*Article* **Evaluating Potential Cetacean Welfare Indicators from Video of Live Stranded Long-Finned Pilot Whales (***Globicephala melas edwardii***)**

**Rebecca M. Boys 1,\* , Ngaio J. Beausoleil <sup>2</sup> , Matthew D. M. Pawley <sup>3</sup> , Emma L. Betty <sup>1</sup> and Karen A. Stockin 1,2,\***


**Simple Summary:** Cetacean strandings occur globally and can impact the welfare as well as the survival of the animals involved. Understanding the welfare status of stranded cetaceans is important to inform appropriate human intervention. However, there is a lack of knowledge on how to assess animal welfare in this context. Here, we used video footage of live stranded animals of four odontocete species to explore which proposed welfare indicators can be assessed at live stranding events. We identified and evaluated potential indicators that could be non-invasively assessed, including 10 non-behavioural and 2 composite behavioural indicators (category of many behaviours). The first data on the fine-scale behaviour of stranded odontocetes and associated human intervention during stranding responses are presented. Our findings suggest that remote assessments of stranded cetacean's welfare states are feasible. These data provide the foundation to develop a systematic, structured welfare assessment framework specific to stranded cetaceans that can inform conservation and management decisions.

**Abstract:** Despite the known benefit of considering welfare within wildlife conservation and management, there remains a lack of data to inform such evaluations. To assess animal welfare, relevant information must be captured scientifically and systematically. A key first step is identifying potential indicators of welfare and the practicality of their measurement. We assessed the feasibility of evaluating potential welfare indicators from opportunistically gathered video footage of four stranded odontocete species (*n* = 53) at 14 stranding events around New Zealand. The first stranded cetacean ethogram was compiled, including 30 different behaviours, 20 of which were observed in all four species. Additionally, thirteen types of human intervention were classified. A subset of 49 live stranded long-finned pilot whales (*Globicephala melas edwardii*) were assessed to determine indicator prevalence and to quantify behaviours. Four 'welfare status' and six 'welfare alerting' non-behavioural indicators could be consistently evaluated from the footage. Additionally, two composite behavioural indicators were feasible. Three human intervention types (present, watering, and touching) and five animal behaviours (tail flutter, dorsal fin flutter, head lift, tail lift, and head side-to-side) were prevalent (>40% of individuals). Our study highlights the potential for non-invasive, remote assessments via video footage and represents an initial step towards developing a systematic, holistic welfare assessment framework for stranded cetaceans.

**Keywords:** animal welfare assessment; behaviour; human intervention; marine mammal; cetacean; management; stranding; wildlife

**Citation:** Boys, R.M.; Beausoleil, N.J.; Pawley, M.D.M.; Betty, E.L.; Stockin, K.A. Evaluating Potential Cetacean Welfare Indicators from Video of Live Stranded Long-Finned Pilot Whales (*Globicephala melas edwardii*). *Animals* **2022**, *12*, 1861. https://doi.org/ 10.3390/ani12141861

Academic Editor: Francesco Filiciotto

Received: 21 June 2022 Accepted: 18 July 2022 Published: 21 July 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

#### **1. Introduction**

The welfare of free-ranging animals is increasingly recognised as important to conservation [1–3]. In addition, there is a growing acknowledgment that human activities may directly and indirectly compromise the welfare of wild animals [4]. However, the conservation of wildlife populations is often a focus of government regulations, policies, and biodiversity plans, and the welfare of the individual animals comprising such populations is often overlooked. This is despite the fact that animal survival, and thus conservation success, is inextricably linked to welfare [1,5].

Although the need to assess wild animal welfare has been highlighted [6–9], there are limited systematic, scientific protocols for such assessments [10]. Furthermore, detailed behavioural and physiological data from species in the wild are often lacking [11], hindering the development of welfare assessments for wild populations [12]. Thus, a first step to progressing systematic and holistic welfare assessment for free-living wild animals is developing methods to capture relevant data. Such data need to be species/taxonand context-specific and should address known or suspected welfare concerns [12,13]. Furthermore, to provide information about the welfare state of the animal, science-based indicators that can be observed and/or measured must be identified [12,14–16].

In the context of free-swimming cetaceans, data on stress hormones [17–19], body condition [20–22], skin disease [23–25], and the impacts of anthropogenic activities on behaviour [26–28] have been collected. However, few studies interpret their findings in terms of welfare or discuss possible welfare implications [9]. During live strandings, cetaceans are subject to both natural [29,30] and anthropogenic stressors [31] that may affect their welfare and survival likelihood [32]. Unfortunately, thus far, such data concerning behavioural and physiological indicators have not been gathered.

Major knowledge gaps concerning the welfare of stranded cetaceans were identified by international experts to be interpreting behavioural and physiological parameters, diagnosing internal injuries, and making end-of-life decisions [32]. Furthermore, these experts stated that major barriers to assessing the welfare of stranded cetaceans related to the limited relevant data collection at strandings and the lack of experts available onsite to interpret parameters and assist in decision making [32]. Notably, the characterisation of stranded cetacean welfare by those experts aligned with contemporary animal welfare science, which interprets interrelated aspects of health, biological function, and behaviour in terms of their impact on the animal's mental state [33,34]. Welfare assessments guided by such characterisations are often facilitated via the use of the Five Domains Model framework for assessing animal welfare [35]. In such a framework, the indicators in domains 1–4 are observed and/or measured, and their cumulative impacts are used to cautiously infer the animal's potential affective state (mental state) in the fifth domain [5,36,37].

Subsequently, these same experts proposed a range of potential welfare indicators for stranded cetaceans [38]. The potential indicators could be grouped into three physical/functional (nutrition, physical environment, and health) and one situation-related domain (behavioural interactions) of the Five Domains Model [35]. The proposed indicators included animal-based parameters, reflecting some aspect of the physical (e.g., body condition), physiological (e.g., respiration rate), or behavioural state of the animal (e.g., vocalisation). Other indicators were resource-/management-based parameters, reflecting aspects of the stranded cetacean's environment (e.g., substrate or duration stranded) or management (e.g., human interaction) that may influence its welfare [39,40]. Resource- /management-based indicators provide welfare-relevant information but do not provide direct evidence of the welfare state and are thus characterised as 'welfare alerting' [12]. Only animal-based indicators can provide direct information about the animal's 'welfare status' and are often preferred in welfare assessments. However, some animal-based indicators may only be 'welfare alerting' in that they can indicate a predisposition for welfare impacts that relate to the animal itself rather than its environment, for example, an animal that is neonatal or unweaned. Welfare alerting indicators are generally more feasible and

reliable to assess across time and different observers and are often non-invasive. They are therefore commonly applied in welfare assessments.

To successfully apply indicators in a welfare assessment framework, the feasibility of measuring the indicators, methods of measurement, and validity for inferring welfare states (i.e., mental states) from observable indicators must be evaluated [10,41,42]. In this study we examined the feasibility of assessing animal-based and resource-/managementbased indicators proposed by experts at live cetacean stranding events [38]. Furthermore, since experts highlighted the need for assessments to be undertaken by remote, skilled personnel [32], we evaluated which indicators can be observed and/or measured using video footage gathered at live strandings.

Specifically, this study evaluates the use of video footage to (1) identify potential animal and resource-/management-based welfare indicators that could be feasibly measured, (2) examine why certain proposed welfare indicators cannot be identified and/or feasibly measured, and (3) assess whether indicators observed from pilot whales can be quantitatively evaluated via video. There are currently no ethograms available for stranded cetaceans, and there is limited detail on the types of human intervention employed at stranding events. Therefore, we sought to identify and characterise all stranded cetacean behaviours (animal-based indicators) displayed and to provide the first description of the types of human intervention (resource-/management-based indicators) that occurred at these same stranding events. Additionally, we examined whether the features of the stranding circumstances affected the prevalence, frequency, or duration of behaviours displayed by stranded pilot whales.

#### **2. Materials and Methods**

#### *2.1. In-field Data Collection*

Due to the stochastic nature of strandings, video footage was collected opportunistically at 14 live stranding events between August 2010 and March 2022 around New Zealand (Table S1). Filming occurred with 53 live stranded cetaceans involving four species of odontocete: long-finned pilot whale (*Globicephala melas edwardii*), pygmy killer whale (*Feresa attenuata*), Cuvier's beaked whale (*Ziphus cavirostris*), and Gray's beaked whale (*Mesoplodon grayi*). Most stranding events (11 events, 49 individuals) involved long-finned pilot whales, herein referred to as pilot whales. Accordingly, the analyses presented here focus only on pilot whales. Additional data from the other examined species are included in the initial ground-truthing only to identify and characterise all animal behavioural and human intervention indicators.

The camera set-up varied based on the opportunistic nature of events and equipment availability (Table S1). When the researchers could attend a stranding, two GoPro Hero 7 Black video cameras (GoPro Inc., San Mateo, CA, USA) were mounted on wooden stakes anchored into the ground at 1–2 m from the focal animal. Each camera was mounted at a height of 50–100 cm, positioned cranio-laterally, and angled caudally (0–45◦ ) towards the tail flukes for each focal individual. These recordings were made at 720 p and 60 fps with a wide-angle view, allowing for the focal animal's entire body to be observed bilaterally. Where researchers were unable to access the animals prior to re-floatation, footage was acquired from stranding personnel, including Department of Conservation (DOC) rangers (the government agency responsible for the management of stranding events), marine mammal medics, and the public using camera phones, GoPro cameras, or other similar video cameras. In such circumstances, the videographer stood 1–2 m from the animal and, when possible, positioned themselves cranio-laterally to the focal individual's head, angling the camera caudally towards the tail flukes. The videographer alternated positions around the animal to enable the entire body to be observed bilaterally. In other cases, individuals were filmed from a lateral position, capturing the entire body on one side. The filming duration was dependent upon battery availability, time of day, and the stranding response procedures in progress.

#### *2.2. Selection of Potential Welfare Indicators*

Based on the opinions of an international panel of experts in cetacean biology, veterinary medicine, and/or animal welfare [38], we developed a list of theoretically observable/measurable parameters that could be used as potential welfare indicators for stranded cetaceans. The list also included parameters that were deemed observable/measurable from an initial viewing of the video footage collected during the stranding events (Table S1). The indicators and composite behavioural parameters (category including many different behaviours, each of which would be considered an indicator) were organised into the three physical/functional domains (nutrition, physical environment, and health) and the situation-related domain (behavioural interactions) of the Five Domains Model for welfare assessment [35] (Table 1). Within each domain, indicators were further split into animal-based indicators that may directly reflect animal state ('welfare status'), and 'welfare alerting' indicators (both animal- and resource-/management-based), which provide relevant information about the animal or its environment that may affect its state [12] (Table 1).

**Table 1.** Proposed animal welfare indicators, or composite indicators \*, organised into the three physical/functional domains and one situation-related domain of the Five Domains Model for welfare assessment [35]. Within each domain, indicators are organised according to the type of information they provide about the animal's state. See text for details of each indicator and how it was measured or scored.


#### *2.3. Data Scoring*

Each video file, for all species, was examined manually at 0.8× speed by the lead author (RMB) at least twice to identify all observable indicators for each focal individual. A subset of videos was examined by two independent observers to ensure consistency in indicator classification. For each animal, information was collated about which indicators could be assessed (Table 1) and, for those indicators that can vary bilaterally, whether they could only be assessed on the left, right, or both sides. The reasons that particular indicators could not be observed and/or measured for each individual cetacean were also noted. Since 92.5% (*n* = 49) of individual focal stranded cetaceans were pilot whales (Table S1), only data related to that species were subsequently analysed and are presented here. To be considered feasible, an indicator had to be fully assessable (across the whole body) and prevalent, being observed in at least 40% of the pilot whales.
