Trust and scientific publication: <i>AJB</i> policy for digital images

Science by its nature and in practice requires a balance between trust and skepticism. The latter comes into play as we formulate research questions, analyze our data, and interpret our own scientific results and those of others. Indeed, professional skepticism of the work of others is the basis of critical assessment of manuscripts and published papers. This article discusses the former, trust, specifically the relationship between trust and scientific publication with respect to image manipulation and representation in published works. On behalf of the American Journal of Botany editorial board and staff, I present our newly updated and revised policy on acceptable manipulation of digital images in articles for the American Journal of Botany. There are few endeavors where trust is more essential than in scientific publication. Readers of scientific papers start by assuming the validity of published articles in learned journals. We trust that the authors collected the data carefully and objectively, that they analyzed the evidence appropriately and without bias, and that they did not only include data that support a preferred outcome. We trust that results of a study are reported accurately and completely (NAS, 2009), and that relevant evidence or ideas generated by other workers are cited fully and fairly (ACS, 2006). If this trust is violated, either inadvertently or by intention, the integrity of published research and of individual researchers is called into question, which may undermine the public's confidence in scientists and science. Scientific progress may consequently falter. As individuals engaged in scientific research in service to the public, we therefore have an ethical responsibility to maintain the highest integrity in all aspects of our science and scientific publication (Botanical Society of America, 1997; Ecological Society of America, 2013). The adoption of comprehensive imaging and image-editing software to prepare digital images of research results represents a dramatic change in the publication process from the days of manual preparation of illustrations, and it also presents unique ethical challenges. The traditional techniques and methods survive in hobbyist and art photography circles, but are now largely obsolete in science. For most of us, image preparation and editing is done with Adobe Photoshop (Adobe Systems, New York, New York, USA) or similar software. The simplicity and efficiency of digital-image editing, and the resulting image manipulations, make possible the creation of high-resolution and professional-looking illustrations, at little cost. However, this simplicity also makes it possible and even tempting to modify or adjust digital images to “make them look better.” Some image manipulations are acceptable, if described fully, but other types of modifications and adjustments constitute inappropriate changes to the original data and are categorized as scientific misconduct (Rossner and Yamada, 2004; Society for Scholarly Publishing, 2013). Both acceptable and unacceptable image manipulations can now be reliably detected using features of the imaging software itself, and the field of image forensics has developed to investigate authenticity of digital image content (Swaminathan et al., 2008). Publishers, institutions, and funding agencies go to great lengths to ensure the ethical conduct of researchers (NAS, 2009; European Molecular Biology Organization, 2014; National Science Foundation, 2014; Natural Sciences and Engineering Research Council Canada, 2014; ORI, 2014) and the integrity of scientific literature (COPE, 2014; CSE, 2014; Nature, 2014; Royal Society Publishing, 2014). Nonetheless, the ultimate responsibility for integrity in published science lies with authors, reviewers, editors, and readers (ACS, 2006; Hammes, 2006). Many resources are available to guide authors on best practices in the preparation of digital images (e.g., COPE, 2014; NAS, 2009; CSE, 2014; Nature, 2014; ORI, 2014). These explain and illustrate acceptable digital manipulations, and how to identify and report unacceptable ones. The essential principles of best practice should include the following elements: (1) Authors should understand what constitutes acceptable image data manipulation. (2) Authors should document and report exactly how images were manipulated and also state when they were not manipulated. (3) All authors of a paper should review final images in a manuscript prior to submission for peer review and compare these with the original images to ensure that the visual results are accurately reported. (4) Reviewers should familiarize themselves with acceptable vs. unacceptable image data manipulations to be able to critically assess the quality of image data in the manuscripts that they are reviewing. (5) Editors should independently evaluate image data while taking into account the assessments of reviewers. If needed, editors should ask for additional information from authors (Cromey, 2010; Editorial Policy Committee, 2012; CSE, 2014). Individual journals and publishers have taken slightly different routes for codifying specific guidelines for computer-generated images. Some publishers devise their own, while others have adopted guidelines formulated by other bodies (Rossner and Yamada, 2004; CSE, 2014). We believe that the most comprehensive current guidelines for digital image editing and presentation in journals are those of The Journal of Cell Biology (JCB), published by Rockefeller University Press (JCB, 2014), now used by many journals (Editorial Policy Committee, 2012; CSE, 2014). As a publisher of large numbers of manuscripts containing image data (i.e., micrographs, gels, blots, tomograms) and a long-standing reputation for image reproduction of the highest quality, JCB developed explicit and detailed policies to ensure the integrity of published digital images (JCB, 2014). We believe that the standards and prohibitions described in the JCB guidelines should be common practice, and common sense, for all authors contributing to the American Journal of Botany. We are therefore adopting JCB editorial policies and guidelines for image manipulation, with permission of The Journal of Cell Biology Editorial Office (Rockefeller University Press; http://jcb.rupress.org/). The following guidelines expand upon our current policies for figure construction and digital manipulations (American Journal of Botany, 2014). Our goal is to enable authors to do the right thing, to help authors to understand what constitutes permissible image manipulation for optimal and accurate presentation of their data, and to avoid questionable or unacceptable image manipulations. Our explicit aim is to avoid situations in which the validity of the data, or the integrity of the author, is called into question. Adherence to these guidelines by all prospective authors will ensure that the final published work merits the full trust placed in it by other scientists and by the broader public. Images that will be compared with each other must be acquired and processed under the same conditions. No specific feature within an image may be enhanced, obscured, moved, removed, or introduced. The grouping of images from different parts of the same micrograph or gel, or from different micrographs or gels, fields, or exposures, must be made explicit by the arrangement of the figure (i.e., using dividing lines) and in the figure legend text. Adjustments of brightness, contrast, or color balance are acceptable if they are applied to every pixel in the image and as long as this does not obscure, eliminate, or misrepresent any information present in the original image, including the background. Nonlinear adjustments (e.g., changes to gamma settings, changes in color balance or tonal range based on threshold settings) must be explained in the Materials and Methods. Questions about a manuscript raised during or after review or publication will be referred to the Editor-in-Chief, who will request the original data from the authors for comparison with the prepared figures and who may refer the matter to the Publication Ethics Subcommittee of the AJB Editorial Board. If the original digital image data cannot be produced by an author when asked to provide it, the manuscript may be rejected or acceptance of the manuscript may be revoked. When the manipulation affects the interpretation of the data, the manuscript will be rejected or prior acceptance will be revoked. Cases of suspected misconduct may also be reported to an author's home institution or funding agency, following procedures recommended by the Committee on Publication Ethics (http://www.councilscienceeditors.org/resource-library/editorial-policies/white-paper-on-publication-ethics/) or, for U.S. authors, referred to the Office of Research Integrity (http://ori.hhs.gov/education/products/RIandImages/default.html). We hope that these clear and explicit rules will assist authors as they prepare high-quality illustrations for submission.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.