Open access, i.e., the free online availability and re-usability of scientific publications, leads to greater visibility, impact and equitable accessibility of scientific research results and knowledge for the global scientific community and interested public . In the traditional subscription-based model of scientific publishing, most articles require payment for access, which is particularly disadvantageous for scientific exchange and for people in resource-poor institutions and regions. Early attempts to make scientific journal articles and preprints openly available on the internet began with bottom-up initiatives of scientific researchers in the 1980s and 1990s, including but not limited to the high-energy physics preprint repository arXiv (Sect. ), the (OA since 1998, ) and the (OA since 1999, ). In the early 2000s, further OA journals were established by researchers, learned societies and innovative commercial publishers in the geosciences and life sciences, including , the and its predecessor (1971–2003), the , and . In parallel, major scientific institutions adopted the aims and further developed the concepts of OA publishing. For example, the has been signed by over 800 leading scholarly organizations worldwide, whereby the EGU was one of the first international learned societies among the signatories .

The proportion of OA articles among the large number of scientific journal articles (>2 million per year, published in >20000 peer-reviewed journals, e.g., ; ; ), however, increased only slowly during the first 10 years after the Berlin Declaration because most funds available to cover costs of scientific journal publishing were bound in subscription contracts. Traditional publishers moved only very slowly and reluctantly towards a proper OA publishing market . To accelerate the progress and transform the corpus of traditional pay-walled subscription journals to OA, scholars, scholarly organizations and research funders developed a variety of initiatives at institutional, national and international levels . These and related initiatives aim at redirecting funding from subscription or paywall access to OA in scholarly oriented and cost-efficient, i.e., cost-neutral or cost-saving, ways .

Different approaches are employed to provide free access to scientific content. These approaches vary in terms of timing of open access, copyrights, coverage of publications costs and funding sources. The three major forms are often described by the following categories, which are, however, only loosely defined:

Green OA (OA archiving) requires that authors self-archive their publications in repositories that are funded by their institution or by other sources. Publishers of the journals where the original article was published may impose embargo periods prior to archiving of closed-access articles, and they also may retain the copyrights for the distribution of the work. The green OA approach can help to advance OA, but it has not been proven to offer a viable alternative to traditional journal publishing and quality assurance. It can introduce ambiguities and confusion about the public availability and validity of different versions of scientific studies, where preliminary versions have been self-archived and the final validated version may be available through a subscription journal only. The open science community platform EGUsphere (Sect. ) enables the archiving of preprints as well as their linking and transfer to peer-reviewed scientific journals, including but not limited to the EGU interactive OA journals.

Transformative agreements (TAs) or “publish-and-read” agreements are contracts between scholarly institutions or consortia and publishers to transition publication output to OA; at the same time, these contracts ensure access to content that was previously covered via the institutional subscriptions to the journal. These agreements, thus, lead to OA publication for participating institutions while maintaining access to subscription-based content from other institutions or consortia that have not established OA. Such agreements offer flexible ways for publishers, institutions, consortia and countries of gradually converting subscription-based journals into OA . At the same time and as outlined in the OA2020's mission statement , it remains important to uphold and promote further improvements in scientific publishing at reasonable and competitive costs through innovation and competition. This involves not only traditional subscription publishers through TAs, but also innovative fully OA publishers through equivalent OA publishing agreements (e.g., institutional agreements with Copernicus Publications, Sect. ).

The evolution of the number of TAs in the that start per year and the number of OA articles published on the basis of TAs are shown in Fig. . Some examples of such TAs with traditional subscription publishers and of agreements with full OA publishers are listed in Table . Leading scholarly organizations and consortia in a growing number of countries have successfully negotiated and used TAs with major international scientific publishers to achieve high percentages of OA for their publication output during recent years (Fig. S1 in the Supplement), while some national organizations continue to resist these developments due to concerns over supposed financial sustainability, perceived inequities in cost distribution or speculative fears about future cost increases .

Several countries and scholarly organizations have already achieved OA for more than 90 % of their scientific journal publishing output using TAs with traditional publishers, agreements with proper OA publishers and further OA publishing funds for scientists to flexibly cover OA article processing charges. For example, the German Max Planck Digital Library (MPDL) has converted more than 95 % of the publication output of the Max Planck Society into OA via the German National Consortium DEAL with big traditional publishers (Elsevier, Springer, Wiley) and via individual contracts with numerous OA and learned society publishers .

This development towards a full OA landscape helps to maintain and increase bibliodiversity of scholarly scientific publications, i.e., the variety of publishing formats, models, platforms and outlets used to disseminate scientific knowledge in different disciplines, languages and regions of the world . It also includes the range of financing models and mechanisms (e.g., gold/diamond OA) to make scientific publishing and its access inclusive and equitable. This desired outcome will be reached if OA agreements not only focus on large publishers but are also extended to smaller publishers and new publishing models. This is, for example, practiced by the in the UK that can be considered a role model for how library agencies could and should provide OA publishing agreements to learned societies and small innovative OA publishers like Copernicus. Acknowledging this evolution toward full OA to all scientific literature, Time magazine recognized the advances in OA achieved by transformative agreements as one of the “13 ways the world got better in 2023” .

The average fee per article in the traditional subscription publishing model is approximately EUR 4000, whereas OA journals of similar quality are sustainable with article charges of EUR 2000 or less . This was shown by in a comprehensive analysis of the global scholarly journal publication market with a financial annual volume of ∼ EUR 10 billion, mostly paid by publicly funded academic libraries. Similar differences were found in more recent, less comprehensive studies . In other words, the costs of traditional subscription journals are on average higher by a factor of 2 than required to produce high-quality publications. This is also reflected by the high profit margins of traditional journal publishers, often exceeding 35 %, i.e., higher oligopoly revenues than in many other industries . Thus, concerns that OA threatens the financial viability of the academic publishing system, as, e.g., put forward by , are unfounded because much more public funding than needed for OA publishing is bound in the traditional subscription journal business. The financial benefits and savings through transformative agreements have been clearly demonstrated by the MPDL and the German DEAL consortium. For example, the MPDL managed to convert practically all scientific journal publication output of the Max Planck Society from subscription to OA over a period of 6 years without any increase in expenses in spite of substantial inflation (2018–2024: approx. EUR 14 million unchanged; see slide 13 in ). Moreover, the DEAL consortium managed to obtain OA for all research articles with corresponding authors from German institutions while reducing the overall expenses for Elsevier journals from approx. EUR 50 million in 2015 to approx. EUR 30 million in 2023, corresponding to a 40 % cost reduction . The opportunities and viability of cost saving through OA are confirmed by the following: while nearly 50 % of new peer-reviewed articles are now published OA, 80 % of publisher revenues are still bound in opaque subscription fees . On average, 1 %–2 % of total research budgets by scientific institutions is spent on literature and information provision and publishing . However, the distribution of these funds is important in the OA transition and also implies that financial flows within the institutions should be adjusted accordingly.

OA publishing models have led to valid concerns about the potential impacts of OA on the quality of scientific content . Since APCs are levied for individual articles, unlike in subscription payments which are remunerated to entire journals or journal families, every published OA article enhances the publisher profits. As a consequence, the APC business model led to the launch of a large number of journals motivated purely by commercial interests. These journals solicit a high number of submissions that often undergo only cursory (if any) peer review, resulting in publications of low scientific quality, and are thus frequently referred to as “predatory journals” . We strongly propose that such economically driven aberrations of OA publishing be counteracted by measures of quality assurance providing transparent evidence of rigorous peer review. With the goal of maintaining or even improving scholarly quality assurance, the EGU interactive OA journals and other innovative OA publishing platforms applying various forms of open and/or transparent peer review and different OA approaches (diamond/gold) were launched even prior to initiatives aimed at the OA transformation of traditional subscription journals (Sects.  and ).

The idea of sharing non-peer-reviewed manuscripts, nowadays called “preprints”, within scientific communities reaches back several decades: in the 1960s, the National Institute of Health circulated manuscripts by regular mail within “Information Exchange Groups” . At the same time, researchers in the Soviet Union were encouraged to deposit their papers on VINITI for efficient dissemination . Similarly, in the 1970s, distributed manuscripts on physics- and mathematics-related topics prior to publication, which eventually led them to launch the first preprint server, , in 1991. By now the same concept is applied to numerous other discipline-specific servers

Examples include biorxiv.org (https://www.biorxiv.org/) (2013), socopen.org (https://socopen.org) (2016), psyarxiv (https://psyarxiv.com) (2016), chemrxiv.org (https://chemrxiv.org/engage/chemrxiv/public-dashboard) (2017), paleorxiv.org (https://paleorxiv.org) (2017), LawArXiv (https://osf.io/preprints/lawarxiv) (2017–2021), eartharxiv.org (https://eartharxiv.org) (2017), metaArxiv (https://osf.io/preprints/metaarxiv) (2017), medrxiv (https://www.medrxiv.org/) (2019), techrxiv (https://www.techrxiv.org) (2020); all links last access: 24 January 2025.

In parallel to arXiv, publishers launched more interdisciplinary preprint servers, e.g., (acquired by Elsevier in 2016), by the Nature Publishing Group, by MDPI, by Wiley/AGU and (acquired by Springer Nature in 2022). Preprint servers have also been launched by non-profit organizations such as . To facilitate access and visibility of scientific publications in developing countries, regional initiatives started even before the official open-access initiatives . They include the Scientific Electronic Library Online of OA journals in Latin America and South Africa that was later complemented by . To date, the adoption and utilization of preprints greatly vary across regions and scientific disciplines . Other research communities started promoting the advantages and benefits of preprints such as Accelerating Science and Publication in Biology . In 2001, EGU/Copernicus started the interactive journal discussion forum for Atmospheric Chemistry and Physics to share papers in an interactive discussion and peer review; this concept has by now been adapted in the 18 newer EGU journals (Table ) and complemented by the interdisciplinary preprint repository .

Generally, preprints are indexed in common databases (Google Scholar, Scopus, since 2017) and are thus citable as non-peer-reviewed publications, sometimes also referred to as gray literature. Both benefits and shortcomings of preprints became particularly obvious during the COVID-19 pandemic: on the one hand, the quick dissemination of novel scientific evidence was crucial to collaboratively developing solutions to the global crisis. On the other hand, the public and media are often not fully educated about the non-peer-reviewed, sometimes preliminary status of the results presented in preprints and how to interpret the published information, potentially leading to premature assessments or conclusions .

To overcome the lack of quality assurance for preprints, proposed extending arXiv by a peer-review model, similar to journals; however, this idea did not succeed in the suggested form. Later, similar concepts were termed the “publish-then-review model” or peer review of preprints in “overlay journals”, in which manuscripts on preprint servers undergo peer review (Sect. ). An example of successful overlay journals combining arXiv preprints with an interactive OA publishing concept are the published since 2016 in physics and other fields, with a structure similar to that in the discussion forums of the EGU journals and EGUsphere (Sects.  and ).

When ACP was launched in 2001 as EGU's first interactive scientific OA journal, preprints posted for public review and discussion were labeled as “discussion papers”. This labeling indicated that these papers had already passed some form of basic scientific access review by an editor, optionally supported by referees, before they were accepted for public review and discussion in Atmospheric Chemistry and Physics Discussions (ACPD), the discussion forum of ACP. It is clearly indicated on all relevant web pages and through watermarks in the papers' PDF files that discussion papers are not fully peer-reviewed scientific articles as opposed to final journal papers. For clarification across and beyond the field of geosciences, this was also expressed in an official “EGU Position Statement on the Status of Discussion Papers Published in EGU Interactive Open Access Journals” . Since ACP's launch, the community has readily embraced and accepted this distinction between discussion papers and final journal papers. In fact, ACP and EGU's interactive OA publishing initiative might not have succeeded without this clear labeling – and at that time even quite distinct typesetting – of discussion papers/preprints prior to being publicly revealed. These measures and the introduction of digital object identifiers (DOIs) and related labels effectively dispelled widespread initial concerns and misperceptions about plagiarism of preprints and introduced the novel concept of preprints with public peer review many years prior to the launch of traditional preprint servers in the Earth sciences . By now, most publishers accept submissions of previously preprinted manuscripts for peer review and possible subsequent publication in a peer-reviewed journal.

To date, the more general terms “preprint” and “preprint repository” for the different types of preprints (with and without peer review) on EGUsphere (Sect. ) largely replace the terms “discussion paper” and “discussion forum” across the EGU and in the wider (geo)scientific community. This adaption simplifies the terminology in the (geo)scientific publishing landscape and on the web page structures of the EGU and its publisher Copernicus. We propose, however, that the term discussion paper should not be fully abandoned as a useful label for preprints that underwent a scientific preselection process through access review by a journal editor (Sect. ) and are undergoing full, public peer review (Sect. ), as opposed to other traditional, stand-alone preprints that have not undergone and may never undergo any substantial scientific quality assurance. Such distinction is valuable for our understanding of an epistemic web as a dynamic and interconnected space to trace the creation, sharing and construction of knowledge.

During the past century, different publishing formats have emerged that allow for distributing work prior to publication in scientific journals and to receive feedback by peers. An early example of discussions of unpublished work are the , a journal launched in 1947 by the Royal Society of Chemistry. It publishes research papers presented at “Faraday Discussion Meetings”, together with a record of the questions, discussion and debates that had occurred during the meeting. However, the discussion is limited to the meeting participants only and thus greatly differs from today’s public, interactive discussions on online platforms.

The potential of the internet for interactive discussions among much wider communities was recognized by , who implemented the concept of “scholarly skywriting” into the OA journal Psycoloquy (discontinued as of 2002), which allowed authors to solicit feedback by peers from all over the world on their new ideas and findings. In 2002, Berkeley Electronic Press (Bepress), in collaboration with the California Digital Library, launched the eScholarship Repository to share “working papers” in the humanities and social sciences to allow for soliciting feedback before formal peer-reviewed publication

Bepress was acquired by Elsevier in 2017 .

An overview of all EGU journal titles and the year of their first publication applying the interactive publishing model is given in Table . Some journals were initially launched with closed access (ANGEO: 1983, NPG: 1994, NHESS: 2001) but were subsequently converted into OA journals in 2009, 2014 and 2004, and they introduced interactive, multi-stage public peer review in 2018, 2014 and 2013, respectively. From 2004 onward, all newly launched EGU journals have been full OA journals with multi-stage public peer review and interactive discussion from their start.

Figure  shows the number of submissions, preprints (discussion papers) and published papers in each journal (panels in the first and second rows) and the total of all EGU journals (third row), as well as the cumulative numbers since 2001 in the bottom row. In most journals, there is a continuous increase in the number of papers in each of the three categories. The downward trend in published papers for ANGEO (red lines in Fig. f) is striking. Initially, the publication costs of ANGEO were largely covered by institutional subscriptions. In 2009, ANGEO was converted into an OA journal. As of then, authors had to pay individual APCs, which in many cases may not have been covered by institutional agreements (Sect. 4.1.7). This change in the financing model may have contributed to its decreasing submission rate.

The difference between the number of submissions and the number of preprints in all EGU journals is on average <20 %, reflecting the relatively low rejection rates of submitted manuscripts during access review (Sect. ). The difference between the number of preprints and final published papers is even smaller (∼ 10 %, Fig. ). These low overall rejection rates demonstrate the efficiency of the self-regulating concept that authors are more likely to submit high-quality papers and of the multi-step workflow that filters out weak or out-of-scope papers at the access stage (Sect. ). Thus, manuscripts that enter the discussion stage of a journal and on EGUsphere have a high probability to be accepted for final publication after revision and peer-review completion.

There is a clear difference between the two journal groups without and with an editor decision immediately after the discussion (Sect. ): whereas on average 3 % of papers were rejected in 2024 in the first journal group, this rate is much higher (∼ 10 %) for the second group (Fig. f). This suggests that the majority of authors are able to successfully improve their papers during revision to meet the expected standards. Based on these data, it appears that excessive editorial intervention during the revision process is unnecessary and, in fact, counterproductive. The overall relatively low rejection rates in EGU journals are in agreement with findings that submissions to journals with public peer review are often of higher quality, and authors also make a larger effort to revise the manuscripts carefully and sufficiently since all versions are permanently archived . On the contrary, (very) low rejection rates in OA journals without documentation of peer review may point towards predatory practices; i.e., these journals do not necessarily focus on critical peer review, but are driven primarily by commercial interests and low standards of scientific quality assurance .

An additional consequence of the public peer-review process is the very low retraction rate in the EGU journals (≪0.1 % yr⁻¹). In ACP, 13 papers have been retracted since its launch in 2001 (compared to about 16 000 published papers). Most of them (eight) were labeled as retractions during the early years of the journal, when the concept of discussion papers vs. journal papers was not fully established and accepted by other publishers. These retractions were initiated by authors after unfavorable reviews or rejections of their papers so that they could be submitted to other journals; in later years, such papers were typically labeled as “withdrawn”, consistent with the terminology as used in other journals. The remaining five papers (0.03 %) were retracted from ACP because of invalid results as identified after publication. These trends are in line with results from a comparison of different peer-review models that revealed that during the public peer review, major flaws or fraud are usually caught and are either corrected by the authors during revision or result in a rejection . These retraction rates are lower by at least an order of magnitude as compared to those in journals that do not provide transparency of peer review .

ACP was not only the first but also is by far the largest EGU journal, with ∼ 800 published papers annually during the past decade (Fig. a). While the number of submissions to ACP has somewhat increased since 2016, this increase is not fully reflected in the numbers of published preprints because concomitantly the rejection rate has slightly increased from ∼ 15 % to 20 % (Fig. a). To illustrate the evolution of ACP in more detail, Fig.  shows the trend in the distribution with regards to the , up to two of which are selected by the authors during submission.

The areas “aerosols” and “gases” clearly dominate, consistently representing >70 % of all submissions, whereas the areas “clouds and precipitation”, “dynamics”, “radiation”, “isotopes”, “biosphere interactions” and “hydrosphere interactions” received a minor share of submissions. Areas that describe interactions of the atmosphere with other Earth compartments (biosphere and hydrosphere) contributed < 5 % to the total submissions; these topics are extensively covered in other EGU journals, such as BG, OS and HESS, which have grown during the last two decades. To consolidate the focus of ACP, the subject areas of “isotopes”, “biosphere interactions” and “hydrosphere interactions” were discontinued as of 2023 and “radiation” as of 2025. Isotopes are now covered within the areas of aerosols, gases and clouds, and biosphere–atmosphere interactions and hydrosphere–atmosphere interactions are merged with the new, more comprehensive subject area “climate and Earth system” that is listed as a primary or secondary subject area on ∼ 8 % of all submissions by now. To react to recent developments in the (geo)science community and to explicitly attract submissions, the research activity “machine learning” was introduced in 2022 to supplement the traditional categories “laboratory studies”, “field studies”, “atmospheric modeling” and “remote sensing” that have existed ever since the start of ACP. Two dedicated editors with expertise in machine learning were appointed in 2022 and 18 editors in climate physics and dynamics were appointed in 2023 to handle and attract submissions in these areas.

The 19 EGU journals have several manuscript categories, i.e., different types of articles. The main category in all journals is research article (termed “regular papers” in ANGEO) and review articles (“review papers” or “reviews” in some journals; Table S1 in the Supplement). Currently eight journals accept submissions in the letter category aimed at concise articles targeted to become highlights and to be included in the virtual compilation EGU Letters (Sect. ). The term “virtual” is used here to describe curated, non-indexed compilations of articles that were published in EGU journals. All journals (except GC) allow for the submission and the publication of peer-reviewed comments. These comments provide the option to continue the public, interactive discussion that occurred during the peer-review process of a paper even after its publication as a final journal article. These peer-reviewed comments usually refer to papers published in an EGU journal but could also refer to papers published in another journal. These commentaries have to present sufficiently substantial content that can be publicly discussed, and their preprints undergo peer review and interactive discussion just as all other manuscripts. They often receive a high number of comments from the community, authors and referees (Sect.  and Table S2). In journals with closed, non-public peer review, such post-publication comments are the only way to point out to the scientific community any deficiencies of a paper that were not caught during the closed peer review.

Research articles and review articles comprise by far the largest group of articles as shown for ACP in Fig. ; the proportions in other EGU journals are similar. The manuscript category “technical notes” was introduced in ACP to distinguish regular research articles from submissions that describe mainly new measurement techniques, instruments or models without an in-depth presentation of new results and discussion of their implications for atmospheric science. The launch of the journals AMT and GMD in 2008 provided alternative platforms for such type of studies and allowed ACP to (re)focus on studies highlighting the importance of chemical and physical processes for our understanding of the state and behavior of the atmosphere and processes therein. A strong increase in ACP submissions on local or regional air pollution studies, for example in Asia and other regions with growing atmospheric research communities, triggered the introduction of the manuscript category “measurement reports” in 2020. Such articles often contain valuable datasets at undersampled locations or technical developments of interest to ACP readers but often lack detailed analysis and scientific interpretation and discussion of implications that are required for research articles. In their initial year 2020, most published measurement reports were submitted as a research article and were then re-categorized by the handling editor. By now, this category is more established, and a large proportion of measurement reports are submitted directly in this category. Technical notes and measurement reports contain their category in the paper title (e.g., “Technical note: a new instrument for the detection of …”) to clearly distinguish them from research articles.

Also in 2020, ACP was the first EGU journal that introduced the manuscript category “letters” (Sect. ). Manuscripts in this category and also articles in the category “opinions” in ACP are expected to be ranked as highlight articles upon acceptance for final publication. Unlike highlights selected upon the editor's choice that can be papers of any manuscript category (Sect. ), letters and opinions are already designated by default to become highlight papers at the submission stage. Therefore, the executive editors check the suitability for these two manuscript categories prior to the discussion stage (“Letter manuscript” and “Journal-specific MS type expected to become highlight” in Fig. ). The decrease in the number of highlight articles since 2020 (Fig. b) is likely due to the introduction of an additional step in that year, in which the executive editors have to approve (or reject) the handling editor's highlight recommendation (Sect. ). The proportion of highlight papers in the EGU journals varies greatly (Table S4): on average <10 % of all papers were highlighted in 2024 in most EGU journals. However, the proportion was nearly 30 % in ESD, whereas none were highlighted in GI and SE. In addition to the editorial highlighting, some EGU journals give a publication award for the most outstanding papers, which are selected by the executive editors or an independent committee (e.g., ).

Special issues (SIs) are compilations of articles about a specific topic within the scope of the journal, such as field campaigns, conferences or research themes. Unlike in traditional journals where SIs are published as separate (hard copy) issues, SIs in EGU journals are fully virtual and articles are listed on a dedicated SI web page, in addition to publication on the regular journal web page. SIs are typically open for submission for 1 to 2 years. This way, the publication of SI papers is not delayed by late submissions to the SI. All SI papers undergo the same process of peer review and publication as regular submissions.

Members of the scientific community can propose an SI in any EGU journal. Such requests can be made via a proposal specifying the SI's topic, duration and approximate number of papers, ideally accompanied by a preliminary list of paper titles. The executive editors of the journal check the proposal for its suitability for the journal and may request changes to the scope if needed. A large proportion of all SIs (∼ 230 of >500 since 2001) are organized as inter-journal SIs, in which two or more journals participate (Sect. ). These may include any EGU journal as well as other journals published by Copernicus.

In most EGU journals, SI submissions can be handled by dedicated SI guest editors in addition to regular editorial board members. Guest editors are usually nominated by the SI proposer but they require approval by the executive editors. In 2020, ACP introduced a change in the handling of SIs such that all submissions to SIs are handled by the regular ACP editorial board members (Sect. ), and guest editors are no longer allowed. Two regular ACP editorial board members, who are not closely involved in the activities from which the special issue arises, act as SI coordinators. They oversee the SI in exchange with the executive editors but are not expected to handle the review process of all SI submissions. In addition, one to three SI co-organizers (often the proposers of the SI) are appointed to exchange with the authors and other members of the scientific community; they do not act as SI editors though.

The proportion of SI submissions, preprints and final journal papers in ACP clearly dropped during the last years from ∼ 30 % to ∼ 10 % (Fig. ). This trend may be partially ascribed to the new SI guidelines as introduced in 2020 but possibly also to the COVID-19 pandemic, during which time significantly fewer field studies and conferences took place that are often the scope of SIs. The percentages of preprints and final journal articles are similar, indicating comparable acceptance rates for regular and SI papers. The slightly higher proportion of SI preprints (relative to all preprints, i.e., regular and SI) compared to the proportion of submissions is likely due to the prior approval of the SI by the executive editors resulting in relatively fewer out-of-scope rejections.

The organization and the overall trend in the proportion of SI papers in EGU journals are in contrast to those in journals by some large OA publishers that actively solicit an increasing portion of submissions via SIs that are typically handled by guest editors (often >50 % of all submissions) that frequently result in SIs that contain only very few papers . Publishers often advertise such special issues in terms of both manuscript submissions and guest editorship; incentives for article submissions may even include the offer to publish articles at discounted rates. Such strategies, purely motivated to increase revenue, have led to a huge increase in special issues by large commercial publishers, often resulting in compilations with low scientific standards and little or no peer review. The resulting lack of scientific quality control in such SIs recently prompted the Swiss National Science Foundation to no longer fund any OA papers in special issues in order to reduce the investment of public money in publications of low scientific quality and value . However, such measures seem inappropriate and unjustified for journals such as those of the EGU/Copernicus that apply and demonstrate consistent handling and transparent scientific quality control for all papers.

In 2024, the EGU journals introduced paper collections as an additional way to group papers on a specific topic. Similar to the SIs, collections can be organized across Copernicus journals as inter-journal collections, in which papers are (co-)listed on dedicated collection web pages, in addition to the regular journal pages. Unlike SIs that have a limited submission period (typically 1–2 years), collections do not have a pre-defined end date. Papers will be only included in collections upon their acceptance as final journal papers. This implies that all papers in collections have undergone the regular public peer-review process handled by a regular editorial board member. The article selection for inclusion in a collection is made by the executive editors of the journal.

All editors appointed in EGU journals work on a purely voluntary basis without remuneration, in line with the not-for-profit philosophy of the EGU. Editorial boards are typically fairly large so that each editor is expected to handle a comparably small number of manuscripts per year (e.g., six manuscripts per year in ACP). Editors are selected by the executive editors either upon nomination by colleagues or self-nomination via open calls or spontaneously. They are initially appointed for 3 years but they may stay for several terms if they fulfill their duties over a longer period of time. In several EGU journals, outstanding dedication and performance of editors are recognized by annual awards, e.g., and . Each journal organizes editorial board meetings at least once a year and executive editor meetings as necessary. The executive editors of the EGU journals and the EGUsphere coordinator form the EGU Publications Committee, together with representatives of the EGU Executive Board and Copernicus Publications as ex officio members. Recently, a representative of the EGU Early Career Scientists (ECSs) was appointed as an additional member of the committee to allow for greater representation of early-career perspectives and to enhance engagement with the ECS community.

All EGU journals apply consistent guidelines regarding EGU journal editor obligations. In addition, journals may develop specific guidelines, such as the and (Sects. S1 and S2 in the Supplement), , and .

The inaugural ACP executive editor committee in 2001 consisted of five members (Ulrich Pöschl, Thomas Koop, Ken Carslaw, Bill Sturges, Rolf Sander), three of whom continued in this role for more than 20 years (Ulrich Pöschl, Thomas Koop, Ken Carslaw). In 2022, Ulrich Pöschl and Thomas Koop stepped down from their editorial duties and joined the ACP advisory board. To facilitate continuous and efficient oversight of the journal, it is currently led by two executive editors (Barbara Ervens, since 2019, and Ken Carslaw). Instead of re-creating a larger executive committee, in 2021, the role of Senior Editors was introduced as an intermediate level between the ∼ 160 regular ACP editors and the executive editors. Eight Senior Editors were selected by the executive editors as distinguished and experienced members of the editorial board based on their outstanding commitment and excellence as previously demonstrated in their editorial work. They were selected such that each journal subject area is covered by two Senior Editors according to their scientific expertise and core interests. The Senior Editors take responsibility for the editorial coordination of their subject area(s) and carry out a series of tasks to ensure scientific quality and foster submissions, in collaboration and exchange with the executive editors (Sect. S3). The appointment as an ACP Senior Editor is for 2 years, renewable upon mutual agreement with the executive editors.

Generally, all ACP editors can flexibly choose their workload throughout the year; they are expected to react promptly to the automated editor calls or manual assignments by Senior/Executive Editors (Sect. ). In the initial years of ACP, authors of papers which did not find an editor via automated calls in due time were encouraged to contact an editorial board member of their choice to request the handling of their paper. If the authors were not successful, the paper was automatically rejected. Due to the considerable growth of ACP starting in ∼ 2015, such personal handling requests led to imbalances in the workload of some editors. To facilitate a more even distribution of the workload and to avoid potential conflicts of interest, editors are now explicitly discouraged from responding to such requests. Instead, Executive/Senior Editors assign such papers to editors with low workload or reject them if they are out of the journal scope or of low scientific quality. As editors often hesitate to handle papers of low quality or those at the edge of the journal scope, ACP introduced in 2020 the option of anonymous editor decisions at the access stage (before the preprint is accepted and posted for the public discussion; Sect. ), i.e., allowing paper rejections on the behalf of the journal instead of individual editors. As soon as a preprint is accepted for the discussion stage, the editor's identity is automatically revealed to the authors.

Peer reviewers have become the most limited resource in the entire scientific publication process due to the increase in the number of scientific submissions, publications and journals . The number of review requests per preprint in EGU journals has approximately doubled (from ∼ 5 to ∼ 10) since 2009 (Fig. b, d, f). To enhance the efficiency of referee identification and nomination, the publisher Copernicus continues to develop and improve search tools to suggest suitable referees . Currently, editors can use the following tools that are provided via their editorial interface:

The Copernicus in-house development of cREACTS (“Copernicus REferee ACTivity Score”) ranks potential referees based on the likelihood that they provide a referee report, guided by the referee's statistics in Copernicus journals by considering accepted/declined/fulfilled referee calls during the preceding 12 months.

The processing times for the six steps of the multi-stage peer-review process, as detailed in Sect. –, are displayed in Fig.  for the 19 EGU journals. The first panel shows the time between manuscript submission and its posting on EGUsphere and in the journal discussion forum. The first phase from the manuscript submission to initial decision (dark blue) includes the technical file validation by the Copernicus editorial support team and the assignment of the manuscript to a handling editor. This step takes less than 2 weeks on average for most journals. This time is similar for all journals, independent of whether they employ automated editor calls or use manual editor assignment, as is done in the small journals with ≲150 submissions per year (ESurf, GChron, GC, NPG, SOIL, SE). The initial steps take the longest in the journal GMD (>30 d). This may be due to the fact that its submission requirements are particularly strict as authors have to provide full codes of the underlying models that form the basis of their study (Sects.  and S4.2); such checks and approval often require multiple rounds of file upload and validation. The total processing times, from submission to final paper publication, are the longest for the 10 EGU journals that apply a post-discussion editor decision between the first and second stages of the review process (ANGEO, BG, CP, ESD, GC, GChron, HESS, NHESS, SOIL, TC) with on average 215 d vs. <200 d in journals without this intermediate editor interference (12-month median, December 2024). The overall processing times in EGU journals (∼160–260 d) are comparable to journals practicing traditional peer review without public discussion . However, unlike in journals without public peer review, the preprints intended for peer review and publication in an EGU journal are usually posted within 1 to 2 weeks at most, and their evolution and open discussion can be tracked by the public.

Article processing charges (APCs) are the fees that are charged for OA articles (Sect.  and ). They cover all costs for the production of peer-reviewed OA papers. They include the review support provided by the publisher office, online Supplement, typesetting, English language copy-editing, archiving, and distribution of papers and interactive comments, i.e., the maintenance of websites and servers, as well as electronic copies in open archives. The breakdown of APCs relative to total publications costs in journals (including those by the EGU) of Copernicus Publications is shown in Fig. . The publisher's business profit margin (after taxes) is ∼ 6 %. This is significantly smaller than the margin of many commercial publishers that often exceeds ∼ 30 % . Copernicus Publications re-invests their surplus into training of new staff, journals owned and fully financed by the publisher, and enhancement of services and for outreach activities, operated by their non-profit association Copernicus e.V.

Until 2016, discussion papers (preprints) for EGU journals were typeset by the publisher, and authors were charged APCs upon publication of their discussion paper in a journal discussion forum. This procedure was changed in 2016, in analogy to other preprint servers and publishing platforms: no typesetting is performed and no APCs are charged at the first publication stage. Nevertheless, an initial technical file validation is applied free of charge, including a check of author names, affiliations and general formatting requirements, as well as the addition of the EGUsphere logo and citation to the PDF files. After successful peer review, APCs are charged for the main journal article. Supplemental information can be added free of charge as a separate file (e.g., text, tables, figures), to which no copy-editing is applied by the editorial support. Additional assets can be connected free of charge to the paper, including video supplements or Jupyter notebooks.

Until the end of 2024, APCs were charged per page in most EGU journals (EUR 93 or 77 per page, using the publisher's packages for Word or LaTeX templates, respectively). GMD and CP started in 2021 charging fixed per-article APCs of EUR 1600 (net); OS joined this pilot project in 2022. Since then, the article lengths in these three journals have not significantly changed. As an investment, the EGU supports new journals during their start-up phase, allowing for full waivers in the initial period and for reduced APCs during a transition time. APC discounts (-33 %) were applied to all articles in GChron and WCD until the end of 2024; full waivers were given to GC until July 2025 but full APCs have been charged since then. Both GChron and WCD have been charged full APCs since January 2025. These three newest EGU journals, launched in 2018 and 2019 (Table ), were accepted by Clarivate for the Web of Science in September 2023 and, thus, received their first JIF in 2024. This milestone in the journal development was used up to now as the criterion across the EGU journals to start charging APCs. An alternative criterion may be applied to journals launched in the future, given the limited relevance of the JIF (Sect. ).

Since January 2025, the APCs in all EGU journals have been charged on a per-article basis to make the APC structure more transparent for all authors. The current APC scheme is structured into three journal groups with per-article net APCs of EUR 1800, 1350 and 1980 (Table ). Given that typesetting and copy-editing are major components in the paper production process (Fig. ), the journal groups were created based on the average article length using the statistics during the last 3 years. Short articles such as letters or peer-reviewed comments have approximately the same length in all EGU journals. Therefore, their standard net price is EUR 900 in all journals. As a benefit for its members, the EGU, as the journal owner, grants a 10 % APC discount for papers by corresponding authors (CAs) who are EGU members at the time of manuscript submission. This discount strengthens the EGU's commitment to open science as a member-led, community-driven learned society, distinguishing it from profit-driven interests by commercial publishers. As the largest European geoscientific society, the EGU particularly aims at encouraging scientists from all regions in Europe to participate in the Union's activities, including publishing. Therefore, CAs affiliated in European economically disadvantaged (EED) countries

The classification is based on the list of 46 member states of the Council of Europe as well as Kosovo. EED countries are defined as European countries with a most recent gross national income per capita (GNI, World Bank) that is less than 25 % of the maximum value of all European GNIs. The list is updated quarterly and currently includes Albania, Armenia, Azerbaijan, Bosnia and Herzegovina, Bulgaria, Croatia, Georgia, Greece, Hungary, Kosovo, Latvia, Lithuania, Moldova, Montenegro, North Macedonia, Poland, Romania, Serbia, Slovakia, Türkiye and Ukraine.

Initially, manuscripts under peer review and discussion in an EGU interactive OA journal were termed “discussion papers”, such as in Atmospheric Chemistry and Physics Discussion (ACPD) that started in 2001 (Sect. ). These papers were paginated in a regular volume, issue and page structure and had a DOI that included the journal and pagination. From 2016 onwards, discussion papers were no longer typeset and had a DOI with the structure “journalD-year-number” (with the D to reflect the discussion stage), disclosing the fact that they had been under review for a journal. When such papers were rejected after being posted and discussed in a journal discussion forum, publishers outside the EGU journal family sometimes declined them for peer review in their journals since the DOI implied “previously rejected” or falsely as “already published” journal papers. With the increasing popularity of preprints and publishing platforms for peer review of preprints across the scientific community, the term “preprint” was also adopted for the EGU discussion papers as it more clearly indicates the status of a manuscript prior to potential publication in a journal (Sect. ).

The interdisciplinary repository was launched in 2022, which merges the features of preprint servers with those of the established, well-functioning EGU journals with interactive multi-stage peer review. For the geosciences community, EGUsphere serves as a repository of non-peer-reviewed preprints and preprints that undergo peer review with the intention of publication in one of the EGU journals, as well as conference abstracts and presentations. Unlike on traditional preprint servers, the complete peer-review process of preprints intended for eventual journal publication is guided by a journal editor. Preprints that do not undergo peer review are not handled by journal editors but by preprint moderators. This differs from many other publishing platforms where editorial duties may only include the nomination of reviewers but no other editorial decisions. On such platforms, preprints may only receive reviewer ratings such as “approved” or “non-approved” without a final editor decision (Sect. ).

All preprints on EGUsphere are assigned a DOI with the structure “egusphere-year-number”, regardless of whether they undergo peer review or not. Such a DOI clearly reflects the fact that they are not (yet) peer-reviewed papers but are preprints that may or may not undergo peer review. This change to the DOI setting, as compared to the initial format “journalD-year-number” that included the journal name, eliminates previous problems faced by authors whose papers were rejected and subsequently declined submission to other journals. Only after successful revision and peer-review completion are papers published in the respective EGU journal with a DOI “journal-year-firstpage”. During a transition period until the end of 2024, some EGU journals still offered both options; i.e., preprints were either posted on EGUsphere (DOI “egusphere-year-number”) and linked to the journal or only posted on the journal website with a DOI “journalD-year-number”. As of January 2025, the latter route has been abandoned; i.e., all submissions to EGU journals are automatically labeled as an EGUsphere preprint. Journals indicate the journal relation during the open discussion on EGUsphere stating that the preprint is “under review for journal name”. This label is removed when a preprint reaches its final status (i.e., rejection or journal publication). Final journal publications are ultimately linked to the initial preprint.

Figure  displays the full concept of EGUsphere, including the three preprint options and conference contributions.

Preprint Option I corresponds to the former discussion papers that undergo the six steps of the interactive, public peer-review process as described in Fig. . Such preprints aimed at publication in an EGU journal undergo peer review, handled by a journal editor, according to the requirements and principles of the individual journals (discussion period, additional items during submission, etc.). Also referees are expected to apply the standards of the target journal. At submission, authors have to choose in addition to journal subject areas. These keywords may be eventually used for an interdisciplinary, searchable index for the EGUsphere content. Preprints in the discussion phase are posted on the EGUsphere website and linked to the journal website to enhance their visibility. Upon publication of a paper in an EGU journal, the EGUsphere preprint and the full documentation of its public peer review and discussion are linked. Preprints submitted with the intention of journal publication that are, however, rejected at the access stage may be recommended by the handling journal editor to become a stand-alone preprint (Option III). If authors refuse this option, there will not be any trace of the submission on the journal website or on EGUsphere.

Preprint Option II allows authors to seek peer review of preprints posted on other preprint servers that provide a DOI (e.g., arXiv). Such “external preprints” maintain their original DOI while they undergo public discussion and peer review on EGUsphere, thereby avoiding double preprinting, i.e., an assignment of two different DOIs to an essentially identical preprint. If authors would like to make changes to their original external preprint prior to the public discussion on EGUsphere, they are asked to create an updated version of their preprint on the external server. If their updates lead to a manuscript with similarity of less than 60 % as compared to the latest preprint version, their manuscript may be considered sufficiently different and treated according to Option I.

Preprint Option III represents stand-alone preprints; i.e., they are posted without a journal relation and, thus, are equivalent to preprints on traditional preprint servers (e.g., arXiv). They undergo basic checks regarding their relevance for the Earth, planetary and space science community and concerning general standards of scientific quality, originality, civil discourse and common decency. These checks are performed by , who are often early-career scientists with little experience in publishing. The role as EGUsphere moderator gives them the opportunity to get involved in OA publishing beyond the author role, but without the comparably high responsibilities as expected for journal editors, making such a role a valuable training opportunity. EGUsphere moderators are automatically called upon submission of new stand-alone preprints, similarly to editor calls in journals, but on shorter timescales. If no moderator responds after two calls, the EGU Editorial Manager contacts moderators of the matching topics so that all preprints are usually posted within ∼5 d after submission. The public discussion of a preprint is limited to 5 years, during which authors can post updated preprint versions, with the version number reflected in the DOI, with versions indicated by a suffix (-v2, -v3, etc.). Both the EGUsphere coordinator and moderator are informed of any new community comments, but they are not obliged to actively guide the discussion.

At any time during the discussion period, the authors can change their stand-alone preprint (Option III) into a preprint of Option I. Authors of stand-alone preprints are encouraged to actively solicit feedback from the community, which may help them to prepare a more elaborate manuscript for potential journal publication if they receive encouraging and constructive community comments on the technical soundness and relevance of the article. If authors choose to submit their manuscript for peer review to a journal outside the EGU family, eventually a link to the external journal article will be included on the page of the original EGUsphere preprint (Option III).

Currently, 98 % of all submissions to EGUsphere are preprints aimed at journal publication (Option I, discussion papers), reflecting the high standing and popularity of the EGU journals. This preference is expected because trends on other preprint servers show that ≳70 % of all preprints initially posted there are eventually published in a journal . While peer review of preprints is practiced on other publishing platforms with varying success, rigor and popularity (Sect. ), EGUsphere takes advantage of the journal infrastructure, including experienced academic editors and a well-functioning, efficient workflow to ensure high scientific quality and the high reputation of the EGU journals in the geoscience community. Unlike other publishing platforms, EGUsphere combines the various preprint options and their straightforward conversion into each other, the documentation of their public peer review and discussion, and the distillation of highlight articles in a single publishing framework. This structure avoids the need for overlay journals or external peer-review platforms that lead to a disconnection of the preprints, their public discussion and peer review, and final journal publication, which often results in low popularity and efficiency .

In 2020, one part of EGUsphere was already launched as a repository for conference material, including conference abstracts and presentations for conferences organized by the EGU, such as the annual (more than 20 000 participants in 2024). All conference abstracts submitted since 2015 have been retroactively included in EGUsphere. Before the conference, attendees can upload additional documents as supplemental material to their abstracts that can then be commented on by the conference participants until a limited period after the conference. The material is then permanently archived as a supplement to the conference abstract if the authors agreed on a CC BY or CC Zero license.

In summary, EGUsphere is an interdisciplinary repository for preprints and conference contributions in the geosciences. Its preprint repository hosts traditional preprints (Option III) but predominantly preprints with interactive discussion and peer review that are intended to lead to publication in an EGU journal (Options I and II).

Going beyond other repositories and preprint servers, EGUsphere provides a seamless connection from preprints to journal articles and highlight selections as well as conference contributions. It offers functionalities for community-based commenting, discussion and public peer review as well as other forms of open peer review and open science, including open data and open source as outlined below (Sect. ).

Additional features from other open science initiatives and publishing platforms can be easily implemented. For example, informal community review can be introduced for stand-alone preprints that are not undergoing peer review in an interactive OA journal. Such preprints can be ranked in a similar way as, e.g., on publishing platforms managed by F1000, using labels such as approved/approved with reservations/not approved to reflect recommendations by community members engaging as reviewers without having been formally appointed by an editor. The outcome of such self-organized community review may encourage or discourage authors to pursue formal peer review in one of the EGU interactive OA journals or any other journal, which can build on the community ratings and comments already available on EGUsphere (overlay journals, Sect. ). Beyond that, EGUsphere offers opportunities to explore the benefits and shortcomings of new technologies for scientific review and quality assurance, in particular related to the rapid recent evolution of artificial intelligence and machine learning (AI/ML) tools. EGU recently published a statement on the use of AI-based tools for the presentation and publication of research results in Earth, planetary and space science . Concerns have been raised that AI-generated papers may undermine the integrity of scientific publishing by introducing content that lacks originality and quality . In this regard, the publication of preprints and their public peer review on EGUsphere enable the identification and disclosure of potential misuse of AI-generated content in the scientific discourse, including manuscripts and review reports.

The interactive OA publishing approach (publish-then-review model) effectively reduces ethical and legal concerns regarding the upload of unpublished material into AI/ML tools such as large language models (LLMs) for reviewing purposes because preprints posted on EGUsphere are freely available for re-use under the CC BY license. AI tools may be applied to produce summary reports aimed to identify deficiencies of manuscripts such as unclear or incomplete descriptions of methodologies, results and conclusions . Such reports, clearly labeled as AI-generated, could accompany preprints on EGUsphere to provide optional input for community and peer review, whereby community members and referees can decide to use or discard the AI-generated information. If properly applied, such use of AI may benefit and enhance the efficiency and rigor of scientific quality assurance . Note, however, that the involvement of AI in the peer-review process cannot replace human referees because human scientists must remain the actual “peers” of human authors.

Inter-journal special issues (SIs) are organized across two or more journals on topics that are within the scope of these journals. Any EGU journal but also other journals by Copernicus Publications may be part of an inter-journal SI. The handling of SI manuscripts in each of the journals is determined by the journal-specific SI guidelines (Sect. ). Averaged over all EGU journals, nearly half (46 %) of all special issues are inter-journal SIs, whereas in some EGU journals (e.g., ACP, AMT), they represent the majority of special issues (Fig. a and Table S5). About 60 % of all inter-journal SIs are organized between two journals, but the number of journals participating in an inter-journal SI has been (so far) as high as six (Fig. b). All papers, i.e., preprints and final journal papers, of inter-journal SIs are listed on the websites of every journal that participate in the SI, thereby enhancing the visibility and interdisciplinary character. The strong journal overlap in special issues is illustrated in Fig. c. A similar concept of inter-journal special issues or special collections was introduced in AGU journals in 2012 , i.e., several years after the EGU journals started this concept.

Some journal combinations, e.g., ACP and AMT, are quite frequent due to the complementary and overlapping journal scopes: 95 % of all inter-journal SIs in AMT (72 of 76) are co-organized with ACP. Since ACP has organized more inter-journal SIs in total (104), these SIs correspond to 69 % of all inter-journal SIs in ACP. The wide variety of journal combinations demonstrates that the inter-journal SIs provide an efficient and popular way to create interdisciplinary compilations, in which all articles undergo the same transparent peer-review process. The linking of papers to SIs across journals is not only applied to EGU journals but also extends to some other Copernicus journals, including (AR), (ASCMO), (ESSD) and (HGSS). The particular role of ESSD (launched in 2009) is evident since it has inter-journal SIs with nearly all EGU journals. ESSD focuses on publication of comprehensive data documentation, including their sources, codes and algorithms (, and Sect. ), while the interpretation and discussion of their implications are included in the articles of the discipline-specific EGU journals.

SIs allow submissions only during a limited time period (typically 1 to 2 years). The inclusion in a particular SI is performed upon request by the authors during manuscript submission. As of 2025, theme collections can be organized on specific topics with the scopes of one or more journals, in addition to special issues. Such collections are also organized as inter-journal compilations and are open without a defined end date to group papers on specific topics. The linking of papers to collections is performed upon final paper acceptance upon selection by an executive editor; thus, no preprints are included in collections.

The (EG) was established in 2017 and comprises 507 review articles (as of December 2024) that have been published in one of the 19 EGU or other Copernicus journals since 2001 (Fig. ). All 19 EGU journals consider the manuscript type review articles (Table S1); review articles in EGU journals are expected to summarize the status of knowledge on a particular topic and outline future directions of research within the scope of the journal. Occasionally other manuscript types may also be considered for inclusion in the Encyclopedia of Geosciences if they contain a sufficiently broad overview of a relevant topic (e.g., some ACP Opinion articles and NHESS Invited Perspective articles). Articles that fulfill these criteria are added to the Encyclopedia of Geosciences after their public peer review and acceptance for publication according to the criteria in the given EGU journal (Sect. ). During or after the peer-review process, authors interested in having their review article included in the encyclopedia are requested to contact the editors of the encyclopedia. Alternatively, the handling journal editor or possibly a referee can also make the recommendation to the editors of the encyclopedia.

The editorial board of the Encyclopedia of Geosciences is composed of one editor from each contributing journal. Articles in the encyclopedia keep the DOI of their original journal article. This way, they are linked to the full documentation of their review process and also benefit from the journal community and reputation. To emphasize its interdisciplinary encyclopedia character, searches can be performed based on 11 and 183 . In addition, readers can also find articles based on author names, title or words in the abstract, just like for any EGU journal.

More than a decade ago, laid out the potential of the EGU's multi-stage, public peer-review concept for a third stage to create interdisciplinary compilations of specific journal articles of highest quality and particular significance, as opposed to (potentially) high-JIF journals where such selection criteria are often hidden. , launched in 2023, is a compilation of letter-style articles that underwent peer review in one of the eight EGU journals (ACP, BG, ESurf, ESD, GC, NPG, OS, SOIL) that currently consider the manuscript category letters (Sect. ). Letters are concise, engaging articles (≤2500 words) reporting exceptionally important results and significant scientific advances in geoscientific research that are of high general interest to the entire geoscientific community and/or to the broader public (and media). Letters include both of the following characteristic features:

Important discoveries and research highlights in geoscientific research.