Development of a welfare assessment tool for tourist camp elephants in Asia

Introduction

Around 15,000 captive Asian elephants live under human care across 13 range countries, predominantly in India, Thailand, and Myanmar. These elephants are used for various purposes, including tourism, education, religious ceremonies, logging, transportation, and forest patrolling (Menon & Tiwari, 2019). Thailand alone is home to around 3,800 captive elephants, with 2,700 employed in 250 tourist venues, engaging in shows, riding, bathing, feeding, and observation activities (Bansiddhi et al., 2018). This represents more than twice the number of tourist venues for all other range countries combined (IUCN/SSC Asian Elephant Specialist Group, 2017). Increasingly, international organizations and animal activists are raising concerns over the welfare of elephants used in tourism (Duffy & Moore, 2011; PETA, 2017; Carr & Broom, 2018; Baker & Winkler, 2020), with research studies finding problems with long chaining hours, inadequate exercise, lack of socialization and foraging opportunities, improper equipment use (e.g., ankus misuse, ill-fitting saddles), and provision of high-calorie supplements that leads to obesity (Magda et al., 2015; Norkaew et al., 2018; Bansiddhi et al., 2019a, 2019b; Norkaew et al., 2019a, 2019b; Brown et al., 2020; Supanta et al., 2022). While accredited Western zoos must adhere to strict welfare standards, stronger and more enforceable regulations are needed to support evidence-based management standards for ensuring the ethical treatment of tourist camp elephants across Asia (Bansiddhi et al., 2020; Szydlowski, 2022). Thus, a holistic, evidence-based welfare assessment tool is needed to identify potential welfare risks, inform management decisions, and document welfare changes over time as management changes are made (Barber, 2009; Kagan, Carter & Allard, 2015; Sherwen et al., 2018; Von Fersen et al., 2018; Jones et al., 2022).

As reviewed by Ghimire et al. (2024), dozens of assessment tools have been developed to monitor the welfare of animals in zoological collections, varying in practicality, reliability, and effectiveness. The comparison of these tools highlighted how challenging it is to comprehensively score factors associated with animal care and welfare because they are multifaceted and often species-specific (Mason & Mendi, 1993; Fraser et al., 1997). It was also clear that having well-validated measures and a clear grading system is key (Hampton et al., 2023). The most holistic tools rely on both animal- and resource-based measures to record positive and negative affective states (Kagan, Carter & Allard, 2015; Sherwen et al., 2018; Von Fersen et al., 2018; Benn, McLelland & Whittaker, 2019; Jones et al., 2022; Baumgartner et al., 2024). Animal-based measures such as behavior, health, and mental state provide more direct insight into individual animal welfare and how they experience the environment (Whitham & Wielebnowski, 2009, 2013; Hemsworth et al., 2015; Sherwen et al., 2018; Benn, McLelland & Whittaker, 2019). The use of resource measures can identify conditions necessary to support welfare, such as proper housing, food, and water, and is based on the assumption that if appropriate resources are provided, animal welfare will be good (Whay, 2007; Appleby, Olsson & Galindo, 2018). Access to adequate nutrition, enriched environments, and opportunities for behavioral expression are key drivers of an animal’s mental state (Mellor & Beausoleil, 2015). The key is ensuring that conditions exist so that captive animals have “agency,” or the potential to make meaningful choices with some control over important outcomes in life (Kagan & Veasey, 2010).

The most common welfare assessment tools used today were designed for use with multiple species, but there is a growing interest in developing species-specific tools that integrate a comprehensive set of measures that apply to that species (Jones et al., 2022). A few elephant-specific tools have been developed for zoo animals, but comprehensive animal- and resource-based measures are often lacking (Ghimire et al., 2024). From welfare surveys of hundreds of elephant venues throughout southeast Asia, World Animal Protection has declared that most elephants are kept under poor welfare conditions (Schmidt-Burbach, Ronfot & Srisangiam, 2015; Schmidt-Burbach, 2017, 2020). However, those evaluations were based only on resource-based criteria, with low scores given to the presence of an ankus (also known as a bullhook or guide) or if the elephant participated in tourist activities; there were no animal-based measures to corroborate scores of poor welfare. Recently, an assessment tool relying on animal-based measures was developed for routine use by elephant keepers to examine day and nighttime behaviors as indicators of well-being (Yon et al., 2019). Zoo-centric tools can be challenging to apply across tourist camp contexts, however. Daily routines in most camps involve tourist activities not common in zoos, such as shows, riding, walking, bathing, and feeding (Bansiddhi et al., 2018). In Asian range countries, one or two mahouts are generally assigned to one elephant and are responsible for its care and handling (Kontogeorgopoulos, 2009; Vanitha, Thiyagesan & Baskaran, 2011; Mumby, 2019), whereas zoo elephants are generally managed in groups and by a team of keepers (Association of Zoos & Aquariums, 2012; European Association of Zoos & Aquariums (EAZA), 2020). Elephants at most accredited zoos are managed in protected contact where humans and elephants are separated by a barrier (Laule & Whittaker, 2009), while most camp elephants are managed in free contact and share the same space with mahouts and tourists (Bansiddhi et al., 2018). The contact system affects the types of management needed to ensure human and elephant safety, as few tourist venues have enclosures similar to zoo exhibits. Restraint methods like chaining and the use of an ankus are common in Asia (Bansiddhi et al., 2019a), but less so in zoos (European Association of Zoos & Aquariums (EAZA), 2020). Many large camps manage more than 30 elephants (Bansiddhi et al., 2018), so recording behavior every 30 min throughout the day and night or conducting individual behavior observations over days (Yon et al., 2019) is not practical in the context of these tourist camps. Consideration of these fundamental differences is needed in assessing welfare, given the wide variation in management systems used across venues in Asia.

In Western zoos, tracking welfare often utilizes technologies like closed-circuit TV (CCTV), radio frequency identification (RFID), and global positioning system (GPS) monitoring using sophisticated computerized tracking software (Whitham & Wielebnowski, 2013; Whitham & Miller, 2016; French, Mancini & Sharp, 2018; Coe & Hoy, 2020). These methods are impractical for use in elephant camps for various reasons because of limited experience in using these tracking technologies, lack of research staff for data input and analysis, budget constraints, and language barriers, as most programs are only in English. Moreover, CCTV monitoring could raise public privacy concerns in tourist camps. Reliance solely on mahout evaluations also has limitations compared to zoo keepers, who are often more familiar with welfare criteria and methodology (Whitham & Wielebnowski, 2009). While mahouts are familiar with elephants, they may not consistently identify or understand stereotypic or other abnormal behaviors (Fuktong et al., 2021) and present positive biases due to the fear of losing jobs. Thus, expert knowledge should be integrated into the assessment process.

Our goal was to develop a practical, reliable, valid, and low-cost welfare assessment tool for tourist camp elephants in Asia. Validated measures from peer-reviewed literature and expert consultancy were incorporated into the tool criteria, while content validity and reliability tests were conducted to ensure consistency and accuracy in welfare assessments across multiple raters and venues in Thailand (Jones et al., 2022; Malkani, Paramasivam & Wolfensohn, 2022). Ultimately, it is hoped that the tool will encourage stakeholders to participate regularly in comprehensive evaluations to identify welfare risks and solutions in an under-regulated industry. It also will be key to engaging policymakers to standardize good welfare practices for captive elephants throughout Asia. Lastly, the tool can be used to track welfare changes over time, enabling institutions to make more informed decisions as to what changes are needed to ensure the welfare of elephants across a diverse array of management practices (Sherwen et al., 2018).

Materials and Methods

The study protocol was approved by the Institutional Animal Care and Use Committee, Faculty of Veterinary Medicine, Chiang Mai University, Chiang Mai, Thailand (FVM-ACUC; S7/2566) and The Human Research Ethics Committee, Faculty of Veterinary Medicine, Chiang Mai, Thailand (HS5/2566).

Welfare assessment framework

The Five Domains Model developed by Mellor et al. (2020) was selected as the assessment framework and modified to encompass four domains as described by Racciatti et al. (2022): (1) Nutrition, (2) Environment, (3) Health, and (4) Behavior and Mental State.

The steps used to develop the elephant welfare assessment tool are shown in Fig. 1. First, a review of the scientific literature was conducted to identify relevant elephant welfare measures using PubMed and Google Scholar databases. Search terms included ‘elephant’, ‘Elephantidae’, ‘Loxodonta’, and ‘Elephas’ used in conjunction with ‘behavior’, ‘nutrition’, ‘husbandry’, ‘health’, ‘welfare’, ‘living conditions’, ‘tourist activities’, ‘housing’, and ‘mental state’. Guidelines and standards created for elephant tourist camps (Association of British Travel Agents Animal Welfare Guidelines and Usage of Captive Elephants in Malaysia by the Malaysian Association of Zoological Parks and Aquaria) were also reviewed to aid in defining criteria related to resource-based measures primarily.

Discussion

This project aimed to develop a welfare assessment tool for elephants managed in non-zoological settings, specifically those at tourist camps in Asia. While many Western zoos are governed by accredited organizations and follow strict standards and guidelines, including requirements for routine welfare assessments (e.g., AZA, EAZA, BIAZA, WAZA, ZAA), such oversight is absent for Asian tourist camps (Bansiddhi et al., 2020). Thus, a tool is needed to evaluate the welfare of animals living under conditions that are not generally experienced by those in zoo environments. To that end, the EWAT was designed to be reliable and practical under field conditions and consisted of 18 resource- and 18 animal-based measures to provide information on physical, environmental, behavioral, and social states, as well as husbandry practices used in tourist camps of varying sizes, elephant numbers, and tourist activities.

A range of welfare assessment tools, many developed using the Five Domains Model framework, are utilized by zoological organizations worldwide to conduct regular assessments at individual and/or institutional levels (see review, Ghimire et al., 2024). The most holistic tools rely on both animal- and resource-based measures and record positive and negative affective experiences (Baumgartner et al., 2024; Sherwen et al., 2018). Consequently, the adoption of the Five Domain Model was pivotal to developing our tool specific to elephants in Asia. Like Racciatti et al. (2022), we combined the Behavior and Mental State domains because behavioral indicators directly or indirectly assess some affective states. Veasey (2020) has stated that behavior associated with survival, along with species-specific social and cognitive opportunities, are crucial for evaluating the psychological priorities of captive elephants. Thus, this domain was intended to capture behavioral outputs as indices of elephants’ perceptions of external circumstances. This integrated approach makes it easier for assessors to focus on and score measures associated with behaviors within a single domain.

The allocation of measures in this new welfare assessment tool addresses a range of biological requirements and is based on recent animal welfare studies (Bansiddhi et al., 2018; Williams et al., 2018; Bansiddhi et al., 2019a, 2019b; Norkaew et al., 2019a; Yon et al., 2019; Brown et al., 2020; Veasey, 2020). However, measures such as safety from predators (Sherwen et al., 2018) were omitted because they have a minimal impact on captive elephants in tourist camps. Financial constraints, impracticality in resource-constrained settings like tourist camps, and the aim to create a cost-effective tool also led to the exclusion of physiological measures of glucocorticoid (GC) and immunoglobulin A concentrations (O’Brien & Cronin, 2023), which are frequently used to assess the impacts of management factors on health and welfare of zoo and tourist camp elephants (Brown et al., 2019; Edwards et al., 2019; Kosaruk et al., 2020). Likewise, the decision to omit other invasive methods like blood chemistries or technology-based measurements of heart and respiratory rates (Wolfensohn et al., 2018) underscores the commitment to noninvasive and easy-to-apply methodology. However, it is recognized that these techniques can provide valuable insights into the biological functioning of animals, and so should be included whenever possible, if logistically feasible.

In Western zoos, welfare assessments often rely on the use of technologies and keeper knowledge (Whitham & Wielebnowski, 2009; Whitham & Miller, 2016); however, the practical constraints of tourist camps necessitated a focus more on direct observations and interviews with mahouts. Reliance on keeper ratings to detect subtle behavioral changes can present problems with subjectivity, personal biases, and a lack of knowledge of behavioral diversity (Sherwen et al., 2018; Brouwers & Duchateau, 2021). Furthermore, a lack of understanding and awareness of animal welfare among mahouts in range countries makes it particularly challenging (Ward et al., 2020). To address this issue, our tool integrates the perspectives of the elephant’s mahout and an expert rater familiar with welfare metrics and methodology to ensure a more comprehensive and objective evaluation. Additionally, interviews are conducted in the language of the mahout (in this case, Thai), which is vital as most do not speak English. Interviewers can be veterinarians, veterinary assistants, or researchers familiar with elephants. The tool prioritizes ease of use to accommodate the varying schedules and management practices of tourist camps, and to allow mahouts to be actively involved in the assessment process during tourist activities. Similarly, clear communication and coordination with elephant camp managers and raters is essential, especially given the sensitivity of discussing elephant welfare in front of visitors.

Numerical scoring systems are common in welfare assessment tools as they allow for easy data analysis (Jones et al., 2022), with scales typically going from 0 to 2 to 0 to 8 (Kagan, Carter & Allard, 2015; Sherwen et al., 2018; Von Fersen et al., 2018; Benn, McLelland & Whittaker, 2019; Baumgartner et al., 2024). Our tool uses a 0–2 Likert scale, similar to several other welfare tools (Sherwen et al., 2018; Benn, McLelland & Whittaker, 2019; Baumgartner et al., 2024). From a practical perspective, numerical scales are easy to use and allow for rapid assessments. Subjectivity is always a concern, however, especially when using small scales. For example, a 5 m chain length scores a 0, a 6 m chain scores 1; chaining for 12 h scores a 1, while 13 h receives a 0. A revised tool may need to implement a finer scale for measures where little scientific evidence exists regarding how they impact physical or psychological welfare. To that point, selected indicators must be scientifically validated as much as possible or rely on the opinion of field-based experts (Barber, 2009; Melfi, 2009; Kagan, Carter & Allard, 2015). Our expert panel consisted of veterinarians and researchers working with zoo and tourist camp elephants across Western and non–Western countries, who provided input on content clarity, interpretability, appropriateness, and relevance to tourist camps. The calculation of CVI provided a statistical dimension to the experts’ opinions, with six experts considered sufficient to calculate a CVI score (Polit & Beck, 2006; Polit, Beck & Owen, 2007; Yusoff, 2019). The resulting high levels of consensus among experts reflected in S-CVI/UA and S-CVI/Average scores demonstrated the robustness of the tool development and assessment process. Some measures were revised or omitted after expert review and follow-up team discussions. Notably, the opportunity to mate, mahout job satisfaction, and duration of the mahout-elephant working relationship were removed, while the type of substrate, environmental complexity, use of ankus, and elephant-tourist relationship underwent revisions. The opportunity to mate was removed because experts suggested that females do not always have a choice, with anecdotal reports of forced mating occurring. Although recognized as important to elephant welfare, mahout job satisfaction was removed because experts were concerned mahouts might feel uncomfortable expressing a lack of job satisfaction. Thus, the measure would not be accurate. The high turnover of mahouts throughout Thailand also limits the usefulness of the metric for the duration of the mahout-elephant working relationship. The highest score for ‘substrate’ was switched from sand to a choice of multiple substrates because of the limited use of sand in relatively small enclosures (Bansiddhi et al., 2018) and because giving elephants an option to select a preferred substrate at any given time is ideal (Meller, Croney & Shepherdson, 2007). Environment complexity was updated to include mahout-provided enrichment such as food puzzles, balls, and tires. Recognizing the importance of the ankus for the safety of mahouts and tourists when elephants are managed in free contact, a score of 2 was updated to state that the ankus can be carried but only used in emergencies if elephant actions threaten human life. Inclusion of the animal-based measure Wound Score indirectly rates ankus use, as most wounds are caused by misuse of that tool (Bansiddhi et al., 2019a). Finally, a score of 2 for the ‘elephant-tourist relationship’ was revised to reflect scenarios where there is no interaction between elephants and tourists, for example, in camps that only allow observation.

Another aim of this study was to test the reliability of the tool at several tourist camps in Thailand. The ICC values for three raters scoring 10 elephants ranged from 0.78–0.90, which is considered good reliability (Czycholl et al., 2016; Koo & Li, 2016; Czycholl, Klingbeil & Krieter, 2019). This can be attributed to the expertise of the raters, who had 3 to 13 years of elephant experience. Mahouts also contributed invaluable insights based on their working knowledge of elephants. Possible challenges may be encountered in other countries like Nepal, where elephants are managed by two mahouts with distinct responsibilities (Mumby, 2019), which could introduce variability in data collection. Elephants were assessed two times for test-retest reliability, which also showed good reliability (0.77–0.91). The assessor, a veterinarian with more than 5 years of experience, performed two assessments within 7 days to limit the influences of seasonality or husbandry changes on elephant behavior and health. It is reasonable to assume that, due to the short timeframe, the assessor could have remembered some scores from the previous assessment; however, remembering all scores of 36 measures for 10 elephants was highly unlikely. While we successfully demonstrated inter-rater reliability and test-retest reliability of the protocol, constraints in staff availability led to test-retest reliability being evaluated by only one assessor. The simplicity in the implementation of the tool likely played a role in achieving good reliability. A straightforward and user-friendly tool enables assessors to apply the measures consistently, regardless of variations in their backgrounds or experience levels. This ease of use minimizes potential errors and misunderstandings during data collection, which can adversely affect reliability scores. In our study, the simple design of the tool allowed both experienced veterinarians and mahouts to efficiently and accurately assess the elephants, contributing to the high inter-rater and test-retest reliability observed. Additionally, in situations with staff limitations—such as having only one assessor for test-retest reliability—the tool’s simplicity ensures that reliable data can still be collected without the need for extensive training or adjustments.

Although the tool was tested by research veterinarians from Chiang Mai University, the tool is simple enough for it to be used by tourist camp staff and mahouts for regular assessments. Plans are underway to involve in-house staff in future reliability testing after they receive training on the welfare assessment process using EWAT. As demonstrated by Rodríguez Ruiz & Heredia Rico (2013), training increases the reliability of results. It reduces protocol application time, so we acknowledge the critical role of proper training in ensuring the quality and accuracy of assessments, as assessor familiarity with elephant behavior is essential, particularly for accurately identifying stereotypic or other abnormal behaviors and conducting physical examinations.

Conclusions

Welfare assessment tool development is a continuous process, evolving through refinement and improvement over time as the tool is implemented. In the development of EWAT, the level of detail in the assessment and grading system was defined with consideration for constraints with limited assessment time at each venue and restricted access to observe individual animals involved in tourist activities to avoid interference with venue management. The EWAT represents the first context-specific welfare assessment tool specifically designed for tourist camp elephants. It consists of 36 reliable and valid measures covering aspects of the Five Domains Model with a step-by-step paper-based data collection process. The tool is intended for regular use by elephant veterinarians and mahouts to ensure consistent and ongoing monitoring of welfare as part of in-house assessments. Additionally, it can serve as a useful research tool for elephant biologists and ethologists to track the effects of interventions aimed at improving elephant welfare. A pilot study utilizing the EWAT is currently underway, which should help refine the tool, and provide insights into the tool’s ability to assess welfare at individual and camp levels. Welfare score comparisons will also be conducted across domains, parameters such as age, sex, origin, tourist activity, camp size, and daily tourist visits, and between resource- and animal-based measures to facilitate further critical analysis.

Other welfare measures could be used to further validate this tool in subsequent studies. Analyses of GCs are commonly used to validate elephant welfare measures (Williams et al., 2018), although it is important to note that GCs are not related only to stress, but are metabolic hormones that respond to stimuli, negative and positive, to maintain homeostasis (Pokharel & Brown, 2023). IgA is a novel welfare and health indicator that fluctuates in positive and negative welfare states (Yeates & Main, 2008), and so can be equally as difficult to interpret as GCs (Edwards et al., 2019). However, both can be measured non-invasively in feces, so incorporating GCs and IgA analyses into the welfare assessment tool could provide additional insight into the biological effects of tourist camp management factors. However, as summarized by Pokharel & Brown (2023), it is important to consider factors such as age, sex, season, and context for proper interpretation of GC data, all of which can affect and limit the usefulness of this measure, especially for single samples.

Finally, a global effort is required to bring greater awareness and educate people in range countries about animal welfare and the importance of maintaining healthy populations of endangered flagship and umbrella species like elephants. Public perceptions of captive elephants, often shaped by anthropocentric views and a lack of knowledge about animal welfare issues, remain a concern. Institutions and researchers should actively engage with the public, fostering better communication and understanding to help people make informed decisions about which tourist venues to visit. Tourist camps primarily focus on business and economic profits, whereas zoos are more oriented toward animal welfare, species conservation, education, and research. Activities such as shows, riding, logging, participating in religious festivals, and involvement in polo tournaments, as well as management practices about the ankus, chaining, training methods, and weaning provoke international concern. However, the outright dismissal of such practices could lead to tension between local communities and outside experts, and not all activities directly harm elephants (Brown et al., 2020; Kongsawasdi et al., 2021). Establishing collaborations among all stakeholders is vital for informed tool application and subsequent management adaptations. Such collaborations will also create an environment of open communication about animal welfare among tourist venue staff, benefit staff education, raise awareness, and foster a positive staff culture around animal welfare. The government can play a pivotal role by adopting strong welfare standards, requiring the use of welfare assessment tools and camp audits, and imposing penalties for noncompliance or incentives for compliance. This collaborative approach will ensure that the welfare of elephants is not only assessed but effectively addressed through informed and evidence-based actions.

Supplemental Information