Openness of Spanish scholarly journals as measured by access and rights

Metrics on open access (OA) availability of content published in scholarly journals (i.e. content licences, copyright ownership, and publisher‐stipulated self‐archiving permissions) are still scarce. This study implements the four core variables of the recently published Open Access Spectrum (OAS) (reader rights, reuse rights, copyright, and author posting rights) to measure the level of openness in all 1,728 Spanish scholarly journals listed in the Spanish national DULCINEA database at the end of 2015. Data exported from the database and used as variables for the analysis were: journal research area, type of publisher, type of access, self‐archiving and reuse policy, and type of Creative Commons (CC) licence used. Out of the total number of journals (1,728), 1,285 (74.5%) published their articles OA immediately after initial publication and thus received the maximum OAS score for reader rights; 37.5% of all journals used CC licences, and 79.5% allowed self‐archiving in some form. In 72% of journals, authors retained or publishers granted broad rights, which included author reuse and authorization rights (for others to reuse), whilst 13.5% did not specify any terms for copyright transfer. Similar studies could be carried out on other countries as this would enable comparisons of the general adoption and form of openness in different parts of the world.

subscription access journals now often have the option of independently disseminating their own copy of the article on other websites, as authors usually retain at least some self-archiving rights (Laakso, 2014). Archambault et al. (2014) found that around half of all recently published articles in scholarly journals could be retrieved for free through the internet in some form thanks to a combination of OA journal publishing, author self-archiving, and other mechanisms uncategorized in the study.
Although these changes have actively shaped the practical field of scholarly communication, there has been a persistent lack of standardized tools and measurement methods to study the degree of openness in scholarly journal publishing at regional and global scales. Although individual measurements and estimations have been made, such as those referenced in the previous paragraph, many studies have failed to consider the extent to which the emerging openness in science is actually supported by sustainable policies rather than enabled by breaches of copyright. An initiative to remedy the lack of proper measurement tools was supported by the Scholarly Publishing and Academic Resources Coalition (SPARC), the Public Library of Science (PLOS), and the Open Access Scholarly Publishers Association (OASPA), and resulted in the Open Access Spectrum (OAS) (SPARC, 2015). A recent article by Chen and Olijhoek (2016) documented the history of the OAS and described the method and experiences. When it was first designed, an initial batch of over 1,000 journals was scored, and the openness information was manually entered into a database that can be accessed through a public website: the Open Access Spectrum Evaluation Tool (OASET) (oaspectrum.org, 2016).
The objective of this article is to build upon existing research on access and rights issues in scholarly journals by taking a close look at Spanish journals as the population and empirical material.
Of interest are active scholarly journals owned by public or private Spanish organizations that publish peer reviewed papers in Spanish or other languages.
In addition to a brief description of the journals, the study introduces a new analysis based on the structure of the OAS.
This provides insight into the implementation of editorial policies that define reader access, publishing licences, self-archiving policies, and reuse rights.
The OAS and the numerical scoring system (OASET) were used as the main instruments to gauge the level of openness in Spanish journals. Use of the OASET will make it easy to replicate the study in future comparisons using the same population of journals or journals from other countries. To guide the study, two main research questions were defined. RQ1. What is the current composition of Spanish scholarly journals? We gathered data on: • Types of publishers.
• Distribution across research areas.
• Publishing and access models. RQ2. What is the current degree of openness in Spanish scholarly journals according to the criteria of the OASET? We established values for the following variables and assessed them according to the OAS/OASET: • Reader rights.
• Author posting rights.
• Automatic posting (not measured as not significantly applicable).
• Machine readability (not measured as not significantly applicable).

Reader rights
In this context, 'reader rights' is synonymous with readers being granted access to articles on a journal's webpage and covers the kind of (potential) limitations that are related to such access. Of the four OAS variables included in this study, this is certainly the one that has garnered the most attention in practice and research to date. As mentioned in the introduction, there has been strong growth in the number of full OA journals (Crawford, 2015). In evaluations using established citation indicators, it was found that the quality of full OA journals has been increasing, and such journals are already among the highest ranking in Life Sciences and Medicine (Gumpenberger, Ovalle-Perandones, & Gorraiz, 2013). However, reader rights/access to read is not something that is either on or off but can be in between, as also indicated by the OAS variable in question. For example, many subscription access journals make their articles free to read after a set embargo period, for example, 6 or 12 months. This is known as delayed OA (Laakso & Björk, 2013). In other journals, only some of the

Key points
• The Open Access Spectrum Evaluation Tool can be used to measure and compare the openness of large journal collections.
• The 1,728 Spanish journals included in the study make use of the full scales of the four core Open Access Spectrum criteria.
• In all, 70% of Spanish journals publish freely onlinefar higher than world averages. articles are free to read. Most of the main publishers' subscription access journals allow authors to publish their individual article OA in exchange for a one-time fee. These are known as hybrid OA journals (Laakso & Björk, 2016). In a recent development in hybrid OA, organizations that subscribe to journals from a publisher also obtain a quota or full compensation for hybrid OA fees. These 'offset agreements ' (e.g. JISC, 2016) can be assumed to have increased uptake in recent years and will continue to do so.

Author posting rights
Author posting rights (i.e. self-archiving rights) have been studied from several perspectives. The main difference between studies is whether the emphasis is on publisher-or journal-level of analysis or expanded to consider the volume of articles published in the journals.
Similar to this study, a country-specific scholarly journal overview incorporating an analysis of retained author rights has been carried out in France (Dillaerts & Chartron, 2013). The authors analysed the self-archiving policies of journals included in the French national Héloïse index (Heloise.ccsd.cnrs.fr, 2016), an initiative that was inspired by the Spanish DULCINEA index (Accesoabierto.net, 2016) that will be the focus of this study. The Héloïse index covers 32 publishers and a total of 299 journals. Of all these journals, 88% allowed self-archiving of some version of a published article (as per March 2013) (Dillaerts & Chartron, 2013). From a global perspective, Laakso (2014) conducted an analysis of the copyright agreements of the 100 largest scholarly journal publishers in Scopus (by annual article volume) based on information on their websites. The results revealed that over 80% of all published articles globally could be made available on institutional repositories as accepted manuscripts 12 months after the original publication (Laakso, 2014).
To highlight the complex nature of author rights in OA journals specifically, Singson, Sevukan, and Murugaiyan (2015) studied the self-archiving and licensing clauses of 132 Directory of Open Access Journal (DOAJ)-registered OA journals on library and information science through the SHERPA/RoMEO database and by visiting the journals' websites. Authors found that only 23% of journals allowed post-print self-archiving and noted that a drawback of SHERPA/RoMEO is that it is not updated as frequently as could be desired, which can lead to the need to consult other sources for comprehensive selfarchiving and licensing information. In a longitudinal analysis covering 2004-2015, the policies of the 107 publishers that were first entered into SHERPA/RoMEO in 2004 were studied by Gadd and Troll Covey (2016). The initial 107 publishers included many of the large global publishers. The authors found that while a higher proportion of publishers are now explicitly coded 'green' (i.e. they allow some form of self-archiving), the conditions for doing so (how, where, when) have been defined and restricted to a very high degree. The authors also found that self-archiving policies have become more restrictive in tandem with publishers introducing paid OA options (hybrid OA journals).

Copyright ownership
The issue of copyright ownership is very relevant, even in an age with many immediate OA journals. It may even require a stronger focus, as issues such as reuse and redistribution need to be clearly defined for authors, publishers, and the general public. Hoorn and van der Graaf (2006) reviewed the various approaches to copyright ownership adopted by OA journals and surveyed authors of articles published in full OA journals within biomedicine to determine their opinions on copyright transfer. The survey results mainly show that rights management is a complex issue with divisive answers to many of the questions relating to commercial exploitation and reuse of published contents. Ludewig (2014) studied the copyright and OA landscape among scholarly journals in Mediterranean Europe (France, Greece, Italy, Portugal, Spain, and Turkey) through a survey answered by 187 publishers in 2012. A total of 110 out of 174 (63.2%) publishers did not require their authors to sign a copyright agreement, and 19 (10.9%) publishers only required authors to transfer non-exclusive rights to publish the article. Schlosser (2016) recently reviewed the copyright information of 385 journals from 83 library publishing programmes in the USA. Of the 385 journals, 285 (74%) included a copyright ownership statement at journal or content (article) levels. Of those 285 journals, 261 journals presented consistent and intelligible copyright ownership information. Of those 261,166 journals (64%) stated that authors retain copyright, and the publisher is only assigned some non-exclusive rights required for publication. The general conclusion was that there is still a lot of room for improvement in the presentation and consistency of how copyright information is presented in journals and individual articles.

Reuse rights
The most beneficial licence for scientific progress, while also acknowledging the interests of individual authors and publishers, is still actively debated (see e.g. Graf & Thatcher, 2012;Morrison & Desautels, 2016), and many different practices exist among publishers. Relatively little has been researched on this aspect, probably because the practice of permissive licences is still emerging, and aggregated information is not readily available. Previous studies on the Spanish scholarly journal publishing landscape areas, types of publishers, publishing models, and self-archiving permission data extracted from the DULCINEA database in October 2013. DULCINEA is an index containing active scholarly Spanish journals. It provides their identification data, type of access, self-archiving, and licensing policies. DULCINEA classifies journals in accordance with the SHERPA/RoMEO colours taxonomy (Sherpa.ac.uk, 2016). The DULCINEA database (http://www. accesoabierto.net/dulcinea/) is updated and supported by the Acceso Abierto a la Ciencia (accesoabierto.net) working group and is hosted by the University of Barcelona. Journals that are included must be active, owned by a public or private Spanish Institution, and contain peer reviewed research papers in Spanish or another language. Inclusion is based on these criteria and does not take into account quality control of contents. DULCINEA was created in 2008, with initial data collected by an online survey sent to editors and publishers. The data was then incorporated manually or provided automatically by third parties (as others can suggest journals for inclusion), in which case, it is validated by the web administrator. In December 2015, the database contained records of 1,728 journals.
In July 2011, the Spanish government introduced a mandate to make publications stemming from publicly funded research available through OA within 12 months of publication. Borrego (2016) evaluated the effectiveness of the mandate in the first quarter of 2014, 2.5 years after its implementation, by taking a random sample of articles from the Web of Science (WoS) that mentioned the Spanish Ministry of Science and Innovation in the grant information. A total of 478 (58.4%) of 818 articles were found OA in some form, which is fairly close to the figure of around 50% global OA presented by Archambault et al. (2014). This suggests that OA mandate compliance was still lacking as most countries did not have a national-level OA mandate in place. Having an OA mandate in place creates long-term demand for compatible publication outlets and transparent rights information. Directories such as SHERPA/RoMEO or DULCINEA can help authors, librarians, and repository managers to check journals' reuse rights and self-archiving allowance policies.
Torres-Salinas, Robinson-García, and Aguillo (2016) conducted a longitudinal study of the output of researchers affiliated with Spanish institutions, using WoS data for 2005-2014. The authors discovered a persistent trend: Spanish authors publish more in full OA journals than the global average. The global average for 2014 was 10% of articles published in full OA journals, while Spanish authors published 13% of their articles in such journals. The 2-3% increase was persistent during the study period. An analysis of average, subject-normalized impact placed Spanish research output in available OA (articles that could be found in a variety of journal types, i.e. immediate OA, embargo, and hybrid OA) below the global average for all observed years.
According to the authors, the fact that most of these articles are published in national, Spanish-language, low-impact journals explains this result.

Journal data
Journals were extracted directly from a MySQL database containing DULCINEA's master records on December 2015, at which time it contained 1,728 journal records. The export was stored in a CSV file, which was used to create graphs and tables in Microsoft Excel (Microsoft Office Professional Plus 2016). Table 1 contains a summary of variables from the dataset that were included in our analysis and their potential values.

Descriptive analysis
In addition to updating the previous dataset analysed in Melero et al. (2014) by including the new journals introduced to the database and the two new years of development for the entire

Measuring degree of openness
To evaluate the degree of openness of journals, we used four categories from the HowOpenIsIt Guide (SPARC, 2015) and the quantitative scales published as part of the OASET (oaspectrum. org, 2016). We excluded the evaluation of 'automatic posting' and 'machine readability' from the analysis because it is difficult to find journals in our population that specify these issues in their websites, and very few journals in our study would meet either of these criteria. Taking into account the omitted categories, the maximum score a single journal could obtain was 72 points.

Statistical analysis tool and methods
SPSS statistical package v.23 by IBM was used for all analyses.
Crosstabs tables were grouped by subject, type of publisher, copyright holder, and CC licence, and the chi-squared test was applied to evaluate potential associations or dependence among groups. Spearman's correlation factors were calculated to measure the strength and direction of associations between variables.

RESULTS AND DISCUSSION
First, the obtained results were compared with a previous study of Spanish journals (Melero et al., 2014). Then, we analysed the new data and variables introduced as part of this study. Finally, we applied the OAS framework to the population in order to evaluate the openness of Spanish journals.
To provide a high-level descriptive overview of the data and show how the Spanish journal landscape evolved in the 2-year period between the studies, Table 2 compares the new results with those of the previous study on the DULCINEA database (Melero et al., 2014). The most notable changes over the last 2 years relate to the type of access, as the percentage of restricted access journals has decreased; the specification of rights, which has increased; and the percentage of journals that allow immediate self-archiving, which has also increased (from 65 to 72.6%) in direct relation to the increased specification of author's rights on the journals' websites (from 73 to 86.5%). The percentage of journals coded 'white' according to SHERPA/ RoMEO's criteria has remained almost the same, while the proportion of journals coded 'blue' and 'green' has increased because publishers have improved the specification of their self-archiving policies. Regarding which version authors are allowed to selfarchive, 69% of journals allowed the deposit of the version of record (VOR) immediately after publication, 18% allowed the author's postprint at acceptance, 22% the preprint at submission, and almost 7% the author's post-print or VOR after an embargo.

Journal subject areas and types of publishers
Strong relationships were found between life sciences journals and the association/learned society publisher type, between humanities and social sciences journals and university/research institution publishers, and between health sciences journals and commercial publishers (p < 0.001). Most social sciences (60%) and humanities journals (68%) were published by academic and research institutions, and most health sciences journals by private commercial publishers (48%) or by academic/professional learned societies (41%) (Fig. 1). Over 50% of journals within the health sciences were published by Elsevier.

Journal subject areas and reader access
Most Spanish scholarly journals were accessible online free of charge (Fig. 2). There was a strong relationship between health sciences journals and restricted access, and social sciences and free/gratis access (p < 0.001).
Slightly over 40% of journals published by commercial publishers were freely accessible online. This is possible because agreements existed between publishers and scholarly associations or learned societies to distribute the electronic version of the journals for free. Some journals could be freely accessed after an embargo. Examples include journals published by Elsevier that are associated with medical societies (Allergologia et Immunopathologia, Medicina Intensiva, and the Revista Española de Medicina Nuclear e Imagen Molecular, among others). There were also a few hybrid OA journals, mostly published by Springer (12), Taylor & Francis (6), Cambridge University Press, and EPI SCP (1).

Self-archiving conditions and comparison between subject areas
Most journals allowed self-archiving of the VOR immediately, in parallel with publication of the journal article. In all, 80% of all journals allow archiving (72.6% immediately and 6.8% after an embargo). However, when split into subject categories, it is noticeable that only 42% of health journals allow immediate archiving and 24% publication after an embargo. Figure 3 provides an overview of the key document versions and when they can be self-archived.
There was a statistically significant relationship between subject area and self-archiving allowance (p < 0.001), and a strong relationship between health science journals and prohibited selfarchiving or self-archiving after an embargo, and social sciences and humanities journals with a self-archiving allowance (Fig. 4).
This is in accordance with previous comments regarding subject and type of publisher.
Regarding the RoMEO classification of journal colours     (p < 0.001), with a strong positive association between health sciences journals and the white code (34% of total health journals).

Copyright holder by journal subject area and publisher type
Authors were only allowed to retain copyright in 14% of the journals; of those journals, the percentages by discipline were as follows: life sciences, 6%; engineering + physics + mathematics, 5%; humanities, 35%; health sciences, 6%; and social sciences, 48%.
Although we assigned a unique category for authors who were listed as the copyright holders, we found different ways of expressing this: • Authors retain copyright and grant the journal the right of first publication. • Authors retain copyright and grant the journal right of first publication, with the work simultaneously licensed under a CC Attribution.
• The authors hold copyright.
• The authors transfer to the journal the rightnot exclusiveto reproduce and distribute the article.
• The authors hold copyright and assign the publisher the exclusive right to distribute the paper for 1 year after publication.
• The publisher, when the access is by subscription. The authors, when they pay article processing charge (APC).
• The society; however, authors can request to hold the copyright and grant a publication.
The association between subject area and copyright holder was found to be statistically significant according to Pearson's chisquared test at a level of p < 0.001. The strongest positive associations were between health sciences and commercial publishers, when publishers are the copyright holders, and between humanities and non-profit academic publishers, when the latter are the copyright owners. Some societies and professional associations have their journals published by commercial publishers but do not utilize standard commercial copyright transfer policies because the society is the owner of the journal and determines its own policy. This is the case of the Anuario de Psicología Jurídica published by Elsevier. This journal is a gold OA journal, and the authors transfer copyright to the Colegio Oficial de Psicólogos de Madrid.
Another example of an OA journal published by a commercial publisher (Springer) is SERIEs -Journal of the Spanish Economic Association. In this example, the Association covers the APCs, and papers are distributed under a CC-BY-ND licence. Authors can retain copyright and only assign to Springer the exclusive right to any commercial use of the article. In around 80% of journals published by universities, authors transfer copyright to the institution. In this regard, the behaviour of commercial and academic publishers does not differ: both request copyright transfer.

Copyright holder, reader access, and selfarchiving
There was a close relationship between who owns the copyright and the free online availability of the journals (Fig. 6). The strongest positive associations were found between authors (owners) and RoMEO's taxonomy, most Spanish journals were blue (60%). The relationship between colour and copyright holder was significant (p < 0.001, Fig. 7), and the strongest associations were between authors as copyright owners × green journals, academic publishers × blue journals, and commercial publishers × white journals.

CC licences across publisher types and reader access
A total of 643 of the journal population use a CC licence. BY-NC-ND is the most popular classification (55% of those 643 journals distribute their content with a CC licence, followed by CC-BY and CC-BY-NC). The relationship between licence and journal subject area (Fig. 8) was not found to be significant (p > 0.05). Out of the total population of journals using a CC licence, the order by use of any CC licence was: social sciences (53%), humanities (30%), health sciences (8%), life sciences (6%), and engineering + physics + mathematics (4%). The limitation of no commercial redistribution was found in 78% of licences, likely due to publishers guarding against third parties reusing published papers for profit.
A strong relationship was found between type of licence used and type of publisher (p < 0.001). Regarding the type of licence and who is the copyright holder, the statistical analysis showed a relationship between the type of licence and the owner of the rights (p < 0.001). In 319 (49.6%) of the 643 journals that used CC licences, the copyright holder was the university/research institution; in 23%, it was the authors; in 19.6%, the association/learned society; in 4.8%, the commercial publisher; and in 3.0%, governmental organizations. When journals used CC-BY, the authors were mainly the owners of the rights (51%) followed by private publishers. Spearman's correlation analysis revealed significant correlations among these ordinal variables (access, CC licence, and RoMEO colour). CC licences were ordered from CC-BY (1) to CC-BY-NC-ND (6), access from free access (1) to restricted access (3), and RoMEO colours from white (1) to green (4). Taking this into consideration, there was a positive correlation between access and CC licence (both variables increased and decreased in parallel) and a negative correlation between CC licence and RoMEO colour.

OAS analysis
In this analysis, the OAS, with quantitative scores defined by the OASET, was applied to all Spanish journals listed in the DULCI-NEA database.  (7).
The rest are published by small independent private publishers. Figure 9 shows a histogram of the total scores for all journals (as percentages of the maximum of 72 points per journal). The median is 55.6 and the mean 52.1. In our study, 68% of journals received a total score equal to or higher than 50%. Most journals (86%) ranked within the fourth quartile (Q4) are published by universities or research institutions. Journals with a total score of zero (11.5%) are restricted to subscribers, and there is no information about reuse and/or depositing rights on their websites.
The distribution of the OAS score is not the same across subjects (p < 0.05) because subject, type of publisher, and access type are related, as revealed in the previous analysis. A total of 662 (38%) out of 1,285 journals received a perfect score of 20 for reader rights but do not allow any reuse beyond fair use or any other limitations or exceptions (all rights reserved); 22.5% allow reuse (no remixing or further building upon the work) and use CC-BY-NC-ND; 5.9% allow reuse, remixing, and further building upon the work, subject to certain restrictions and conditions, and use CC licences BY-NC or CC-BY-SA; and 133 journals (7.7%) had a maximum score for readers rights and reuse rights and use a CC-BY licence ( Table 4). Out of all 1,285 gratis-OA journals (i.e. those that achieve a perfect score of 20 for reader rights), 61 (3.5%) received a perfect score of 16 for copyright (authors hold the copyright); 1,039 (60.1%) are freely available on the internet and the publisher grants broad rights; and in 185 (10.7%), the publisher holds the copyright, with no reuse beyond fair use (Table 5). These results are based on the first condition of an OA journal: free access to its contents immediately after publication. Other combinations can be seen in Table 5. (https://openaccess.mpg.de/Berlin-Declaration), an OA contribution must satisfy two conditions: first, it must grant users 'a free, irrevocable, worldwide, right of access to, and a licence to copy, use, distribute, transmit and display the work publicly and to make and distribute derivative works, in any digital medium for any responsible purpose, subject to proper attribution of authorship', and second, 'a complete version of the work and all supplemental materials, including a copy of the permission as stated above, in an appropriate standard electronic format is deposited (and thus FIGURE 9 Open Access Spectrum Evaluation Tool score distribution for Spanish journals (as percentages of the maximum of 72 points total per journal).  OASET, Open Access Spectrum Evaluation Tool. OASET, Open Access Spectrum Evaluation Tool. 70% of journals freely available on the internet. This is an exceptionally high n5umber compared with the global average, which does not even reach a third of this number, depending on the reference index that is used. If we contrast reader rights with self-archiving rights, 64% of journals are gratis-OA and also allow the deposit of some versions of articles. This percentage represents periodicals that meet the widely recognized Budapest Open Access Initiative (BOAI) definition of OA (Budapestopenaccessinitiative.org, 2016). However, a lower percentage (36%) is found when we compare readers' rights with reuse rights. This is due to the division of reuse and self-archiving rights in the OAS. In reality, there is an intersection between both rights when the authors become users but also have the right to post their works.
Regarding the type of publisher and research areas, most journals are focused on the social sciences and humanities, followed by the health sciences. Social sciences and humanities journals are mainly published by higher education and research institutions, and health sciences journals are generally published by commercial publishers, learned societies, or professional associations, which in some instances have agreements with private publishers to distribute their publications.
There are challenges involved in collecting and curating data on journal policies at a global scale, whether we are talking about SHERPA/RoMEO or the OASET. Even with the DULCI-NEA database as a starting point, we encountered similar challenges to those stated by Chen and Olijhoek (2016): editorial policies are not easy to find, and sometimes, they are not even stated on the website; they change, and editors do not tell aggregators about the changes. This makes it difficult to maintain updated records. There are also inconsistencies in rights permissions and the use of open licences. When discovered, DULCINEA administrators alert journal publishers or editors to amend these kinds of contradictions, but changes are not always made immediately, and in some cases, publishers do not respond.
On a global scale, the reporting of access and rights by journals is still problematic. The OAS and the OASET are valuable initiatives to help increase the availability and accuracy of such information in a standardized way. This study is the largest practical implementation of the OAS that examines not only an individual outlet but a large cohort of journals. It is the only study that incorporates a complete national population of journals. National indexes of journals' editorial policies are still rare, and without the data in DULCINEA, it would have been impossible to conduct this study. The long-standing problem with SHERPA/RoMEO is that ever-changing journal policies are not updated frequently, and there is a lack of clear responsibility for submitting information and updates. By decentralizing journals to national level but still retaining RoMEOcompliant coding and data structures, the population becomes much easier to curate. We encourage actors overseeing numerous journals, publishers, and national science policy-makers to make use of the OAS as a tool for monitoring the development of openness among journals.