Is YouTube promoting the exotic pet trade? Analysis of the global public perception of popular YouTube videos featuring threatened exotic animals.

The exploitation of threatened exotic species via social media challenges efforts to regulate the exotic pet trade and consequently threatens species conservation. To investigate how such content is perceived by the global community, mixed model sentiment analysis techniques were employed to explore variations in attitudes expressed through text and emoji usage in public comments associated with 346 popular YouTube® videos starring exotic wild cats or primates in 'free handling' situations. Negative interactions between wild cats and primates with other species were found to be associated with both text and emoji median sentiment reduction, however were still accompanied by a median emoji sentiment above zero. Additionally, although a negative trend in median text sentiment was observed in 2015 for primates, an otherwise consistent positive median text and emoji sentiment score through time across all IUCN Red List categories was revealed in response to both exotic wild cat and primate videos, further implying the societal normalisation and acceptance of exotic pets. These findings highlight the urgency for effective YouTube® policy changes and content management to promote public education and conservation awareness, whilst extinguishing false legitimisation and demand for the exotic pet trade.

Introduction

Unsustainable trade in wildlife is an important challenge to species biodiversity, where live animals are trafficked for pets amongst other purposes [1]. Exotic pets are animals without an extensive history of domestication or life in captivity that are not traditionally viewed as companion animals [2]. The exotic pet trade encompasses the global exploitation of exotic animals including popular mammalian species such as primates and carnivores (including Felidae) commonly sourced from wild populations, thus imposing a significant threat to species biodiversity, animal welfare and public health [2-4]. Although international wildlife trade regulations are in place through the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES), much of the trade still operates illegally [1, 3, 5]. Therefore, modern drivers and routes of the exotic pet trade must be identified to facilitate better regulation and preserve wild populations of these species.

Consumer interaction with the exotic pet trade has drastically changed in association with widespread access to the internet, wherein social media has been identified as an important driver of the trade through significantly influencing public perception [2]. Social media has changed how society access, consume and share information, wherein exposure to content is individually tailored to enhance user experience [6]. Social media platforms have become a dominant source for news, entertainment and information, therefore understanding factors that generate appeal for these sites is important to enable content regulation [7, 8]. Currently, a considerable amount of published content remains unregulated due to poorly established policy guidelines and limited enforcement disproportionate to the volume of data uploaded [9]. Overall, social media platforms permit public accessibility to unregulated exotic animal content, influencing user perceptions and facilitating accessibility to the exotic pet trade like never before.

Investigating how social media platforms influence public perception is vital to understanding attitudes towards exotic animal content and recognise the impact these platforms have in promoting consumer demand and enabling the globalisation of the market. With more than two billion users worldwide as of 2020, YouTube® is the largest global online video website and is thus considered to be one of the most influential social networking platforms [10, 11]. YouTube® enables users to upload, create, share and engage in content featuring exotic animals, thus promoting global participation [6]. Individuals accessing unregulated videos on YouTube® commonly choose to express their opinions and approval or condemnation of content through written language and emoji usage within the associated comments section [12]. Therefore, these comments may indicate how the public perceives content depicting exotic animals in a variety of ‘free handling’ situations, defined as having a relationship or interaction with humans, and further suggest how this perception is altered throughout the life of a video. Sentiment analysis, being the study of opinions, attitudes and emotions conveyed via text or emoji usage, has been used to evaluate positive or negative intentions associated with viewer comments [7, 13, 14]. It has been suggested that YouTube® is used to influence public perception surrounding ‘cute’ exotic animals and thus promote demand for exotic pets, however the significance of this platform as a driver of the exotic pet trade has not been fully investigated [15]. Therefore, sentiment analysis techniques applied to YouTube® viewer comments could provide an indication of how such content is perceived by the public and highlight areas requiring enhanced regulation to preclude the normalisation of exotic pets.

The aims of this study were to explore public perception of exotic wild cat and primate species in ‘free handling’ situations featured in popular YouTube® videos and investigate variations in perception associated with time, conservation status and interactions with other species through sentiment analysis techniques. Following this, the impact on the exotic pet trade and species conservation will be discussed where YouTube® policies encompassing animal usage will be assessed accordingly.

Materials and methods

Online data collection

Due to their popularity in the exotic pet trade, wild cat and primate species formed the basis of the video collection [2]. Common wild cat and primate species kept as exotic pets were manually researched, predominately through Google® and with reference to the International Union for Conservation of Nature (IUCN) Red List [16], and utilised as search terms within YouTube® (Fig 1). Popular terms which formed the basis of the video search included: ‘cheetah’, ‘tiger’, ‘lion’, ‘leopard’, ‘jaguar’, ‘monkey’, ‘capuchin monkey’, ‘marmoset monkey’, ‘primate’, ‘chimpanzee’, ‘slow loris’, ‘spider monkey’, ‘macaque’, ‘lemur’ and ‘gibbon’. Further variations in search terms included the addition of seemingly popular key phrases, including ‘pet’, ‘cute’ and ‘baby’. A full list of search terms investigated is located within the S1 Table. It is believed the majority of videos which feature this description were captured using these search terms.

Fig 1

Methodological framework for comment extraction and analysis.

Methodological framework for comment extraction and analysis across the 346 YouTube® videos published from 2006 to 2019, sourced in accordance with pre-determined selection criteria. Only comments extracted written in English, or containing terms identified by the sentiment software employed, were analysed for text sentiment. Likewise, only comments containing at least one emoji were analysed for emoji sentiment. Filters were applied to ensure comments analysed for emoji sentiment contained at least one emoji and had an associated positive or negative sentiment score.

Inclusion criteria involved the portrayal of an exotic animal in a ‘free handling’ setting, namely either within a captive (i.e. petting zoo) or domestic (i.e. pet) environment, interacting in some manner with either a human or another species. Videos were only considered if they had accumulated a minimum of one million views at the time of collection to ensure they had been received by a significant audience. Videos were excluded from the study if the associated comments section had been disabled. Videos published between May 2006 and October 2019 were included within the study.

Once selected, the following information was manually extracted and entered into a Microsoft Excel spreadsheet: Video web address (URL), search term used, species identified, IUCN Red List status, CITES Appendices listings, setting (captive vs pet) and species interactions present. Species were determined through analysis of the animal(s) featured in comparison to images and information sourced through online research (via Google®) and the IUCN Red List, unless specified within the video title or description. In regards to the primate videos, not all macaque and gibbon species were readily identifiable due to limitations imposed by video quality and duration. The identifiable species for both were classified as either ‘Vulnerable’ or ‘Endangered’, hence the genera were generally represented as being ’Endangered’ to reflect the most severe classification.

All unnatural interactions between the exotic animal featured and other species (including humans) were classified as positive, negative or neutral. Interactions involving elements of aggression or fear were categorised as negative, while situations wherein the animal and other species were seen next to each other but not directly interacting (i.e. sleeping next to a woman talking) were considered neutral. Interactions where people were cuddling exotic animals, treating them like kids (i.e. monkey dressed up like a baby drinking a milk bottle) or otherwise promoting anthropomorphism were categorised as being positive.

Data management and analysis was conducted in R v. 3.6.1 [17]. Additional datum was extracted through the tuber package [18] to fulfil the following categories: Video ID, comments, date published, date downloaded, view count, like count, dislike count and comment count. Code was designed to bypass the 100 comment limitation imposed by the tuber::get_comments() function to enable all comments to be extracted. All public primary (main comments within the comment section) and secondary (comments posted in response to primary comments) comments were extracted, including those posted by the account that shared the video. The data was arranged to ensure each row represented an individual comment for analysis using the tidytext package [19], accompanied by the language detected through the textcat package [20]. Video information and comments were extracted from May 2006 to October 2019. In accordance with the aims of the study, the comments produced over time for each video were evaluated for sentiment based on language (English only) and emoji usage. 1-grams were used due to the small number of words within most comments. Tokens were neither stemmed nor lemmitised as it was considered the sentiment software would cope with the common forms of most words.

A multiple-response approach was applied to analyse species frequencies, taking into consideration the number of instances a particular species was featured across all videos within their assigned animal group. For example, when selecting for videos using primate specific search terms some videos included multiple primate species, thus each primate species appearance was listed as an observation. Twelve primate videos were considered to be ‘compilation videos’ as they featured four or more videos compiled together. In these instances, species and IUCN Red List classification were not specified due to ambiguity.

Sentiment analysis techniques were employed to overcome limitations imposed by traditional manual qualitative techniques and enable more efficient analysis of a large dataset [21]. Lexicons derived from the tidytext package were applied to impart values to key words [19, 22]. Due to software limitations, only English comments were analysed for text sentiment. The “bing” lexicon qualitatively analysed terms, denoting words as either ‘negative’ or ‘positive’, and was thus used to analyse word frequencies. The “bing” lexicon was selected as the binary categorisation enabled the most popular positive and negative sentiment terms to be identified to evaluate word frequencies and cultural trends. Similarly, the “afinn” lexicon provided negative or positive values between -1 and +1 to particular pre-determined sentiment terms to quantitatively analyse variation in median text sentiment across comments. In association with this lexicon, negative values represent negative sentiment terms, zero relates to neutral terms and positive values correspond to positive terms. The “affin” lexicon was chosen as it fulfilled the aims of the study and was the most relevant option, being based on data derived from the social media platform Twitter [23]. Individual sentiment terms were assigned a quantitative value via the “afinn” lexicon to provide an average sentiment score between -1 and +1 for each viewer comment. These average sentiment scores for individual comments were collated to derive an average text sentiment score across all wild cat or primate comments for the year in which they were published. The ggplot2 package [24] was employed to plot variations in median text sentiment over time, in accordance with the year comments were published. Trends in median text sentiment were compared with emoji sentiment and changes in average trends over time were analysed.

Comments containing emojis were extracted and analysed within R [17]. The emoji sentiment lexicon utilised in this paper was based on work conducted by Novak et al. [14] and Peterka-Bonetta [25] as it best suited the aims of the study and provided a sentiment scale comparable to the “afinn” text lexicon used. The code provided by Novak et al. [14] was transferred directly into R and applied to the dataset without altering the sentiment data. The lexicon provided an average sentiment score for each emoji-containing comment as a value between -1 and +1, wherein negative values represent negative emoji usage, zero relates to neutral emojis and positive values correspond to positive emojis. Filters were applied to ensure comments analysed contained at least one emoji and a sentiment value not equal to zero. Changes in median sentiment score associated with emoji usage over time (based on year of comment publication) were analysed in relation to species and conservation status through the ggplot2 package [24]. Trends in median emoji sentiment scores through time were graphically analysed in conjunction with median text sentiment scores.

Mixed-effects models were developed to assess associations between sentiment and video characteristics. The lmer function which employed, which is a general linear mixed model. Therefore the commands used to run the mixed models were ‘lmer(emo_sent ~ interaction_type + (1 | youtube_ID), data = comments_sent_clean)’ and ‘lmer(text_sent ~ year + interaction_type + IUCN + (1 | youtube_ID), data = comments_sent_clean)’. Sentiment for each comment, either emoji or text, were set as response variable and predictors included species, year, interaction type and IUCN red list status. A random intercept for each video was included to handle the repeated measurement of the data. Best models were selected using AIC and forward selection.

Results

Frequency analysis

In total, 346 videos were compiled for analysis with the majority (n = 203, 58.7%) categorised as exotic wild cat videos, based on the prominent species depicted and search terms through which they were sourced. The most popular species recorded was the tiger (Panthera tigris), making an appearance in 48.3% of all wild cat videos (Table 1). With regards to the wild cat videos, tigers attracted the greatest number of comments, followed by lions (Panthera leo) then cheetahs (Acinonyx jubatus), closely reflecting species frequencies within the dataset (Table 1). Conversely, with regards to the primate videos, vervet monkeys (Chlorocebus pygerythrus) attracted the highest number of comments, followed by capuchin monkeys (Cebus apella) and macaques (Macaca spp.). The number of associated comments varied per video, as specified within the S2 Table. An interesting observation was the finding that 18 of the videos had one viewer that posted over 10% of the comments to the video with one viewer posting >2,500 comments (data not shown). The highest proportion of comments from one viewer was 46.5% (33/71). These frequent commenters can obviously influence the sentiment of the individual video. However, as they are only present in relatively few videos, they are unlikely to have affected the overall findings in this study.

Table 1

Species representation and associated conservation status.

The number of instances and percentage of corresponding YouTube® videos published between 2006 and 2019 associated with search terms explored, wherein key exotic wild cat (a, n = 203) and primate (b, n = 143) species were identified. A multiple response analysis was considered as some videos featured multiple species. The frequency describes the number of videos in which each species appeared, whereas the percentage represents the number of videos each species appeared in as a percentage of all wild cat or primate videos. Instances where primate or exotic wild cat videos featured more than one associated species have been included. Species with a frequency equal to or less than one have been excluded from the table.

Species	Frequency	Percentage (%)1	CITES2 Appendix	IUCN3 Red List Status
Wild cats4 (a)
Tiger (Panthera tigris)	98	48.3	I	Endangered
Lion (Panthera leo)	64	31.5	II	Vulnerable
Cheetah (Acinonyx jubatus)	53	26.1	I	Vulnerable
Jaguar (Panthera onca)	13	6.4	I	Near Threatened
Leopard (Panthera pardus)	13	6.4	I	Vulnerable
Caracal (Caracal caracal)	3	1.5	I	Least Concern
Primates5 (b)
Capuchin monkey (Cebus apella)	42	29.4	II	Least Concern
Macaque (Macaca spp.)	28	19.6	I	Endangered6
Vervet monkey (Chlorocebus pygerythrus)	8	5.6	II	Least Concern
Chimpanzee (Pan troglodytes)	7	4.9	I	Endangered
Gibbon (Nomascus spp.)	6	4.2	I	Critically Endangered6
Pygmy slow loris (Nycticebus pygmaeus)	7	4.9	I	Vulnerable
Orangutan (Pongo spp.)	6	4.2	I	Critically Endangered
White faced capuchin monkey (Cebus albifrons)	6	4.2	II	Least Concern
Green monkey (Chlorocebus sabaeus)	5	3.5	II	Least Concern
Spider monkey (Ateles spp.)	5	3.5	I	Endangered
Pygmy marmoset (Callithrix pygmaea)	4	2.8	II	Least Concern
Gorilla (Gorilla spp.)	3	2.1	I	Critically Endangered
Black and white ruffed lemur (Varecia variegata)	3	2.1	I	Critically Endangered
Squirrel monkey (Saimiri spp.)	2	1.4	I	Vulnerable

1Percentage of wild cat (a, n = 203) or primate (b, n = 131, excluding ‘compilation’ videos) videos which featured a particular species

2Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) Appendix entries (2019)

3International Union for Conservation of Nature (IUCN, 2019)

4Other wild cat species featured include serval (Leptailurus serval), puma (Puma concolor)

5Other primate species featured include baboon (Papio spp.), tarsier (Tarsius spp.), owl monkey (Aotus spp.)

6Not all videos with Macaca spp. and Nomascus spp. contained identifiable species. IUCN listing was based on the most severe status of identified species.

Human interactions with wild cats and primates were observed in 90.5% of all videos. The most common non-human species interaction depicted was with domestic dogs, as featured in 17.9% of all videos. The predominant language detected within comments was English (49.4%). Of the comments from which emojis were extracted, emoji usage varied between the animal groups. Overall, 8.6% of wild cat comments contained at least one emoji, compared with only 3.2% of primate comments. Species IUCN Red List conservation statuses ranged from ‘Least Concern’ to ‘Critically Endangered’. Species listed in CITES Appendices I and II were represented within the dataset, indicating imposed trade restrictions [5].

Sentiment analysis

The most popular sentiment terms identified in both wild cat and primate comments were ‘cute’, ‘like’ and ‘love’. ‘Like’ presented as the most frequently used term associated with wild cats, comprising 8.6% (n = 81,102) of all sentiment terms recognised, compared to ‘cute’ constituting 11.5% (n = 33,203) of all sentiment terms identified within primate comments. Videos of most species demonstrated a text sentiment above one (Fig 2). Primate videos were associated with the six highest median sentiment scores, closely followed by leopards (Panthera pardus) which reached the highest score for wild cats. In contrast to most primates, chimpanzees (Pan troglodytes) had a median text sentiment score close to zero and slow loris (Nycticebus pygmaeus) videos had a median sentiment score below zero.

Fig 2

Median text sentiment of 346 YouTube® videos published between 2006 and 2019 associated with search terms explored, wherein key exotic wild cat (a, n = 203) and primate (b, n = 143) species were identified.

Error bars represent first to third quartiles.

Text and emoji sentiment varied little between different IUCN Red List categories for exotic wild cats and remained positive for all categories despite median emoji sentiment being lower than median text sentiment. ‘Least Concern’ IUCN listed primates had the highest sentiment of all categories (Fig 3), while IUCN listed ‘Vulnerable’ primates had the lowest sentiment and was the only IUCN group approaching neutral. This was also the only category with a neutral text sentiment and where emoji sentiment was higher than text sentiment. Videos showing positive interactions between exotic wild cats, primates and other species had a median text sentiment above one and negative interaction videos had a sentiment below zero (Fig 4). All types of interactions had median emoji sentiment above zero with negative and neutral interactions having similar median sentiment, although both were lower than for videos with positive interactions.

Fig 3

Sentiment and IUCN Red List classification.

Median text and emoji sentiment comments score across 346 YouTube® videos of exotic wild cat published between 2012 and 2019 associated with species conservation status as listed on the IUCN Red List (2019). Emoji usage within the comments was only present from 2012. Twelve primate compilation videos featuring four or more compiled videos with various species were not provided with an IUCN Red List Classification due to ambiguity and as such have not been included. Median sentiment score ranged between -1 (negative) and +1 (positive), where 0 is neutral, in accordance with the software utilised. Error bars represent first to third quartiles.

Fig 4

Median text and emoji sentiment of 346 exotic wild cats and primates YouTube® videos published between 2006 and 2019 based on interaction between wild cats or primates with other species.

Error bars represent first to third quartiles.

Median sentiment scores varied between exotic wild cats and primates but they remained above zero through time for both groups, indicating a positive sentiment (Fig 5). For cats, sentiment decreased from 2006 until 2011 where it remained stable until 2019. Primate text sentiment varied until 2011 and decreased from 2010 to 2016 after which it increased to 2006 levels. There was a decrease in primate sentiment in 2015. Emoji sentiment appeared less varied but remained positive in the available period (2012 onwards, data not shown).

Fig 5

Median text and emoji sentiment over time of 346 exotic wild cats and primates YouTube® videos published between 2006 and 2019.

Error bars represent first to third quartiles.

Modelling of the text sentiment showed that average sentiment was 0.6 to 0.7 lower in 2011 to 2019 compared to 2006. Both positive and neutral interaction videos had 1.2 higher average sentiment than videos with negative interactions. Videos with species of IUCN red list category of ‘Least Concern’ had an average sentiment higher than videos with ‘Vulnerable’ to ‘Critically Endangered’ species (Table 2).

Table 2

Variables associated with text sentiment of 346 YouTube® videos of exotic cats and primates uploaded between 2006 and 2019.

Variable		Estimate	Std. error	p-value
Intercept		0.337	0.255
Year	2006	Reference	-	-
	2007	-.0179	0.184	0.187
	2008	-0.062	0.166	0.331
	2009	-0.218	0.164	0.709
	2010	-0.466	0.163	0.183
	2011	-0.681	0.163	0.004
	2012	-0.699	0.163	<0.001
	2013	-0.708	0.163	<0.001
	2014	-0.648	0.163	<0.001
	2015	-0.664	0.163	<0.001
	2016	-0.635	0.163	<0.001
	2017	-0.707	0.163	<0.001
	2018	-0.668	0.163	<0.001
	2019	-0.769	0.163	<0.001
Interaction type	Negative	Reference	-	-
	Neutral	1.177	0.159	<0.001
	Positive	1.198	0.156	<0.001
IUCN listing	Least Concern	Reference	-	-
	Vulnerable	-0.482	0.098	<0.001
	Near threatened	-0.589	0.207	0.027
	Endangered	-0.585	0.090	<0.001
	Critically Endangered	-0.431	0.157	0.006
Random effect		Variance	Std. Dev
	YouTube® video	0.328	0.573

Only English comments are included.

An association between interaction and average emoji sentiment was found with videos including negative interactions having an average sentiment of 0.2 less than videos with neutral or positive interactions (Table 3). Inclusion of any additional variables were not found to improve the model.

Table 3

Variables associated with emoji sentiment of 346 YouTube® videos of exotic cats and primates uploaded between 2006 and 2019.

Variable		Estimate	Std. error	p-value
Intercept		0.145	0.028
Interaction type	Negative	Ref
	Neutral	0.200	0.031	<0.001
	Positive	0.220	0.029	<0.001
Random effect		Variance	Std. Dev
	YouTube® video	0.010	0.101

Emojis were not available until 2012.

Discussion

This study explored viewer attitudes towards exotic species in free handling situations featured in popular YouTube® videos as an indication of public perception and how variation in perception developed in different contexts. Free handling has been defined as exotic animals featuring an unnatural relationship with humans, where direct human interaction was evident in over 90% of videos analysed whilst over 17% of videos also depicted interactions with domesticated dogs. These interactions could be normalising keeping exotic pets in captivity and increase demand for these species in the pet trade [26]. In this study, YouTube® comments serve as a gauge of public perception due to the platforms wide-reaching accessibility and popularity, where several studies have supported the impact social media platforms have on public opinion [27, 28].

The prevailing frequency of positive terms, especially ‘cute’, ‘like’ and ‘love’, within video comments in this study reflects the use of the ‘cute’ factor in the promotion of online content and as a universal marketing tool for exotic pets. ‘Cute’ animal videos have been an integral part of YouTube® since its inception in 2005, where even the first ever video posted featured exotic animals [11]. In accordance with prior research, the ‘cute’ factor enhances desirability and the likelihood of consumers acquiring an exotic pet [4, 29], similar to strategies used in the sale of companion animals or even commercial products, contributing to the normalisation of exotic pets [30]. However, the popularity of the term ‘like’ in this context has a dual purpose as it echoes the participatory culture fostered by YouTube®, wherein users are encouraged to ‘like, share and subscribe’ to content [31, 32]. As 1-gram tokens were employed, the lexicon used could not distinguish between the positive sentiment term and the alternative application of the word, therefore this may provide inaccuracies in the interpretation of user sentiment. Further limitations associated with the sentiment software were identified, including the inability to assess non-English words or to recognise word usage in alternate contexts, such as sarcasm [14, 33]. To overcome these limitations emoji usage was assessed, wherein the overwhelming majority of analysed comments (87.1% wild cats, 89.8% primates) supported positive global public perception. Emojis effectively represent a global language of non-verbal cues used to convey the true meaning and emotion of text on social media, mirroring traditional behavioural indicators which would typically be present in face-to-face communication. For example, popular ‘tongue’, ‘wink’ and ‘smirk’ emojis commonly assist in the universal expression and recognition of sarcasm and therefore amend inaccuracies in text sentiment analysis by acting as metadata [34-36]. Likewise, certain emojis can signify that the user identifies content as being ‘cute’, or to express affection or happiness [37]. Both text and emoji usage displayed a predominately positive viewer response to content as they remained above zero irrespective of species and time, suggesting positive attitudes towards these ‘cute’ exotic animals in free handling situations.

The interaction between exotic wild cats and primates with other species were associated with both text and emoji sentiment in the mixed model (Tables 2 and 3). Negative interactions videos had lower sentiment than videos including neutral or positive interactions, suggesting that viewers are concerned about the welfare of the animals, although it cannot be ruled out that it is based on concern for the human or other interacting species in the video. An example supporting the notion that reduced sentiment is due to concern for the exotic animal is the “Tickling is TORTURE” campaign launched in 2015 aimed to combat slow loris exploitation, which had an evident impact on the perception of exotic animals featured online [38]. The success of this strategy may even be reflected in the negative text sentiment trend associated with slow loris videos (Fig 2) and primate videos overall in 2015 (Fig 5), highlighting the positive impact of public education. However, following this decline sentiment scores began to increase from 2016, possibly suggesting the awareness encouraged by the campaign was either forgotten or not extrapolated to other exotic species. Fluctuating internet trends and ‘viral videos’ need to be closely monitored as they challenge the implementation of strategies to successfully oppose featured content [39]. However, these finding suggests that education of the public on welfare of exotic animals in captivity could be used to influence people’s perception of them as pets.

Videos with ‘Least Concern’ species demonstrated a higher median text sentiment than videos including species in the categories ‘Vulnerable’ to ‘Critically Endangered’ (Table 2), however IUCN listing was not found to be associated with emoji sentiment (Table 3). It should be noted that exchanging IUCN listing with species created very similar results (data not shown). However, the sentiment scores remained consistently positive across all species, despite IUCN Red List classification (Fig 5). Interestingly, the most viewed video included in the study displayed ‘Endangered’ macaques performing for entertainment purposes (S3 Table). Despite their conservation status and evident animal welfare implications, the associated volume of views and overall positive sentiment score suggested this content was positively received by viewers. This evident lack of understanding mirrors findings presented by Stazaker and Mackinnon [40], where public ignorance of species conservation status may translate into creating demand for these animals. It may be argued that portraying exotic animals in free handling situations, as seen in more than 90% of all videos, alludes to the supposed domestication of such animals and consequently endorses exotic pets [26, 41]. The concept of free handling blurs the line between privately and institutionally owned exotic animals as they all feature a relationship with humans in some capacity, thus they are treated in the same manner for the purposes of this paper. Free handling of exotic animals further negates the risk of important zoonoses such as rabies and therefore presents a significant public health risk [42]. Overall, emoji usage analysed within the study demonstrates viewer’s inability to recognise and acknowledge biodiversity and public health consequences associated with the portrayal of exotic animals in free handling situations, heightening the urgency for widespread public education.

The exploitation of wild animal populations for the exotic pet trade has significantly contributed to welfare issues in addition to loss of biodiversity [2, 29]. Due to the consistent positive public perception highlighted across all videos, it is suggested that the public are inadequately educated about the origins of featured animals and associated welfare consequences. This implies that previous programs established by conservation organisations to discourage the online exploitation of wild animals have been largely ineffective. However, in the context of the study it was not possible to accurately assess the welfare status of animals featured in all selected videos accurately as they were typically too short, of poor quality and/or did not adequately display the animal’s living conditions. Education is vital to ensure viewers can identify the inappropriate depiction of exotic animals and are encouraged to report such content under YouTube®’s current guidelines [11]. The positive reception of online content featuring exotic animals interacting closely with humans further enhances demand for the exotic pet trade, therefore effective measures must be taken to appropriately inform viewers about the consequences of their support of this content.

Accessibility to the sale of exotic animals has been revolutionised through the internet, where exotic animal content posted on social media has supported the normalisation, desirability and acceptability of exotic species as pets [30]. Compared to primates, wild cats were the most popular group represented within the dataset. Tigers appeared in 48.3% of all wild cat videos in a wide range of domestic and captive situations despite their ‘Endangered’ conservation status [16, 43], with their popularity on social media supported by Spee et al. [26]. Exotic cats are heavily traded due to their popularity as pets, allure in captivity and demand for their parts, where the internet is involved in facilitating the trade [44]. Although majority of the species featured in this study were listed under CITES Appendix I, wherein trade is only permitted under exceptional circumstances, many have been reported in the live trade [5]. Harrington [45] described popular wild cat and primate species traded and reported via CITES including lions, tigers, leopards, jaguars, servals, squirrel monkeys, marmosets, capuchins and tamarins, many of which were sourced from wild populations. These particular species were also featured in selected YouTube® videos (Table 1), further supporting the link between social media depiction and the exotic pet trade. Trade routes for the importation and export of live animals were identified to include South America, North America, Africa, Asia and Europe, demonstrating the vast reach of the live exotic animal trade, wherein this distribution was reflected in the video origins included within this study (S3 Table) [45]. Regardless of CITES protection status and trade restrictions, there is still an extensive illegal trade present to supply the significant demand for exotic pets, much of which remains unmonitored [29]. Understanding how social media influences accessibility, public perception and consumer demand is essential to enable implementation of effective strategies to monitor the modern trade and exploitation of threatened species.

Content posted on YouTube® is loosely regulated despite the provision of policies outlining expectations and limitations, wherein the site heavily relies on consumers to report (‘flag’) policy violations or illegal content [11]. In the case of animal abuse, reporting under the “violent or graphic content policies” guidelines (S1 Fig) is heavily reliant on an individuals’ ability to detect signs of distress or mistreatment, which may vary. As indicated by this research, the public cannot be relied upon to accurately identify the inappropriate use of exotic animals as much of the content featured was seemingly considered ‘acceptable’ by viewers despite underlying welfare and conservation concerns. The inaccurate interpretation of welfare is further aided by the short duration, poor quality and obscured setting of many videos, as previously discussed. YouTube®’s proprietary algorithm is designed to enhance user engagement by directing users to suggested videos based on their previous viewership, therefore implying that once a user has shown an interest in exotic animal videos they are more likely to be directed towards similar content. The website has been progressively redesigned to focus less on advertising the most popular videos to tailoring suggested content to the individual, however corporations can also bias recommended content [40]. As YouTube® itself plays a crucial role in enabling and encouraging public access to this content, they must also accept social responsibility for creating a culture wherein human engagement with threatened exotic wildlife has become acceptable.

Development or adaptation of artificial intelligence systems which can accurately identify threatened exotic species depicted in public social media content, such as the ‘Wildbook’ software [46], should be explored. This technology could be linked to an automated notification which appears with information pertaining to that particular species’ conservation status and the exotic animal trade before enabling the video to be viewed. The purpose of this system is to identify popular exotic animal usage for research purposes whilst providing credible information to the public, enabling viewers to make a more informed decision before accessing content or supporting the exotic pet trade, similar to Instagram®’s ‘Wildlife Alert System’ designed to combat the use of exotic animal photo props [47]. Alternatively, it is recommended YouTube® employ software to automatically detect key terms such as species names within video titles or descriptions and flag them for immediate review. In association with revised YouTube® policy guidelines prohibiting videos displaying interactions between humans and exotic wildlife, reviewed videos in violation of newly established policies should be promptly removed to discourage normalisation of these exotic animals as pets. Alternatively, the development of an advertisement campaign presenting information surrounding exotic animal exploitation through the pet trade and the importance of species conservation could be automatically applied before all exotic animal videos to discourage support from viewers. However, in relation to Moorhouse et al. [48], it may be beneficial to focus information around the potential zoonotic risks and legal consequences of exotic pet ownership in addition to the conservation implications to better engage viewers. Effective content monitoring can not only assist with gauging public perception to address the evident education gaps, but also help to identify new trends within the exotic pet industry and thus infer conservation and public health risks.

Conclusions

Analysis of text and emoji usage within comments revealed a predominantly positive global public perception in response to the exploitation of exotic wild cats and primates on YouTube®. Although text sentiment was higher for ‘Least Concern’ IUCN Red List species, the overall positive response generated irrespective of conservation status appeared to indicate a lack of public awareness. In response, implementation of targeted YouTube® policies, provision of information for users on animal welfare issues and conservation status as well as accurate content regulation is highly encouraged to cease the illusion of the normalisation of threatened exotic animals as pets and prevent false legitimisation of the exotic pet trade.

YouTube search terms.

Search terms investigated through the YouTube® search engine to source videos for analysis. Only search terms which produced new videos selected for analysis are listed.

(TIF)

Click here for additional data file.

Number of comments across all wild cat and primate videos.

The number and distribution of comments extracted between May 2006 and October 2019 across all wild cat and primate videos included within the study. The number of comments associated with each video and each identifiable species varied.

(TIF)

Click here for additional data file.

Popular videos within the study.

The top ten most popular videos included within the study, according to view count at the time of data extraction, listed with primary species featured and additional video specifications.

(PDF)

Click here for additional data file.

YouTube® policy guidelines.

Excerpts from YouTube®’s “violent or graphic content” policy guidelines available online. The policy sections featured are those pertaining to the use of animals within videos.

(TIF)

Click here for additional data file.

Exotic wild cat video URLs (n = 203).

(PDF)

Click here for additional data file.

7 Sep 2020

PONE-D-20-17651

Is YouTube® promoting the exotic pet trade?

Analysis of the global public perception of popular YouTube® videos featuring threatened exotic animals.

PLOS ONE

Dear Dr. Chaber,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Oct 22 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Jorge Ramón López-Olvera

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

Additional Editor Comments (if provided):

Dear Dr. Chaber,

I have now received the comments from three independent reviewers regarding your submission. While all three found your study interesting and worth publishing, they also found all that the manuscript needed major deep modifications before being considered fully suitable to be published in PLoS ONE.

I recommend you to address all the comments by the reviewers in order to gain acceptance of your manuscript.

Best luck when carrying out your review.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

Reviewer #3: No

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: No

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Overall this is a useful and interesting paper that with some revision should be published. At the moment, the results read as a random list. This is perhaps because overall the literature review is a bit scanty and the literature that was reviewed was not fully developed to create research questions or hypotheses. Thus the results, although using some novel analyses, feel like a descriptive list rather than answering a series of research questions contextualised from the published literature. Once research questions are presented, then the rest should follow. Some small comments.

1. it does not seem in the contents the authors analysed if people wanted the animals as a pet - with this content analysis missing, it thus not so relevant to have a discussion about the pet trade - just because people think the material is cute does not necessarily mean they want one.

2. on that same note, a lot of the animals are probably in a very poor welfare state - this does not seem to have been recorded, but this could be a more serious discussion point since if people are looking at poorly kept sickly animals, this may be more reflected by the comments than the conservation status (see

Clarke, T. A., Reuter, K. E., LaFleur, M., & Schaefer, M. S. (2019). A viral video and pet lemurs on Twitter. PloS one, 14(1), e0208577.

Nekaris, K. A. I., Musing, L., Vazquez, A. G., & Donati, G. (2015). Is tickling torture? Assessing welfare towards slow lorises (Nycticebus spp.) within Web 2.0 videos. Folia Primatologica, 86(6), 534-551.

3. the authors need to be more clear that emojis were not really available on mobile phones until the period stated (about 2012, but actually even later than that on most phones) - so their predictions and analysis should take this into account.

The IUCN Red List status is interesting; most macaques are NOT Endangered and the ones kept as pets are likely to have been (until about last week when the new listings were published) Least Concern. It would be useful if the species could be identified.

Some random (e.g. in the absence of research questions) results come up that are interesting - like human interaction - the authors could make a better case for why this is important and link it better to zoonosis at the end (c.f. Moorhouse, which they cite)

Grammar check needed - which vs that is used incorrectly; the text is largely passive; the term data should be plural -etc.

Reviewer #2: Overall, I think this is a really good paper, clearly written, and an interesting piece of work. Well done. I have a few minor comments that I think you need to address and a general point you may wish to consider. I would recommend this paper for publication with these minor comments addressed.

Minor comments:

Within the methodology:

We need to know your text mining process in more detail, including how you tokenised the text (did you go so far as to explore n-grams before decided on tokens?)

Were your tokens stemmed or lemmatised? Either is fine, but the reader needs to know. (My personal preference is always lemmatisation).

Do you have an estimate of the penetration of your search terms, e.g. was there more out there? I think a general statement of “We feel confident we captured the majority of the videos that feature this description” / “We are sure there could be many more out there that the search algorithms could not identify” is fine.

Within the results:

This pairs with the n-gram comment in above. Your results that ‘like’ is common in the comments, could this be the repetition of ‘like and subscribe’ (the Youtube mantra?). If you only observed 1-gram tokens, you need to discuss this and highlight that ‘like’ here serves a dual purpose and may not be indicative of sentiment.

My general point:

I felt the piece was missing a bit of YouTube culture discussion, particularly as you’ve focussed on high-view count videos. The interaction between the Algorithm, the engagement on the video (the ‘like and subscribe’ problem), and how influencers engage with the material is really interesting. This is all a complicated and perhaps more cultural than you are usually familiar with, but some discussion around this would really be the cherry on the icing for me.

I wasn’t able to spot your data and code files?

I believe that signing reviews leads to a more open and transparent review process. – Jill R D MacKay

Reviewer #3: I've written up my review in markdown format. So that it is appropriately formatted and not filled with markdown syntax I have instead opted to attach the review in html and pdf format to the authors. I hope that they find them useful.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Jill R D MacKay

Reviewer #3: Yes: Mason Fidino

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: D-20-17651_review.pdf

Click here for additional data file.

9 Nov 2020

Reviewer 1

Thank you for your comments and editing suggestions.

This research project was designed as a descriptive observational study and as such no

hypothesis or research questions were developed. We wanted to explore whether there was a

relationship between the variables, however as there is limited published research surrounding

how the public respond to captive wildlife featured in online videos it seemed unreasonable to

formulate a hypothesis as we did not know what to expect. Due to the gap in the literature, the

paper was predominately a descriptive analysis. Further studies have been planned where data

will be collected based on hypotheses.

1. In accordance with research conducted by Nekaris et al. add others, the ‘cute factor’ is

commonly used as a marketing tool for pets, both domestic and exotic, as it increases the

likelihood that a consumer will acquire a pet. The link between the ‘cute’ factor and the desire

for an animal has been well established, therefore we evaluated the use of ‘cute’ and additional

positive sentiment terms to suggest a higher likelihood that these people wanted the animals

shown as a pet, not necessarily to say that they would be able to attain one. These positive

terms were also associated with the keywords used to select videos. The exotic pet trade is

fuelled by such desire and therefore the authors considered it appropriate to suggest a link

between the depiction of ‘cute’ animals on YouTube® with increased demand for exotic pets.

We have included a more thorough description of this point in line 279-281. Although it is a

valid point, full analysis of the content itself, as suggested, was not part of the study and could

not be accurately assessed based on our data.

2. The assessment of animal welfare was initially considered in the early stages of the research.

However, upon analysing videos for the study, it was not possible to accurately assess the

welfare state consistently across all videos as they were generally too short and the content

disallowed the conditions to be assessed properly as certain aspects were not shown on screen.

Therefore, it was difficult to accurately comment scientifically on welfare status, as mentioned

in line 327-330.

3. Thank you for your comment. A statement has been added to the paper to enhance clarity

regarding emoji availability. In regards to the macaque conservation status, not all macaque

species were readily identifiable due to limitations imposed by video quality and duration. The

identifiable macaques were classified as either Vulnerable or Endangered on the IUCN Red List,

hence macaques in the study were generally represented as being 'Endangered' to reflect the

most severe classification, as explained within the methods (line 101-104).

4. In regards to the seemingly random results regarding human interaction, the authors have

edited the discussion to clarify why this was important. Human interaction, either directly seen

within the video or implied as the animals were captive, was an important component of the

study as we aimed to analyse videos showing animals involved in ‘free handling’ situations (see

the discussion for further information on this, line 311-315) and to stress the importance of

zoonoses (where the discussion was edited to highlight this, as suggested). Thank you for your

suggestion to clarify this point and to link the discussion back to the zoonosis risks, this has

been considered and hopefully explained more clearly in the revised discussion.

In regards to the grammar check, we thank you for bringing this to our attention. The paper was

edited carefully to try to overcome this.

Reviewer 2

Thank you for your time and comments.

In terms of how we tokenised the text, we decided to use 1-grams due to the smaller number of

words in most comments. This is now included in line 116-117.

We did not stem or lemmatise tokens, again due to the small number of words in each comment,

however we agree that lemmatisation is a good idea for future investigations. We considered that

the use of the sentiment lexicons would cope with the common forms of the words, as stated in line

117-118.

In regards to the penetration of search terms, we feel confident we captured the majority of the

videos that feature this description. We have added a statement within the paper (line 88-89) as

suggested.

Thank you for your insightful suggestion regarding the reflection of the dual usage of ‘like’. 1-gram

tokens were used and as such we found this comment helpful and have added some discussion

around YouTube culture to the paper (see line 281-286). We are hoping this provides the missing link

between the YouTube algorithm and user engagement.

Data and code files should now be uploaded with the revised paper as a single pdf.

Reviewer 3

1. Thank you for your insight into the social science aspect of the use of ‘perception’ vs ‘opinion’. In

this instance, we believe that the use of either term is effectively equivalent for the purposes of this

study.

2. We agree that further statistical analysis would be helpful in interpreting the results, however as

this was designed as a descriptive observational study we have decided not to go into this detail as it

was beyond the scope of the study. As such, the data collected was not associated with a hypothesis

but was collected to evaluate whether there was a potential connection between exotic animal

content and perception. We have further studies planned where data will be collected based on

hypotheses.

2.1 Thank you for your suggestion. The reviewer raises a valid point. In future, if we were to perform

further statistical analysis, we would take this into account. However, at this stage the study was

designed as a descriptive observational study. The aim was to analysis the relationship between

sentiment (as an indication of public perception) and species conservation status and explore the

association with the exotic pet trade. We wanted to avoid selection bias within the study. Being a

descriptive study, we wanted to observe this potential relationship to suggest that it may be an area

which could warrant further investigation. At this stage, the addition of the suggested model is

outside the scope of this paper.

2.2 We believe this comparison would be interesting, however the aim of the study was to evaluate

exotic animals in ‘free handling’ situations, wherein they featured some sort of interaction with

humans. Therefore, the way we view it, the interaction with humans was more important than the

setting, where we were aiming to imply that the setting was not necessarily as important in how

these animals were viewed in this context (i.e. animals seen in captivity could still be considered

desirable and promote demand for these animals as pets). We agree that this was not clearly

explained within the discussion and therefore have modified it to account for this (line 311-315).

3. Thank you for highlighting the ambiguity associated with this point. We have added some

additional description to explain why emoji analysis was important within this research (see line 288-

296). Firstly, we argue that it can be difficult to gauge emotion from words alone, where emojis act

as metadata. Additionally, emojis represent a global language, with the aim of avoiding cultural bias

and misinterpretations of words detected in non-English languages.

Introduction

1. We agree with the reviewer to make the link between social media and the exotic pet trade

stronger. We have revised the flow of the introduction, whilst ensuring the main points (as pointed

out by the reviewer) have been explicitly addressed.

2. We agree with the suggestion to make the link between YouTube and the popularity of animal

videos and have added a comment to address this.

3. In accordance with your comment, we have included further information as to why sentiment

analysis techniques were chosen and why it was relevant to the study (see lines 64-71).

Line 32: The text has been edited as per the suggestion.

Line 32-34: These changes have been applied.

Line 32-39: A statement has been added to summarise this paragraph.

Line 40-41: We have made changes to the flow of the paragraph, as requested.

Line 42: This line has been edited to avoid confusion.

Line 40-47: A closing point has been added to this paragraph to help strengthen the link between the

exotic pet trade and social media.

Line 51: We have changed ‘insufficiently regulated’ to ‘unregulated’ to avoid confusion.

Line 48-60: A new topic sentence which better addresses the main points of the paragraph has been

included.

Line 61-64: We have altered the layout of the aim accordingly.

Line 64-66: Upon review, we agree that this statement was misplaced. We have moved it into a

more appropriate position within the aim (line 35-36).

Methods

1. Further information regarding why the specific lexicons were chosen has now been included

within the methodology (see line 131-140 and line 155-158). After exploring various dictionaries, we

found that these lexicons provided the most appropriate information for our study.

2. Analysis was predominately based on evaluating sentiment trends generated and depicted

through gglot2. In accordance with this, we firstly noted no statistical difference between the IUCN

Red List Classifications in regards to sentiment (as shown in figure 2a and 2b) which suggested no

clear distinction between the categories (interpreted as lack of public awareness of the significance

of conservation status). In regards to changes in sentiment trends over time, we observed that

sentiment remained consistently positive throughout time and was significantly different from zero

based on the application of 95% confidence intervals (except for primate text sentiment in 2015,

figure 3b).

Line 90: This has now been rewritten.

Line 91: You are correct in that tuber::get_comments( ) typically only extracts 100 comments at a

time, however we designed code to bypass this, enabling the function to effectively be repeated to

extract all comments. A comment has been included within the paper to address this (see line 107-

109).

Line 109-111: Clarification regarding the sentiment parameters and lexicon selection has been

included.

Line 120 – 121: The emoji sentiment lexicon published by Novak et al. was chosen as it provided

suitable code to complement the requirements of the study and was novel work not detailed

elsewhere when the analysis was being carried out. The code itself was not altered but was merely

written into R. A statement has been included in the methodology to describe this.

Results

1. As stated previously, although it would have been interesting to analyse the different response to

captive animals compared to pets, the most important point is that regardless of the setting the

animals were depicted in the interaction with humans (‘free handling’). The idea behind this was

that we wanted to suggest that despite the living situation of the animal (whether it was in a zoo or

as a pet), public perception was similar and both contributed to increased demand for these animals.

This was perhaps not very clear; hence, some revisions have been made to make this point stronger.

2. A table has been added to the supporting information regarding the number of comments

extracted. If a user made multiple comments within the comments section, these were all included

(as all comments were extracted). This includes both primary comments (main comments made) and

secondary comments (comments made underneath primary comments). If the creator commented

within the comments section, these were also included. A description has been included within the

methods (line 109-111).

Table 1: The table legend has been updated to include further information, as requested.

Line 155-160: This information has been moved to the methods section.

Line 192 – 194: Thank you for your suggestion. However, upon analysing the boxplots, the results

were not statistically significant from zero (with all plots intersecting zero), therefore the standard

deviation was not included. The analysis was predominately based on the positive distribution of

data. One of the main findings from the box plots was also surrounding how the classification groups

overlap, suggesting the public either did not recognise the conservation classification associated

with the species shown or did not care. 95% confidence intervals were applied and have been

included within the supplementary code file.

Line 186: Thank you for highlighting this uncertainty. The methods section has been updated to

make the sampling method clearer. A simple random sample was taken from each wild cat species

using the set.seen(n) function in R to produce the 10,000 comments. Sub-sampling involved ensuring

these comments were refined to only those containing at least one emoji (n>0) to provide a more

meaningful analysis.

Line 186-189: This section has been revised.

Line 192-193: This section has also been revised.

Discussion

1. The additional headers within the discussion have been removed to help improve flow, as

recommended. Further editing has been undertaken to try and make the aims clearer to the reader.

2. Thank you for highlighting this. We have aimed to re-emphasise the data and findings within the

discussion to address this.

Line 225: As now stated within the discussion, ‘YouTube® comments serve as a gauge of public

perception due to its wide-reaching accessibility and popularity, where several studies have

demonstrated the impact social media platforms have on public opinion’ (line 271-273). Therefore,

we believe that for the purposes of this study the comments extracted represented public opinion

surrounding the depiction of exotic animals in YouTube videos.

Line 225-230: The beginning of the discussion has been revised to ensure better flow.

Line 233-234: Some of this information has been moved to the methods section as requested,

whereas the interpretation remains within the discussion.

233-241: Thank you for your comment. We have aimed to incorporate the limitations of the

sentiment software with the aims of our study, as suggested. In regards to the identification of

sarcasm, it was difficult to derive a subset of comments which demonstrated this given the nature of

our data, however studies have been referenced regarding which emojis are commonly used in the

expression of sarcasm as a generalised example (line 293-295).

Line 304-331: Thank you for your insight regarding the recommendations made, we were unaware

of projects such as Wildbrook and after some research have decided to include it within the

discussion. Additional simpler changes have also been suggested, however there are likely other

approaches which could be considered also.

Figures

Changes have been made to the figures, as suggested.

Submitted filename: Response to Reviewers.pdf

Click here for additional data file.

30 Dec 2020

PONE-D-20-17651R1

Is YouTube® promoting the exotic pet trade?

Analysis of the global public perception of popular YouTube® videos featuring threatened exotic animals.

PLOS ONE

Dear Dr. Chaber,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

==============================

Dear Dr. Chaber,

I would like to apologise for the delay in rendering a decision on the revised version of your submission PONE-D-20-17651R1-Is YouTube® promoting the exotic pet trade? Analysis of the global public perception of popular YouTube® videos featuring threatened exotic animals. Reaching an opinion on your manuscript has been challenging, since the two reviewers who assessed the revised version radically disagreed on their decision, one of them even recommending rejection. Consequently, I had then to search for a third opinion. As you will see below, major concerns remain preventing the publication of your manuscript. However, the reviewers and myself still feel than your research has potential to become an article worth being published in PLoS ONE. I therefore recommend you to undertake the revisions suggested by the reviewers and thoroughly revise your work to not only address the issues the reviewers have raised, but to substantially modify the study beyond the descriptive analysis, as suggested by the reviewers, including references and following examples such as Fidino et al. 2018 (Fidino, M., Herr, S. W., & Magle, S. B. 2018. Assessing online opinions of wildlife through social media. Human Dimensions of Wildlife, 23(5), 482-490).

If failing to produce such a relevant modification in your manuscript, the most likely outcome of a  second revised version would be rejection. I consequently encourage you to squeeze your database and take the most out of it through roust meaningful analyses following the reviewers' advice in order to gain final acceptance for publication in PLoS ONE.

Best luck when carrying out your review.

==============================

Please submit your revised manuscript by Feb 13 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Jorge Ramón López-Olvera

Academic Editor

PLOS ONE

Additional Editor Comments (if provided):

Dear Dr. Chaber,

I would like to apologise for the delay in rendering a decision on the revised version of your submission PONE-D-20-17651R1-Is YouTube® promoting the exotic pet trade? Analysis of the global public perception of popular YouTube® videos featuring threatened exotic animals. Reaching an opinion on your manuscript has been challenging, since the two reviewers who assessed the revised version radically disagreed on their decision, one of them even recommending rejection. Consequently, I had then to search for a third opinion. As you will see below, major concerns remain preventing the publication of your manuscript. However, the reviewers and myself still feel than your research has potential to become an article worth being published in PLoS ONE. I therefore recommend you to undertake the revisions suggested by the reviewers and thoroughly revise your work to not only address the issues the reviewers have raised, but to substantially modify the study beyond the descriptive analysis, as suggested by the reviewers, including references and following examples such as Fidino et al. 2018 (Fidino, M., Herr, S. W., & Magle, S. B. 2018. Assessing online opinions of wildlife through social media. Human Dimensions of Wildlife, 23(5), 482-490).

If failing to produce such a relevant modification in your manuscript, the most likely outcome of a second revised version would be rejection. I consequently encourage you to squeeze your database and take the most out of it through roust meaningful analyses following the reviewers' advice in order to gain final acceptance for publication in PLoS ONE.

Best luck when carrying out your review.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

Reviewer #3: (No Response)

Reviewer #4: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

Reviewer #3: No

Reviewer #4: Partly

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

Reviewer #3: No

Reviewer #4: No

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: Yes

Reviewer #3: (No Response)

Reviewer #4: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: You have addressed all my comments. I commend you on a very interesting paper and look forward to more - Jill R D MacKay

Reviewer #3: There were two main issues that I brought up with my previous review that the authors did not address.

First, the authors contend that perceptions and opinions are the same. This is wrong. Perceptions are what someone sees, how they interpret, what they smell, etc. Opinions are a view or judgement about a topic, and may not necessarily be based on knowledge of a topic. These words are not interchangeable, they mean different things in the literature and the authors have not convinced me, in this case, that the words are interchangeable.

Second, the authors state that they did not do any statistical analysis because they had no hypotheses to test given that this was a descriptive observational studies. This is an inadequate response. Observational studies, even if they do not have a set hypothesis (or hypotheses) very often require statistics. I believe that these data will require some statistical analysis, especially to control for the large-scale variability in the number of comments between videos. The current analysis, in this case, certainly functions as a "first step" in what the authors should have done for this paper. They plotted out some of the data. However, that is not the end point of this analysis. The unequal sample sizes among videos, combined with the fact that topics of the videos are different, no doubt has a strong influence on a mean response. Nevertheless, the authors have combined all the the data for a number of analyses (e.g., mean sentiment response), which essentially assumes that all content of videos is equivalent.

Even without set hypotheses, inferential statistics can still be extremely useful because they can provide a comparison of effect size among categories or features of a dataset that are more correlated to a given response variable. This would be incredibly helpful for future hypothesis generation. Using multi-model inference, for example, would be a great way to select for variables that explain more variability in the data than others (i.e., provide feature selection, make comparisons, etc.). As an example, even being able to statistically compare the number of comments made among videos of different species (line 185) would be better than simply making a statement about the given sample. In it's current form, I do not know if the results are essentially being overweighed by one or two videos or if their sample is representative of the overall population of videos online.

Reviewer #4: I think the topic of investigation is of broad interest and it would be important to publish these results. That said, more work is necessary, especially in the presentation of the results. Some of the issue raised by previous reviewers are still not fully addressed (statistical analysis, hypotheses), and I provide more suggestions.

- Lines 72-76. I understand here the critique of the reviewers. Even if you want to have a descriptive paper, you inevitably test hypotheses since you expect a difference in perception of exotic pets based on their IUCN status, and you also expect a trend with time. You can justify why in here.

- Lines 189-190. Table S2 does not really show a right-skewed distribution. There is no index of skewness reported. The median value is much lower than the mean, so the majority of videos have lower comments than the mean (so probably left-skewed), and there seems to be outliers on the right. But I would expect a histogram or a skewness index if you talk about skewness.

- Table 1. For some genera, you probably have more than one species (it is not clear if you did not identify them). It should be clear how you handled with this frequent situation as for some genera (e.g. Macaca) you can have IUCN classifications ranging from Least Concern to Critically Endangered. So having Macaca spp. and the relative IUCN Red List category Endangered is probably incorrect unless you considered only Endangered species. That is true for other genera as well as you do not present the species. In the table, there is also a typo (Varecia variegata not variegate). If you have the genus you should also add spp. if you consider more species of the same genus (and of course they should all be with the same IUCN status).

- Lines 210-211. I do not understand this result. You selected the videos with wild cats and primates, so not sure what are these results relative to dogs. Do you mean that wild cats and primates had interactions with dogs in 17.9% of the videos? Can we know more about which species?

- Line 216. Likewise is not appropriate here as IUCN Red List and CITES are not going on the same direction.

- Text sentiment analysis section. I would expect a word frequency analysis, that can be interesting and quick to add, and can show more about the content of the comments.

- Lines 229-231. This statement would require a statistical analysis. You also need to consider the other issue relative to IUCN Red List status, comment to table 1.

- Lines 246-247. Here you need statistical analyses to confirm your statements.

- Line 249. It seems neutral as it is close to 0 (unless it is statistically significant from 0).

- Line 270-271. Here you report the same phrase as in the results but you do not explain why it is important, you do not discuss this result, you just repeat it.

- Lines 279-280. I do not see this from your results.

- Line 299. You do not really discuss the implications of your results. You found that the perceptions were positive, but what does it mean? What does it entail? You should move here the paragraph that is now almost at the end of the discussion

- Line 301. You did not really test this

- Line 306. Did you recognize the species? As there are so many macaques with so many different statuses.

- Lines 359-374. I think this paragraph should go above as this is your main discussion on your findings.

- Line 369. But this is only relative to slow lorises, how can this impact your analysis as you have many species and only 7 videos of pygmy slow lorises.

- Line 420. IUCN Red List status ranged from Least Concern to Critically Endangered, so not all the species you considered are threatened. You should rephrase it.

- Figure 2a. The IUCN categories should be upper case. In the x axis label you say Mean Sentiment Score but you show box plots (i.e. median, quartiles, range, and outliers). I suggest changing the figures to have a better presentation, maybe use theme(bw), can try different themes in ggplot2.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: Yes: Jill R D MacKay

Reviewer #3: No

Reviewer #4: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

18 Feb 2021

Dear Reviewer,

Thank you very much for your valuable comments, we have substantially modified the manuscript following your suggestions and hope this improved version will meet your expectation.

Best regards,

ALC

The reviewer’ comments are addressed below the initial comments.

Reviewer #4: I think the topic of investigation is of broad interest and it would be important to publish these results. That said, more work is necessary, especially in the presentation of the results. Some of the issue raised by previous reviewers are still not fully addressed (statistical analysis, hypotheses), and I provide more suggestions.

=>In order to address these comments we have included mixed models to assess which variables were associated with text and emoji sentiments

- Lines 72-76. I understand here the critique of the reviewers. Even if you want to have a descriptive paper, you inevitably test hypotheses since you expect a difference in perception of exotic pets based on their IUCN status, and you also expect a trend with time. You can justify why in here.

=>We have included more analyses in the paper and have reworded objectives accordingly.

- Lines 189-190. Table S2 does not really show a right-skewed distribution. There is no index of skewness reported. The median value is much lower than the mean, so the majority of videos have lower comments than the mean (so probably left-skewed), and there seems to be outliers on the right. But I would expect a histogram or a skewness index if you talk about skewness.

=>This is correct, we have hence reworded this

- Table 1. For some genera, you probably have more than one species (it is not clear if you did not identify them). It should be clear how you handled with this frequent situation as for some genera (e.g. Macaca) you can have IUCN classifications ranging from Least Concern to Critically Endangered. So having Macaca spp. and the relative IUCN Red List category Endangered is probably incorrect unless you considered only Endangered species. That is true for other genera as well as you do not present the species. In the table, there is also a typo (Varecia variegata not variegate). If you have the genus you should also add spp. if you consider more species of the same genus (and of course they should all be with the same IUCN status).

=>We have clarified the IUCN classifications of this group and included this in the paper (and corrected the spelling mistake; thank you for noticing).But, you are correct, there were several Macaca species, for this situation, IUCN listing was based on most severe status of identified species

- Lines 210-211. I do not understand this result. You selected the videos with wild cats and primates, so not sure what are these results relative to dogs. Do you mean that wild cats and primates had interactions with dogs in 17.9% of the videos? Can we know more about which species?

=>The interpretation is correct, this has now been reworded – line 220 now

- Line 216. Likewise is not appropriate here as IUCN Red List and CITES are not going on the same direction.

=>This has been reworded to correct (line 225)

- Text sentiment analysis section. I would expect a word frequency analysis, that can be interesting and quick to add, and can show more about the content of the comments.

=>To address this we have included the mixed models for text and emoji sentiment

- Lines 229-231. This statement would require a statistical analysis. You also need to consider the other issue relative to IUCN Red List status, comment to table 1.

=>We have reworded this and changed the figure – now starting line 245

- Lines 246-247. Here you need statistical analyses to confirm your statements.

=>This has been reworded (line 276 start)

- Line 249. It seems neutral as it is close to 0 (unless it is statistically significant from 0).

=>This has been deleted and results presented based on the mixed models

- Line 270-271. Here you report the same phrase as in the results but you do not explain why it is important, you do not discuss this result, you just repeat it.

=>Added interpretation - Now line 305

- Lines 279-280. I do not see this from your results.

=>These results are included in line 230

- Line 299. You do not really discuss the implications of your results. You found that the perceptions were positive, but what does it mean? What does it entail? You should move here the paragraph that is now almost at the end of the discussion

=>We have moved the paragraph as suggested (line 345) and combined with discussion as per mixed method models

- Line 301. You did not really test this

=>This is true, we have now deleted this statement

- Line 306. Did you recognize the species? As there are so many macaques with so many different statuses.

=>The results are based on the identified species (as per mentioned above)

- Lines 359-374. I think this paragraph should go above as this is your main discussion on your findings.

=>Moved as suggested - line 344

- Line 369. But this is only relative to slow lorises, how can this impact your analysis as you have many species and only 7 videos of pygmy slow lorises.

=>This was probably not clear from the results included. We suggested this due to the results in the new and more detailed figures included now. This includes figure 2 (slow loris the only species with negative sentiment) and figure 5 (reduction in primate sentiment in 2015).

- Line 420. IUCN Red List status ranged from Least Concern to Critically Endangered, so not all the species you considered are threatened. You should rephrase it.

=>This comment is correct; however, the results show that all groups on the IUCN red list have a positive median sentiment. We have tried to clarify this.

- Figure 2a. The IUCN categories should be upper case. In the x axis label you say Mean Sentiment Score but you show box plots (i.e. median, quartiles, range, and outliers). I suggest changing the figures to have a better presentation, maybe use theme(bw), can try different themes in ggplot2.

=>Figure has been changed (now fig 3)

Submitted filename: Review comments round 3.docx

Click here for additional data file.

8 Mar 2021

PONE-D-20-17651R2

Is YouTube® promoting the exotic pet trade?

Analysis of the global public perception of popular YouTube® videos featuring threatened exotic animals.

PLOS ONE

Dear Dr. Chaber,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

==============================

I have now received the comments by one of the reviewers who assessed the first revision of your manuscript, and while he founds that the manuscript has improved considerably, he still points out some minor issues to address before considering your submission fully publishable in PLoS ONE.

I suggest you to undertake the suggested modifications in order to gain final acceptance for publications. Best regards,

==============================

Please submit your revised manuscript by Apr 22 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Jorge Ramón López-Olvera

Academic Editor

PLOS ONE

Journal Requirements:

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Additional Editor Comments (if provided):

Dear Dr. Chaber,

I have now received the comments by one of the reviewers who assessed the first revision of your manuscript, and while he founds that the manuscript has improved considerably, he still points out some minor issues to address before considering your submission fully publishable in PLoS ONE.

I suggest you to undertake the suggested modifications in order to gain final acceptance for publications. Best regards,

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #4: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #4: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #4: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #4: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #4: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #4: The authors did a good job in addressing the comments. In my opinion, the paper is now almost ready to be published. I just have a few more suggestions:

- In the abstract you need to integrate your new findings.

- Line 113. Here you have an open bracket, need to edit the sentence somehow.

- Line 116. Data is the plural of datum, please check throughout the manuscript that the verbs are correct.

- Lines 175-179. Add the command used to run the mixed model. Also, it is not clear to me what you used as fit function for your response variables. Did you use general or generalised linear mixed models? Also, you do not add a random effect to deal with the repeated measurement. You add a random effect to deal with the fact that your data are taken from different videos. The concept of repeated measurement and the concept of random effect are not the same.

- Line 190. add spp. after Macaca.

- In the results, your modelling sections are now disconnected from the other sections of the results. It is a bit confusing. I suggest to integrate them better, you do not need separate sections.

- Line 253. It should be 0.6. Also, it should be 2011 as 2010 is not significantly different than 2006.

- Line 312. Missing s in slow loris.

- Line 321. It is significant so might want to rephrase with something "stronger" than appeared to have.

- Lines 343-347. This paragraph needs some link to literature or to be integrated into another paragraph. It is now too short and unreferenced to be a separate paragraph.

- Lines 324, 344. It is weird to discuss "data not shown". It would be better to add them as supplementary materials.

- Lines 430-432. It seems that Least Concern animals have more positive texts associated to the videos than the other IUCN categories? Please check again that your discussion and conclusion reflects your new results.

- Figure legends. Quantiles should be quartiles in the figure legends (and check the rest of the MS)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #4: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

25 Mar 2021

Response to Reviewer 4

Abstract: The abstract has been altered to incorporate the new findings.

Line 113: This was corrected.

Line 116. The terminology was checked accordingly and we believe it to be correct now.

Line 175-179: The command used to run the mixed model has now been added to the methods section (line 181-183). We used the lmer function which is a general linear mixed model. Furthermore, we understand that repeat measures and random intercepts are not the same, but we found no evidence for the need of a temporal effect.

Line 190: Thank you for pointing this out, it has been corrected.

Results: Thank you for your suggestion, the individual sections have been reduced to integrate the results better.

Line 253. Thankyou for noticing this, it has been corrected.

Line 312: This has also been corrected.

Line 321. This has been rephrased with more certainty.

Lines 343-347. This paragraph has now been integrated into the results section where it appears to be more suitable.

Lines 324, 344: Data can be obtained from the authors as required upon request, hence was not included within the supplementary information.

Lines 430-432: The higher median text sentiment in ‘Least Concern’ species was mentioned earlier in the text (line 333-335), wherein the conclusion has now been reworded to make this clearer to the reader.

Figure legends: Thank you for pointing this out, ‘quantiles’ has been changed to ‘quartiles’ throughout the manuscript.

Thank you for your time and suggestions to help improve our manuscript.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

29 Mar 2021

Is YouTube® promoting the exotic pet trade?

Analysis of the global public perception of popular YouTube® videos featuring threatened exotic animals.

PONE-D-20-17651R3

Dear Dr. Chaber,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Jorge Ramón López-Olvera

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Dear Dr. Chaber,

I have now received the comments on your last submission from the reviewer who assessed the last version and i am pleased to inform you that your manuscript can be now considered suitable for publication in PLoS ONE.

Thank you very much for considering our journal to publish your research and congratulations for the work done and the results obtained.

Best regards,

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #4: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #4: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #4: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #4: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #4: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #4: I had a quick check at the sections the authors modified and I am happy with the edits. I think the paper is interesting and it is now coherent and ready to be published.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #4: No

1 Apr 2021

PONE-D-20-17651R3

Is YouTube promoting the exotic pet trade? Analysis of the global public perception of popular YouTube videos featuring threatened exotic animals.

Dear Dr. Chaber:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Jorge Ramón López-Olvera

Academic Editor

PLOS ONE