Electronic Letters are short, timely responses to selected articles published in Science. Recently published E-Letters appear below. To submit an E-Letter on one of our articles, go to the article and follow the "Submit an E-Letter" link in the "Article Tools" area.

E-Letter responses published in the past 21 days:

Read E-Letter responses published in the last 1, 2, 3, 4, 5, 6, 7, 14, 21 days

6 E-Letter responses published for 4 different topic sources.

Articles			E-Letter Responses
	reports: Sequential Processing of Lexical, Grammatical, and Phonological Information Within Broca’s Area Sahin et al. (16 October 2009) [Abstract] [Full text] [PDF]		Sequential Information Processing and Limited Interaction in Language Production Matthew Goldrick, et al.   (4 December 2009) Read every E-Letter response to this article
	perspective: The Speaking Brain Hagoort and Levelt (16 October 2009) [Full text] [PDF]		Sequential Information Processing and Limited Interaction in Language Production Matthew Goldrick, et al.   (4 December 2009) Read every E-Letter response to this article
	p-forum: Energy and Technology Policies for Managing Carbon Risk Patrinos and Bradley (21 August 2009) [Full text] [PDF]		Response to G. A. Schmidt and M. E. Singer's E-Letter Aristides A. N. Patrinos   (4 December 2009) Broaden Policies to Manage Carbon Risk Gavin A. Schmidt, et al.   (4 December 2009) Read every E-Letter response to this article
	perspective: Barcoding of Plants and Fungi Chase and Fay (7 August 2009) [Full text] [PDF]		Response to J. Schultz and M. Wolf's E-Letter Merk W. Chase, et al.   (10 December 2009) ITS Better Than Its Reputation Matthias Wolf, et al.   (10 December 2009) Read every E-Letter response to this article

reports: Sequential Processing of Lexical, Grammatical, and Phonological Information Within Broca’s Area Sahin et al. (16 October 2009) [Abstract] [Full text] [PDF] Sequential Processing of Lexical, Grammatical, and Phonological Information Within... Sequential Information Processing and Limited Interaction in Language Production 4 December 2009 Matthew Goldrick, Associate Professor of Linguistics Northwestern University, Gary S. Dell, Judith Kroll, Brenda Rapp Respond to this E-Letter: Re: Sequential Information Processing and Limited Interaction in Language Production N. T. Sahin et al. report data indicating that the processing components of speech production are distinct in their content and sequentially ordered in time ("Sequential processing of lexical, grammatical, and phonological information within Broca's area," Reports, 16 October 2009, p. 445). Sahin et al. and the accompanying Perspective by P. Hagoort and W. J. M. Levelt ("The speaking brain," 16 October 2009, p. 372) claim that these data specifically support production theories in which the operations of these processing components are discretely separated and temporally non-overlapping. However, this full discreteness claim is not required by the Sahin et al. data and conflicts with an extensive body of research that has already demonstrated its inadequacy: reaction time and error data from neurologically intact monolingual speakers (1); aphasic language impairments (2, 3); and bilingual language processing (4). This body of evidence also indicates—consistent with Sahin et al.'s results—that a full interactivity claim is likewise incorrect (1, 2, 4). To account for the inadequacy of both fully discrete and fully interactive theories, language production architectures must allow for interaction while incorporating critical constraints on the degree of coupling between processing components (2, 3). Such architectures respect the sequentiality of processing different types of information (in line with Sahin et al.'s results) without sacrificing the ability to account for interactive effects. Rather than viewing results such as those of Sahin et al.'s through a Boolean lens—supporting either fully interactive or fully discrete models—a more profitable approach would acknowledge the continuum of possibilities between these extremes. Matthew Goldrick Department of Linguistics, Northwestern University, Evanston, IL 60208, USA. Gary S. Dell Psychology Department, University of Illinois at Urbana-Champaign, Champaign, IL 61820. USA. Judith Kroll Department of Psychology, Pennsylvania State University, University Park, PA 16802, USA. Brenda Rapp Department of Cognitive Science, Johns Hopkins University, Baltimore, MD 21218, USA. References 1. M. Goldrick, Lang. Cogn. Proc. 21, 817 (2006). 2. G. S. Dell, M. Schwartz, M. Martin, E. M. Saffran, D. A. Gagnon, Psych. Rev. 104, 801 (1997). 3. B. Rapp, M. Goldrick, Psychol. Rev. 107, 460 (2000). 4. J. Kroll, S. Bobb, Z. Wodniecka, Biling. Lang. Cogn. 9, 119 (2006).

perspective: The Speaking Brain Hagoort and Levelt (16 October 2009) [Full text] [PDF] The Speaking Brain Sequential Information Processing and Limited Interaction in Language Production 4 December 2009 Matthew Goldrick, Associate Professor of Linguistics Department of Linguistics, Northwestern University, Evanston, IL 60208, USA, Gary S. Dell, Judith Kroll, Brenda Rapp Respond to this E-Letter: Re: Sequential Information Processing and Limited Interaction in Language Production N. T. Sahin et al. report data indicating that the processing components of speech production are distinct in their content and sequentially ordered in time ("Sequential processing of lexical, grammatical, and phonological information within Broca's area," Reports, 16 October 2009, p. 445). Sahin et al. and the accompanying Perspective by P. Hagoort and W. J. M. Levelt ("The speaking brain," 16 October 2009, p. 372) claim that these data specifically support production theories in which the operations of these processing components are discretely separated and temporally non-overlapping. However, this full discreteness claim is not required by the Sahin et al. data and conflicts with an extensive body of research that has already demonstrated its inadequacy: reaction time and error data from neurologically intact monolingual speakers (1); aphasic language impairments (2, 3); and bilingual language processing (4). This body of evidence also indicates—consistent with Sahin et al.'s results—that a full interactivity claim is likewise incorrect (1, 2, 4). To account for the inadequacy of both fully discrete and fully interactive theories, language production architectures must allow for interaction while incorporating critical constraints on the degree of coupling between processing components (2, 3). Such architectures respect the sequentiality of processing different types of information (in line with Sahin et al.'s results) without sacrificing the ability to account for interactive effects. Rather than viewing results such as those of Sahin et al.'s through a Boolean lens—supporting either fully interactive or fully discrete models—a more profitable approach would acknowledge the continuum of possibilities between these extremes. Matthew Goldrick Department of Linguistics, Northwestern University, Evanston, IL 60208, USA. Gary S. Dell Psychology Department, University of Illinois at Urbana-Champaign, Champaign, IL 61820. USA. Judith Kroll Department of Psychology, Pennsylvania State University, University Park, PA 16802, USA. Brenda Rapp Department of Cognitive Science, Johns Hopkins University, Baltimore, MD 21218, USA. References 1. M. Goldrick, Lang. Cogn. Proc. 21, 817 (2006). 2. G. S. Dell, M. Schwartz, M. Martin, E. M. Saffran, D. A. Gagnon, Psych. Rev. 104, 801 (1997). 3. B. Rapp, M. Goldrick, Psychol. Rev. 107, 460 (2000). 4. J. Kroll, S. Bobb, Z. Wodniecka, Biling. Lang. Cogn. 9, 119 (2006).

p-forum: Energy and Technology Policies for Managing Carbon Risk Patrinos and Bradley (21 August 2009) [Full text] [PDF] Energy and Technology Policies for Managing Carbon Risk Response to G. A. Schmidt and M. E. Singer's E-Letter 4 December 2009 Aristides A. N. Patrinos, President Synthetic Genomics, Inc., Washington, DC 20024, USA Respond to this E-Letter: Re: Response to G. A. Schmidt and M. E. Singer's E-Letter We agree that it would have been best to not have conducted this experiment in altering the atmosphere that has been occurring since the start of industrialization. The scientific consensus to which we refer in our Policy Forum ("Energy and technology policies for managing carbon risk," A. A. N. Patrinos and R. A. Bradley, 21 August 2009, p. 949) is based on the Intergovernmental Panel on Climate Change 4th Assessment Report, and recognizes that inertia in the climate system and in the existing capital structure means that it will be some time before concentrations are stabilized. We also acknowledge that there is an emerging consensus that after having achieved 450 parts per million, the international climate policy should aim to achieve a reduction in atmospheric concentrations. However, we do not think this changes the message in our Policy Forum. Aristides A. N. Patrinos President, Synthetic Genomics, Inc., Washington, DC 20024, USA. Energy and Technology Policies for Managing Carbon Risk Broaden Policies to Manage Carbon Risk 4 December 2009 Gavin A. Schmidt NASA Goddard Institute for Space Studies, New York, NY 10025, USA, Mark E. Singer Respond to this E-Letter: Re: Broaden Policies to Manage Carbon Risk In the Policy Forum by A. A. N. Patrinos and R. A. Bradley ("Energy and technology policies for managing carbon risk," 21 August 2009, p. 949), they claim that there is a scientific consensus that greenhouse gas stabilization levels of a CO2 equivalent of 450 to 550 parts per million would avoid serious impacts from climate change. However, stabilization at these levels would lead to global temperatures that would be unprecedented for millions of years with unknowable long-term consequences for sea levels and ecosystems. On the contrary, most climate scientists would much prefer to have no further increases in CO2 at all, and many have called for reducing levels below current concentrations in order to minimize the risks. The stabilization scenarios being discussed in current climate negotiations are therefore not based on a scientific consensus, but rather the current political and economic estimate of what is achievable. We agree with Patrinos and Bradley that a broad range of actions will be required to reduce emissions to the levels implied by these scenarios. Gavin A. Schmidt NASA Goddard Institute for Space Studies, New York, NY 10025, USA. Mark E. Singer Wilmette, IL 60091, USA.

perspective: Barcoding of Plants and Fungi Chase and Fay (7 August 2009) [Full text] [PDF] Barcoding of Plants and Fungi Response to J. Schultz and M. Wolf's E-Letter 10 December 2009 Merk W. Chase Jodrell Lab, Royal Botanic Gardens Kew, Kew, Richmond, Surrey TW9 3DS, UK, Michael F. Fay Respond to this E-Letter: Re: Response to J. Schultz and M. Wolf's E-Letter The following points on M. Wolf and J. Schultz's comments toward our Perspective on DNA barcoding ("Barcoding of plants and fungi," 7 August 2009, p. 682) are relevant. Firstly, this was not an empirical paper; we were merely reporting progress in several areas and the DNA regions that had been selected for groups of organisms. We were not expressing our opinions on the selection of these markers, but rather making clear their limitations and differences from the application of CO1 (the widely accepted DNA barcoding marker in most groups of animals). We do know the reasons why the nuclear ribosomal non-transcribed spacers of the large subunits (commonly known as nrITS) in plants were not selected as standard barcoding markers, despite their well known, conserved secondary structure (particularly ITS2), and relatively higher rates of substitution versus plastid DNA regions (1, 2). The Plant Working Group (PWG) of the Consortium for the Barcode of Life (CBOL) (3) selected two plastid regions, as we reported. In the earlier stages of identifying appropriate loci for use as universal land plant barcodes, nrITS was eliminated because (i) pseudogenes are frequently present in many taxa, making amplification more complicated, although this problem can be overcome in some cases by the addition of dimethyl sulfoxide in the polymerase chain reaction stage (4); (ii) many taxa maintain multiple functional copies. Effective concerted evolution only occurs in the flowering plants among land plants, and even there it does not always homogenize the many tandem copies of nrITS. This means that with current sequencing technologies amplification products must be cloned before sequencing, making the whole process more complicated and difficult to automate; (iii) GC-rich regions are frequently present, which, although important in conserving secondary structure, also make sequencing more difficult and lowers the quality of sequence output in many groups of plants, raising concerns over the accuracy of automated identifications. Wolf and Schultz point out that in particular cases, such as in senedesmacean green algae (6), microarray approaches using ITS2 are extremely good at distinguishing closely related species that differ little genetically. The important point they overlook in making this comment is the difference between a generalized DNA barcoding approach that can be used with an unknown fragment from any group of land plants and a specific approach for identifying species in a group of organisms known to be characterized by low levels of DNA variation. In the former case (the whole of land plant flora, more than 400,000 species), microarrays cannot be applied, and ITS2, despite its well-known levels of variability and success in narrow case studies, is not amenable to the general approaches of automated amplification and sequencing used for DNA barcoding, such as for CO1. If an unknown botanical specimen is first placed in a specific genus with the use of the plastid matK/rbcL combination identified as best for use on land plants, and if it is known that this genus suffers from low levels of interspecific variability, such that matK/rbcL cannot distinguish species in the genus, and if it is important to determine which species among these it is, then a microarray could be developed to supplement the matK/rbcL approach, provided the economic impetus is present to enable development of this relatively expensive method. The PWG of CBOL was perfectly aware of the limitations imposed by the choice of matK/rbcL, but given the problems inherent with using nrITS as a universal barcoding marker, they clearly had no choice but to eliminate this region. Wolf and Schultz ignore the many problems of using nrITS for universal barcoding and focus only on its use as a supplemental marker in the specific groups where its evolution makes it a useful marker. Mark W. Chase and Michael F. Fay Jodrell Lab, Royal Botanic Gardens Kew, Kew, Richmond, Surrey TW9 3DS, UK. References 1. B. G. Baldwin, Mol. Phylogenet. Evol. 1, 3 (1992). 2. V. van den Berg et al., Am. J. Bot. 92, 613 (2005). 3. CBOL Plant Working Group, Proc. Natl. Acad. Sci. U.S.A. 10.1073/pnas.0905845106 (2009). 4. M. W. Chase et al., Ann. Bot. 92, 107 (2003). 5. J. C. Engelmann et al., Mol. Ecol. Res. 9, 83 (2009). Barcoding of Plants and Fungi ITS Better Than Its Reputation 10 December 2009 Matthias Wolf Department of Bioinformatics, Biocenter, University of Würzburg, Am Hubland, 97074 Würzburg, Germany, Jörg Schultz Respond to this E-Letter: Re: ITS Better Than Its Reputation Cytochrome oxidase I (COI) is a heavily promoted marker for species barcoding. In their Perspective "Barcoding of plants and fungi" (7 August 2009, p. 682, published online 30 July 2009), M. W. Chase and M. Fay argue that, for plants and fungi, COI might not solve all the problems of species identification. They suggest the internal transcribed spacer (ITS) as the best alternative for fungi, but note the biggest problem of this and other regions (for example, matK, rbcL): There is no barcode gap, which makes identification of species with these markers less definitive. This undervalues ITS2, which shows potential. The ITS2 sequence-structure information provides a binary feature—compensatory base changes (CBCs)—that correlates with the biological species concept (1). This eliminates the barcode gap problem, which is caused by the continuity of sequence divergence. With a single CBC present in an ITS2 sequence-structure pair, the confidence level of species delineation is at least 93%, even for closely related species classified within the same genus (2). This holds true for fungi and plants as well as for algae, protozoa, and animals. In addition, J. C. Engelmann et al. conducted an experiment in which they successfully used ITS2 microarrays to separate species with sequence identities up to 97% (3). This case study demonstrated the potential of ITS2 phylochips in biodiversity analyses. Finally, due to its high sequence variability and conserved core of secondary structure, ITS2 has a wider taxonomic coverage than usually assumed (4, 5). All of these points indicate that ITS2 is more than just a barcode marker. Matthias Wolf and Jörg Schultz Department of Bioinformatics, Biocenter, University of Würzburg, Am Hubland, 97074 Würzburg, Germany. References 1. A. W. Coleman, Protist 151, 1 (2000). 2. T. Müller et al., RNA 13, 1469 (2007). 3. J. C. Engelmann et al., Mol. Ecol. Res. 9, 83 (2009). 4. A.W. Coleman, Trends Genet., 19, 370 (2003). 5. J. Schultz et al., Nucleic Acids Res., 34, W704 (2006).