylogenetic trees, mitochondrial DNA sequences of most species form single clusters distinct from those of other species. As a result “mtDNA data and traditional taxonomic assignments tend to converge on what may be “real” biotic units in nature”  (Avise and Walker 1999 Proc Natl Acad Sci USA 96:992). Although Avise and Walker’s original observation was largely based on terrestrial, temperate zone vertebrates, growing barcode libraries demonstrate similar patterning in diverse invertebrates, vertebrates, and protists in marine and terrestrial environments, and in tropical and temperate zones, and in at least some fungi and plants (see last week’s post on COI barcodes in red algae) 

What underlies the usual patterning of small differences within and large differences among most animal species? The unsolved puzzle is how to reconcile these two findings. Large differences among closely-related species indicates mitochondrial DNA undergoes rapid sequence evolution, and there are reasonable mechanistic explanations for why this might be so. On the other hand, rapid sequence evolution should also lead to accumulation of sequence diversity within species over time and in those with large populations. Instead the data shows a relative absence of variation within most species, including those thought to be ancient and those with enormous population sizes. I will set aside two of the usual suspects: population bottlenecks and small effective population size. Population bottlenecks are implausible given the diversity of species showing this pattern. Postulating a small effective population size is a restatement of the finding of absence of variation, not an explanation. 

This table-napkin analysis leads me to selective sweeps as pruning mitochondrial diversity within species (eg Bazin et al 2006 Science 312:570, see also editorial and reader commentary). If selective sweeps restrict mitochondrial diversity, then the question becomes what is being selected for? Environmental adaptation seems unlikely, as restricted variation is seen in species that are as best one can tell morphologically and ecologically unchanged (eg see earlier posts on horseshoe crabs, salamanders). It might be there is little tolerance for genetic variation due to interactions of mitochondrial proteins with other cellular components, but if so there should be species with genetic stasis in mitochondrial DNA, just as there are many species with apparent morphologic stasis. However, in simple distance trees most species show roughly similar genetic distances. 

I am intrigued by a time series of influenza A hemagglutinin gene evolution which reflects competition between virus and host and wonder if there might be some kind of competition that helps drive mitochondrial sequence evolution forward and at the same time suppresses variation. It is exciting there will be an EMBO workshop “Molecular Biodiversity and DNA Barcodes” May 2007 in Rome which may help answer scientific questions posed by DNA barcode data.

Red seaweeds, kingdom Rhodophyta, are “weird, wonderful, and extremely ancient” organisms distantly related to plants (Tudge 2000 The Variety of Life). Multicellular red algae arose at least 1.2 billion years ago, predating the earliest multicellular animals by 600 million years. Visual identification is challenging, as “morphology can be highly variable within and between species, and conspicuous features with which they can be readily identified are often lacking. In addition, highly convergent morphology is commonly encountered. …Identification is further compounded by the complexities of red algal life histories, many of which have a heteromorphic alternation of generations. Different life history stages of species have frequently been described as separate species and have only been linked through observations of life histories in culture and use of molecular techniques” (Robba et al 2006 Am J Botany 93:1101). Red seaweeds call out for DNA-based identification.

In a preliminary appraisal, COI barcodes “resolved accurately and unequivocally species identities” in three morphologically confusing species complexes of red algae (Saunders 2005 Phil Trans Royal Soc B 360:1879). In September 2006 Am J Botany, researchers from Natural History Museum, London, and University of Bristol, analyzed COI sequences of 79 specimens from 37 red algae, chosen to span 6 orders and represent species “extremely difficult to distinguish morphologically”. All species could be discriminated and intraspecific variation was generally much lower (0.0-0.7%) than interspecific differences (5.2-27.3%). Large intraspecific distances suggested the presence of cryptic species or incipient speciation in two cases. COI outperformed a commonly used genetic marker, the plastid Rubisco spacer, in revealing cryptic diversity. As is generally observed, “a good taxonomic foundation coupled with extensive sampling of taxa is essential for the development of an effective identification system.”  A holiday wish for red algae identifiers: a well-curated DNA barcode library.

Labelling specimens helps make natural history museum collections valuable. Once specimens have been carefully examined by an expert, a species label can be applied. Labelled specimens can then be re-found and re-examined. Unsorted specimens lacking species designations are more raw material than scientific resource.  

As in all sciences, taxonomic knowledge undergoes continuous revision. Changes in names and understanding of species boundaries may mean that specimen labels need to be updated in light of current knowledge. Partly to compensate, there is enormous effort to link current taxonomic understanding to historical species descriptions. Nonetheless, the accuracy of specimen labels may decay over time. In one study, a revision of predatory flies (Euscelidia) (Meier and Dikow Conservation Biol 2004 18:478), 83% of 1383 specimens from 19 collections were found to be incorrectly identified.

DNA sequences offer a simple approach to help keep specimen labels up to date. DNA sequences are an intrinsic, unvarying characteristic of a specimen. A DNA sequence from a standardized locus (ie a DNA barcode) can serve as a permanent “name” for a specimen. Results so far with more than 20,000 invertebrate and vertebrate species show it is generally straightforward to use a COI barcode to assign specimens to known species. Future taxonomic revisions may change species names or boundaries, but that will not change DNA barcodes of specimens or the clustering patterns of barcode sequences. Thus it should be simple to use a specimen’s barcode sequence “name” to search a regularly revised public database for the current species name it corresponds to. A public database of sequences, specimens, and associated data as is BOLD can undergo continuous refinement, whereas revising labels in every collection around the world is impractical. Given that assigning specimens to species involves sorting among millions of species names, this approach does not have to give 100% resolution to species level to be valuable. It will be useful both for those species with unique barcode clusters or characters and for those sets of closely-related species with overlapping or indistinguishable barcodes.

Finally, on a related note I observe that diagnostic keys are unwieldy and not easily amenable to computerization. For example, Dragonflies of North America (Gainesville: Scientific Publishers, 2000) by Needham, Westfall, and May covers 350 species in 939 pages, or about 2.7 pages per species, which by my count is typical for diagnostic keys. At this rate, a world key for the 5500 known Odonates would run to 15,000 pages. The recondite language that is required to describe morphologic detail in this and other keys makes the tools inacessible except to highly-trained persons.  On the other hand, anyone can interpret a DNA sequence. It is exciting that taxonomists and others are increasingly taking up the challenge of translating taxonomic knowledge into a much more widely-accessible format, namely DNA barcode libraries.

Freshwater snails are intermediate hosts for schistosomiasis and flukes, trematode parasites that infect approximately 10% of world’s human population. Freshwater snails are also indicator species for water quality. Snail identification is essential for reducing disease burden and monitoring water quality.

Researchers at the University of Frankfurt (November 2006 BMC Evol Biol 6:100) compared efficacy of morphologic and DNA-based taxonomy in freshwater snails in the genus Radix. Regarding Radix species in northwestern Europe, “species determination by shell morphology is difficult [and] unreliable…intraspecific variability of the putatively distinctive anatomical measurements largely overlaps among species” and identifications are “further complicated by recent nomenclatorial revisions”. 

In their report, Pfenninger, Cordellier and Streit analyze morphology, mitochondrial COI and nuclear ITS-1 sequences, and describe breeding experiments with Radix snails collected at 60 sites throughout Europe. Using mtCOI sequences they found five MOTU (molecular operational taxonomic units), defined as “terminal clades with bootstrap support of 90% or more”. Populations of these MOTU overlapped broadly in geographic range and none corresponded to described species. Nuclear ITS sequences analyzed in a subset of specimens produced MOTU congruent with those generated by mtCOI.

ALL crosses between individuals from the same MOTU population were viable, whereas NONE of crosses between individuals from different MOTU produced eggs. In morphometric analysis, Radix MOTU overlapped as shown at left, and in rearing experiments, shell shape changed in 4 of 5 populations, demonstrating phenotypic plasticity of putative morphologic characters. In northwestern European Radix snails, DNA trumps morphology.

This work follows what might be a “best practices” pathway for single-locus mtDNA species discovery, aka DNA barcoding applied to species discovery:

1. COI sequence clusters (MOTU), found in analyzing  multiple specimens from geographically widespread locations, are proposed as putative species.

2. COI clustering is congruent with nuclear sequence data.

3. COI clusters show corresponding biological differences, such as morphologic characters, behavioral differences, or breeding incompatibility.

In some cases a virus or bacteria is recognized to be the causative agent even though not all of Koch’s postulates have been fulfilled. In a similar way in some cases it might be desirable to recognize mtDNA clusters as representing species without fulfilling all of the above criteria.

Growing barcode libraries challenge understanding. There are already about 200,000 mtCOI barcodes from about 25,000 species in BOLD, the Barcode of Life Data Systems Database. The burgeoning data sets hint at insights into biological diversity, revealed by looking at many species at once. A map of counties of the United States shows both large and small scale patterns, shaped by history, geography, and politics.  Viewed through the lens of mitochondrial variation, what would a map of species show? Are differences among and within species similar in birds and butterflies? Do species boundaries differ in marine vs terrestrial species, or in tropical vs. temperate zones?