The authors have declared that no competing interests exist.
Conceived and designed the experiments: AC LW. Performed the experiments: AC. Analyzed the data: LW. Contributed reagents/materials/analysis tools: AC LW ED NB. Wrote the paper: LW AC ED NB.
Current address: Unité Mixte de Recherche AMURE, Institute Universitaire Européen de la Mer (IUEM), Université de Brest, Best, France
Cultural ecosystem services are defined by people’s perception of the environment, which make them hard to quantify systematically. Methods to describe cultural benefits from ecosystems typically include resource-demanding survey techniques, which are not suitable to assess cultural ecosystem services for large areas. In this paper we explore a method to quantify cultural benefits through the enjoyment of natured-based tourism, by assessing the potential tourism attractiveness of species for each protected area in Africa using the IUCN’s Red List of Threatened Species. We use the number of pictures of wildlife posted on a photo sharing website as a proxy for charisma, popularity, and ease of observation, as these factors combined are assumed to determine how attractive species are for the global wildlife tourist. Based on photo counts of 2473 African animals and plants, species that seem most attractive to nature-based tourism are the Lion, African Elephant and Leopard. Combining the photo counts with species range data, African protected areas with the highest potential to attract wildlife tourists based on attractive species occurrence were Samburu National Reserve in Kenya, Mukogodo Forest Reserve located just north of Mount Kenya, and Addo Elephant National Park in South-Africa. The proposed method requires only three data sources which are freely accessible and available online, which could make the proposed index tractable for large scale quantitative ecosystem service assessments. The index directly links species presence to the tourism potential of protected areas, making the connection between nature and human benefits explicit, but excludes other important contributing factors for tourism, such as accessibility and safety. This social media based index provides a broad understanding of those species that are popular globally; in many cases these are not the species of highest conservation concern.
Spatial assessments of ecosystem services (ES), aiming to support management of our natural environment, are increasingly common in science (e.g., [
One of the cultural ES that receives a large and growing demand from our industrializing society is the enjoyment of nature through recreation and tourism activities [
People find some species more attractive than others [
In this paper we aim to construct an attractiveness index to assess the nature-based tourism potential for PAs, based on Red-Listed species occurring in an area. Our social media-derived index is intended to help understand how species contribute to the tourism potential of African PAs; an index that can be used—along with other tools- by conservationists, scientists and practitioners. Our index is based on minimal, freely available data and could therefore be used in large scale quantitative ES assessments when detailed surveys are not an option.
We base our analysis on the species listed on IUCN’s Red List of Threatened Species [
Information on African PAs was derived from the World Database of Protected Areas [
To derive a quantitative proxy on attractiveness of different species for tourism, including factors of popularity, charisma and ease of observation, we counted the number of photos of IUCN-listed species posted on the internet. The ease of wildlife to be observed includes aspects of visibility, diurnal versus nocturnal species, occurrence, and ease of identification. The number of online images for each species was counted based on a search on the binomial Latin species name in the Application Programming Interfaces (or APIs), which are available from the search engines Google, Bing, and Flickr. Flickr is the most established photo-sharing and management site and is now over 10 years old. APIs are freely available tools for searching web resources and allow for searching of images using all of the common search patterns and parameters. The domains in which the searches are made varies by provider; Google and Bing return the number of images that are available on the entire internet, whereas Flickr only returns the number of images on Flickr itself. Therefore the number of images for each species varies considerably between the different providers. Our main concern was to only include images in our analyses that actually depict the searched species. For this reason we had to abandon an original idea to search for images using the species’ common name in English, as this resulted in poor search results for species with common words in their English common name, such as ‘Red Kite’ or ‘Wild Cat’. After visual interpretation, we found that images resulting from a search for binomial names in Flickr showed the highest number of correct species depicted on the images compared to results from the Google or Bing APIs. Therefore, only the image counts resulting from the Flickr API search using binomial names are used in our analyses. We searched the contents of the photos as described in their label irrespectively of location where the picture was taken. We are aware that photos posted online might not have been shot in the wild, such as animals in zoos, but even in that case people only take photographs of the species they consider to be attractive.
Two corrections to the search results are made. First, one issue with searching the Flickr API is that the number of photos of species which have the same genus name as species name (e.g.
Second, we attempted to exclude photos of species without a clear link to tourism, in particular images of very common species in gardens, pests, or domesticated pet species. These species have many images in Flickr but it is unlikely that for these common species people would travel to PAs to view them. For example the Mallard (
The species attractiveness index (SAI) per PA is subsequently calculated by combining the IUCN Red List species range data, PA locations, and online image counts (
We subsequently explored if species with different conservation statuses contribute differently to the total attractiveness for tourism of a PA. IUCN groups Red List species that occur in the wild by the following increasing ranks of conservation concern: Least Concern (LC), Near Threatened (NT), Vulnerable (VU), Endangered (EN) and Critically Endangered (CE). Evaluated species lacking assessment data are labeled by IUCN as Data Deficient (DD), whereas non-evaluated species are null in the dataset. Using a non-parametric Kruskal-Wallis test, we tested for differences between the Red List status and the number of posted images (with H0: no differences in distribution of photo counts are observed among the included Red List status classes CR, EN, VU, NT and DD).
There are a large variety of PAs. IUCN has developed guidelines to group PAs according to their (assigned) conservation management objectives (see
Areas managed for: | |
---|---|
I | Strict protection (1a Strict Nature Reserve, 1b Strict Wilderness area) |
II | Ecosystem conservation and protection (i.e. National Park) |
III | Conservation of natural features (i.e. Natural monument) |
IV | Conservation through active management (i.e. Habitat species management area) |
V | Landscape/seascape for conservation and recreation (i.e. protected landscape/seascape) |
VI | Sustainable use of natural resources (i.e. managed resource protected area) |
Intersecting the selected IUCN Red List species range data with the location of the African PAs resulted in a list of 2473 species that are assumed to occur within these PAs. Most of these species are mammals (513), followed by bird species (365), amphibians (439), fish (253), reptiles (129), coral species (269) and the remaining 236 species of jellyfish, snails and plants. The Lion is by far the highest scoring species by photo counts, followed by the African Elephant, Leopard and Cheetah (
The highest 55 ranked species contribute to 72% of the total image counts.
Binomial name | Common name | Photo count | Taxonomic Class | Red List status |
---|---|---|---|---|
Lion | 18574 | Mammalia | VU | |
African Elephant | 8375 | Mammalia | VU | |
Leopard | 6737 | Mammalia | NT | |
Cheetah | 5998 | Mammalia | VU | |
Red Kite | 3739 | Aves | NT | |
Ring-tailed Lemur | 3666 | Mammalia | NT | |
Chimpanzee | 3159 | Mammalia | EN | |
Hippopotamus | 2730 | Mammalia | VU | |
Black-tailed Godwit | 2635 | Aves | NT | |
White Rhinoceros | 2614 | Mammalia | NT | |
Eurasian Curlew | 2435 | Aves | NT | |
African Penguin | 2344 | Aves | EN | |
Lemon Shark | 2165 | Chondrichthyes | NT | |
Sperm Whale | 2152 | Mammalia | VU | |
African Wild Dog | 1970 | Mammalia | EN | |
Grey Crowned-crane | 1917 | Aves | VU | |
Gentoo Penguin |
1835 | Aves | NT | |
Tiger Shark | 1594 | Chondrichthyes | NT | |
European Rabbit |
1535 | Mammalia | NT | |
Barbary Macaque | 1499 | Mammalia | EN | |
Lesser Kestrel | 1498 | Aves | VU | |
Grevy's Zebra | 1267 | Mammalia | EN | |
Black Rhinoceros | 1229 | Mammalia | CR | |
European Roller | 1212 | Aves | NT | |
Mandrill | 1200 | Mammalia | VU |
* Occurs on an island in the sub-Antarctic Indian Ocean, an administrative part of South-Africa
** Occurs in northern Africa
The findings presented in
For the 5288 African PA entries in the WDPA, the SAI values were calculated based on the photo count of the overlapping IUCN Red List species (
The 20 highest ranked PAs by SAI are labelled.
All scores per PA and Flickr photos counts per species are published at this spatial user interface,
We explored what type of species show up most often according to our search criteria in Flickr and therefore contribute most to a high SAI at the PA level. We found that the number of photos posted is positively correlated with the extent of species range, but only to a small extent (rho 0.41, or when 0 counts excluded rho 0.29, for p = 0.05). We also looked for associations between the IUCN Red List status and number of image posts. Based on a Kruskal-Wallis test, differences between the Red List status and the number of photos were found. The photo counts are heavily skewed towards zero, 1441 of the 2473 species were not found at all on Flickr. When excluding species for which no images were found (i.e. are not contributing the SAI), significantly more images were found for species categorized as Vulnerable and Near Threatened compared to the low scoring Data Deficient category (with p = 0.05). On average the Data Deficient species had the lowest number of photos posted (with an average count of 37 per species), whereas the Vulnerable (219) and Near Threatened (173) had the highest number of posted pictures on average. Note that these numbers are generated by a small number of highly popular species (
Our approach to defining species attractiveness can be seen as a species richness index weighted by species attractiveness for each PA. The left graph in
When we compare our SAI results among the different IUCN classes for PAs, we find that the IUCN classes Ia (average of 11737 summed photos, or 0.17 when normalized) and VI (37629 photos on average, or 0.57 when normalized) are significantly (p = 0.05) different from the ‘not listed,’ class, used as reference, see
PAs managed as Strict Reserves, category Ia, score lowest on their total attractiveness for nature-based tourism. The boxes indicate the quantiles and the line the median. The Not Applicable class includes PAs managed as UNESCO Man and Biosphere Reserves, World Heritage Sites, Wetlands of International Importance through the Ramsar Convention and Special Protected Areas of Marine Importance through the Barcelona Convention.
We developed this method to explore if preferences for species could be included in a quantitative indicator to describe nature-based tourism potential at a large scale. One of the problems of producing metrics on tourism
For 69 of the PAs, we collected visitor data (for around the year 2009) and linearly regressed that data with the photo count-based SAIs, and found that that relation was not significant (p > 0.05, R-square of approximately zero). However, Tsavo East, Addo Elephant, Maasai Mara, and Zambezi PA which rank within the top 20 in the SAI list, also appear in the top 20 of highest reported visitor numbers in our limited dataset of 69 PA (in places 6, 8, 9, 19 respectively). We are well aware that actual tourism levels in an area are affected by a range of other factors besides the presence of attractive species [
To explore the validity of the use of species photo-counts in our SAI, we also counted photos posted in Flickr for all non-African species included in the IUCN Red list. That global species count resulted in high scores for widely known charismatic and iconic species, see
The SAI of PA is calculated based on a summation of photo counts per species. We decided not to use any data transformations to calculate the SAI to 1) directly follow posted image counts as we could not justify any transformation choice, and 2) to keep ‘outlier’ effects. While smoothing data through transformations might reduce the impact of errors in photo counts, we are dealing with attractive species of which some have exponentially higher public attraction, i.e. these species would be outliers in the dataset. Our results show that only 2% of the studied species contribute 72% of the contributed photos, largely influencing the highest scoring PA. In the online interface we show the different SAI results when applying different transformations (log, sqrt, ranks, classes) and summation of species counts per PA (see
The input data used for the development of our index are the best available large scale datasets, but have known limitations, accuracies and uncertainties. For example, for many species maps in the IUCN database, the extent of occurrence that is shown is considerably larger than its realized range (area of occupancy), that leads to errors of commission when we intersect IUCN species range data with WDPA PA polygons. The degree to which this affects our results depends on the species—the more common species tend to have more generalized polygons than the more threatened species. Even though many very common species (often species of Least Concern for conservation) where excluded from our attractiveness assessment, we expect that our SAI is therefore over-assessing species richness within PA, even leading to situations where marine species ranges overlap with land (for example case for the PA ‘Coutada No. 9’ in Mozambique). Besides that, the SAI is based on binary range data; i.e. it does not include an assessment of the size of the population inside these boundaries, a factor that could also influence tourism attractiveness. The WDPA which was used to locate PAs also has known issues relating to areas boundaries, duplicates and missing IUCN management categories etc. Both the WDPA and IUCN Red List of Threatened Species are neither perfect or complete datasets, however both are the most comprehensive and scientifically rigorous information about the of state and distribution of species [
Even though the use of social media to capture social preferences to assess a cultural ES is promising, issues about data quality, validation, and representativeness remain. The work presented in this paper on the use of online-posted photos to quantify species contributions to nature-based tourism could be seen as a first step in this. The index could be improved by more precise selection of images to able to also include species of LC typical of particular natural habitats and hence exclude the generalist and urban LC species.
If well-designed, equitably managed PAs could provide a powerful solution for maintaining ES, conserving biodiversity, and addressing the needs of human communities [
However, it is worth mentioning that in our analysis strict reserves (a PA with IUCN category I) scored significantly lower on our tourism species-attractiveness index compared to the PAs that are categorized as areas for sustainable use (IUCN VI). This means that strict reserves have fewer and/or lower scoring attractive species compared to PAs with a human use. This finding highlights an opportunity to explore the tourism potential for areas that already have a sustainable use objective (IUCN VI), without interfering with strict biodiversity protection strategies in place in IUCN I areas. At the same time, this finding also highlights the challenge to PAs to design management strategies aimed at safeguarding both biodiversity and ES.
As our tourism-index by itself cannot be used to directly inform PA management, the index combined with information on visitors’ willingness to pay (WTP), access to the PAs, safety of the visited country, local infrastructure and others, might be interesting for processes influencing PA management at larger scale, such as targeted fundraising based on the species attractiveness, or adjusted conservation strategies for less attractive species [
The presented species-attractiveness index contributes to the quantification of the potential for natured-based tourism in Africa, using the photos posted in social media to weight individual species attractiveness. Social media is becoming a rich source of data on the public’s behaviour, ideas and values, and therefore a new and promising way to assess subjective measures such as cultural ES. With this paper we hope to promote debate and move this area of work forward. The proposed method uses three global data sources which are freely accessible and available online which makes the index attractive for large scale quantitative ES assessments, including global and continental level ES accounting and ES modeling. The index links species presence to the tourism potential of PAs, making the connection between nature and human benefits explicit. Yet it still excludes other important contributing factors for realized tourism such as accessibility and tourist facilities. Using a social media based index in the first place gives insight into what species are most popular globally; these are in many cases not the species with highest conservation concern. This finding highlights the challenge to PAs to design management strategies aimed at safeguarding biodiversity and ES.
The authors would like to thank IUCN and UNEP-WCMC for making their valuable data publically available for non-commercial use. The very helpful critical and constructive comments of three reviewers improved our work considerably.