The pace of climate change combined with a need to address societal and ecological impacts with limited resources mean that climate change research is fast-moving and interdisciplinary. Some fields, such as biological conservation, can be considered triage disciplines that require efficient and rapid decision making (Bottrill et al., 2008). To this end, OS principles can help to minimize scientific uncertainty while increasing collaboration potential. For example, OS encourages data and code sharing (Ram, 2013), assists the peer-review process with fully-reproducible manuscripts (Lowndes et al., 2017), and reduces time to publication with preprints and open access (OA) journals (Vale, 2015). Most scientists agree that publicly-funded research should be freely available (Dallmeier-Tiessen et al., 2011) and several institutions have successfully implemented OS practices to share data and research in open-access archives. For instance, research on climate-driven thermal bleaching events in coral reef ecosystems has benefited hugely from open access to NOAA's large-scale monitoring data (e.g., NOAA CoralWatch; Harris et al., 2017). Although comprehensive open data policies have been implemented by some governments (e.g., USA; Obama, 2013) and journal groups (e.g., Nature editors, 2018), journal policies on data sharing are typically insufficient for adequate reproducibility (Stodden et al., 2018). Nonetheless, these examples demonstrate importance of adopting open data principles; comprehensive uptake of these practices will substantially enhance the application of academic research to climate change issues.

Academic and non-academic communication of climate change may be especially important for developing nations. Most climate change research is published through institutes within the developed world (McSweeney, 2015), yet the greatest impacts will be observed in some of the least developed and most vulnerable regions of the world (IPCC, 2014; Blasiak et al., 2017). Inability to access subscription-only publications may inhibit science-based policy in developing countries. For example, inaccessibility of primary research has contributed to low citation rates in policy plans for tropical marine protected areas, implying that environmental management may fall behind current scientific knowledge (Cvitanovic et al., 2014). With the rise in usage of publication repositories such as Sci-Hub (https://en.wikipedia.org/wiki/Sci-Hub), which enable users to download PDF versions of paywalled articles, there is clearly a widespread demand for OA research (Bohannon, 2016; Himmelstein et al., 2018).

Open science practices can result in greater public engagement (Wang et al., 2015) and, through OA publications, increase citation rates (“the OA citation advantage”) (Lawrence, 2001; Eysenbach, 2006). Using Scopus citation data, we show that the proportion of OA studies increased substantially over time in publications containing “climat^* change” in their title, abstract, or keywords between 2007 and 2016 (Scopus; www.scopus.com), accounting for only 4% in 2007 and increasing to 25% in 2016 (Figure 1). However, this varied by journal rank (JR). We categorized journals into four groups, using JRs that are 3-year weighted citation rates obtained from SCImago Journal Rankings (see Figure 1 caption for category breakdown; SCImago¹). For the low JR category, OA publications in 2016 accounted for < 20%, while the medium category had the largest OA proportion at 30%. High and very high categories had 23% and 26% OA, respectively. Popular OA journals such as PLoS ONE and Scientific Reports comprised 71 and 24% of OA publications within their JR groups (medium- and high-ranked, respectively), and 15 and 3% of all publications within their groups, respectively. Across all journal ranks, OA climate change studies were cited more than closed studies (Figure 2A), indicating that adopting OA could lead to earlier and increased citations of climate change research, and thus accelerate scientific progression by building upon existing science at a faster rate (Eysenbach, 2006; Lowndes et al., 2017). Though we used SCImago Journal Rankings to keep consistency with the Scopus citation database, such citation-based metrics are coarse measures of journal research quality, and do not represent research impact for individual papers (Lariviere et al., 2016) or non-academic audiences.

Beyond academic citation advantages, OA climate change research can have a greater societal impact when studies are communicated to non-academic audiences by mainstream news and social media, as well as used by policymakers (Wang et al., 2015; Bornmann et al., 2016). In “mentions” of climate change studies in online news sources, Twitter feeds, and policy documents (www.altmetric.com), we show that OA studies were communicated more frequently (Figures 2B–D), likely due to those studies being more accessible to non-academic audiences. Despite the positive OA effect, the most widely-communicated papers were high impact and closed access papers (e.g., 88% of studies with >100 news mentions were closed access). High-ranking journals such as Nature and Science are often promoted with academic press releases, highlighting how paywalls can limit public understanding and engagement of academic knowledge (Parker, 2013). Nonetheless, higher news and Twitter activity for OA studies—irrespective of journal rank—supports a longstanding perception that open research is more widely disseminated and discussed online (Wang et al., 2015; Côté and Darling, 2018).

Policy documents cited open studies more often than closed, and this difference was consistent across JRs (Figure 2D). Thus, when policymakers lack institutional access to paywalled journals, the OA effect may result in greater uptake of primary research into policy. However, because Altmetric tracks major policy groups in North America and Europe (Bornmann et al., 2016), we note that these policy trends may be biased toward academic authors working for international organizations (e.g., Food and Agriculture Organization of the United Nations, World Bank, Intergovernmental Panel on Climate Change). While our results show a positive trend toward OA (Figure 1) and higher OA mentions in policy documents (Figure 2D), important research still remains behind paywalls and there is evidence that subscription-only publishing models can limit the uptake of current scientific knowledge by policymakers (e.g., Cvitanovic et al., 2014; Fuller et al., 2014; Rafidimanantsoa et al., 2018). For example, OA may be especially important for small-scale, low-impact studies which are relevant for local policy but do not receive much media attention.

Core OS principles are simply the open sharing of data, code, and papers throughout the research process (Hampton et al., 2015; McKiernan et al., 2016). Such practices have reformed entire disciplines (e.g., preprints in mathematics, open genome data in genetics; Nielsen, 2011), but the transition to OS for climate change research is incomplete. For climate change scientists, who must respond to evolving environmental changes with research that has considerable societal impact, the open sharing of data, code, and research outputs could be transformative (e.g., Lowndes et al., 2017). Because of the success of OS in other fields, tools for OS are already freely available (Table 1). For example, several preprint and data repositories target climate change fields (e.g., MarXiv for marine science), while existing version control and coding tools have been adapted for an OS workflow in environmental research (e.g., RStudio and Github, Lowndes et al., 2017).

Despite the clear benefits of OS in enhancing research output and communication to stakeholders, considerable barriers to OS uptake persist, including closed publishing, fear of being “scooped,” and clarity of data ownership (Nosek et al., 2015). Research outputs—usually publications—are already required by most granting agencies, where OA publishing costs are typically covered by grants and institutions (Dallmeier-Tiessen et al., 2011). Furthermore, most climate change research is funded by developed countries yet may focus on climate issues in developing countries that often lack the institutional capacity for journal subscriptions and OA fees (van Helden, 2012; McSweeney, 2015). Thus, to incentivize OS climate change research, we propose funding bodies should require grant holders to openly publish datasets, papers and code, and mandate active dissemination of climate change findings to stakeholders rather than passive dissemination by publication.

Scientists across disciplines have argued, convincingly, for improving research practices by adopting OS principles (Hampton et al., 2015; Nosek et al., 2015; McKiernan et al., 2016). We extend these arguments to show that adoption of OS practices, such as OA publications, OS workflows, and sharing data, is particularly needed to improve the academic and societal impact of climate change research. Given that global efforts to combat climate change impacts will require both rapid collaborative research and communication among academics, policymakers and the public, climate change research is in urgent need of strong OS stewardship.

Journal citations and mentions were extracted from Scopus (www.scopus.com) and Altmetric (www.altmetric.com). We provide our queried search terms and R coding scripts at github.com/travistai2/open-science-cc.

TT conceived the idea. TT and JR contributed equally to data analysis and writing.