Harnessing public perception of cheetah reintroduction in India to facilitate management through participatory decision-making

Abstract

Scientists, conservationists, and decision-makers are debating the restoration of the cheetah population in India, aiming to protect biodiversity and habitats, while several experts express concerns and question the benefits and drawbacks of the experiment. This study represents a pioneering effort to understand public perception of cheetah reintroduction in India, utilizing the tools of Natural Language Processing to evaluate widespread discourse from social media platforms (Twitter and Instagram) and online sources (Google and news media). Public attitudes and levels of awareness were examined by employing correspondence analysis, co-occurrence network mapping, multidimensional scaling, and sentiment analysis. The results indicate that the social media community appears enthusiastic about the initiative, with 91.8% of the dataset reflecting positive emotion, particularly clustering around terms like ‘wildlife conservation’ and ‘fastest animal’. In contrast, Google and news media groups displayed a slight skepticism, with 46.2% exhibiting cynical attitudes, highlighting concerns about the project’s outcomes. Public discussions were found to be concentrated in Rajasthan, Madhya Pradesh, and Uttar Pradesh. The study underscores the importance of integrating opinions and different perspectives into the conservation initiative. This approach will help wildlife managers understand both supportive and opposing responses, share important information with key stakeholders, and adapt effectively to achieve sustainable results in difficult situations.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-25548-6.

Introduction

The Indian Government’s plan to reintroduce the cheetah has sparked debate among scientists, bureaucrats, and decision-makers about the benefits and drawbacks of restoring the large carnivore, which has been extinct from the Indian landscape for over five decades¹. The cheetah was historically abundant across the Indian subcontinent, except on coastlines, high-altitude mountains, and northeastern plains. During the 16th − 18th century, Mughal emperors Akbar and Jahangir used the species to hunt gazelles and other antelopes. However, the continued removal of cheetahs from the wild led to an all-time low population by the early 18th century². The last known three individuals were hunted by Maharaja Ramanuj Pratap Singh Deo in 1947³. In 1952, the Indian government formally declared the species ‘extinct’ in India, calling for their protection and proposing a risky effort to save them. Since then, a large group of environmentalists and wildlife enthusiasts have pushed for the return of the species to its known homeland.

The natural history of cheetahs dates back around 10,000 years. The genetic data of the species have been documented to monitor their population genetics, genomic inheritance, genetic diversity, evolutionary history, and trade routes, emphasizing the importance of ecological stability and habitat protection in conservation efforts^4–10. It has been demonstrated that surviving Iranian cheetahs are the last representatives of the Asiatic subspecies (Acinonyx jubatus venaticus)¹¹. Recent studies by^12,13 revealed that genetic distinctions between East African cheetahs (Acinonyx jubatus raineyi) and Southern African cheetahs (Acinonyx jubatus jubatus) were substantial, with significant genetic variation between Asiatic cheetahs (Acinonyx jubatus venaticus) and all the African subspecies. Negotiations to import the Asiatic cheetah (Acinonyx jubatus venaticus) to India began in the 1970 s with the Shah administration of Iran. However, due to their limited numbers in Iran, the African counterparts (Acinonyx jubatus jubatus) were selected for the reintroduction program¹⁴. It was hypothesized that the ‘Cheetah Project’ would conserve biodiversity, including grassland species like Indian wolves (Canis lupus pallipes), caracals (Caracal caracal schmitzi), and Great Indian bustards (Ardeotis nigriceps), increase prey base, and improve habitat (open-dry forest or savanna ecosystems) rehabilitation programs. The project also aimed to enhance the livelihoods of the local communities through social security, ecotourism, and rural development¹⁴. Following the plan, teams of wildlife biologists, veterinarians, and government officials from both India and the source countries translocated eight cheetahs from Namibia on September 17, 2022, and twelve cheetahs from South Africa on February 18, 2023, to Kuno National Park (KNP) in Madhya Pradesh, as part of India’s efforts to revitalize and diversify its biodiversity wealth. While few conservationists and decision-makers are optimistic about this mission, some experts have raised concerns about the exercise, citing issues such as long-term viability, territorial displacement, breeding concerns, stray dog attacks, and potential conflicts with other large carnivores. According to some, India lacks a suitable prey base, habitat, and open space for wild African cheetahs (Acinonyx jubatus jubatus). The project was also criticized as an expensive initiative, as the funds are not being used to address local problems, such as relocating Asiatic lions (Panthera leo persica) from Gir to KNP, which was initially proposed for their second home. It was also claimed that the project lacks clarity regarding fund estimates and operational planning, stating the government should refocus its efforts and allocate resources efficiently to implement more realistic and sustainable conservation measures^15,16.

Conservation practices involve managing natural resources within an intricate human context¹⁷. Although reintroduction is a useful conservation strategy, it is rarely conducted in spaces totally devoid of people. Therefore, a plan to include people directly and/or indirectly affected by a reintroduction should be in place¹⁸. Thus, such ambitious projects require careful consideration of the local stakeholders in the long-term success of the project. These are the stakeholders who will gain from the exercise while also bearing its costs, both economic and ecological. Ignoring the local community’s concerns may lead to overlooked opportunities or project rejection, undermining its goals¹⁹. So, if the views are not addressed, it may result in backlash and pose a significant threat to conservation efforts. Public perception is, therefore, critical to project success as communities hold significant power in shaping the implementation of conservation policies¹⁸.

A more inclusive view of nature conservation should consider people’s perceptions, attitudes, and knowledge when formulating wildlife management plans. This approach ensures sustainable conservation plans, transparency, and accountability. Public awareness plays a crucial role in understanding people’s perspectives and human dimensions. Information from print and electronic media (e.g., books, newspapers, television, and radio) influences public feelings and attitudes toward wildlife conservation and management. The coverage and portrayal of how the news media present wildlife issues directly impact how the public feels about them²⁰. Nowadays, the widespread use of social media has significantly shaped people’s perceptions. More than two billion people use social media sites like Facebook, Twitter, Instagram, etc. on a global scale, accounting for 67% of internet users and 29% of the world’s total population²¹, and the number is continuously rising.

Considering this exponential growth of human interactions on social media platforms, mining textual data to extract meaningful insights can be a potent tool for critically interpreting public perceptions and patterns related to a particular topic. Natural Language Processing (NLP) is a subfield of artificial intelligence that uses computer programming to alter language and understand texts. It is primarily employed in companies where consumer perceptions of goods and services are assessed to improve quality. It is particularly useful in wildlife and conservation sciences, where public participation is crucial. Previous literature utilizing NLP in the field of wildlife sciences and conservation has focused on newspapers, reports, radio, and television^22–30. Studies using social media platforms like Twitter, WeChat, Facebook, Flickr, and Instagram were also conducted^26,31–36, which hold implications for wildlife conservation and policy building. Consequently, the goal of the present study was to use social media platforms like Instagram, Twitter, online news articles, and Google, and employ NLP algorithms to analyze sentiments and mine people’s opinions on the Cheetah Reintroduction Plan. It is believed that monitoring public opinion from social media would strengthen and advance the management of the human-wildlife interface and contribute to creating a coherent conservation framework.

Materials and methods

Content and sentiment analysis tools of NLP were used to determine the answer to our research question: “What are the attitudes and sentiments of the Indian public about the Cheetah Reintroduction Plan?”.

Data collection and processing

To collect data, choosing the best platform was crucial in mining the public’s opinions regarding the ‘Cheetah Reintroduction Plan’. Textual datasets were extracted from two well-known social media websites, Instagram and Twitter, including the Google search engine and online news sources, using the data miner add-on for Google Chrome (https://dataminer.io/services) on November 10, 2024. Materials in the English language were selected, and regional languages were deleted to ensure broader accessibility of data and uniform consistency of information. In data cleaning and pre-processing steps, the standard terms of articles, prepositions, and special symbols like hashtag (#), exclamation sign (!), at sign (@), etc., were removed. Spam, duplicates, names of people, irrelevant numerical values, common words, URLs, and common repetitive materials were filtered out. The datasets were also refined using lemmatization, and noise reduction to remove non-contextual materials and select materials only related to the cheetah reintroduction program for the analysis, and used to prepare the word cloud (Fig. 2a). The final datasets were processed into KH Coder Ver. 3, VOS viewer (version 1.6.15), and R (version 4.2) to analyze the terms. Our study uses a combination of text mining, correspondence analysis, multidimensional scaling, and co-occurrence network analysis to examine the Indian public discourse on the reintroduction exercise, as depicted in Fig. 1. Location coordinates from the extracted textual datasets were compiled in a global context with a primary focus on Indian perception, ensuring diverse perspectives and to comprehend the intensity of geographical distribution of public discussions from different regions (Fig. 2b and c).

Fig. 1.

Comprehensive Workflow for Analyzing Public Perceptions of Cheetah Reintroduction through Text Mining. This figure outlines the multifaceted approach undertaken in this manuscript, integrating diverse elements of text-mining investigations. The process commences with the acquisition of data from a plethora of sources, subsequently progressing through sequential phases of preprocessing, text mining, and multifarious data analysis methods, encompassing clustering, ordination, and sentiment analysis, to meticulously explore the varying perceptions surrounding cheetah reintroduction.

Fig. 2.

The figure elucidates the distribution of data, text preprocessing, and class differentiation among the designated research groups. A. illustrates a word cloud, wherein the dimension of each word is proportional to its frequency of occurrence. B. portrays a geographical representation of the global data acquired, utilizing a colour gradient ranging from red (indicating high frequency) to grey (signifying lowest frequency). (C) represents the distribution of data across India, with a corresponding colour code depicting data frequency. (D) scatter plot generated through correspondence analysis, enabling the identification of target research groups. In the scatter plot, individual bubbles denote the occurrence of each keyword, with the size of the bubble representing the frequency of the word. The map was generated using ArcGIS version 10.6. The species photograph was taken by Mr Subharanjan Sen and edited manually in Adobe Photoshop 7.0.

Text mining and correspondence analysis

Text mining and Correspondence Analysis (CA) were employed to study the ranking of high-frequency words and phrases. Descriptive statistics in the KH Coder estimate the term frequency (TF) distribution and the document frequency (DF) distribution. The TF command displays a frequency table showing the occurrences of the words for analysis. The DF command assembles the number of documents or cases for each word, allowing the units to identify a case in the data. CA in KH Coder is a graphical technique to represent qualitative data in a 2-dimensional scatter plot between two groups of variables for helpful interpretation. CA compresses the textual data information into smaller components to inspect the relationship between the variables and understand the similarities and dissimilarities in the terms. It deals with term frequency by using chi-square distances to calculate the association between the frequencies in each cell, exploring identical appearance patterns³⁷. In CA, factor scores and are calculated from a frequency distribution to maximize the correlation coefficient which is written as follows³⁸,

1

For our present analysis, CA paved the way for identifying research groups from the dataset. Consequently, two target groups were created to explore the goal.

Co-occurrence network and multidimensional scaling analysis

A Co-occurrence Network (CON) map was prepared in KH Coder and VOS viewer with high-frequency co-occurring words connected through edges. CON graphs contain three key elements: nodes, edges, and sub-graphs. Nodes represent the extracted words from the data; edges show how the nodes relate at the two ends of the edge; and sub-graphs depict a group of nodes strongly connected by edges³⁹. In the social network, the degree of a node reflects the importance of the node’s status in the whole network⁴⁰. The higher the degree of a node, the higher its priority. The degree of a node is defined as:

2

where represents the degree of the -node, indicating the number of nodes directly connected to other nodes. The connection between the two terms shows their co-occurrence relationship. A higher co-occurrence frequency represents a stronger correlation between words and a greater edge weight between nodes⁴¹. The strength of a node in a network is given as:

3

represents the neighbouring nodes of the node , and represents the comprehensive characteristics of the node degrees and ⁴². In a network, the clustering coefficient of words suggests the probability that words will become neighbours and describes the network’s closeness. A higher clustering coefficient implies greater closeness between the adjacent nodes, reflecting the public’s discourse regarding the matter⁴¹. The extracted terms are drawn in 2-D bubbles in a network plot based on the Fruchterman-Reingold layout algorithm⁴³, indicating the relative frequency of the terms. The thickness of the connecting edges suggests associative strength between the terms, measured as the Jaccard coefficient. We chose betweenness centrality while determining sub-graphs, as it gives information about any node or edge’s significance for depicting network connectivity.

A Multidimensional Scaling (MDS) plot was also conceptualized in the 2 and 3 dimensions of the extracted words to study the combinations of texts with similar patterns in both research groups. The similarities among the word pairs are given as distances, measured using the Jaccard coefficient, as it emphasizes whether specific words co-occur. The output displays word clusters employing Kruskal’s algorithm, which initiates the networking with the smallest edge and subsequently adds the next smallest edge without creating a loop. MDS helps to create visualizations of effective words based on similarity ratings, defined by their relative frequency. Words concentrated together, and coloured by groups, reveal important themes within the data⁴³. Thus, CON and MDS maps offer powerful visualizations of crucial contents across large datasets, showing relative relationships between texts or topics.

Sentiment analysis

Sentiment Analysis (SA) is a computational treatment of opinions and sentiments. Sentiment classification techniques can be roughly divided into machine learning, lexicon-based, and hybrid approaches in binary mode. The classification levels include document-level (which classifies an opinion document expressed as positive or negative sentiment, considering the whole document as a basic information unit), sentence-level (which classifies sentiment expressed in each sentence by identifying it as subjective or objective), and aspect-level (which classifies sentiment concerning specific aspects of entities)⁴⁴. R program software (version 4.2) was employed to analyze the opinions by assigning scores to each input file text for the target groups using the syuzhet package (https://github.com/mjockers/syuzhet). Texts were classified into three categories based on their emotional polarity scale score: positive, negative, and neutral. So, when SA was performed on a text, each relevant term received a score based on its proximity to a positive or negative word in the library. Different emotions like ‘anger’, ‘joy’, ‘sadness’, ‘anticipation’, ‘fear’, ‘trust’, etc., and the sentiment spectrum from positive to negative were scored and plotted in a graph for visualization and better understanding.

Results

The discourse analysis was finally initiated with 955 tweets from Twitter and Instagram and 966 Google and news media pieces.

Correspondence analysis

The output of the correspondence analysis shown in Fig. 2d reveals the relationship between the qualitative data. Individual clustering and segregation of qualitative data from both research groups were observed, which shows different perceptions between them. The distribution of words was relatively concentrated; most were close to their origin. Google and news media, and the social media group are held closely at their origin, indicating a strong connection within themselves but not far enough to describe an explicit disassociation between them. The keywords shown in the perceptual map result from the frequency of occurrence, created with the top 150 distinctive words filtered by a chi-square value of 70 in the KH Coder. Recurrent words in the preface of Google and News Media are ‘controversial’, ‘reintroduction’, ‘challenge’, ‘extinct’, ‘kuno’, ‘concern’, etc. The recurrent words in the preface of Twitter and Instagram are ‘beauty’, ‘congratulation’, ‘majestic’, ‘ecosystem’, ‘forest’, ‘historic’, ‘welcome’, and ‘speed’.

Co-occurrence network analysis

The distribution of public discussions regarding the ‘Cheetah Reintroduction Plan’ in a 2D space has been depicted in Figs. 3a and 4a, where the connection between the two terms shows their co-occurrence relationship.

Fig. 3.

Representing the clustering Co-occurrence Network for Google and News Media research group. (A) Co-occurrence Network map was devised from subgraph centrality and clustering coefficients of keywords for the Google and News Media research group in KH Coder. Bubbles indicate the frequency of each keyword in the plot. (B) VOS visualization network map of the Google and News Media research group was fashioned from 41 most relevant terms and 10 clusters with a total of 209 links and a total link strength of 1266. (C) Multidimensional scaling plot for the Google and News Media research group in 2D cartesian space constructed in KH Coder with six lead concept clusters. Bubbles indicate the frequency of each keyword in the plot.

Fig. 4.

Representing the clustering Co-occurrence Network for Twitter and Instagram research group. (A) Co-occurrence Network map was devised from subgraph centrality and clustering coefficients of keywords for the Twitter and Instagram research group in KH Coder. Bubbles indicate the frequency of each keyword in the plot. (B) VOS visualization network map of the Twitter and Instagram research group was created from 73 most relevant terms and 6 clusters with a total of 1444 links and a total link strength of 11,544. (C) Multidimensional scaling plot for the Twitter and Instagram research group in 2D cartesian space constructed in KH Coder with six major concept clusters. Bubbles indicate the frequency of each term in the plot.

Google and news media research group

The TF and DF distributions in the descriptive statistics command of the KH Coder generated 965 words with a mean of 5.04 and 5.01, respectively (Supplementary Fig. S1, S2, S3) to analyze the CON and MDS maps. Figure 3a represents a network map of high-frequency terms based on subgraph centrality for the Google and News Media research group. Several positive and negative terms were fused together. ‘cheetah’, ‘reintroduction’, ‘madhya’, ‘pradesh’ ‘kuno’, ‘africa’, ‘extinction’, and ‘wildlife’ were the primary words used in the map with maximum centrality. Word clusters with strong connections were denoted with a darker edge, such as ‘pm’, ‘release’, ‘big’, ‘cat’, ‘madhya’, ‘pradesh’, ‘gandhi’, and ‘ready’, with one of the highest clustering coefficients. The arrangements in the matrix imply that the news media have covered the entire chain of events regarding the relocation exercise, starting from their origin and history in Namibia to their extinction in India, to their new home in KNP, Madhya Pradesh, the historic reintroduction program, people witnessing their first sighting a decade after implementation, to habitat preparation of a new home in Gandhi Sagar Wildlife Sanctuary. Some prominent clustering coefficients for pairs included ‘madhya-pradesh’, ‘big-cat’, ‘pm-release’, and ‘return-controversial’ were 0.91, 0.58, 0.44, and 0.25, respectively.

Positive keywords such as ‘wildlife’, ‘habitat’, ‘reintroduction’, ‘conservation’, ‘species’, ‘welcome’, and ‘effort’ are present in higher numbers than negative terms such as ‘death’, ‘controversial’, and ‘extinction’. ‘cheetah’ (f = 918) and ‘reintroduction’ (f = 291) are the most frequently occurring terms used in conjunction with ‘bring’, ‘expert’, ‘reintroduction’, ‘kuno’, and ‘africa’. The presence of words like ‘tiger’, ‘wildlife’, ‘species’, ‘animal’, ‘habitat’, ‘cub’, ‘conservation’, and ‘death’ in the network matrix represents continuous reporting of expert discussions or concerns regarding the ecological impact of this reintroduction program on other species and the entire biological system. ‘plan’ and ‘return’ are one of the most significant terms, interlinked with ‘reintroduction’ and ‘controversial’, which explains the divided views of scientists regarding the reintroduction program. Intricate networking design is being witnessed among terms like ‘expert’, ‘kuno’, ‘home’, ‘wildlife’, ‘birth’, ‘home’, ‘effort’, and ‘conservation’, expressing robust interconnections. The word ‘death’ (f = 41) appears several times, indicating that the reported deaths of the cubs in the enclosure had been a prominent topic of concern.

The VOS clustering has highlighted the characteristic elements to ensure clarity of the most important themes regarding the subject matter (Fig. 3b, Supplementary Table S1). The primary terms on the map were ‘india’, ‘cheetah reintroduction’, and ‘namibia’. ‘cheetah’, ‘extinction’, ‘decade’, ‘new home’, and ‘national park’ were clustered together. The word groups ‘cheetah reintroduction’, ‘south africa’, ‘new home’, ‘return’, ‘extinction’, ‘namibia’, ‘india’, ‘history’, ‘cheetah return’, and ‘decade’ were interconnected with mostly positive connotations.

Twitter and Instagram research group

Similarly, CON and MDS plots were created using 2693 terms with a TF distribution mean of 3.93 and a DF distribution mean of 3.58 for the bigger Twitter and Instagram research group (Supplementary Fig. S4, S5, S6). Various positive and negative words were interconnected with one another in the network map based on subgraph centrality (Fig. 4a). Primary terms with maximum centrality include ‘cheetah’, ‘namibia’, ‘home’, ‘madhya’, ‘pradesh’, ‘kuno’, ‘conservation’, ‘extinction’, ‘species’, ‘world’, and ‘reintroduction’. Word clusters with robust connections were shown with a darker edge with the highest clustering coefficients, such as ‘quarantine’, ‘enclosure’, ‘big’, ‘cat’ ‘beauty’, ‘visit’, ‘plan’, ‘graceful’, ‘welcome’, ‘madhya’, ‘pradesh’, ‘flight’, ‘majestic’, and ‘animal’. Some of the highest clustering coefficients included pairs between ‘enclosure-quarantine’, ‘fastest-animal’, ‘special-flight’, ‘graceful-visit’, and ‘wildlife-conservation’ were 0.50, 0.44, 0.43, 0.40, and 0.34, respectively. The combinations of these terms show that people have been updating their media profiles regarding such extensive discourses.

These associations convey the degree of awareness among the people of how the newly released cheetahs will be kept under supervision by the authorities and portray their enthusiastic reactions to the species’ arrival in Indian territory. Positive keywords dominate the matrix, with ‘conservation’, ‘environment’, ‘nature’, ‘fastest’, ‘happy’, ‘special’, ‘welcome’, ‘home’, and ‘historic’ among others. Only a lesser number of negative terms are present, which include ‘extinction’, and ‘hunt’. ‘cheetah’, ‘namibia’, ‘conservation’, ‘home’, and ‘reintroduction’ are some of the frequently occurring terms coupled with ‘madhya pradesh’, ‘kuno’, ‘project’, ‘wildlife’, ‘habitat’, ‘nature’, ‘majestic’, ‘area’ respectively. The names of additional wildlife species, such as ‘leopard’, ‘asiatic lion’, and ‘tiger’, were visible on the network map, indicating information about lion translocation in Kuno and the presence of other predatory animals in the area. A very complex and intricate networking design was visible amongst ‘forest’, ‘welcome’, ‘today’, ‘land’, ‘beauty’, ‘graceful’, ‘plan’, ‘visit’, ‘madhya’, and ‘pradesh’, suggesting a high association between them. The goals of the reintroduction project were also well known and can be comprehended by pairing connections between ‘wildlife’, conservation’, ‘extinction’, ‘species’, ‘effort’, ‘habitat’, and ‘environment’.

The VOS viewer enables the visualization of the most pertinent themes with the majority of positive words, like ‘graceful cheetah’, ‘new home’, ‘environment’, ‘fastest land animal’, ‘return’, ‘history’, ‘effort’, ‘conservation’ and ‘welcome cheetah’ for the Twitter and Instagram group (Fig. 4b, Supplementary Table S2). Fewer negative words like ‘extinction’ and ‘hunt’ were also listed. Some of the high-frequency terms in the VOS map include ‘cheetah’ (f = 927), ‘india’ (f = 307), ‘national park’ (f = 201), and ‘namibia’ (f = 154).

Multidimensional scaling

The distribution of public communications regarding the ‘Cheetah Reintroduction Plan’ in a 2D (Figs. 3c and 4c) and 3D (Supplementary Fig. S7, Fig. S8) space has been investigated using the classic metric MDS algorithm.

Google and news media research group

The MDS plots (Fig. 3c, Supplementary Fig. S7) enabled us to identify key concept clusters of major recurring words, each with a different colour code and frequencies ranging from 250 to 750. Cluster 01 (teal, central) discusses the main theme of the cheetah reintroduction project with the most significant terms such as ‘project’, ‘africa’, ‘cheetah’, ‘reintroduction’, ‘return’, and ‘animal’. Cluster 02 (yellow, right) indicates the relocation challenges, with terms such as ‘relocation’, ‘cub’, ‘home’, ‘gandhi’, and ‘challenge’ signifying concerns related to the adaptation of translocated species. Cluster 03 (purple, top-left) highlights the idea of conservation by emphasizing terms like ‘conservation’, ‘controversial’, ‘extinction’, ‘effort’, ‘fly’, and ‘forest’. Cluster 04 (pink, bottom-left), focusing on expert discussions reflecting ecosystem dynamics, comprehended from ‘wildlife’, ‘translocation’, and ‘habitat’. Cluster 05 (blue, top-right) is associated with media discussions understood from ‘birthday’, ‘welcome’, ‘pm’, ‘release’, ‘roam’, and ‘enclosure’, suggesting celebratory and publicized narratives. Finally, Cluster 06 (orange, left) documents all the past and current events of the exercise. These associations explain that news articles about such topics were regularly reported.

Twitter and Instagram research group

The MDS plots (Fig. 4c, Supplementary Fig. S8) describe the key clusters of the Twitter and Instagram research group, with occurrences ranging between 300 and 1200. Major concepts were characterized by Cluster 01 (teal, central), demonstrating discussions about cheetah reintroduction in KNP, indicated by terms like ‘kuno’, ‘cheetah’, ‘fastest’, ‘wildlife’, ‘forest’, ‘translocation’, ‘conservation’, ‘effort’, and ‘environment’, among others. Cluster 02 (yellow, right) displays the theme of the species’ adaptation and community interactions inferred from ‘roam’, ‘soil’, ‘adapt’, ‘roar’, ‘activity’, ‘long’, ‘visit’, ‘history’, and ‘mitra’. Cluster 03 (purple, left) focuses on the ecological significance and benefits upon the arrival of the species, denoted by ‘grassland’, ‘population’, ‘landscape’, and ‘area’. Cluster 04 (pink, bottom-left) emphasizes the cultural and symbolic importance of the species implied from ‘graceful’, ‘regal’, ‘wilderness’, and ‘majestic’. Cluster 05 (blue, bottom-right) showcases the policy and implementation highlights. Lastly, Cluster 06 (orange, center-top) captures public happiness of the celebratory and historic event surrounding the project, with terms like ‘community’, ‘congratulation’, ‘wish’, ‘wonderful’, ‘initiative’, and ‘event’.

Figure 5 shows the frequency bar graph of the most recurrent terms utilized in the CON and MDS plots of both research groups. Among many others, the combined frequency of the following terms is presented: ‘reintroduction’ (f = 364), ‘namibia’ (f = 237), ‘extinction’ (f = 159), ‘madhya’ (f = 186), ‘pradesh’ (f = 181), and ‘kuno’ (f = 346).

Fig. 5.

Graph illustrates the frequency of the most recurrent terms within the co-occurrence network map, distinguishing between the Google-News-Media group (depicted in blue) and the Twitter-Instagram group (depicted in pink).

Sentiment analysis

When the attitudes of the target groups were examined, it was found that both groups had fairly favourable attitudes regarding the reintroduction project. The Twitter and Instagram communities contribute the most positive emotions, accounting for 91.8% (Fig. 6, Supplementary Fig. S9, S10). The group appeared to be thrilled with the initiative based on their scores for trust (44.0%) and joy (37.0%). However, the group also displayed more depressive feelings (8.5%) compared to the other group. A sense of distress and anguish was also noticed, as they scored higher levels of anger (11.3%), fear (22.4%), and anticipation (53.0%) sensations. Experts have expressed concerns about the reintroduction plan in interviews that have been covered and reported by the mainstream news media as stories; nevertheless, how those interviews were presented contributed to a false impression by giving the substance a negative tone. A feeling of anticipation (30.2%) was also computed for Google and news media, indicating apprehension over the situation (Fig. 6, Supplementary Fig. S9, S10).

Fig. 6.

Percent sentiment scores of target research groups depicted through a bar graph. Turquoise-coloured bars represent sentiment scores for Google and News Media, and Pink-coloured bars for Twitter and Instagram. Google and News Media have the highest counts of anticipation, positive, and negative feelings. Twitter and Instagram have the highest counts of positive, anticipation, and trust feelings.

Discussion

The rapid development of social media platforms has significantly impacted the acquisition of information about wildlife and assessing people’s perceptions. Social websites like Instagram and Twitter have become popular among smartphone users, attracting more readers and spreading awareness about biodiversity conservation and environmental protection⁴⁵. The current study was aimed at examining people’s perspectives and awareness towards a topic of biological conservation conducted on an intercontinental basis, focusing on the vulnerable cheetah (Acinonyx jubatus jubatus). The news of the species’ translocation in India sparked public attention, and people viewed it as an emotion-evoking issue, as the species had gone extinct due to anthropogenic factors, thereby amplifying public engagement with the subject.

Our study demonstrated numerous discussions worldwide on the cheetah reintroduction plan, particularly among Indian citizens. The discourse was initiated in 2009 when experts from representative countries gathered to discuss the matter¹. The discussions accelerated when efforts were made to bring the species back to India to build the founder stock population and promote grassland ecosystem conservation. The word arrangements in both research groups displayed how people were aware of the entire sequence of events, as depicted in Fig. 5. Google and the news media group discussed the African species, their Namibian habitat, their new home in Kuno, extinction reasons, their adaptable capabilities in a changing environment, current conservation efforts, translocation effects, the mortality of new-born cubs, probable ecological impact, and debate over prioritizing other wildlife conservation projects. Twitter and Instagram groups revealed people across India have celebrated the successful reintroduction of the species and welcomed its return. Discussions regarding transportation, releasing them into predator-proof enclosures, new environment, its characteristics, hunting techniques, etc., also took place. Some critics cited historical incidents of hunting, poaching, and gameplay of the species causing extinction, and hinted at current potential loss due to inadequate protection measures. Experts have also raised concerns over the lack of suitable habitat, limited prey density, carrying capacity, demographic management, adaptability, human-wildlife conflict, disease risks, species’ spatial ecology, and abrupt fatalities^15,46,47.

Our study is a pioneering effort to examine the public discourse of the exercise, which offers unique insights into societal attitudes toward this conservation initiative. The results support the assertion of⁴⁸ that the Indian public’s enthusiasm has greatly aided the success of the program, understood from the word clusters in the network maps (Fig. 4a and b) expressing joy, pride, celebrations, and displaying the strongest positive emotions (Fig. 6). While species translocation has been explored in various global contexts, such as wolves (Canis lupus) in the Greater Yellowstone Ecosystem and Banff National Park⁴⁹, Iberian lynx (Lynx pardinus) in Spain and Portugal⁵⁰, European bison (Bison bonasus) in several European countries⁵¹, Oryx dammah in Tunisia⁵², and Oryx leucoryx in Oman⁵³, this study is the first of its kind to assess how the Indian public perceives such efforts. We found that Rajasthan, Madhya Pradesh, Uttar Pradesh, and Maharashtra had the highest geographical intensity of conversations (Fig. 2c), with international discourse from North America, the United Kingdom, South Africa, and Namibia (Fig. 2b). Support from local communities and positive public sentiments significantly enhance the feasibility of translocation projects^17,18,54. In line with previous research, our results reinforce the concept that, under circumstances where public acceptance is high, translocation can be an effective conservation strategy. While challenges remain, a favourable perception towards the program supports the broader notion of sustainable management of translocation plans. A key aspect of our study is the usage of NLP for capturing and interpreting large-scale, widespread, and real-time public emotions, ensuring an unbiased, representative, and scalable method for understanding perceptions. Similar approaches have been utilized in other wildlife conservation studies, such as for the Indo-Pacific humpback dolphin (Sousa chinensis)³³, human-African wild dog (Lycaon pictus) interactions³⁴, gray wolves (Canis lupus), coyotes (Canis latrans), opossums (Didelphis virginiana), and raccoons (Procyon lotor)^26,36.

Science, practice, and policy will shape conservation targets in the future⁵⁵. A more inclusive representation of views is required for a more inclusive approach to conservation, with scientific values grounded in pragmatism and understanding⁵⁵, justifying nature’s protection and restoration needs. Cooperation by sharing skills and knowledge, community feedback through stakeholder meetings, concern-addressing mechanisms, citizen consultations, transparent communication systems, community incentives, and monitoring shifts in public attitudes are essential for the project’s long-term success, as these enhance local support for conservation efforts. India has conducted successful species relocation programs in the country itself, such as tigers (Panthera tigris) in Sariska⁵⁶, and Panna Tiger Reserve⁵⁷, gaurs (Bos gaurus) in Bandhavgarh⁵⁸, and Sanjay-Dhubri National Park⁵⁹. The Cheetah Reintroduction in India is one such transcontinental project that gave voice to various experts with a common goal of species protection and stabilizing natural ecosystems. By considering data from different aspects, wildlife managers can make informed decisions, even in challenging circumstances⁶⁰. Through the present analysis, we highlight the urgent need to include diverse observations to generate new solutions, supporting the national conservation ethic, rather than engaging in conflict. The project should be given an adequate opportunity to monitor and adapt; action plans should be robust and effective to prevent additional fatalities. The best scientific, collaborative, and evidence-based conservation management practices should be applied in future decisions involving health monitoring, habitat protection, wild prey, population connectivity, socio-spatial tactics, cheetah-human interactions, and translating public opinion into actionable policy decisions. The outcomes of the project will serve as a case study for future translocation initiatives worldwide, and the lessons learned will guide prospective transcontinental introduction programs.

Conclusion

A broader perspective on nature conservation has emerged that considers people’s opinions, attitudes, and knowledge when creating plans for managing wildlife. The rapid expansion of social media and the internet provides enormous opportunities for boosting public awareness of wildlife conservation practices. The purpose of the current study was to evaluate mass perceptions and the degree of public awareness of the cheetah reintroduction plan using tools of NLP. We recommend investigating more public views on the ground and encouraging engagement in conservation efforts to reduce gaps among scientists, the public, and the decision-makers. Such studies will aid in the planning process and ensure that projects align with local values and priorities. The public support revealed through this study will assist in identifying further support and opposing viewpoints, monitoring, and adapting novel strategies for implementing participatory conservation actions based on evolving sentiments. The results will aid in formulating regulations and guidelines by wildlife management departments, minimizing the likelihood of conflict situations, enhancing public understanding of wildlife conservation, and enhancing environmental education on critical matters.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

D.C. and T.M. performed the conceptualization, visualization, data curation, formal analysis, validation, writing - original draft, and editing. D.C., T.M., and B.S.A. performed visualization, supervision, manuscript review, and editing.

Data availability

The text data utilized for the NLP analysis were sourced from open-access resources. The records of comment occurrences depicted on the network maps can be made available upon request to the corresponding author.

Declarations

Competing interests

The authors declare no competing interests.