These examples are likely to have fueled misperceptions on topical issues across a wide audience. They may therefore have contributed to, or have the potential to contribute to, poor policy decisions and actions leading to costly and inappropriate management of fisheries and marine conservation, foregone sustainable benefits, and other negative outcomes.
The criticisms highlighted here are based on the published rebuttals and take account of published responses by the authors of the examples to those rebuttals.
The use of marine protected areas (MPAs) for fisheries management is a contentious policy issue addressed by several papers of concern. Whereas MPAs are important for biodiversity conservation [70–72], a problem with some publications advocating more applications of MPAs as a fishery management tool is that they have been misleading in terms of the benefits and costs of MPAs on fisheries yield. For example, White et al. [73] concluded that MPAs can achieve “….higher profit than attainable under conventional management”, but Hart and Sissenwine [74] questioned some of the assumptions in the paper and pointed out that it potentially misleads the reader by reporting yield in numbers of fish rather than weight or volume as is customary, thereby ignoring the reality that larger fish of a species are usually more valuable than smaller fish.
Three more recent contentious MPA papers are included as examples of problematic scientific papers (Table S1). The first, Cabral et al. [21], estimated that increasing the current global network of no-take MPAs would allow future catches to increase by at least 12 %. Various assumptions made in the paper were challenged by Hilborn [23] and Ovando et al. [24]. Almost a year after publication, the article was retracted when it was shown that the article’s editor had a conflict of interest as a frequent collaborator with several of the authors and that the paper had overestimated the benefits of the MPAs [22].
Sala et al. [25], using the same model as [21], reported that a considerable increase in the protected areas of the oceans would have substantial benefits for biodiversity conservation and carbon storage, and result in improved yields from fisheries. However, Hilborn and Kaiser [29] and Ovando et al. [31] pointed out several flaws in the as- sumptions and analyses, while Hiddink et al. [30] reported that the authors had over-estimated the impacts of trawling on the release of CO2 by several orders of magnitude. The authors of Sala et al. [25] discuss these criticisms of their study and challenge several of them in their responses [26–28] but, importantly, acknowledge the need for further research and analyses to assess the implications of some of the key assumptions of their model for the results and conclusions obtained.
Medoff et al. [32] reported that the establishment in 2016 of the world’s biggest fully protected MPA, in waters around Hawaii, had led to a gradient in CPUE for two species of tuna near the boundary of the reserve. The authors concluded that large MPAs protect migratory fish species and benefit their fisheries. Hilborn and Hampton [33] pointed out that the analyses did not consider the very low catch of tuna species in the area before the MPA was created and that the authors had also failed to take into account the assessed increase of yellowfin tuna in the West Pacific that had started two years before the MPA was created. In addition, Hampton et al. [34] investigated the change in abundance of tunas after the establishment of the Phoenix Islands Protected Area, where fishing pressure had been much higher than around Hawaii, and concluded that large protected areas for tunas and other highly migratory fishes were unlikely to increase abundance or benefit fisheries.
In examples on the status of fish stocks, Myers and Worm [15] claimed that the biomass of many large predatory fish had been reduced to about 10 % of their pre-industrial levels, with potentially serious consequences for ecosystems. Walters [42], Hampton et al. [41] and Polacheck [14] contested those findings, reporting that the extent of depletion had been exaggerated through the use of CPUE exclusively, while ignoring other indices and alternative interpretations of early CPUE declines. On the same theme, Baum et al. [43] and Baum and Myers [44] described what they considered to be the collapse and possible extirpation of shark populations in the northwest Atlantic Ocean and Gulf of Mexico. Both papers were criticized by Burgess et al. [46], who argued that the extent of decline had been overstated through the use of limited and inadequate data sets as well as disregard by the authors of other data sets and assessment results.
Worm et al. [35] projected a global collapse of all taxa then being fished by 2048. There is evidence that the suggestion of a collapse by 2048 was included in the primary publication to obtain popular media attention [17]; their projection has been frequently quoted in both the academic and popular media. Several fisheries scientists challenged that result (Supplementary Information), including Branch [38] who pointed out a number of methodological problems with the paper. A working group representing the various different perspectives reconsidered the global status of fisheries using actual stock assessment results and concluded instead that a mixture of sustainably harvested and over-exploited stocks would eventuate [75]. The original paper, with the incorrect forecast of collapse, had been cited 5712 times as of 27 July 2023 while Worm et al. [75], with the revised analyses, had only been cited less than half as many times (citations from Google Scholar).
Pauly et al. [47] suggested that the mean trophic level of catch could be used as an index of ecosystem health and that the mean trophic level of the global catch had declined. This result has been widely promulgated by various NGOs. However, this paper's findings have been criticized for various methodological problems, including underlying assumptions and failing to take increasing catches of lower trophic level fish into account.
Also on the theme of ecosystem impacts, the authors of Gremmillet [51] on interactions between seabirds and fisheries in South Africa were required to submit an Erratum [52] after being criticised for several incorrect and potentially damaging statements on the management and social and economic features of the fishery [53]. Another example of seabird-fishery interactions is Sherley et al. [55], which reported positive effects on penguin reproductive success of experimental closures to pelagic fishing in the neighbourhoods of penguin breeding islands. However, Butterworth [76,77] reported that the method used did not adequately account for pseudo-replication in the data and thereby overestimated the precision of the results. The same methods were subsequently used by Sydeman et al. [56] to argue for continuation of island closures. Both papers were challenged in a rebuttal of the Sydeman et al. paper by Butterworth and Ross-Gillespie [59]. Final resolution of the methods and results came after a panel of six qualified international scientists appointed by the South African Government conducted an intensive on-site review and confirmed that additional analyses were needed to address the pseudo-replication problem [60]. The Panel concluded further that the impacts of fishing closures around the breeding islands were likely to be small compared to the rate of decline of the penguin population, and that additional research was necessary to determine the factors driving the greater part of the decline.
Another well-publicised example is the paper by Kroodsma et al. [62] claiming that fishing activity covered at least 55 % of the world’s oceans, four times the land area covered by agriculture. However, Amoroso et al. [64] pointed out that the coarse spatial scale of the analysis had led to the original paper overestimating the footprints of all fishing and trawling by factors of >10 and >5 respectively.
In one of a series of contentious papers, Munday et al. [65] reported findings that future ocean acidification could seriously disrupt the ability of clownfish larvae to differentiate between cues used in locating suitable sites for settlement. In another paper, Dixsonet al. [66] reported that juvenile corals and fish are attracted by cues from non-fished areas dominated by corals but are repelled by cues from reefs that are fished and dominated by seaweeds. Doubt was raised about the validity of those results by Clark et al. [67] and in other papers, which reported contrasting results, and that the sizes of the effects that had been reported in those studies were highly unlikely. Munday and Dixson et al. [68] challenged the Clark et al. paper but Dixson et al. [66] was retracted by Science in 2022. There are ongoing concerns about a number of the papers in this suite [78].
Each of the examples presented here is unique, and the criticisms raised in rebuttals differ from paper to paper. Nevertheless, they all exhibited the common problem of weaknesses in assumptions, for example the assumption in Cabral et al. [21] and Sala et al. [25] that for some key species in their analyses, the whole global range of the species was connected through larval and adult dispersal [23]. In Worm et al. [35], the definition of a fisheries collapse and the assumption that catch is a proxy for abundance were primary criticisms, in addition to their erroneous projections of estimated trends [38]. Methodological problems in these examples included: inappropriate treatment of spatial structure or dynamics (e.g. Cabral et al. [21], Sala et al. [25], Kroodsma et al. [62]); inconsistent accounting of fishing effort across the different outcomes analyzed (Sala et al. [25]); inadequate interpretation of CPUE information (Myers and Worm [15]); and use of limited and inadequate data sets (Baum et al. [43], Baum and Myers [44]). In Sherley et al. [55], errors in the statistical analyses were identified as a problem [59,60], while in Gremillet et al. [51]], incorrect statements and lack of understanding created a negatively biased impression of aspects of the local small pelagics fishery and its management [52,53].
Impacts of erroneous and potentially misleading papers
For fisheries to be able to make their potential contributions to food security, nutrition and livelihoods in a sustainable way, it is critical to understand what management methods have worked in different ecological, social and economic circumstances and what methods should best be used where improvements are most required [7]. Similarly, reliable information is required on the true impacts of fisheries on en- environments and ecosystems at a resolution enabling effective decisions that provide adequate protection without unnecessary impacts on livelihoods and other benefits of fisheries. The examples discussed here could result in incorrect information and poorly informed scientific conclusions and recommendations that lead to inappropriate management and policy decisions.
Some of these papers also fall under what could be considered scientific neocolonialism, which strives to control development agendas based on a particular world view, and is often detrimental to local communities and other directly affected stakeholders (e.g. [89,90]).
Striving for 30 % no-take or highly protected areas, a global one-size fits all approach, is one such example that has been directly or indirectly supported by some of the published papers presented here. Sydeman et al. [56] provides another instance. In addition to drawing on scientific results that had been criticized on technical grounds and were simultaneously being debated by the relevant South Africa fisheries scientific working group, the authors are all from northern hemisphere institutions and, based on the Acknowledgements, there appears to have been minimal consultation with South African fisheries experts or stakeholders prior to publication of the paper. Nevertheless, referring to a complex issue that directly affects two African countries, South Africa and Namibia, the publication included strong recommendations with ecological, social and economic implications. Editors are not in a position to control practices such as this, but need to be sensitive to the multi-dimensional risks of misleading information that they might presend. They should, at least, ensure that reviewers include local experts with the necessary backgrounds to evaluate all aspects of a paper, although even that can be difficult to achieve.
The media have an important role in amplifying and sharing scientific findings with the wider public and thereby potentially contributing to consideration of this information in policy-making. However, the inadvertent or deliberate spreading of misleading information from scientific publications adds to the potential damage from these documents. Media reports on scientific publications are typically based on press releases prepared by the authors of publications, often in collaboration with media experts. Headlines such as “Protecting 5 % More Of The Ocean Can Increase Fisheries Yield By 20 %” (Forbes12), “Bottom trawling releases as much carbon as air travel, landmark study finds” (The Guardian13) and “Trawling for fish may unleash as much carbon as air travel, study says” (New York Times14), spread false impressions amongst trusting readers. These and other cases of dissemination by the media of misleading results from examples discussed here are provided in Supplementary Table S2.
Recommendations
The previous section contains a number of detailed recommendations arising from this assessment, divided into those aimed at scientists, those relevant to the publishing process, and those directed at the users of science, in this case those involved in the policy-making process.
The overarching recommendation for all those involved in the production and dissemination of scientific results and information is that every effort needs to be made to ensure the reliability of the information produced and used and claimed to be scientific. This group includes scientists, employers and funders of science, publishers and those involved in the publishing process, and the media. Adherence to this is essential for ensuring that trust in science and scientific information is maintained and strengthened amongst all participants in the policy-making process and the public at large.
Journal publishers have ultimate control of the journals, and there-fore of papers published under their names. They need to take the lead towards ensuring the reliability of the papers they publish, giving particular attention to likely high impact papers, especially where policy issues are involved. They have the means and a responsibility to implement the recommendations above relating to peer-review and quality assurance, or to implement whatever other changes are needed to achieve the necessary improvements.
The recommendations for those involved in policy-making centre on implementing procedures to ensure that all information provided under the name of science is scrutinised by a broad group, considered by the stakeholders to be balanced and with the necessary expertise to review the information for accuracy and relevance. The outputs from that process should represent the best available scientific evidence, which should then be made available to all stakeholders in a form that can be readily understood.
The 2030 Agenda for Sustainable Development and the Sustainable Development Goals [9], as well as the myriad of other local, national and foundation of reliable, objective and relevant scientific guidance. The challenges to meeting the Sustainable Development Goals are enormous, often involving difficult decisions on trade-offs as countries implement their particular visions and approaches for achieving them. Failures and weaknesses in scientific support can lead to poor choices, and hinder or undermine progress towards them. This reinforces the fundamental responsibility of scientists and publishers of science to do all they can to ensure quality and accuracy.
We consider that the implementation of the recommendations above would contribute to meaningful improvement in the overall quality of scientific publishing and publications, and reduce the risks of publication and dissemination of erroneous research that can lead to inappropriate policy recommendations.
