Why language barriers matter in conservation science


Nowadays, almost on a daily basis, we see papers and databases that claim to cover “the globe” or “the entire world”. When I bump into these statements I always get mixed feelings.

I say mixed feelings because, on the one hand, I value those broad-scale studies and data. I am working on a global-scale dataset myself, and am also very interested in how biodiversity is distributed across the globe, where it is changing, what is causing the loss and how our countermeasures might be effective, etc. So inevitably I always start looking into the details of those studies with some excitement.

However, as I go through the paper or database, and particularly when I reach the map of where the data used originate, often proudly shown as Fig. 1 or in the supplement, I can’t help but become skeptical about the paper; is this really ALL the data they were able to collect for this study?

In some cases, I actually know the answer is no. As someone originally from Japan, I am relatively familiar with papers and datasets published in Japan, and I sometimes know for a fact that at least some of those I am aware of have not been used in the study. Quite often the reason behind this is, I suspect, language barriers, i.e., most papers and datasets published in Japanese language are not effectively compiled or used in global studies. But admittedly I understand only two languages, and couldn’t be sure how generally applicable my observation on Japanese-language studies was.

So I decided to ask native speakers of 14 other languages, from Spanish and Portuguese to Turkish and Persian, what they thought about language barriers in conservation science. One great advantage of being based in a place like Cambridge is that I was able to find those native speakers easily, mostly among my (former) colleagues. Many of them, as expected, agreed that language barriers can be a big problem in their countries.

So in our latest paper just published in PLOS Biology, I decided to further explore and discuss potential consequences of language barriers in science, focusing particularly on environmental sciences. The paper is open access and short, so please do have a read, but here is an even shorter summary:

  • Language barriers can limit the transfer of scientific knowledge in two directions: (i) when compiling knowledge in global assessments and (ii) when applying knowledge to local environmental issues, often tackled by local practitioners and policy makers.
  • For (i) we showed that up to 36% of conservation-related scientific documents in 2014 were published in non-English languages.
  • Ignoring non-English literature can also cause several biases, such as the overrepresentation of positive/statistically significant results and underrepresentation of information from countries where English is not widely spoken or knowledge generated by practitioners.
  • For (ii) we showed, as an example, that 54% of protected area directors in Spain identified languages as a barrier to the use of scientific papers for their management.
  • We provide a number of suggestions for tackling this issue. Effective approaches for compiling non-English knowledge include: involving native speakers of non-English languages in projects, and increasing the visibility of non-English literature through the development of databases for non-English journals. To facilitate the multi-lingualization of knowledge currently available in English, we propose producing translations of published papers made available as lay summaries on journals’ websites or supporting information (as we have for this paper in the form of Alternative Language Summaries), preprints and postprints.

This paper is not the first to discuss the consequences of language barriers in science. For example, see this book and this paper for a general overview, or this paper and this for discussions more specific to conservation. Nevertheless, I feel that this issue has rarely been tackled by conservation communities, despite the fact that today there are countless “global” projects collecting information globally and many conservationists try to identify and remove barriers between science and practice. To achieve a less biased synthesis of scientific knowledge, and its smooth application to on-site conservation, I think it’s about time someone spoke up and tackled this issue, and am hoping to be one of them.

So for example, if you have a figure that shows a biased distribution of data for your meta-analyses and are about to write in your paper’s draft, “data are mostly from West Europe and North America“, do consider searching for information in other languages. It might be hard for those in the West to imagine, but in Japan about 1,300 people regularly make presentations in Japanese at the annual meeting of the Ecological Society, and many of these works are published in Japanese. To take another example, in China a colleague of mine has just identified 58 journals published in Chinese (yes, FIFTY EIGHT JOURNALS) that are potentially relevant to conservation. Overlooking these information sources is, in my opinion, simply a loss to international communities.

If you have, on the other hand, conducted research relevant to countries where English is not widely spoken, and are committed to making a real difference in on-site conservation there, do consider making the knowledge you obtained available in relevant local languages. Many practitioners and policy makers whose mother tongue is not English are likely to understand English to varying degrees, but they may not have time to use dictionaries to read entire papers written in English; they might instead just rely on other, more readily understandable sources when faced with the need to make decisions immediately. Proficiency in English may improve in many countries in future, but such a change needs time. Meanwhile, most environmental issues are urgent. In order to resolve these urgent issues, we cannot afford to wait for societies to become more English proficient.

Just to emphasise, I fully recognise the importance of having a lingua franca in this era of globalisation, and the contribution of English to science through, for example, the centralisation of scientific knowledge and its role in making global communications possible. I do not mean that we should move away from using English through a wider use of other languages. What I actually want to argue is that we need scientific information to be made available to its users, such as practitioners and policy makers, in multiple languages.

It is also evident that we are already aware of the importance of cultural diversity in the conservation of biological diversity, as shown by the fact that one of the Convention on Biological Diversity’s Aichi Targets uses linguistic diversity as an indicator. So it only seems to make sense for us to also consider more seriously the implications of having linguistic diversity for the conservation of biodiversity.

Overcoming language barriers should not be considered as an extra burden for conservation scientists. It is certainly a challenge, but a challenge worth taking on, because once achieved it could benefit both those synthesising knowledge globally and those tackling biodiversity conservation locally.


Thriving economies, disappearing languages

Dodo, Tasmanian tiger and passenger pigeon. I am sure you can see what’s in common among these species, particularly if you are a conservationist. But even for non-experts, this is probably easy to answer – they are all extinct species.

But what about Eyak or Ubykh? I suspect few people would know what they are and what has happened to them. They are human languages that have recently gone extinct. Eyak was an Alaskan language and when the last native speaker, Marie Smith Jones, died in January 2008, with her died the language itself. Ubykh was a Turkish language and similarly, the last speaker Tevfik Esenç died in 1992.

About 7,000 languages exist around the world, but many are seriously threatened with extinction. For example, Armando Perea Carijona (below) is one of the six remaining speakers of Carijona, a Colombian language.
Photo by Rodrigo Cámara Leret (Section for Ecoinformatics & Biodiversity, Department of Bioscience, Aarhus University)

Also, if you take a quick look at a language database called Ethnologue, sadly it is not hard to find languages with only one remaining speaker, like Lakondê in Brazil, Luo in Cameroon, and Taushiro in Peru. Yes, only ONE speaker. The world now has 7 billion people, but only one person can speak that language fluently. I have been using this database for a while, and have actually experienced seeing a language previously described as “with one speaker” being newly categorised under “no known speaker” after an update of the database. So language death is happening even at this very moment, at an alarming rate. Even within this year, we have already seen the death of the last speaker of another language, Klallam in the US.

There is a huge debate about why we should or should not be concerned about the loss of languages (see for example some chapters in this book). But one consensus is that the loss of human languages is often associated with the loss of human cultural diversities. Consequently many international organisations, like the CBD, IUCN, WWF and UNESCO are now actively engaged in the conservation of cultural and linguistic diversity.

Species and languages are of course different, but at the same time they show interesting similarities, such as patterns in spatial distribution (Grenflo et al 2012). Approaches used in ecology have frequently been applied to understanding the dynamics of languages as well (Solé et al 2010). So, languages are threatened and in need of conservation. Knowing that we may be able to contribute to this problem by applying approaches we are using every day, it was quite natural for me to start working on this topic, even though I am not an expert in linguistics.

So in this paper recently published in Proceedings B, we decided to assess the global distribution and drivers of language extinction risk, using macroecological approaches. We first collected data on three characteristics of languages: geographical range size, the number of speakers, and changes in the number of speakers. This is based on the IUCN criteria for assessing species extinction risk, but there is both empirical and theoretical evidence supporting that these three characteristics also represent the extinction risk of languages. We thus mapped the global distribution of each, and identified what explains the spatial patterns in these extinction risk components.

What did we find? I believe the most important finding of this paper is the dominating impact of economic growth on language speaker declines. Unlike patterns in species declines, we found that recent speaker declines in the world’s languages have occurred particularly at higher latitudes, such as in the US, Canada, a part of Europe and Russia, and economic growth measured by GDP per capita, among many other environmental and socio-economic factors, was most strongly associated with this spatial pattern. Earlier studies have indicated the importance of economic growth and associated globalization in the process of language endangerment, mainly based on local case studies. In our paper, we showed that this is a global phenomenon.

Another key finding was that there are two types of hotspots for threatened languages. In economically-developed regions, such as North America and Australia, local languages are now rapidly declining through direct and indirect impacts of national economic growth. So these regions need immediate attention if their languages are to be conserved. In contrast, we also found that many small-range and small-population languages still persist in the tropical and Himalayan regions, as these regions have topographically heterogeneous, productive environments as well as rapidly growing human populations, both of which are expected to allow small languages to evolve and persist. But many countries in these regions are now experiencing rapid economic growth, which means that in the near future languages in these regions are likely to face an elevated risk of extinction. Clearly, in the mid/long-term, we also need to pay much more attention to situations in these regions.

There are many other (hopefully, interesting) findings from this work. Luckily this paper is open access, so if you are interested, please do download and read the paper.

I am extremely happy to finally see this paper in publication. I’m particularly glad because this paper is the outcome of a long collaboration with many people. Bill’s paper on language extinction risk, published in 2003, formed the base of this work. Heidi worked on the global congruence of linguistic diversity and biodiversity for her Master’s thesis in Cambridge, and it was Edouard who started working on the changes in speaker numbers of the world’s languages. Then, few years ago, we started collaborating with experienced macroecologists in Denmark and the UK, and finally, this paper is out!

Hopefully, this paper is just a starting point for our project. We have many other ideas, and I hope to be able to talk about our next output here in the near future.

Does heterogeneity always benefit biodiversity?

About two months ago, our paper from one of my collaborative projects got published in PLOS ONE*. I know it’s rather late, but I’ve finally found time to write up a summary of the paper, with some additional background and details.

*Katayama, N, Amano, T., Naoe, S., Yamakita, T., Komatsu, I., Takagawa, S., Sato, N., Ueta, M. and Miyashita, T. (2014) Landscape heterogeneity – biodiversity relationship: effect of range size. PLOS ONE 9: e93359.

Being an ecologist, I’ve always found the word ‘heterogeneity’ to appear frequently in what I see and read (but realised after moving to the UK that this word is not that commonly used by non-ecologists!) Either way, the heterogeneity of habitats and landscapes is undoubtedly key to biodiversity. In fact, thousands of papers have reported that heterogeneity leads to increased species richness.

But how general is this relationship? A recent global meta-analysis has successfully shown that the overall mean effect of heterogeneity on species richness is positive across taxa and regions. This conclusion is reasonable, but we became interested in the other side of the same issue; when we look at individual studies, the relationship can be highly variable, ranging from very negative to very positive effects. So, what factors explain this context dependency?

We developed a hypothesis, which attributes the varying effect of heterogeneity to differences in species composition and historical processes that have formed it. That is, in historically-heterogeneous regions, species adapted to spatially- or temporally-variable environments are expected to be widespread, creating the overall pattern of heterogeneity-richness association. On the other hand, this means that in such regions species adapted to homogeneous environments may have already become extinct or be distributed only over a narrow range.

We tested this hypothesis using avifauna in Japan. Japan is a mountainous country with > 60% forest cover, but a mosaic of landscapes have long been maintained, over thousands of years, by its topographic heterogeneity and more recently, farming activities. So it’s an excellent study system to test this hypothesis.



Mosaic landscapes in Japanese farmland

The results showed a clear support for our hypothesis. Total bird richness was the highest at the intermediate level of forest cover (in Japan, this is a good measure of mosaic landscapes, called Satoyama), showing a positive heterogeneity-richness relationship (panel (a) below). This relationship was, however, mostly due to the same relationship observed in wide-ranging species (b), and narrow-ranging species actually showed the opposite pattern, i.e., they were the most numerous in open or forest habitats (c).


The same was true even when looking at patterns in the abundance of each species; many wide-ranging species (e.g., Japanese bush warbler and meadow bunting) showed the highest abundance in mosaic landscapes while most narrow-ranging species (like ruddy kingfisher and black-browed reed warbler) did not.


Meadow bunting

These results have given us two insights about the heterogeneity-richness relationship.

First, the effect of heterogeneity clearly depends on species composition. Even if an overall pattern shows a positive association, responses by wide-ranging species may be masking those by rare species favouring homogeneous habitats. We cannot assume that maintaining heterogeneity always benefits biodiversity. Instead, we should manage landscapes specifically for the target species or systems. I think this is similar to the conclusion regarding the effectiveness of Entry Level vs Higher Level Stewardship for farmland biodiversity conservation in the UK.

Second, as we hypothesised in this study, the context-dependent relationship between heterogeneity and species richness may be explained by how species composition in the region has been formed historically. In this project, we are keen to test the hypothesis in historically homogeneous regions and are also trying to develop a framework for understanding this relationship in a more general way.

Going back to Japan, recent anthropogenic impacts have been transforming landscapes dramatically, which may alter the historically-developed relationship between heterogeneity and species richness. For example, recent loss of Satoyama landscapes in Japan has caused a range contraction in grey-faced buzzards, a well-known ‘mosaic-habitat’ species. So in the near future, if this trend continues, we might start seeing the positive effect of heterogeneity even in rare, narrow-ranging species.


Grey-faced buzzard

Flowering early or shifting north

This is my first blog entry in English. I’ve been blogging in Japanese for over 8 years (!) and I’ve come to realise that it is a very effective way of conveying things to many people, including those I’ve never met. Having tweeted in English for 2 years, I’ve decided that I want to write in English and make it available to more people. I probably won’t blog very often, but I’m hoping to write something at least when my new papers come out. As many of you would know, there are so many things we can’t say within a short manuscript. I want to use this space to write what I couldn’t write in the paper.

Today I will write about my new paper*, which has just been published in Proceedings B. The main finding of this paper is that the two types of plants’ responses to warming climate, range shifts and flowering changes, are linked in a complementary manner.

*Amano, T., Freckleton, R.P., Queenborough, S.A., Doxford, S.W., Smithers, R.J., Sparks, T.H. and Sutherland, W.J. (2014) Links between plant species’ spatial and temporal responses to a warming climate. Proceedings of the Royal Society B: Biological Sciences 281: 20133017.

It is well known that under warming climates some species shift their ranges poleward while others don’t. What causes this varied response among species is still under debate. In this study we propose that phenological changes may be a reason. Using historical records of distribution (Atlas of the British & Irish Flora) and flowering dates (Nature’s Calendar) of over 200 British plant species, we showed that species that have not moved northward in the past few decades have instead tracked warming temperatures (or “conserved” their climatic niche) by advancing their flowering dates. The link was particularly strong in annual plants, probably because flowering at the right timing is more critical for an annual plant as it only gets the one chance to reproduce.

For example, Rue-leaved Saxifrage (Saxifraga tridactylites) has advanced its flowering dates by 11 days during the past 50 years and has successfully tracked increasing temperatures during flowering time.


Changes in the first flowering dates of Rue-leaved Saxifrage, based on the species-level index developed in Amano et al (2010)*.

* Amano, T., Smithers, R.J., Sparks, T.H. and Sutherland, W.J. (2010) A 250-year index of first flowering dates and its response to temperature changes. Proceedings of the Royal Society B: Biological Sciences 277: 2451-2457.

Consequently, this species’ range has not moved north (mean latitude actually moved south by 0.07 degrees) over the same period.


Changes in the geographical range of Rue-leaved Saxifrage. Blue, red and black dots show observations only in 1987-1999 (new colonization), only in 1930-1960 (local extinction), and in both periods, respectively.

In contrast, Blue Fleabane (Erigeron acer) has only slightly advanced its flowering dates (by 3 days over the past 50 years) and thus, experienced increasing temperatures during flowering time.


Changes in the first flowering dates of Blue Fleabane.

Consequently, the mean latitude of this species’ range has moved north by 0.13 degrees over the same period.


Changes in the geographical range of Blue Fleabane.

Many studies have reported species’ range shifts and phenological changes, but most of the studies have treated these two types of species’ responses separately. Our study indicates that the two well-known responses to warming climates are linked. We believe that this finding helps us understand the complex consequences of climate change for biodiversity.

As always, this work was a collaboration with many great coauthors, and I am particularly happy that this work has finally been published, as it took me almost 4 years to get it out as the second paper from my phenology project. I still clearly remember the day in 2009 when Tim, one of our coauthors, visited Cambridge to give a talk on phenological changes. I had been working on the modelling of bird population counts, and after his talk, Bill, Tim and I discussed the potential application of the modelling approach to phenological data. In 2010, we published our first paper using records from the UK Phenology Network. This paper was more or less a methodological paper, but produced estimates of changes in first flowering dates for 405 British plant species. So it was only natural that we then wanted to do something using these estimates.

Given that we can understand both phenological changes and range shifts within the same framework, i.e., they are both a consequence of climatic niche conservatism (tendency for species to maintain their climatic niche over time), it seemed logical for us to hypothesise that these two responses are linked in a complementary manner. Luckily (and as a positive surprise for me), we (I mean, Bill) could easily find data on historical range shifts for most of the species, based on the Atlas of the British & Irish Flora, and processed by Doxford and Freckleton (2012).

These data were key for this second paper, as it was essential that we had these historical data on both phenology and the distribution of many species if we were to test our hypothesis—that there is a link between plants’ spatial and temporal responses to climate change. So, needless to say, this study would not have been possible without the long history of botanical studies and a wealth of citizen science projects in the UK.

We admit that the detected link was relatively weak, but believe that the main novelty of this paper lies in showing the potential link between the two-well known species responses to climate change. So we are very keen to test this link in other taxa and systems. I want to end this blog entry with a special thanks to the countless people who have contributed to the UK Phenology Network and the Atlas of the British & Irish Flora, as well as the Woodland Trust and NBN Gateway for providing the data.