A paradoxical knowledge gap in science for critically endangered fishes and game fishes during the sixth mass extinction

Despite unprecedented scientific productivity, Earth is undergoing a sixth mass extinction. The disconnect between scientific output and species conservation may be related to scientists studying the wrong species. Given fishes have a high extinction rate, we assessed the paradox between scientific productivity and science needed for conservation by comparing scientific output created for critically endangered fishes and game fishes. We searched 197,866 articles (1964–2018) in 112 journals for articles on 460 critically endangered fishes, 297 game fishes, and 35 fishes classified as critically endangered and game fish—our analysis included freshwater and marine species. Only 3% of the articles in the final database were on critically endangered fishes; 82% of critically endangered fishes had zero articles. The difference between the number of articles on game fishes and critically endangered fishes increased temporally with more articles on game fishes during the extinction crisis. Countries with 10 or more critically endangered fishes averaged only 17 articles from 1964 to 2018. Countries with the most critically endangered fishes are most in need of science. More scientific knowledge is needed on critically endangered fishes to meet the challenges of conserving fishes during the sixth mass extinction.


Scientific Reports
| (2021) 11:8447 | https://doi.org/10.1038/s41598-021-87871-y www.nature.com/scientificreports/ conservation actions or better outcomes for species 18 ; however, scientific knowledge is an important precursor to conservation 19,20 . Yet, there remains a mismatch between increasing extinction rates and the amount of scientific knowledge created for species conservation. The increases in extinction rate and scientific knowledge created for species conservation is a unique paradox. The paradox could be a function of scientific knowledge not being used to reduce extinction rates, scientific knowledge not created on the species most in need, or a combination of both. Here we focus on whether the science knowledge created on fishes is for the species most in need of conservation. We used a bibliometric approach to measure the scientific output on fishes from 1964 (beginning of the IUCN Red List) through 2018 by comparing scientific output created for critically endangered fishes (defined by the International Union for Conservation of Nature [IUCN] as facing an extremely high risk of extinction in the wild in the immediate future), game fishes (defined by FishBase as a species targeted by anglers for recreational purposes), and fishes classified as critically endangered and game fish. We specifically addressed the following questions: is scientific knowledge created on fishes meeting the most pressing needs related to conservation of fishes in the context of mass extinction, and does the scientific knowledge created for critically endangered fishes match the geographical location for highest conservation need?
Scientific knowledge is the property of all humankind and there has been increased accountability of economic and social returns from public investments in science 16 . Evaluating the scientific enterprise can assist in addressing whether science is meeting societal needs. The science of science 16,17 takes advantage of large data sets, such as the Web of Science, to evaluate the practice of science in search of widespread patterns such that science can effectively address problems that are important to society. Assessing the science on fisheries has been conducted by others [21][22][23] ; however, our study is unique because we focus on scientific output from 1964 to 2018 as it relates to meeting the unprecedented challenges of conserving fishes during the sixth mass extinction.

Results
Of the 31,980 articles that we evaluated for the 792 species of interest (i.e., 460 critically endangered fishes, 297 game fishes, and 35 fishes classified as both), only 3% were on critically endangered fishes despite there being 163 more critically endangered fishes than game fishes in the fish-species database. The first articles on critically endangered fishes appeared in 1971 and the highest number of articles (N = 107) occurred in 2011 (Fig. 1). A change point in the number of articles on critically endangered species occurred in 1997 (1993( -2003. Prior to 1997, the average number of articles on critically endangered species per year was 1.8 (1.0-2.7 CI), subsequently, the average number of articles per year increased to 36.2 . However, during the last decade the number of articles has remained level (slope = − 0.08, P = 0.97, df = 8) and the average number of articles forecasted in 2028 was 68.4 (33.4-103.4 PI; Fig. 1). Eighty-two percent of the critically endangered fishes had zero articles (table S1). Five percent of the critically endangered fishes constituted 85% of the articles, of those articles, 73% were on sturgeon species (order Acipenseriformes; Fig. 2). Six species (Huso huso, Acipenser persicus, A. sinensis, A. gueldenstaedtii, A. naccarii, and Xyrauchen texanus) composed 54% of all the articles on critically endangered fishes (Fig. 2).    Fig. 1), which is 20.3 times higher than the number of articles forecasted for critically endangered fishes. Twenty percent of the game fishes had zero articles (table S1). Five percent of the game fishes composed 66% of all articles, with 65% of those articles on Oncorhynchus mykiss, Salmo salar, Gadus morhua, Salmo trutta, and Ictalurus punctatus (Fig. 2).
Ten percent of the articles were for species listed as both critically endangered and game fish. Fish species in both categories had slightly more articles than critically endangered fishes for all years and were noticeably lower than game fishes (Fig. 1). The first articles for fishes in both categories was in 1966, and the highest number of articles was 170 in 2017. A change point in the number of articles on fishes in both categories occurred in 1996 (1988( -2002. Prior to 1996, the average number of articles on fishes classified as both per year was 23.0 (15.9-30.2 CI), subsequently the average number of articles per year was 112.3 (96.5-128.2 CI). During the last decade the number of articles has increased (slope = 4.8, P = 0.007, df = 8) and the average number of articles forecasted in 2028 was 196.6 (165.8-227.5 PI; Fig. 1). Sixty-two percent of the articles for fish species listed as both critically endangered and game fish in both categories were on Cyprinus carpio and Anguilla anguilla (Fig. 2). The classification of Cyprinus carpio as critically endangered is geographic specific because it is based on a critically endangered subpopulation in the Danube River, otherwise the species is considered vulnerable by the IUCN.
Tanzania had the highest number (N = 60) of critically endangered fishes, with Turkey, China, Mexico, Uganda, Greece, Cameroon, and the United States in the next category with 20-35 critically endangered fishes (Fig. 3). Sixteen countries were in the 10-19 category and 66 countries were in the 3-9 category. Twenty-three percent of the countries had zero critically endangered fishes (Fig. 3).
Tanzania, Cameroon, Philippines, Madagascar, Laos, Syria, and Indonesia had 10 or more critically endangered fishes and zero articles on critically endangered fishes (Figs. 3 and 4). There were 114 countries (59%) that had at least one critically endangered fish and zero articles (

Discussion
We contend that science on fishes is not meeting the most pressing needs related to conservation of fishes in the context of the sixth mass extinction. We demonstrated a vast mismatch between the growth in scientific output on fishes and the science for fishes most in need of conserving.
Much of the scientific output on fishes in the last five decades has been on game fishes. Furthermore, much of the research for critically endangered fishes and game fishes was skewed toward a few species and countries. We predicted that the sum of articles for game fishes would exceed critically endangered fishes but were surprised to find that the difference between the number of articles per year for game fishes and critically endangered fishes increased temporally and that the number of articles on critically endangered fishes plateaued during the last decade. We are at a crossroads in science needed for conservation of critically endangered fishes and argue for choosing the path of scientific productivity that matches the science needed for conservation practice.
Some disconcerting discoveries included that 82% of the critically endangered fishes in our database had zero articles and 23 species comprised 85% of the articles. Although a change point was detected and the rate of articles on critically endangered fishes has increased since 1997, the rate appears to have stabilized in the last decade and is woefully inadequate for science needed on all critically endangered fishes. We found that species classified as critically endangered and game fish had a slightly higher number of publications through time when compared to species classified solely as critically endangered, which indicates that having a recreational or commercial value may increase scientific output. Our results are conservative because we searched articles from the categories "Fisheries" and "Biodiversity and Conservation" in Web of Science and article titles for species names. Thus, the total number of articles for fishes within the categories used in this study were likely higher. In addition, our approach using InCites to select the journals for the Web of Science search resulted in mostly journals where the language was English, which may bias results 25,26 . Nevertheless, our results provide a reliable index of the fish species and countries that monopolize the scientific literature and illustrates the need for increased scientific output on critically endangered fishes.
We found that scientific knowledge created for critically endangered fishes does not match the geographical locations with the highest conservation need. That is, 59% of the countries with at least one critically endangered fish had zero articles. Moreover, three countries represented 69% of all the articles on critically endangered fishes. To analyze the disconnect between the countries with the most critically endangered fishes and countries that are conducting the most science on critically endangered fishes, we used a conservative approach and excluded the 15% of articles that did not have at least one author within the species range. We recognize that by excluding articles published by authors outside of a species range we are underrepresenting the number of articles on critically endangered fishes within a country; however, if we included those articles, it would over represent articles in the countries where the authors were located. The search for author location in relation to the species geographical range did give insight into how much research on critically endangered fishes is conducted by scientists outside a species range. Our analysis indicated that it is common for at least one author to occur in the geographical range of the critically endangered fish species. Thus, parachute science seems to be limited in www.nature.com/scientificreports/ the study of critically endangered fishes, which is not always the case in conservation science 27 . We continue to urge international collaborations because they are highly valuable for conservation efforts. However, researchers from outside countries need to involve scientists who are geographically matched to the area where the species in need of conservation occurs in order to ensure that science being conducted is relevant to the conservation planning processes [27][28][29] . We recognize the need for science on game fishes, much of which is necessary to prevent overharvest of gamefish fisheries 24,30 . Our intention here is not to devalue the need for science on game fishes but to raise awareness for the need for science on critically endangered fishes relative to game fishes in the extinction crisis. Many of the top 25 game fish species in our analysis, for number of articles, are also considered a popular commercial species (e.g., Oncorhynchus mykiss, Salmo salar, Gadus morhua, Salmo trutta) and have a geographical range that includes North America. Thus, the model used to fund science for game fishes will likely not be transferable to critically endangered fishes at a global scale.
A fish-species approach was selected here to match the context of the sixth mass extinction and better understand the disparity in research productivity versus societal needs relative to conservation of critically endangered fishes. We argue that understanding the basic ecology of critically endangered species is vital to their conservation and single-species can be used as an umbrella species to conserve many co-occurring species 31,32 , which is valuable in increasing recognition for the freshwater biodiversity crisis 33,34 . Criteria for an umbrella species are rarity and sensitivity to human disturbance, which are identical to characteristics used for critically endangered species in this study 31 . Gaining support for critically endangered species will remain challenging because environmental issues often rank behind other problems faced by the public, such as the economy and health care, despite the growing public concern about the environment 29,35 . Funding is the limiting factor that prevents more research on critically endangered fishes and is most likely the cause of the dichotomy between science productivity for game fishes and critically endangered fishes observed in this study. Even in countries with high gross domestic product, such as the United States, funding for endangered species is inadequate 36 , and funding is often not related to priority needs for endangered species 37 . Moreover, charismatic species receive more funding and conservation attention without having recreational or commercial value 38 .
Much attention has been devoted to conservation practice and connecting science with policy 28,39,40 , but it has been argued that conservation practice is regularly not effective because conservation researchers often study issues not relevant to conservation practice or focus on easy conservation problems 39,41 . We contend the contemporary scientific knowledge created on fishes will not meet the scientific needs for conservation of fishes in the sixth mass extinction because the scientific output is not the science needed. This is corroborated by the United States National Research Council's statement that "…when science is gathered to inform environmental decisions, it is often not the right science" 41 . The results presented here underscore the immediate need for conducting the right science, which we define as science on critically endangered fishes. Conservation action based on rigorous science can prevent extinctions 19,20 . Moreover, addressing the geographic disconnect between researchers and areas with the highest conservation need for critically endangered fishes could result in more successful conservation actions 28,42 .
How do we increase funding for science to enhance our knowledge on critically endangered fishes? Numerous methods already exist that governmental agencies use to acquire funding for research on game fishes, such as excise taxes on fishing equipment, license sales, and Fisheries-based Ecotourism (FbA). A similar approach could be adopted for critically endangered fishes, or at a minimum, a portion of the proceeds from taxes, license sales, and FbA on game fishes could be allocated to critically endangered fishes. Tourism revenue from birdwatch has been highly successful in enhancing funding for conservation of critically endangered bird species 43 .
Decision analysis tools currently exist for allocating funds among endangered species that could possibly be adopted and adapted to balance funding between critically endangered fishes and game fishes 36,44 . Furthermore, given that monetary resources are limited to fund research on critically endangered fishes, one approach to prioritizing the allocation of funds would be to start where the largest disconnect between the number of critically endangered fishes and scientific knowledge exists 45 . If this approach were to be adopted, then funding for research on critically endangered fishes would be prioritized in Tanzania, Cameroon, Philippines, Madagascar, Laos, Syria, and Indonesia.
If history is used to inform the future, governmental agencies will not provide the resources required to ensure that science is focused on critically endangered fishes during the sixth mass extinction. We believe a global effort is necessary to diversify the portfolio of science on fishes and reverse the paradoxical knowledge gap. It will take partnerships between professional societies and non-governmental organizations to make a meaningful shift in the amount of science conducted on critically endangered fishes. Professional societies that facilitate the publication of science on fishes could partner with non-governmental organizations that focus on critically endangered species to assist in fund raising, prioritizing funding, and prioritize publishing science on critically endangered fishes. We posit four actions to promote and fund science on critically endangered fishes. First, the World Council of Fisheries Societies fosters coordination among scientists, professional societies, nongovernmental organizations (e.g., IUCN, Fisheries Conservation Foundation [which currently partners with the American Fisheries Society], World Fish Migration Foundation, The Nature Conservancy, World Wildlife Fund), corporations (e.g., Google Sustainability Initiative), and citizen science organizations (e.g., Citizen Science Association) to promote and fund science on critically endangered fishes. Second, the World Council of Fisheries Societies and IUCN partner to provide an online marketplace where scientists can identify and apply for funding opportunities on critically endangered fishes (e.g., Save Our Species program, which established a program to save Pseudobarbus burchelli in the Cape Region of South Africa). Third, professional societies could work with The World Bank's Environment, Natural Resources, and Blue Economy Global Practice Initiative to increase and prioritize science on critically endangered fishes in low-and middle-income countries. Fourth, professional societies could partner to reduce barriers to publishing by creating an no-fee, open-source journal www.nature.com/scientificreports/ called Endangered Fishes. Only partnerships at a global scale will ensure that we invest in the science that is needed for critically endangered fishes in the sixth mass extinction.

Materials and methods
Names of freshwater and marine fish species were selected from two publicly available databases: the International Union for Conservation of Nature (IUCN) Red List of Threatened Species 46 and the International Game Fish Association (IGFA) 47 (fig. S1). Data for critically endangered fishes on the IUCN Red List in 2018 were collected using search filters in the advanced tab from the IUCN website-data gathered included genus, species, common name, and geographic range. Subspecies and species nova (newly identified species only described by a Genus; e.g., Aplocheilichthys sp. nov. Baringo) were excluded from the list of critically endangered species. Genus, species, and common name for game fishes recognized by the IGFA and listed by 24 were collected from the IGFA website to create the game fishes database. Critically endangered fishes and game fishes databases were merged to create a fish-species database ( fig. S1). Thirty-five species were classified as critically endangered and game fish (table S1) and were included in the category "Both" (table S1). Scientific names of fishes in the fishspecies database were standardized to use the contemporary accepted scientific name provided by FishBase 48 .
In addition, we included common names listed from all three data sources (IUCN, IGFA, and FishBase) and synonyms (which included historical binomial nomenclature) from FishBase ( fig. S1). The final fish-species database included 460 critically endangered fishes, 297 game fishes, and 35 species classified as both, thus 792 species were used in the journal search (table S1). InCites Journal Citation Reports (Clarivate Analytics) was used to select the journals ( fig. S1). All journals in the categories "Fisheries" and "Biodiversity and Conservation," as defined by InCites Journal Citation Reports in 2018, were selected for the analysis, excluding the Journal of Shellfish Research because we focused on vertebrates (table S2). Web of Science Core Collection was used to select all articles from 1964 (first year of IUCN Red List) through 2018 from the 112 journals identified in the categories "Fisheries" and "Biodiversity and Conservation." We selected 1964 as the start date because the IUCN Red List of animals was first published in 1964. Metadata from 197,866 articles were downloaded and converted to a searchable format (i.e., CSV file) using R statistical software 49 and the package bibliometrix 50 .
The titles of 197,866 articles were searched for critically endangered fishes and game fishes using the fishspecies database ( fig. S1). We only searched the titles of articles because searching abstracts and keywords within articles often included articles where a fish species was named but was not the focal species in the article, thus overrepresenting research conducted on a fish species. In addition, the inclusion of abstracts and keywords was not consistent among journals in the Web of Science and the inclusion of abstracts only occurred for more recent articles. Therefore, we used titles as an index to science productivity on critically endangered fishes and game fishes. We assumed the temporal use of species names in titles was similar between articles on critically endangered fishes and game fishes. Articles with two species names in the title were counted as one article for each species. There were cases where a common name of a fish was also the common name for another taxa. For example, an article on beluga whale Delphinapterus leucas would be selected because beluga for beluga sturgeon Huso huso would occur in the fish-species database. We removed all non-fish species from the database, which provided a final database of articles where critically endangered fishes, game fishes, and species classified as both were searched for in the title of the article (fig. S1).
The SiZer package 51 in R was used to determine if a change point occurred in the number of articles through time for critically endangered fishes, game fishes, and both. If a change point was estimated, linear models were used to estimate the average rate of articles published per year before and after the change point. The 5-year moving average and 10-year forecast for the number of articles was estimated using the forecast package in R 52,53 . The auto.arima and checkresiduals functions were used to determine the best model. The ARIMA (0,1,1) with drift was the best model for all time series. All prediction intervals (PI) and confidence intervals (CI) are 95%.
The global distribution of critically endangered species was determined by searching each critically endangered fish species on the IUCN Red List website and recording the country(ies) listed in the geographic range-194 countries were included in the analysis. Global distribution of published articles on critically endangered fishes was conducted only for articles where the authors were within the geographical range of the species. Thus, we matched author location with the geographical range of the critically endangered fishes by using the list of countries where each critically endangered fish was located, and cross referenced every publication on the species list to determine if the country of at least one of the authors was on the list. If none of the authors were within the species geographical range, it was considered outside of the species range and filtered out for the global distribution of articles analysis. Eighty-five percent of the articles on critically endangered fishes had an author whose address was within the species range. This is a conservative approach to the global distribution of articles on critically endangered fishes, but better matches the magnitude of research conducted within the country of the critically endangered fishes and did not overrepresent countries where authors are working on critically endangered species outside their country.
Global distribution maps of critically endangered fishes and articles on critically endangered fishes were created using ArcMap 10.7.1 (ESRI 2019) and the outline of the countries was WGS 1984 from the ArcGIS Hub. We converted the projection to Eckert III. Breaks for symbology were determined by using Jenks natural breaks classification method with five categories. The use of Jenks natural breaks allows for representation of categories based on gaps identified from frequency histograms to maximize the variance among categories and minimize the variance within categories 54,55 . Additionally, a sixth category was added to both maps to delineate locations with zero critically endangered fishes and articles.