We now live in the Anthropocene, an age where human activities have strikingly altered biodiversity and the valuable services it provides to humans (McGill et al. ). We also live in the “Information Age” where a deluge of information is generated from digital technologies and novel data sources (Bennett et al. ). Leveraging the digital technologies and the wealth of user‐generated internet data can help address the biodiversity “crisis” (e.g., through species monitoring) (Di Minin et al. ; Van der Wal & Arts ; Ladle et al. ). Digital conservation is the emerging field of conservation science where digital technologies and novel data sources are being used to help stem biodiversity loss (Van der Wal & Arts ).Thus far, digital conservation has, among other topics, focused on citizen science (Pimm et al. ), novel monitoring tools (Saito et al. ), and cloud applications (Chapron ). The use of data mined from social media platforms, instead, is still limited in conservation science and practice (Di Minin et al. ).
Ecotourism potentially plays a key role in generating political support for protected areas (Di Minin et al. ). At the same time, ecotourism may also increase human pressure and impact on biodiversity (Buckley et al. ). Understanding people's preferences and engagement for biodiversity and biodiversity‐related experiences is essential to inform conservation management (e.g., management plans) and marketing (e.g., fundraising, Smith et al. ), particularly in protected areas (Buckley ). Preferences for biodiversity and biodiversity‐related activities have traditionally been studied using revealed and stated preference methods (Adamowicz et al. ; Louviere et al. ), which can, however, be costly (e.g., traditional surveys, both in terms of time and resources) and spatially and temporally limited (Richards & Friess ). Therefore, it is important to explore if novel data sources, such as social media data, may provide substantial and cost‐effective data at unprecedented spatial and temporal resolution that could be used to monitor human activities and people's preferences for nature‐based activities (Di Minin et al. ).
The use of social media is increasing dramatically worldwide (Mayer‐Schönberger & Cukier ) and millions of users on platforms such as Facebook, Instagram, and Twitter generate billions of posts every year (Kwak et al. ). User‐generated data mined from social media include (geotagged) text, pictures, or videos. In comparison to traditional survey‐based methods, social media data could provide cost‐effective information on activities and preferences of people visiting protected areas (Di Minin et al. ). Social media could also overcome limitations related to sample size, time and location constrains, nonresponse bias, and self‐reported errors (Mayer‐Schönberger & Cukier ). However, some concerns on the use of content mined from social media to inform conservation science should be considered (Arts et al. ). Concerns, for instance, are linked to data quality (Kitchin ), potential location inaccuracy of posts, biased behavior on social media, and the representativeness of the population using social media (Tufekci ).
The goal of this study is to explore whether social media data can be used as an information source to cost‐effectively assess tourists’ preferences for biodiversity and biodiversity‐related activities when visiting protected areas. In particular, we aim to (1) investigate the socio‐demographic background of social media users; (2) understand what nature‐related content tourists share on social media; and (3) assess whether social media data reflect tourists’ preferences for biodiversity as obtained from traditional surveys. To do this, we use Kruger National Park (KNP) in South Africa, as a case study. We chose KNP because it is one of the most visited parks in the world, playing a key role in protecting biodiversity (SANParks ). There, we carried out a traditional visitor survey, and compared the results from the survey to the content posted on Instagram (
KNP (Figure ) receives over 1.5 million tourists every year (SANParks ). KNP is one of the top destinations where to spot charismatic megafauna (e.g., the Big Five––see Di Minin et al. ), but it also attracts tourists interested in less‐charismatic biodiversity, biodiversity‐related activities, and sense of place experiences (Hausmann et al. ). Mobile phone coverage in the park is better around the main tourists’ camps (
In our study, Instagram and Flickr were chosen because both platforms were found to be used by tourists to share geotagged pictures in KNP, and because they provide application programming interfaces (APIs) for data access. Both platforms are popular channels for sharing and accessing media, including pictures, videos, and text. Instagram is generally used to capture, post, and share real‐time memories through a mobile application. On the other hand, Flickr is popular among photographers and is used to upload good quality pictures taken with professional cameras. We collected publicly available geotagged posts (including picture and text) by using the API of Instagram (
The content of the pictures was manually classified (Figure S1, Appendix S1) and double‐checked for consistency (see Appendix S1 for details). We adopted a classification approach based on the main subjects showed in the pictures (Richards & Friess ). In particular, pictures were coded based on the presence/absence of six main categories, namely, biodiversity, landscape, human activity, posing, accommodation, and food (Table ). In addition, we collected further details about biodiversity (i.e., the specific taxonomic group and species names) and human activity (i.e., the type of activity in which people were engaged). We also calculated the number of pictures belonging to more than one category. Pictures that were not publicly available (e.g., removed by the user and protected by privacy) or showing irrelevant subjects (e.g., advertisements) were discarded.
Description of categories used to classify the pictures posted on both Instagram and Flickr. Pictures may fall into more than one category| Category | Description |
| Biodiversity | Animal species visible in the picture and plant species as main subject of the picture (iconic trees and flowers) |
| Landscape | Pictures showing wide views of an area of land, with visible horizon |
| Human activity | People engaged in recreational activities, including objects that were directly involved in those activities (e.g., cameras, bicycles and cars used for game drives) |
| Posing | People looking at the camera (e.g., selfie), with recognizable faces |
| Accommodation | Pictures showing touristic infrastructures (e.g., lodges and boardwalks) |
| Food | Food or drinks showed in the picture |
The logical framework of this study is showed in Figure and explained in detail below. We implemented a survey in KNP during August 2014. A total of 563 national and international tourists, older than 18 years, were surveyed in the park. Tourists were approached randomly at the main tourists’ camps and surveyed using face‐to‐face interviews. We asked respondents to indicate their usage of different social media platforms and the type of nature‐related content shared. In addition, tourists were asked about specific preferences for biodiversity groups, and to provide information about personal socio‐demographic background (see Table S1, Appendix S1, for more details). We defined preferences as what tourists would particularly like to see or experience when visiting KNP. For the purpose of this study, we consider biodiversity as both plant and animal species, as well as habitat types. The sample of national and international tourists interviewed was representative of the tourists entering KNP according to official visitation statistics. (For more information, see Hausmann et al. ).
First, we used information obtained from the survey to evaluate social media usage among tourists (Figure ). We then used Pearson correlation tests performed using the R software (Version 3.2.3) (R Development Core Team ) to explore which factors (socio‐demographic background and biodiversity preferences of respondents) affected social media usage of tourists (Table S1 in Appendix S1). We used Fisher z‐transformation to detect statistical differences between coefficient values obtained by looking at different groups of tourists, such as gender and nationality.
Second, we used social media data to calculate the frequency of each category (Table ) of pictures in Instagram and Flickr. We also calculated the observed preference for biodiversity by looking at the representation (i.e., number of pictures posted), and the proportion of likes (average like/picture/group) for each taxonomic group. As highly popular users (e.g., famous people) may get high likes/picture thereby biasing results related to likes, users with the highest number of average likes/picture were discarded from the analyses (Appendix S1). We used two samples Z‐test, and Cohen's d effect size, to compare proportions of pictures posted on Instagram with those posted on Flickr, for each category/biodiversity group.
Third, we compared preferences for biodiversity observed on social media (representation, users, and likes) with results obtained from the survey. We used a two‐sample Kolmogorov–Smirnov test to explore whether two cumulative frequency distributions were identical. As representation of a subject on social media may be skewed by few highly active users (Li et al. ), we also considered the number of active users (i.e., users sharing at least one picture/group) in the test. We performed the tests both separately for Instagram and Flickr, and by combining information from the two platforms.
More national tourists (56%) and men (57%) were interviewed. The average age was 41 years, and more than half of the respondents had at least a bachelor's degree (64%). The majority of respondents (80%) declared to be a member of at least one social media platform or network. Among them, the majority was registered on Facebook (74%), followed by Twitter (25%) and Instagram (16%) (Figure a). Among all members, 72.5% declared to actively use social media to tell about their nature‐based experiences while visiting protected areas. Pictures were the most frequent type of media shared (Figure b). Among all platforms, Instagram (94%) and Flickr (92%) had the highest proportions of users sharing nature‐based pictures while visiting KNP.
Enlarge this image.Tourists’ membership to different social networks (a) and type of media used (b) to share about nature‐based experiences in protected areas.
The most important variable defining the use of social media among tourists was age, with usage decreasing significantly with older respondents (Pearson correlation: r563 = –0.31, P < 0.0001; Figure S2, Appendix S1). While younger tourists were more likely to be using Twitter, Instagram, and Facebook, the use of Flickr was not significantly correlated with age (Table S2, Appendix S1). Moreover, the use of social media decreased significantly with higher income of tourists (r563 = –0.13, P < 0.01), especially among Facebook and Instagram users. No significant differences were found between national (n = 341) and international (n = 249) tourists (Fisher‐Z = –0.509, P = 0.305), or between men (n = 321) and women (n = 242) (Fisher‐Z = –0.239, P = 0.406).
Large‐bodied mammals (95%) and landscapes (67%) were the preferred groups among respondents. Appreciation of other, less‐charismatic, biodiversity groups increased with higher experience (higher number of times visiting protected areas in Africa before) of tourists (Table S3, Appendix S1). Appreciation of birds, vegetation, and arthropods increased with the increase in age (respectively, r563 = 0.24, r563 = 0.12, and r563 = 0.18, P < 0.01). Interest in small‐bodied mammals, amphibians, and arthropods was also positively correlated with membership on Flickr (Table S2, Appendix S1).
Biodiversity and landscape pictures were most frequently posted on both Instagram and Flickr (Figure a; see Table S4 for complete list). Biodiversity pictures were more frequent on Flickr than they were on Instagram (Z = –36.59; P < 0.05; Cohen's d = –0.88). On the other hand, pictures including people, such as human activity (Z = 22.33; P < 0.0001; Cohen's d = 0.82), and posing (Z = 28.88; P < 0.01; Cohen's d = 1.49), were more frequent on Instagram than they were on Flickr (see Figure S3 in Appendix S1 for additional results). Tourists shared more often pictures that included biodiversity and landscape (7.2 %), and biodiversity and human activity (3.1%), compared to other combinations (Figure b).
Enlarge this image.Content of pictures shared on social media, including (a) proportions of pictures posted for each category by tourists visiting Kruger National Park during 2014; and (b) relationship between the most frequent categories (percentage of pictures containing more than one category) among Instagram pictures (pictures are from the authors). For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.
Large‐bodied mammals were the most frequent group represented on both platforms (Figure a). Among large‐bodied mammals, elephant (Loxodonta africana), lion (Panthera leo), giraffe (Giraffa camelopardalis), and white rhino (Ceratotherium simum) (see Table S5, Appendix S1 for the complete list) were the most represented species (Figure S4, Appendix S1). Pictures showing other taxonomic groups, especially birds (Z = –27.47; P < 0.01; Cohen's d = –1), arthropods (Z = –22.512; P < 0.01; Cohen's d = –1.5), and reptiles (Z = –6.84; P < 0.01; Cohen's d = –0.44), were more frequent among Flickr pictures. Small‐bodied mammals (24%) and birds (23%) were most liked among Flickr users, while large‐bodied (14.6%) and small‐bodied (14.2%) mammals were most liked among Instagram users (Figure b).
Enlarge this image.Comparison between surveyed and observed tourists’ preferences for biodiversity in Kruger National Park. In panel (a), observed preferences refer to the proportion of pictures shared on social media per group; in panel (b), observed preferences refer to the proportion of average likes/picture shared on social media. Lists showing how ranking of biodiversity groups changes in each data source are also provided when considering both pictures and likes.For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.
Compared to preferences expressed in the survey, large‐bodied mammals were overrepresented on both Instagram and Flickr (Figure a). Observed preferences for landscape in Instagram, and for both birds and arthropods in Flickr, were also higher on social media compared to the survey. Tourists’ preferences for all other groups were underrepresented on social media, particularly for small‐bodied mammals and amphibians (Figure a). Ranking of groups may change when looking at different platforms and survey results. However, large‐bodied mammals remain the preferred group across all data sources.
Proportion of likes appeared more evenly distributed among groups (Figure b). When looking at the likes, large‐bodied mammals and landscape were less preferred on social media compared to the survey. On the other hand, observed preferences for some less‐charismatic groups, especially small‐bodied mammals on Flickr, were higher in terms of likes, than interests in them as expressed in the survey. In particular, birds were highly appreciated on Flickr, while reptiles, amphibians, and arthropods received higher attention on Instagram.
When comparing distributions of surveyed and observed preferences for biodiversity, no significant difference was found between the populations of tourists (Table ). In other words, the distribution (of both pictures and likes) of preferences observed on social media was similar to the distribution of preferences expressed by tourists during the survey. Therefore, tourists’ preferences obtained by applying traditional surveys or by mining data from social media are from the same population. (See Appendix S1 for additional results on species preference comparison.)
Comparison of data mined from Flickr and Instagram with results obtained from the survey about tourists’ preference for biodiversity groups| Two‐sample Kolmogorov–Smirnov test | |||
| D | P‐value | ||
| Representation (total pictures/group) | 0.500 | 0.283 | |
| Flickr | 0.375 | 0.660 | |
| Instagram+Flickr | 0.500 | 0.283 | |
| Active users (one picture/user/group) | 0.500 | 0.283 | |
| Flickr | 1.000 | 0.0001 | |
| Instagram+Flickr | 0.500 | 0.283 | |
| Likes/picture/group | 0.375 | 0.660 | |
| Flickr | 0.250 | 0.980 |
Full list of frequencies is provided in Table S4, Appendix S1. Null hypothesis assumes identical cumulative distribution.
We demonstrated that geotagged content mined from social media can be used as a reliable alternative to traditional survey‐based methodologies to explore tourists’ preferences for biodiversity in protected areas. In addition, we found that combining data from different social media platforms can better assess the heterogeneous preferences of ecotourists for biodiversity (e.g., Di Minin et al. ). Furthermore, our findings show that social media data can also be used to understand which activities people engage with when visiting protected areas. While we acknowledge some limitations, we argue that the geotagged content of pictures can be used as a rapid and low‐cost way to explore preferences for biodiversity, and inform protected area management.
Previous studies used data mined from single platforms, such as Flickr (Wood et al. ; Mao ; Richards & Friess ; Willemen et al. ), Twitter (Roberge ), or Panoramio (Casalegno et al. ). However, we found that people use social media platforms differently according to their interests and background. For example, we found that while most of the pictures posted on Flickr were showing biodiversity, Instagram was also popular for sharing pictures about people, such as activities and posing. Moreover, while Instagram users appeared to be younger, Flickr users were nature enthusiasts with specific interests in some less‐charismatic biodiversity groups. Accordingly, our results suggest that different social media platforms may be used by different groups of tourists. Flickr appears more popular among more experienced tourists, who, according to previous studies (e.g., Di Minin et al. ), are less interested in charismatic megafauna. On the other hand, Instagram may reflect preferences of less‐experienced tourists, who are indeed interested in charismatic megafauna and in experiencing nature through biodiversity‐related activities (Hausmann et al. ). Therefore, it may be advisable to mine data from different social media platforms when exploring preferences for biodiversity and ecosystem services in the future.
As a caveat, we found that detectability of species groups (e.g., nocturnal small‐bodied mammals, or elusive and rare species) may result in underrepresentation (observation bias) of some biodiversity groups, limiting the ability of social media to fully capture tourists’ preferences. In this case, looking at pictures’ appreciation from the broader network (e.g., likes) may help address this gap. However, the use of likes to explore general preferences may be affected by social contexts and behavior on the web, which may not necessarily reflect the actual appreciation of the content shared. Integrating information with traditional surveys may help overcome some of the limitations related to social media, such as selection of content (not all that is experienced is also shared on social media), sampling biases (users providing information are self‐selected), geotagging inaccuracy of posts (Crampton et al. ), or restriction of use (not all data are publicly available) (Tufekci ). Future studies could also evaluate how expectations and preconceived images of a destination potentially affected our results (Kim & Stepchenkova ; Lo & McKercher ). Novel methodologies, such as deep learning algorithms (Taigman et al. ), may provide future solutions how to analyze the content and profile of the users more efficiently (Di Minin et al. ).
The use of social media data can potentially have important implications in informing visitor and protected area management. As even best funded conservation authorities may lack resources (both human and financial) to carry out up‐to‐date surveys required to inform protected area management, our study highlights that social media data may provide a rapid and cost‐efficient alternative to surveys. Particularly, protected area managers may take advantage of social media data for real‐time understanding of the ecological and social processes underpinning protected area management. Compared to snapshot‐type visitor surveys, for example, continuous monitoring of social media would allow to better understand spatio‐temporal changes in visitor preferences, cultural services (e.g., sense of place; Hausmann et al. ), as well as the profile of tourists visiting the area. Analysis of the continuous social media data feed would also allow identifying emerging activities or other spatial or temporal patterns, which cannot be captured by predefined surveys. The potential of social media is even broader for practical protected area management. Content analyses could be used as a dynamic data source to understand stakeholders’ (e.g., tourists and local people living nearby the protected area) sentiments (Hauthal & Burghardt ) toward management actions (e.g., culling of animals, renovation of infrastructures, and human–wildlife conflict) and to enhance adaptive management (Wells & McShane ). Moreover, protected area managers may use geotagged social media data to monitor threatened species (e.g., location and population dynamics) and threats to biodiversity (e.g., spatial occurrence of invasive alien species). Content shared on social media may also reveal real‐time management issues, such as traffic hotspots (e.g., most traveled roads) or species particularly exposed to human disturbance (e.g., breeding sites close to trails and roads), which could be addressed real‐time to minimize visitors’ impact on biodiversity. Finally, promoting the use of specific hashtags (e.g., #fire, #flooding, #trafficjam, #nesting, and #roadkill of particular species) may help cost‐effectively monitor specific management concerns, particularly in large, under resourced, protected areas.
Nature‐based tourism is growing worldwide (Balmford et al. ) along with the use of social media (Kaplan & Haenlein ). The availability of information available is likely to grow in the future along with its potential for use in conservation science (Di Minin et al. ). Our findings suggest that social media has the potential to be used in place of traditional surveys as a representative source of data to assess preferences for biodiversity and activities in protected areas. The same methods used here can be repeated to cost‐efficiently inform protected area management and marketing elsewhere in the world. Collaborations between conservation agencies and main social media platforms should also be promoted as a means to freely monitor social media users visiting protected areas, their preferences, and needs in order to develop real‐time solutions that can enhance visitors’ experience and protected area management.
A.H. was supported by the Amarula Trust funding to the Amarula Elephant Research Programme. A.M. thanks the ERC‐StG, Grant 260393 and the Academy of Finland Centre of Excellence Programme 2012–2017, Grant 250444. E.D.M thanks the Academy of Finland 2016–2019, Grant 296524, for support. H.T. thanks DENVI doctoral programme at University of Helsinki for support. All authors thank Kone Foundation for support. Research was carried out by permission of South African National Parks and Humanity & Social Sciences Ethics Committee at University of KwaZulu‐Natal.
© 2018. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.