Language barriers in global bird conservation

Multiple languages being spoken within a species’ distribution can impede communication among conservation stakeholders, the compilation of scientific information, and the development of effective conservation actions. Here, we investigate the number of official languages spoken within the distributions of 10,863 bird species to identify which species might be particularly affected by consequences of language barriers. We show that 1587 species have 10 languages or more spoken within their distributions. Threatened and migratory species have significantly more languages spoken within their distributions, when controlling for range size. Particularly high numbers of species with many languages within their distribution are found in Eastern Europe, Russia and central and western Asia. Global conservation efforts would benefit from implementing guidelines to overcome language barriers, especially in regions with high species and language diversity.


Introduction
Earth's biodiversity is under threat. Human population growth and associated activities are causing the loss of natural ecosystems and species habitats at an unprecedented rate [1,2], with at least one million species currently threatened with extinction [3]. This accelerated loss of biodiversity and the fact that many species and threats extend beyond country borders has stimulated the generation of guidelines for effective transboundary collaboration on international agreements, such as the Convention on Biological Diversity and the Convention on Trade in Endangered Species of Wild Fauna and Flora [4,5]. Many species, in particular vertebrates, have wide distributions spanning multiple countries with different social characteristics. An example of this is the area surrounding the Mediterranean sea, where a diversity of threatened migratory species are distributed across countries with different social and cultural backgrounds, from Western and Southern Europe to North Africa and Western Asia [6,7]. However, existing guidelines for transboundary collaboration rarely consider differences in cultural backgrounds among countries, which can create both challenges and opportunities in conservation [4,8,9]. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 An aspect of culture that has fundamental consequences for conservation is the variety of languages that people speak. Language differences across the distribution of a species can generate a number of challenges for conservation (summarized with examples in Table 1). First, multiple languages being spoken within the distribution of a species can create a barrier to the effective collection and compilation of scientific information relevant to conservation, which is often scattered across languages [8]. For example, comprehensive ecological knowledge of understudied seasonal migratory birds in Brazil could only be achieved by combining information from Brazilian citizen science platforms available only in Portuguese with information from global, English-language, platforms [10]. Second, language differences within the distribution of species can also impede effective agreements between stakeholders in conservation decisions. For example, differences in the use of vocabulary even within the same language influenced the perception of the public on the importance of hedgehog eradication as a conservation measure in Scotland [11]. Such an effect could be magnified further when stakeholders speak different languages. Third, language differences can affect the generation and quality of collaborative conservation projects. For example, overcoming language barriers was recognized as a fundamental step for the generation of effective conservation measures for threatened bird species in the Julian alps, the Bavarian-Bohemian Forest [12] and the Mediterranean sea [7]. Table 1. Potential challenges to conservation outcomes caused by language barriers.

Consequences of language barriers Examples
Scientific research Inaccessibility to scientific information (e.g., peerreviewed papers and databases) Thirty six percent of 75,513 scientific documents on biodiversity conservation published in 2014 were written in non-English languages [8].
The majority of research on fengshui forests has been published in Chinese, and thus is not globally accessible [33].
Several studies have assessed the relationships between species diversity and linguistic diversity at local [13] to global scales [14,15]. However, despite growing evidence of the conservation consequences of language differences within species distributions, it remains unknown where such negative consequences of language barriers might be expected, and for which species. Here we investigate the number of languages spoken within the distribution of each of 10,863 extant bird species and discuss the ramifications of this for conservation. We focus on birds because (i) many bird species migrate, with their distribution spanning multiple countries, (ii) a wealth of ecological knowledge, especially detailed information on distribution is available [16], and (iii) a large number of transboundary conservation projects already exist [6,17]. We specifically aim to identify species with many languages within their distribution, and regions with high richness of such species, where language barriers could impede conservation.

Bird species data
We obtained species distribution maps for the birds of the world from Birdlife International and Nature Serve [16]. We considered parts of each species distribution coded as "extant" for

Policies
Inaccessibility to policy documents One third of the management plans identified for an assessment of strategic planning, zoning, impact monitoring, and tourism management at Natural World Heritage Sites were excluded from the analysis due to language barriers [43].
Barrier to effective policy agreements among countries (e.g., bilateral agreements) Having a common official language between countries with established policy agreements had a statistically significant effect on reducing the tonnage of waste shipments [44] Language barriers between scientists and policy makers Language disparities can lead to poor communication between scientists and policymakers [45].
Time available to read papers, difficulty in understanding technical language and reading in English have been recognized as a barrier to access scientific literature for Brazilian policy makers [46].

Conservation activities
Barrier to developing collaborative actions Language barriers impede collaborative conservation activities between countries in the Mediterranean [7].
Differences in language represent a challenge for decision-making leading to agreements on multi-year resource allocation at National Parks in two transboundary regions in Europe, Italian-Slovenian and German-Czech borders [12].

Extra costs (time/finance) for the multilingualisation of materials (websites, leaflets etc)
Educational kits with information about the ecology and conservation of Critically Endangered spoon-billed sandpipers were translated to five different languages to improve the outreach of the conservation message (https://www.eaaflyway.net/spoonbilled-sandpiper-teaching-kit-available-for-free-download/). [12].

Inaccessibility to relevant information
Differences in language in National Park planning documents represented a challenge for collaborative approaches for decision-makers in two transboundary conservation regions in Europe [12].
The dissemination of indigenous ecological knowledge and established conservation strategies in Australia is hindered by the fact that indigenous ranger groups, especially those in remote regions, who often do not speak English as their first language, are forced to adopt non-Indigenous forms of monitoring [47].
presence and "native" and "reintroduced" for origin. In the case of migratory species, all seasonal sections of the distribution were considered. Additionally, we obtained information on taxonomic classification, threat status and type of migratory characteristics for each species from Birdlife [16]. Species were divided by conservation status [ie., threatened (Vulnerable-VU, Endangered-EN and Critically Endangered-CR) [18] and not threatened (Least Concern-LC and Near Threatened-NT)] and migratory status [(i.e., Full migrants or not (the latter comprising non-migratory, altitudinal migrants and nomadic species)], and results were aggregated for these groups and for each bird species. The area of each species range distribution was calculated (in km 2 ) using the "st_area()" function in a PostgreSQL database in Post-GIS version 3.0.2 [19].

Data on languages of the world
We compiled information on the official and most spoken languages of each country of the world. Official languages are the ones used by a country or jurisdiction for governmental and legal purposes while the most spoken language is the one that the largest proportion of the population of a country or jurisdiction speak. We used the World Fact Book from the United States Central Intelligence Agency [20] as a primary source, but additional sources were used as needed (S1 Table). For the official languages we listed all the languages that each country states as official. Spain (five official languages), Ethiopia (five) and South Africa (11) were the only countries with more than four official languages so for those countries the top four official languages with the highest number of speakers were used. This information was gathered from additional sources (S1 Table). For Antarctica no official language was assigned. We also used the World Fact Book to identify the most spoken language in each country or jurisdiction. If this information was not available, the language recorded as "lingua franca" in the World Fact Book was selected. For Antarctica and Kashmir no language was assigned as most spoken.

Calculation of number of languages in species distribution and bird species richness
First, we determined the identity of the countries each species distribution overlaps with and the official and most spoken languages of those countries. Those languages were assigned to each species. Then we estimated bird species richness using a global 30 km × 30 km grid. This has been identified as an optimal resolution for reducing the effects of commission errors (where species are thought to be present but are not) when working with global species distribution maps [1]. Grid cells that straddle more than one country were split through the country borders into sub-units for each country. The number (i.e., species richness) and identity of the species present in each grid cell was determined. For the earth and countries outline we used the Natural Earth public domain map dataset (http://www.naturalearthdata.com/).

Mapping areas of high numbers of bird species with many languages within their distribution
Finally, by using the identity of the species present in each grid cell and the information on the number of languages spoken in the distribution of each species, we calculated the mean number of languages spoken in the distributions of the species present in each grid cell (S1 Fig).
Using this information, we were able to identify areas in the world with high numbers of bird species with many languages within their distribution. Spatial data were analyzed in a Mollweide equal area projection in ESRI ArcGIS version 10.4 [21] and PostGIS version 3.0.2 [19], and statistics were calculated in R statistical language version 3.5.1 [22].

Statistical analysis
To investigate factors explaining the number of languages spoken within each species' distribution, we performed generalized linear mixed models (GLMMs) assuming a negative binomial distribution with the number of (either official or most spoken) languages spoken within each species' distribution as the response variable, log 10 -transformed distribution range size (km 2 ), migratory status (non-migrant as the reference category), and conservation status (Least Concern (LC) as the reference category) as the explanatory variables, and the order of each species as a random factor. The GLMMs were implemented using the package lme4 in R [23].

Results
On average, seven official languages are spoken within a species' distribution, 16 for migratory species and three for threatened species. Additionally, 75.6% of the 10,863 extant bird species, 93.6% of the migratory species, and 55.5% of the threatened (59% of Vulnerable (VU), 52.5% of Endangered (EN) and 47.9% of Critically Endangered (CR)) species have two or more official languages within their distributions (Fig 1).
There is a strong positive relationship between the number of languages spoken within each species' distribution and range size, and species with wide distributions have as many as 100 official languages spoken within their distribution (Fig 2 and S2 Table). When controlling for the range size effect, threatened (CR and EN) and migratory species have significantly more languages spoken within their distributions, compared to non-threatened (LC) and non- migratory species (Fig 2, S2 Table). For example, Critically Endangered species with many languages within their distribution include Balearic shearwater (Puffinus mauretanicus, 25 languages), sociable lapwing (Vanellus gregarius, 22 languages), and Rüppell's vulture (Gyps rueppelli, 20 languages) (Fig 2B). The results vary between taxonomic groups, with species in some orders, such as Strigiformes (owls) and Psittaciformes (including parrots, parakeets, lorikeets and macaws), having comparatively few official languages within their distribution (seven and three on average, respectively), with others, such as Ciconiiformes (including storks, herons, bitterns, ibises and spoonbills) and Charadriiformes (including waders, gulls and auks), having especially many languages (19 and 17 on average, respectively; Fig 2C). The  Table).
English, Spanish, French and Portuguese are the four languages associated with the most species; this pattern was consistent for all species, threatened species, and migratory species (S3 Table, S4 Fig). Across all bird species, 45% have some area of their distribution associated with Spanish, 38% with English, 27% with Portuguese and 22% with French. For migratory species 67% were associated with English, 61% with Spanish, 42% with French and 38% with Portuguese. Finally, for threatened species, 23% were associated with Spanish, 16% with English, 16% with Portuguese and 12% with French (S3 Table). However, 899 species associated with Spanish are not associated with any other languages and thus, when only species associated with two or more languages were assessed English was the language associated with the most species, for all species and for threatened species (S3 Table). Geographically there is variation in the distribution of the species associated to the top six languages; In south America many species are associated with English, Spanish and Portuguese, in Africa with English, Kiswahili, Portuguese and French, and in Southeast Asia with Mandarin (Fig 3; see https:// translatesciences.shinyapps.io/bird_language_diversity/ for the results of other languages).
Especially many species with high numbers of languages spoken within species distributions were found in central and southern Africa, India, southeast China, eastern Europe, and Russia (Fig 4A). A large number of threatened species with high language richness were found in Western and Central Asia as well as southern Russia (Fig 4B). A similar pattern was found for migratory species with eastern Europe also being a hotspot of species with high language richness (Fig 4C). The results remained qualitatively the same when using the most spoken language, instead of official languages, in each country (S2-S5 Figs).

Discussion
Our research provides important insights into where in the world and for which species conservationists are especially required to make extra efforts to overcome language barriers to improve bird conservation. Our results reveal that threatened (CR and EN) and migratory species have more languages spoken within their distribution, when controlling for range size (S2  Table). Additionally, our results show that 217 bird species have 50 languages or more spoken within their distributions (S4 Table) and that more than two thirds of all species, half of the threatened and almost all migratory species are associated with two or more languages (Fig 1). These results, together with the multiple ways that language barriers can affect conservation (Table 1), highlight the potentially serious consequences of language barriers in bird conservation, especially for migratory and threatened bird species. For example, the distribution of the common pochard (Aythya ferina), which is classified as Vulnerable by the IUCN, spans 108 countries in Europe, Russia, Asia, and north Africa, where a total of 75 official languages are spoken. This means that scientific information on this species (including peer-reviewed papers and grey literature) can be scattered across those different languages [24,25], and successful conservation of the species may depend on effective collaboration and policy agreements among people with diverse linguistic and cultural backgrounds. Species in the orders Ciconiiformes and Charadriiformes have an especially high number of languages spoken within their distributions. For example, the Critically Endangered spoon-billed sandpiper (Calidris pygmaea) has nine different languages spoken within its distribution. For this species, educational kits with information about the species ecology and its conservation have already been translated to five different languages to improve the outreach of the conservation message (https:// www.eaaflyway.net/spoon-billed-sandpiper-teaching-kit-available-for-free-download/), demonstrating the work required to address language barriers in conservation. The conservation of species associated with many languages will likely require such coordinated efforts among stakeholders with different cultural and linguistic backgrounds, for example through incorporating action plans to overcome language barriers in relevant policy agreements, such as those in the Convention on Migratory Species [26].
Even though one third of the bird species globally have English spoken within a part of their distribution, other languages are also associated with a large number of species in certain regions, such as Spanish and Portuguese in South America, Kiswahili in Africa, and Mandarin in South East Asia. These languages could be key to conservation research, policies, and practices in those regions. For example, important information related to species ecology and conservation is often available in non-English languages [8], as researchers in countries where English is not widely spoken publish their research not only in English but also in other languages [27]. Scientific information that is available only in non-English languages is however usually omitted when conducting conservation research and generating conservation plans [28]. The omission of such non-English-language information can bias inferences of ecological analysis [29] and cause gaps in the spatial and taxonomic coverage of scientific evidence that is useful for conservation [24], which in turn can cause suboptimal conservation decisions. Effective conservation of bird species would require synthesizing scientific information and transferring generated knowledge in these key languages, and our results provide practical information on which species would benefit from multilingual assessments and which languages are key to those species (S4 Table and  Overcoming language barriers will play an important role in areas with a large number of species with many languages spoken within their distribution. These regions include central and southern Africa, India and Southeast Asia as well as Kazakhstan, southern Russia and Western Asia for threatened and migratory species. Challenges for bird conservation in these regions include a need to reconcile perspectives and interests among extremely diverse stakeholders, as species in these regions have, on average, up to 84 different languages spoken within their distributions. Establishing cross-national associations, such as the European Bird Census Council (https://www.ebcc.info/), in these regions would be an effective approach for coordinating monitoring and conservation efforts and achieving consensus decisions among countries. Other areas where species associated with particularly many languages are found are Europe, north Africa, western Asia and north Russia; again up to 84 languages, on average, are spoken within the distribution of species found in those regions. Although these regions did not show the highest richness of such species, this does not diminish the importance of proactively accounting for language barriers in conservation initiatives in these areas. For example, the United Nations Barcelona Convention has developed guidelines on conservation of Mediterranean seabirds that promote coordinated actions between countries with different language backgrounds. The Mediterranean Small Island Initiative also aims to facilitate collaborations between ten different countries in the region [6]. Such initiatives would benefit from the creation of guidelines to overcome language barriers between the parties involved [7].
While our analysis is, to our knowledge, the most comprehensive assessment of the identity and number of languages spoken through the distribution of bird species globally, the way this association was measured has some caveats. The presence of a species on a particular country does not imply that all of the official languages spoken in that country are spoken within that species distribution. Additionally, the fact that multiple languages are spoken through the distribution of a species does not imply that all the scientific information is being generated in different languages through the species distribution. Education systems in many countries promote learning of multiple languages (including English) that are different from the official or most spoken one in the country [30,31]. Future research is needed to understand the ability of people to work across language barriers and how it varies geographically, and also to identify particular species with low compatibility between the languages spoken within their distribution, areas with an especially large number of such species, and languages that generate such incompatibility.
The global community has a joint responsibility to address the biodiversity crisis and avoid further species extinctions, which requires an effective transfer of knowledge and information between countries with diverse linguistic and cultural backgrounds. In Table 1 we identified four different pathways through which language barriers affect conservation: 1) scientific research, 2) policies, 3) conservation activities and 4) general public. In Table 2 we provide a set of potential solutions to overcome such ramifications of language barriers in conservation. A way to improve the transfer of scientific research and to overcome language barriers when generating and executing conservation policies is to promote the multilingual transfer of relevant information, ideally through a clear, concise and easy-to-use translation protocol [32], especially for species with many languages spoken within their distributions and in areas where those species are found. This can be done by, for example, providing translations of Table 2. Potential solutions to overcome multiple consequences of language barriers for species conservation.

Pathway Potential solution
Scientific Research Engage with scientist with different linguistic and cultural background, particularly from countries/territories where the species of interest is present.
Actively search for information sourced from multiple languages, especially languages associated with the species of interest.

Policies
Promote the multilingual transfer of relevant information when generating conservation policies, ideally through a clear, concise and easy-to-use translation protocol.
Translate relevant policy documents in multiple, relevant languages when generating species conservation policy agreements.

Conservation Activities
Generate transboundary conservation projects involving people with different linguistic and cultural background.
Exchange personnel involved in conservation activities of the same species across countries/ territories with different linguistic and cultural background.

General Public
Understand how different cultures might pose a threat to species and devise solutions specific to each culture.
Disseminate critical conservation information on the targeted species in a relevant language to the general public.
https://doi.org/10.1371/journal.pone.0267151.t002 relevant scientific papers and policy documents in multiple, relevant languages. Using information sourced from multiple languages, especially languages associated with the species being assessed, and actively engaging with scientist and politicians with different language and cultural background would also increase the access to otherwise omitted information. This improves the quantity and quality of the knowledge on the ecology and conservation of the species, which in turn facilitates the generation and execution of more effective conservation policies. On the other hand, stimulating multilingual conservation activities, such as the ones implemented in the program "Birds without borders" (https://www.birdlife.org/africa/ projects/conservation-migratory-birds-cmb), as well as promoting the translation of critical conservation information on target species into clear and brief documents for the general public would improve the success of conservation actions and the outreach of information on how to avoid the extinction of those species. Our analysis has shown species and areas with significant challenges of language barriers to conservation and we have provided some potential solutions for these challenges. To implement these solutions and overcome these barriers there is a need for political will, local support and sufficient resourcing [4,8].
Supporting information S1 Table. List of official and most spoken languages for each country in the world. (XLSX) S2 Table. Results of negative binomial generalized linear mixed models (GLMM) to explain the number of (official or most spoken) languages spoken within bird species distribution (the response variable) using the three explanatory variables: log 10 -transformed distribution range size (km 2 ), migratory status (non-migrant as the reference category), and IUCN threat categories (Least Concern as the reference category). The order of each species was also incorporated in the models as a random factor. Statistically significant p-values (p < 0.05) are indicated in bold.