Sir — H. C. J. Godfray (Nature 417, 17–19;2002) points out in his Commentary, quiterightly, that taxonomy and systematics, thenaming and classification of living things,have an image problem among fundingbodies and the community in general. He says the discipline faces a crisis: muchinformation on living things needs to becaptured, organized and disseminated, but funds often go into higher-profileprojects such as the Sloan Digital SkySurvey and the Human Genome Project.

What do these projects offer thattaxonomy does not? One thing is arelatively centralized model of data organization and retrieval. One can enter a web portal and easily find and downloadan authoritative account of star positionsor gene sequences. No equivalent portalscurrently exist for taxonomic information(although initiatives are in place to providethem: see Nature 418, 362–363 & 367;2002). To create such a portal, Godfraysuggests that taxonomy needs to movefrom the current dispersed, decentralizedapproach to a unitary, centralized one.

Although we agree that Godfray’smodel would provide many advantages tooutside users, we believe that it misses onecrucial problem. A taxon — the base unit

of a taxonomy — is an hypothesis, not anobservation or “fact”. It is not like a star ora base pair, and taxonomy is not like a starsurvey or genome project. Because of this,solutions cannot be so easily transferredfrom these domains to taxonomy.

Taxonomists routinely filter largearrays of observations through anextensive knowledge base to decide thatthis group of specimens or that set of prior taxa comprises a new taxon.Crucially, another taxonomist using the same knowledge base may validlyarrive at quite different conclusions, and it may take time for thorough testing of the alternative concepts to arrive atconsensus (itself subject to futurechallenge and refinement).

Taxonomy is a complex and dynamicscience, not a simple (albeit technicallydifficult) collation of facts like thenucleotide sequences of a genome survey.It is probably only at the most complexlevels of genetics, in dynamic proteomicsor developmental biology, that anything like the complexity ofhypothesis-generation and testinginvolved in routine taxonomy andsystematics is encountered.

Any proposal for changing the

processes of taxonomy must account forthis. Godfray’s model, in which new names require validation by a centralcommittee before acceptance, could leadto authoritarianism and a stifling ofinnovative taxonomic viewpoints. Noother hypothesis-driven field of sciencewould accept such a straitjacket.

The heart of the problem is the centralduality and tension in taxonomic names.On the one hand, names are convenientshorthand representations of scientifichypotheses, and as such should be asvolatile as hypotheses in any other field —proposed, used, modified and thenperhaps discarded as evidence dictates. On the other hand, taxonomic names arekeys to biological information, uniquecodes that unlock the library of knowledgeabout those organisms, and for thispurpose they must be non-volatile or the links will be lost. A solution mustadequately recognize these dual roles anddecouple the system that allows maximumfreedom of hypothesis-generation fromthe system that provides names for users.Kevin Thiele, David YeatesCSIRO Plant Industry and Entomology,Commonwealth Scientific and Industrial ResearchOrganisation, Canberra, Australia

correspondence

NATURE | VOL 419 | 26 SEPTEMBER 2002 | www.nature.com/nature 337

Tension arises from duality at the heart of taxonomyNames must both represent a volatile hypothesis and provide a key to lasting information.

No alternative to animaltests for behaviourSir — Your News story “Race is on to findalternative to animal tests” (Nature 418,116; 2002) discussed the possibility that in vitro methods could replace animal tests used in the toxicologic screening ofchemicals. However, it is highly unlikelythat in vitro approaches could be used toreplace in vivo animal tests in theassessment of neurotoxicity. For instance,behavioural assessment in rodents is a powerful experimental tool in theevaluation of the effects of chemicals on neural function. Pharmaceuticalcompanies and regulatory agenciesconsider behavioural toxicology animportant part of the screening of newpharmaceutical products and potentiallyneurotoxic chemicals.

It is true that in vitro experimentalapproaches can be used in the investi-gation of the cellular and molecularmechanisms of behaviour. But drug-induced alterations in parameters such as anxiety, motor activity, and learning can only be assessed by in vivo tests onmammals. It is not possible to replace

such tests with non-animal methods using, for example, human cells or yeast.Rafael RoeslerGenotoxicity Laboratory, Centre for Biotechnology,Federal University of Rio Grande do Sul, 91501-970 Porto Alegre, RS, Brazil

Laws stay constant butthe world changesSir — Efforts by Stephen Schneider andRichard Moss to quantify uncertainty(Nature 418, 476–478; 2002) succeededwith the Intergovernmental Panel onClimate Change (IPCC) Working Group 1climate modellers but failed with WorkingGroup 3’s emissions modellers. As a leadauthor on the IPCC’s 2000 Special Reporton Emissions Scenarios, I would like toelaborate on why this happened.

The first group of scientists know thatthe laws of physics operative today will bethe same laws of physics operative in 2100.The second group, of economists andanalysts, know with equal certainty thatthe geopolitical, socioeconomic andtechnological forces operative today arelikely to be radically different in 2100.

Stuart R. GaffinCenter for International Earth Science InformationNetwork (CIESIN), Columbia University, 61 Route9W, PO Box 1000, Palisades, New York 10964, USA

Food labels should statethe benefits of GMOsSir — You state in the News story “Europegets tough on labelling genetically modifiedfoodstuffs” (Nature 418, 114; 2002) that anangry food industry claims that stricterlabels would imply a health risk.

On the contrary. The GMOs potentiallypresent in trace amounts have been farmore thoroughly assessed for safety thanthe rest of the food, but the labelling tells usnothing of this. The European Union hasselected GMOs for special labelling becausemany people are uncertain of the benefitsand risks. Surely, then, labels should conveythe positive (and, where appropriate,negative) environmental, nutritional oreconomic attributes of the modification?This would help consumers make informeddecisions to buy or avoid products.Renton Righelato63 Hamilton Road, Reading RG1 5RA, UK

© 2002 Nature Publishing Group