Introduction
Recent decades have seen increased interest in the assessment of the welfare of wild animals under human care (Daigle & Siegford Reference Daigle and Siegford2014; Clegg et al. Reference Clegg, Borger-Turner and Eskelinen2015; Mellor & Beausoleil Reference Mellor and Beausoleil2015; Wolfensohn et al. Reference Wolfensohn, Sharpe, Hall, Lawrence, Kitchen and Dennis2015; von Fersen et al. Reference von Fersen, Encke, Hüttner and Baumgartner2018; Fischer et al. 2021; Jones et al. Reference Jones, Sherwen, Robbins, McLelland and Whittaker2022; Chavarría et al. Reference Chavarría, Vásquez-Vargas, Calderón, Matamoros, Leitón, Fernández and Vargas2023). Currently, only a limited number of comprehensive welfare assessment tools exist for cetacean species. Among these, the C-Well which is based on the Welfare Quality® Framework (Clegg et al. Reference Clegg, Borger-Turner and Eskelinen2015), and the Dolphin-Wet which follows the Five Domains model (Baumgartner et al. Reference Baumgartner, Hüttner, Clegg, Hartmann, Garcia-Párraga and Manteca2024) have been specifically developed for the most commonly managed cetacean species under human care: the bottlenose dolphin (Tursiops truncatus). However, the behaviour, social structure, and adaptation of animals, including cetaceans, to captive environments vary even among closely related species (Mason Reference Mason2010; Serres et al. Reference Serres, Hao and Wang2019). For example, bottlenose dolphin or beluga whale (Delphinapterus leucas) are quite gregarious species with herd size around the several hundred mark and are characterised by high ratings in inter-individual behaviours, such as affiliative/social/contact behaviours, aggression, and object manipulation/play behaviours. On the other hand, species such as the Indo-Pacific finless porpoise (Neophocaena phocaenoides) are more often encountered singly or in pairs, and they are characterised by low ratings in inter-individual behaviours, and object manipulation (Nakahara & Takemura Reference Nakahara and Takemura1997). Therefore, species-specific tools need to be developed to ensure the welfare of the individual animals is accurately assessed. The development of a comprehensive animal welfare assessment protocol requires a detailed, well-justified, and clearly outlined step-by-step process to avoid potential issues related to the applicability and reliability of the results (Hampton et al. Reference Hampton, Hemsworth, Hemsworth, Hyndman and Sandøe2023). In addition, an animal welfare assessment tool must be standardised, practical, and allow for comparison across different time-periods and individuals (Jones et al. Reference Jones, Sherwen, Robbins, McLelland and Whittaker2022). Only a limited number of studies have provided a comprehensive insight into the framework development process (Beausoleil et al. Reference Beausoleil, Fisher, Littin, Warburton, Mellor, Dalefield and Cowan2016; Sherwen et al. Reference Sherwen, Hemsworth, Beausoleil, Embury and Mellor2018; Allen et al. Reference Allen, Allen, Ballard, Drouilly, Fleming and Hampton2019; Baker et al. Reference Baker, Ayers, Beausoleil, Belmain, Berdoy, Buckle, Cagienard, Cowan, Fearn-Daglish, Goddard, Golledge, Mullineaux, Sharp, Simmons and Schmolz2022; De Ruyver et al. Reference De Ruyver, Baert, Cartuyvels, Beernaert, Tuyttens, Leirs and Moons2023; Serres et al. Reference Serres, Boys, Beausoleil, Platto, Delfour and Li2024), including the recent welfare assessment tool for the Yangtze finless porpoise (YFP) (Neophocaena asiaeorientalis asiaeorientalis) (Platto et al. Reference Platto, Serres, Normando and Hao2025).
Among the different animal welfare assessment tools applied to a variety of animal species (Barnard & Hurst Reference Barnard and Hurst1996; Hosey Reference Hosey2005; Dawkins Reference Dawkins2006; Whitham & Wielebnowski Reference Whitham and Wielebnowski2013), the Five Domains model (FDM) represents the most used framework due to its versatility and adaptability to different contexts (Kagan et al. Reference Kagan, Carter and Allard2015; Sherwen et al. Reference Sherwen, Hemsworth, Beausoleil, Embury and Mellor2018; Mellor et al. Reference Mellor, Beausoleil, Littlewood, McLean, McGreevy, Jones and Wilkins2020). The FDM is a tool designed to recognise that an animal’s welfare state is shaped by its subjective experiences (Fraser Reference Fraser2008; Mellor et al. Reference Mellor, Patterson-Kane and Stafford2009; Hemsworth et al. Reference Hemsworth, Mellor, Cronin and Tilbrook2015; Mellor Reference Mellor2016), and it is structured to differentiate between internal and external physical and functional conditions that lead to both positive and negative affects (Hosey Reference Hosey2005; Dawkins Reference Dawkins2006; Mellor et al. Reference Mellor, Patterson-Kane and Stafford2009, Reference Mellor, Beausoleil, Littlewood, McLean, McGreevy, Jones and Wilkins2020; Whitham & Wielebnowski Reference Whitham and Wielebnowski2013; Sherwen et al. Reference Sherwen, Hemsworth, Beausoleil, Embury and Mellor2018). When using this framework for the assessment of animal welfare, four physical domains are considered (nutrition, physical environment, health, and behavioural interactions) and their impact on a fifth domain (mental experiences). Moreover, assessing animal welfare requires a wide range of indicators to be taken into consideration (Mason & Mendl Reference Mason and Mendl1993; Botreau et al. Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissier2007a,Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissierb; DEFRA 2012; Salas et al. Reference Salas, Manteca, Abáigar, Delclaux, Enseñat, Martínez-Nevado and Fernández-Bellon2018). In this regard, the FDM identifies the indicators in two distinct categories: Welfare Status which refers to animal-based measures reflecting observable or quantifiable conditions of an individual (e.g. body condition, species-specific behaviours); and Welfare Alerting which can be either resource- or animal-based indicators, and show the potential or capacity of the environment and care system to impact upon the welfare of the animal (e.g. providing food, available shelter). Additionally, Welfare Status and Alerting indicators are categorised based on their valence, reflecting either a positive (Welfare Enhancement) or a negative impact (Welfare Compromise) on the animal’s overall welfare (Mellor et al. Reference Mellor, Beausoleil, Littlewood, McLean, McGreevy, Jones and Wilkins2020).
Once welfare indicators have been selected and validated, a standardised scoring system must be established (Welfare Quality® 2009). The scoring method must align with the intended objectives of the protocol, specifying how the outcomes will be utilised, and ensure consistent application across different assessments, allowing for comparable results across diverse situations and individuals (Jones et al. Reference Jones, Sherwen, Robbins, McLelland and Whittaker2022). Nevertheless, creating a scoring system poses a lot of challenges, particularly considering that all scoring approaches have inherent limitations, and are shaped by ethical and methodological assumptions that affect the interpretation of outcomes (Mellor & Reid Reference Mellor and Reid1994; Fraser et al. Reference Fraser, Weary, Pajor and Milligan1997; Hemsworth et al. Reference Hemsworth, Mellor, Cronin and Tilbrook2015; Sandøe et al. Reference Sandøe, Corr, Lund and Forkman2019; Mellor et al. Reference Mellor, Beausoleil, Littlewood, McLean, McGreevy, Jones and Wilkins2020; Hampton et al. Reference Hampton, Hemsworth, Hemsworth, Hyndman and Sandøe2023). For example, some frameworks assign numerical values using ordinal scales such as 0–5 (Hampton et al. Reference Hampton, Hyndman, Laurence, Perry, Adams and Collins2016), 1–8 (De Ruyver et al. Reference De Ruyver, Baert, Cartuyvels, Beernaert, Tuyttens, Leirs and Moons2023), –1–1 (King et al. Reference King, Matson and DeVries2021), and 0–2 (Clegg et al. Reference Clegg, Borger-Turner and Eskelinen2015; Baumgartner et al. Reference Baumgartner, Hüttner, Clegg, Hartmann, Garcia-Párraga and Manteca2024), while others use letter grades (A: none to E: severe for Welfare Compromise, and 0: none to +++: high level for Welfare Enhancement) (Mellor & Reid Reference Mellor and Reid1994; Mellor & Littin Reference Mellor and Littin2004; Sandøe et al. Reference Sandøe, Forkman and Christiansen2004; Williams et al. Reference Williams, Mellor and Marbrook2006; Littlewood & Mellor Reference Littlewood and Mellor2016; Mellor et al. Reference Mellor, Beausoleil, Littlewood, McLean, McGreevy, Jones and Wilkins2020), or categorical terms like ‘None’ to ‘Extreme’ (Beausoleil et al. Reference Beausoleil, Fisher, Littin, Warburton, Mellor, Dalefield and Cowan2016). These different approaches are not directly comparable and may imply different levels of precision or severity, even when referring to similar welfare impacts. Such variability highlights the challenge of standardising welfare interpretation across frameworks and underscores the risk of misinterpretation if scoring scales are not explicitly defined and calibrated.
The Yangtze finless porpoise (YFP) (Neophocaena asiaeorientalis asiaeorientalis), subspecies of the narrow-ridged finless porpoise (Neophocaena asiaeorientalis), is deemed to be the only remaining freshwater cetacean in the Yangtze river (Gao & Zhou Reference Gao and Zhou1993; Turvey et al. Reference Turvey, Pitman, Taylor, Barlow, Akamatsu and Barrett2007). Drastic declines in population over the last thirty years have seen the establishment of ex situ breeding programmes (Mei et al. Reference Mei, Zhang, Huang, Zhao, Hao and Zhang2014), including the Yangtze Cetacean Breeding and Research Centre (YCBRC) that currently holds 12 YFPs. The development of welfare assessment tools for the YFP is crucial for the success of these programmes and can be useful for welfare assessment for similar subspecies of the genus Neophocaena (Neophocaena phocaenoides, and Neophocaena asiaeorientalis sunameri) under human care. In a previous study, a list of indicators that could provide information on YFP welfare was validated through expert opinion and was used to build the Yangtze Finless Porpoises-Welfare Assessment Protocol (YFP-WAP) (Platto et al. Reference Platto, Serres, Normando and Hao2025). The current study represents the second phase of a wider research project on the development of a YFP welfare assessment tool which includes three phases (1: establishing the indicator; Platto et al. Reference Platto, Serres, Normando and Hao2025; 2: developing the scoring system; and 3: validation/feasibility and implementation of the protocol). This paper focuses specifically on the second phase, aiming to create a scoring method for assessing YFP welfare based on previously validated indicators.
Materials and methods
The YFP-WAP Scoring Method
The FDM (Mellor Reference Mellor2017) was chosen as a guide for the development of the YFP-WAP and the scoring method was adapted from Serres et al. (Reference Serres, Boys, Beausoleil, Platto, Delfour and Li2024). The FDM utilises a five-point Likert grading system for Welfare Compromise, and a four-point Likert system for Welfare Enhancement. While comprehensive, this approach can be relatively labour intensive and time consuming, leading to alternative, simpler scales being proposed to seek to mitigate these challenges (Cook Reference Cook2018; Kubasiewicz et al. Reference Kubasiewicz, Rodrigues, Norris, Watson, Rickards, Bell and Burden2020). In addition, when the FDM grading system was applied to the YFP-WAP, it became evident that the scoring method did not allow for the equal contribution of each indicator to the final result within each valence group and domain. Specifically, indicators with higher scores, whether positive or negative, showed a disproportionate influence on the final outcome. This imbalance led to the development of a new scoring system that would provide a more balanced welfare assessment.
Various scoring methods have been explored in the literature, including those where indicator scores are weighted and combined to produce an overall result (Botreau et al. Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissier2007a,Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissierb; Sherwen et al. Reference Sherwen, Hemsworth, Beausoleil, Embury and Mellor2018; Benn et al. Reference Benn, McLelland and Whittaker2019; Yon et al. Reference Yon, Williams, Harvey and Asher2019). For the development of the current scoring system, the scoring of Welfare Enhancement (WE) and Welfare Compromise (WC) were conducted separately, in accordance with prior frameworks based on the FDM. This distinction between these two aspects is critical because WE and WC represent fundamentally different scenarios: WE reflects “an opportunity” for the animal to improve its welfare, and experience positive affective states, while WC refers to “something causing a problem” (Mason & Veasey Reference Mason and Veasey2010; Green & Mellor Reference Green and Mellor2011). According to Botreau et al. (Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissier2007a,Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissierb) to avoid compensation issues, aggregation should be limited to the valence group level (WC and WE), allowing only indicators of the same valence to compensate for one another. This approach enhances traceability, and avoids the problem of compensation, where negative and positive scores could cancel each other out, masking real welfare conditions (Botreau et al. Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissier2007a,Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissierb).
Moreover, the sum of indicator scores was chosen as the most suitable approach for the YFP-WAP, a method already widely used in other animal welfare assessments (Bartussek Reference Bartussek1999; Horning Reference Hörning2001; Scott et al. Reference Scott, Nolan and Fitzpatrick2001; Bracke et al. Reference Bracke, Metz, Spruijt and Schouten2002; Serres et al. Reference Serres, Boys, Beausoleil, Platto, Delfour and Li2024). In this approach, all indicators are weighted based upon their impact on animal welfare, with lower weights assigned to measures with reduced impact on the individual’s welfare (Botreau et al. Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissier2007a,Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissierb). The use of numerical calculations ensures consistent scoring while avoiding spurious averages of welfare scores, with the final welfare outcome being converted into non-numerical values. Although the sum of scores method has limitations, such as the potential for compensation (Spoolder et al. Reference Spoolder, De Rosa, Hörning, Waiblinger and Wemelsfelder2003), by addressing the challenges of the original FDM approach, and implementing a clear, balanced scoring method, this protocol offers a comprehensive tool for assessing the welfare of animals in a way that is both practical and scientifically robust.
Indicator scores
During the first phase of the project (Platto et al. Reference Platto, Serres, Normando and Hao2025), and in accordance with the methods of Botreau et al. (Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissier2007a,Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissierb) and Serres et al. (Reference Serres, Boys, Beausoleil, Platto, Delfour and Li2024), each indicator was assigned an intensity level (High, Medium, or Low), reflecting its impact on the YFPs’ welfare. Indicators were also assigned a valence score which weighted the degree of WE or WC, using the scale previously developed by Mellor et al. (Reference Mellor, Beausoleil, Littlewood, McLean, McGreevy, Jones and Wilkins2020). By using these pre-assigned intensity levels and valences, scores for Welfare Enhancement and Compromise for each indicator could be readily determined. Specifically, WE and WC scores were designated by using a 4-point Likert scale: WE, +++ (High), ++ (Medium), + (Low), and 0 (no impact on the animal’s welfare); WC, D (High), C (Medium), B (Low), and A (0: no impact on the animal’s welfare). For example, the ‘Water Quality’ indicator has a pre-assigned ‘High’ intensity level for the condition ‘unacceptable’, corresponding to a score of D for the WC. Conversely, if rated as ‘acceptable’ it is assigned a ‘Medium’ intensity level corresponding to a score of ++ for WE. Furthermore, an indicator may affect multiple domains and, therefore, could be categorised as Welfare Status (WS) in its primary domain and Welfare Alerting (WA) in other domains, with different scores reflecting its distinct impact across those domains. For example, the ‘General Health’ indicator was classified as WS in Domain 3 (Health), as it directly reflects the individual’s health state, and as WA in Domain 1 (Nutrition) and Domain 4 (Behavioural Interactions), since health can influence the YFPs’ appetite and social interactions. This systematic approach facilitated the creation of a comprehensive table that includes WS and WA indicators, categorised within their respective valence groups of WE and WC, along with their corresponding scores, all organised within their specific domains (see Table S1; Supplementary material; Platto et al. Reference Platto, Serres, Normando and Hao2025).
Normalisation process to defining Welfare Compromise and Enhancement ranges
Previous studies on animal welfare assessment have commonly used ordinal scales to assign scores reflecting the intensity or severity of welfare indicators and their impact on animal well-being (Beausoleil et al. Reference Beausoleil, Fisher, Littin, Warburton, Mellor, Dalefield and Cowan2016; Hampton et al. Reference Hampton, Hyndman, Laurence, Perry, Adams and Collins2016; Littlewood & Mellor Reference Littlewood and Mellor2016; King et al. Reference King, Matson and DeVries2021; Baker et al. Reference Baker, Ayers, Beausoleil, Belmain, Berdoy, Buckle, Cagienard, Cowan, Fearn-Daglish, Goddard, Golledge, Mullineaux, Sharp, Simmons and Schmolz2022; De Ruyver et al. Reference De Ruyver, Baert, Cartuyvels, Beernaert, Tuyttens, Leirs and Moons2023). However, relying solely upon numerical scores may create a misleading impression of precision and can lead to inappropriate statistical treatments, such as averaging scores or performing calculations that do not accurately capture the qualitative nature of the data (Mellor & Beausoleil Reference Mellor and Beausoleil2015; Sandøe et al. Reference Sandøe, Corr, Lund and Forkman2019). Nonetheless, numerical data can be useful for representing the degree of WC or WE in specific contexts, and for facilitating the aggregation of multiple indicators within domains. Therefore, in the present study, numerical scores were applied to support internal aggregation within domains, but the final welfare outcomes were converted into non-numerical grades to offer a clearer representation of welfare status.
Furthermore, as each domain in this study includes a different number of indicators, summing raw scores directly could introduce bias by disproportionately weighting domains with more indicators. To address this, the YFP-WAF framework incorporates a normalisation process that adjusts WC and WE scores to account for differences in the number of indicators across domains. In this regard, each indicator’s non-numerical scores were easily converted into numerical values ranging from 0 to 3: 3 (High: +++; D), 2 (Medium: ++; C), 1 (Low: +; B), and 0 (corresponding to A or 0).
The initial step of the new scoring system involves defining the range for each numerical score category, enabling the back-conversion of final scores into non-numerical values. The pre-assigned numerical values for each indicator are converted into percentages using the following approach: the highest possible scores, +++ (3) or D (3), are assigned a percentage value of 100%, serving as a reference point for determining the percentage equivalents of lower scores through proportional scaling.
This system allows determination of the ranges for each score category: 33% indicates the maximum of the percentage range for the value 1 (B and + which fall in the percentage range 1–33%); 67% indicates the maximum of the percentage range for the value 2 (C and ++ which fall in the percentage range 34–67%); and 100% represents the maximum of the percentage range for the value 3 (D and +++ which fall in the percentage range 68–100%). These percentage ranges allow the conversion of final welfare assessment scores from percentages back into their original non-numerical values, ensuring consistency in interpretation (Table 1).
Table 1. Welfare Compromise (WC) and Enhancement (WE) scores in their numerical and non-numerical forms, and associated percentage ranges as part of the proposed standardised scoring system to assess the welfare of Yangtze finless porpoises (Neophocaena asiaeorientalis asiaeorientalis)
Normalisation process of the sum of the scores of Welfare Alerting and Welfare Status indicators
The normalisation process is also applied to the sum of indicators within the WS and WA categories. As previously stated, this process is designed to normalise the sum of scores, ensuring that each indicator contributes equally to the final welfare assessment, regardless of the number of indicators within each domain. The following procedure will elucidate the transition from scoring the indicators in the YFP-WAP to determine the final scores for WS and WA. The first step involves determining the maximum scores for each domain across the four welfare components: welfare enhancement-status (WE-S), welfare enhancement-alerting (WE-A), welfare compromise-status (WC-S), and welfare compromise-alerting (WC-A). The maximum score refers to the highest possible sum of scores for all indicators included within a specific category (i.e. welfare status or welfare alerting) under the valence (WE or WC). In contrast, the cumulative score represents the total of the numerical values actually assigned to the indicators in that specific category (observed scores), based on the outcomes observed during the welfare assessment. These values reflect the animal’s WS or WA at the time of the evaluation.
In the current paper, for illustrative purposes, only WC-S and WE-S indicators within the domain D1-Nutrition are presented in Table 2. As shown, the maximum score for WE-S indicators is 12, whereas the WC-S maximum score ranges from 10/12 (this variation is due to the body condition scoring indicator, which includes two indicators - poor condition and obese condition- in the WC that are evaluated separately, and may yield different maximum scores depending on the scoring pathway followed). The corresponding cumulative scores reflect the actual values recorded during the assessment for each indicator (observed scores), as detailed in Table 2.
Table 2. Example of sum of the scores of Welfare Status (WS) indicators categorised under Welfare Enhancement (WE) and Welfare Compromise (WC) valences in Domain 1–Nutrition as part of the proposed standardised scoring system to assess the welfare of Yangtze finless porpoises (Neophocaena asiaeorientalis asiaeorientalis)
* WC-S for body condition scoring has two indicators (poor condition/obese condition) which are scored differently therefore giving a different final maximum score.
The maximum values for each category (12 for WE, and 10/12 for WC) are assigned the percentage value of 100%. Once this reference point is established, all lower scores can be converted into percentages using a proportional calculation. For example, referring to Table 2, if the Cumulative Score for Welfare Enhancement-Status (WE-S) indicator is 6, and the Category Maximum score for WE-S is 12 (100%), the associated percentage can be determined by dividing the cumulative score for that category by its maximum possible score, and then multiplying the result by one hundred. The cumulative score refers to the sum of all individual item scores within the category, while the maximum possible score is the highest total that could be achieved if all items in the category received full marks. Subsequently, Table 1 can be referenced to determine the percentage range within which 50% falls and identify the corresponding final non-numerical score. In this case, a Welfare Enhancement-Status score of 50% falls within the range of 34–67%, corresponding to a final score of ++. This process can be systematically applied across all categories to determine the final scores for all WS (Welfare Status) and WA (Welfare Alert) indicators for each valence and domain (the example for D1-Nutrition for WE-S and WC-S is shown in Figure 1). This proportional approach ensures that all welfare scores, whether WE or WC, can be consistently normalised into percentage values, facilitating cross-domain comparisons and final welfare assessments.
Figure 1. Decision trees illustrating all possible outcomes for (a) welfare enhancement status indicators (WE-S) and (b) welfare compromise status indicators (WC-S) for one hypothetical Yangtze finless porpoise (Neophocaena asiaeorientalis asiaeorientalis), referring to the example illustrated previously for the Domain 1-Nutrition. *Body Condition Scoring (BCS) in WC-S scores differently depending on the condition: poor condition with score 3, and obese condition with score 1. This leads to two different possible final maximum outcomes: 10 or 12.
In the context of welfare assessment for captive animals, the number of individuals being monitored may be limited; however, detailed scoring is crucial for promptly identifying potential issues. Therefore, additional aggregation of WS and WA scores for each domain is generally not required, as visualising both aspects separately can provide distinct and valuable insights for management decisions.
Aggregation of Welfare Alerting and Welfare Status scores within their valence
If further aggregation of the scores is required by the facility, WA and WS scores (within the same valence) can be combined into final unified WE and WC scores within each domain. This can be achieved using the system outlined in Table 3: the top row of the WC and WE tables contain scores for WA indicators, while the bottom row includes scores for WS indicators. The values within these tables are derived by summing the scores from both WA and WS indicators, using the numerical scale defined in Table 1. For instance, in the WC table, combining WC-S (welfare compromise-status) A (numerical score: 0) with WC-A (welfare compromise-alerting) A (numerical score: 0) results in a final score of 0. Similarly, combining WC-A B (numerical score: 1) with WC-S A (numerical score: 0) leads to a final score of 1. The process is very similar for the welfare enhancements status and alerting categories. Following this approach, the tables are completed with the assigned numerical values, with maximum values reaching 6 for both welfare compromise and enhancement scores (Table 3).
Table 3. Sums of scores for Welfare Status and Welfare Alerting indicators under the Welfare Enhancement (0, +, ++, +++) and Welfare Compromise (A, B, C, D) valences. Numerical values used in the calculations are shown in parentheses alongside the corresponding non-numerical WE and WC grades as part of the proposed standardised scoring system to assess the welfare of Yangtze finless porpoises (Neophocaena asiaeorientalis asiaeorientalis)
To convert the numerical values within the tables into percentages, the process outlined previously will be followed. Each table has a maximum numerical value of 6, which is assigned the percentage of 100%. The percentage associated with each score was calculated by dividing the obtained score by the maximum possible score (6) and multiplying the result by 100. For example, summing WC-A (0) and WC-S (1) gives a total score of 1. Dividing 1 by 6 and multiplying by 100 gives a result of ≈17% (Table 4).
Table 4. Calculation of percentages based on the sum of scores for Welfare Status and Welfare Alerting indicators under the Welfare Enhancement (0, +, ++, +++) and Welfare Compromise (A, B, C, D) valences as part of the proposed standardised scoring system to assess the welfare of Yangtze finless porpoises (Neophocaena asiaeorientalis asiaeorientalis)
Ultimately, the percentage ranges presented in Table 1 facilitate the conversion of each calculated percentage of the above tables into a final non-numerical score for both welfare compromise and welfare enhancement across each domain (Table 5).
Table 5. Conversion of the percentages of the Welfare Status and Welfare Alerting indicators under the Welfare Enhancement (0, +, ++, +++) and Welfare Compromise (A, B, C, D) valences as part of the proposed standardised scoring system to assess the welfare of Yangtze finless porpoises (Neophocaena asiaeorientalis asiaeorientalis)
Visualisation of the welfare scores
The Welfare Enhancement and Compromise scores for each Welfare Status and Welfare Alerting indicator can be visualised via a colour-coding system (Figure 2[a], [b]) which allows creation of a four domains plot.
Critical Scoring pre-assessment
In order to help the facilities to have a pre-screen of the welfare status of their animals prior to starting a full YFP-WAP assessment, a ‘Critical Scoring’ table was prepared. The ‘Critical Scoring’ table includes a total of 14 indicators, eleven of which are animal-based measures, and three resource-based covering the four domains (Table 6, and Section 1 of the Supplementary material). The indicators were selected based upon their higher impact (welfare compromise; WC) on the welfare of the porpoises (for detailed information on indicators’ pre-assigned scores, see Table S1 in the Supplementary material, and Platto et al. Reference Platto, Serres, Normando and Hao2025). If at least one indicator with a negative or absent condition is recorded in each domain during the pre-screening, it will indicate that the facility presents a ‘Critical Scoring’ situation that must be immediately addressed prior to any further assessment. The ‘Critical Scoring’ tool is not meant to be a substitute for the entire YFP-WAP assessment, but only to precede it in order to immediately address very critical situations that could cause irreversible damage to the animals.
Table 6. Critical Scoring checklist including a total of 14 indicators (11 animal-based measures, and 3 resource-based) covering the four domains. The indicators are selected for their highest negative impact (WC) on the welfare of the Yangtze Finless porpoises (Neophocaena asiaeorientalis asiaeorientalis). The ‘Scoring’ column includes the highest non- numerical and numerical (between parenthesis) scores for each listed indicator. The column ‘Check-list’ is reserved for the assessment of the listed indicators
* This behavior should not be assessed based on one single observation but within 24 hours.
Fictional YFP individual welfare assessment
A fictional YFP individual will be used as an example to illustrate how to use the YFP-WAF tool and how to score the welfare indicators. The date and location of the assessment were noted. The individual is an adult YFP female, approximately three months pregnant and sharing the pool with her one year old female calf. The adult female has optimal food intake and responds well to the presence of trainers as well as their commands. Most of her time was spent swimming alone in a counter-clockwise pattern interrupted by echelon swimming with her female calf. The trainers provide toys such as balls and swimming noodles (floating strands of foam) after each feeding time and the adult displays occasional play behaviour with her calf. The water is clean, and routine water analyses are carried out every day. The fish preparation room is cleaned after each feeding, and fish quality assessed every week via random fish sampling. The trainers have more than five years’ experience working with YFP and there is a well-established training programme in place to ensure variability regarding their daily interactions with the animals. Routine medical examinations, including ultrasound, take place once weekly to assess pregnancy status while blood tests are carried out every six months for all the animals.
Results
Scoring method for a fictional YFP individual welfare assessment
An initial pre-screening using the critical scoring table was conducted but did not reveal any critical situations. Consequently, based on the description of the fictional YFP individual provided above, a full assessment was carried out using the YFP-WAP, and the scores were entered into the FDM table to generate the final welfare scores (see Table S2; Supplementary material).
Calculation of the Welfare Enhancement and Compromise for the welfare category and welfare valence for the fictional YFP individual
The welfare enhancement scores for the YFP female were relatively high (> 40%) across all domains (D1, D2, D3, and D4) where indicators were applicable. Welfare Compromise scores were relatively low, not rising above 18%. For each domain, percentages for the welfare categories and valences were calculated (Table 7; percentage scores column). By using the conversion (Table 1), the percentage scores obtained are converted into non-numerical scores (final scores column), with the final results shown in Table 7.
Table 7. Percentages and final non-numerical scores for Welfare Enhancement -Status (WE-S), Welfare Enhancement-Alerting (WE-A), Welfare Compromise-Status(WC-S), and Welfare Compromise-Alerting (WC-A) indicators across each domain (D1: Nutrition; D2: Physical Environment; D3: Health; D4: Behavioural Interactions) as part of the proposed standardised scoring system to assess the welfare of Yangtze finless porpoises (Neophocaena asiaeorientalis asiaeorientalis)
The final scores are converted into a colour coding plot by using the colour coding observed in Figure 2, that more readily allows a visual summary of the final results of the YFP-WAP in a plot (Figure 3).
Figure 2. Colour coding scheme representing (a) non-numerical scores for Welfare Compromise (A, B, C, D) and Welfare Enhancement (0, +, ++, +++) and (b) final colour-coded plot showing indicators of Welfare Enhancement-Status (WE-S) and Welfare Enhancement-Alerting (WE-A), as well as Welfare Compromise-Status (WC-S) and Welfare Compromise-Alerting (WC-A), categorised under their respective valence scores (0, +, ++, +++; A, B, C, D) across the four domains of the YFP-WAF: D1-Nutrition, D2-Physical Environment, D3-Health, and D4-Behavioural Interactions.
Figure 3. Colour plot that shows the final scores of the YFP-WAP for the fictional animal. Each triangle represents one of the four domains (D1-Nutrition, D2-Physical Environment, D3-Health, and D4-Behavioural Interactions) and includes indicators of Welfare Enhancement-Status (WE-S) and Welfare Enhancement-Alerting (WE-A), as well as Welfare Compromise-Status (WC-S) and Welfare Compromise-Alerting (WC-A).
The current study does not require additional aggregation between Welfare Status and Alerting indicators for each valence. In the case of the facility requiring a further aggregation, the scores initially obtained and shown in Table 7, can be aggregated by using Tables 3–5 to obtain a unified score for WE and WC for each domain (Table 8). Domain 2-Physical Environment only has welfare alerting indicators for each valence therefore no further aggregation is possible for this domain as the scores remain the same. By following the same colour coding shown in Figure 2 a final plot can be obtained (Figure 4).
Table 8. Unified scores (columns on the right in white) for Welfare Compromise (WC) and Welfare Enhancement (WE) across each domain (D1: Nutrition; D2: Physical Environment; D3: Health; D4: Behavioural Interactions) as part of the proposed standardised scoring system to assess the welfare of Yangtze finless porpoises (Neophocaena asiaeorientalis asiaeorientalis). The Domain 2 - Physical Environment only has welfare alerting indices, and therefore no further aggregation is possible
Figure 4. Colour plot that shows the aggregated scores of the YFP-WAP for the fictional animal. Each triangle represents one of the four domains (D1-Nutrition, D2-Physical Environment, D3-Health, and D4-Behavioural Interactions) and includes indicators of Welfare Enhancement (WE) and Welfare Compromise (WC).
Welfare assessment scoring interpretation of the fictional YFP individual
The welfare assessment conducted on January 25, 2023, on a pregnant and lactating adult, female YFP indicated the following welfare enhancement scores: +++ for D1-Nutrition, D3-Health, and D4-Behavioural Interactions, and ++ for D2-Physical Environment. Conversely, for Welfare Compromise (WC), three domains (D1, D2, and D4) received a B score, while D3-Health was assigned a C. Overall, this individual is experiencing a relatively good welfare state. Her reproductive status, notably pregnancy and lactation, might have influenced the individual’s welfare with slight alterations to her nutritional habits creating a degree of discomfort, thereby influencing the WC in all domains. In addition, the D4 for the WC indicators shows lack of opportunity for choice as well as the presence of directional swimming, both of which have had an influence on the final score for this valence. The presence of these two indicators within the WC valence underscores an urgent need for intervention by the facility, via the introduction of specific apparatus that enables porpoises to exert agency. This might also reduce the presence of stereotypical behaviours, as has been observed in other facilities where dolphins were given the ‘opportunity of choice’ (i.e. the opportunity to select the pool, opt for isolation from other individuals, decide whether to engage in training sessions, and choose whether to play or interact with environmental enrichment devices; Markowitz & Aday Reference Markowitz, Aday, Sheperdson, Mellen and Hutchins1998).
Discussion
Scoring system
The FDM is a practical framework for developing animal welfare assessment tools based on Welfare Compromise (WC) and Enhancement (WE). It highlights welfare issues and risks that may require further investigation (Sherwen et al. Reference Sherwen, Hemsworth, Beausoleil, Embury and Mellor2018). In this study, a standardised scoring system was developed for the YFP-WAP tool, based on the FDM, to assess the welfare of YFP and provide a visual representation of welfare outcomes to support decision-making in managed facilities (Malkani et al. Reference Malkani, Paramasivam and Wolfensohn2022). To accurately determine the relative contribution of each indicator to its respective domain is highly challenging, making the overall welfare assessment complex (Mellor et al. Reference Mellor, Beausoleil, Littlewood, McLean, McGreevy, Jones and Wilkins2020). Until further information becomes available, in the current protocol all the indicators are weighted equally in the scoring process, with weighting linked only to the intensity of their impact on welfare. Since the importance of many indicators has not been fully established, and welfare-status indicators are deemed to provide more direct information regarding an individual’s welfare compared to welfare-alerting indicators (Botreau et al. Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissier2007a,Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissierb; Hampton et al. Reference Hampton, Hemsworth, Hemsworth, Hyndman and Sandøe2023), these two categories were scored separately in order to minimise biases in welfare interpretation. Additionally, because WE and WC are also scored separately, this protocol assesses how different aspects of the four domains contribute positively or negatively to overall welfare. However, it does not determine how these inputs interact to produce an overall net welfare state (Hampton et al. Reference Hampton, Hemsworth, Hemsworth, Hyndman and Sandøe2023). This approach ensures transparency and prevents compensation between domains, accurately reflecting welfare improvements or deterioration, and avoiding misrepresentation (Botreau et al. Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissier2007a,Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissierb). Aggregation could mask specific domain issues. For example, a high D4-Behavioural Interactions domain score might compensate for a lower D2-Physical Environment domain score, potentially under-representing critical welfare concerns. Moreover, to enable comparability across domains, and correct for differences in the number of indicators per domain, a normalisation process was applied prior to visualisation.
Although the system is designed to be comprehensive, careful interpretation of scores is still necessary, taking domain-specific factors into account. Over-reliance upon numerical scores without understanding the underlying conditions may lead to misinterpretation. To maintain scientific rigour in animal welfare assessment, it is essential that objective and transparent methods are deployed to empirically investigate how animals are affected by human interactions and management. This principle guided the development of the YFP-WAP scoring system, ensuring clarity in the assessment and evaluation for future users (Sandøe et al. Reference Sandøe, Corr, Lund and Forkman2019; Hampton et al. Reference Hampton, Hemsworth, Hemsworth, Hyndman and Sandøe2023).
Fatal scoring pre-assessment
In order for successful implementation of the YFP-WAP tool, prior preparation is a key stage when it is used for the first time. Therefore, a first screening of the porpoise individuals would be more appropriate by using a ‘Critical Scoring’ system to facilitate practicability. In fact, before embarking on the full YFP-WAP assessment, it is crucial that a preliminary evaluation is implemented to identify critical welfare issues that require immediate attention and intervention, thereby preventing potentially fatal outcomes. The ‘Critical Scoring’ system is not a substitute for the full YFP-WAP assessment but has the primary objective of proactively identifying and addressing life-threatening conditions. The indicators included in the ‘Critical Scoring’ were selected because of their crucial negative impact on porpoises. Situations including unacceptable water quality, or the presence of a number of skin conditions represent direct threats to the animal’s overall health and immediate corrective actions should be taken to alleviate these critical issues, preventing escalation to a fatal conclusion.
The ‘Critical Scoring’ system allows for a more efficient resource allocation. Specifically, caretakers and managers can prioritise efforts and resources towards resolving the issues that pose the greatest risk to animals’ well-being, by pinpointing the most pressing welfare concerns. This prioritisation is crucial in those settings that have limited resources, where addressing the most critical concerns first ensures that interventions have the greatest possible impact on improving animal welfare. However, a positive evaluation using the ‘Critical Score’ system does not equate to the animal being in an acceptable welfare situation, so the YFP-WAP should also always be performed after a successful ‘Critical Score’ evaluation, as the latter cannot operate as a substitute for the former.
Animal welfare implications
The development of a standardised scoring system for assessing the welfare of YFP under human care represents a significant advancement in cetacean welfare science. This protocol provides a systematic and quantifiable approach to ex situ management, ensuring welfare considerations are central to husbandry, veterinary care, and facility design. The standardised welfare assessment can improve the overall management of cetaceans under human care by enabling objective evaluation of physical and behavioural needs, supporting individualised care plans, and enhancing species-specific enrichment programmes. Additionally, it facilitates data-driven adjustments to management strategies, optimising living conditions and animals’ overall well-being. While the current system offers a robust welfare assessment protocol, ongoing refinements are essential. For example, incorporating AI-driven monitoring, and real-time behavioural and physiological data collection can enhance accuracy and predictive capabilities. Although designed for the YFP, this protocol can be adapted for other Neophocaena spp by modifying indicators to account for environmental conditions and species-specific needs. Furthermore, it also serves as a framework that, with further adaptations, can be applied to other cetacean species under human care. By embedding welfare assessments into facility operations, this protocol ensures a scientifically sound and ethically responsible approach to the care of YFP. Future advancements should focus upon technological integration and interdisciplinary collaboration to further enhance welfare monitoring and management strategies.
Challenges and limitations
It is important to acknowledge the challenges and limitations of the approach used in the present study to develop a scoring system for the YFP-WAP. The knowledge of the compensation processes and relative weight of indicators and domains is currently limited. Therefore, it was decided to keep the scoring neutral and to treat all aspects equally. However, welfare may actually be more complex, and this scoring only reflects a specific vision of it, and not the real biological processes that create welfare states (Botreau et al. Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissier2007a,Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissierb). In addition, welfare-alerting indicators, which only highlight a risk for welfare, represent the majority of the indicators within the YFP-WAP. Therefore, caution should be exercised in interpreting the final outcome of the assessment. Future studies may contribute to help validate more animal-based indicators that could also be included in the tool.
Conclusion
The development and implementation of the YFP-WAP tool marks a significant step forward in the assessment of YFP welfare under human care. By adopting the FDM as its foundation, the YFP-WAP provides a structured, transparent, and systematic protocol that evaluates both welfare compromise and enhancement (Botreau et al. Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissier2007a,Reference Botreau, Bonde, Butterworth, Perny, Bracke, Capdeville and Veissierb; Mellor et al. Reference Mellor, Beausoleil, Littlewood, McLean, McGreevy, Jones and Wilkins2020). This approach ensures a balanced and comprehensive understanding of how various indicators influence the well-being of these animals. Furthermore, the decision not to aggregate scores across domains prevents the potential masking of critical welfare issues, ensuring that domain-specific challenges are accurately reflected (Sandøe et al. Reference Sandøe, Corr, Lund and Forkman2019). This transparency strengthens the tool’s utility in identifying areas requiring targeted intervention and supports informed decision-making processes for animal welfare management. The inclusion of a ‘Critical Scoring’ pre-assessment system enhances the practicality and effectiveness of the YFP-WAP by identifying and addressing life-threatening welfare concerns prior to the full assessment. By focusing on indicators with the most significant negative impact, the ‘Critical Scoring’ enables efficient resource allocation and prioritises urgent interventions, particularly in resource-limited settings. This ensures timely actions to mitigate risks and improve welfare outcomes. Despite its strengths, the YFP-WAP tool faces challenges and limitations that must be addressed in future research and refinement. The current lack of knowledge regarding the relative weight and interrelationships of indicators and domains necessitates a neutral scoring approach. While the current scoring system ensures fairness and consistency, it may oversimplify the complexities of animal welfare and the biological processes that underpin it. Consequently, users must interpret the results with caution, recognising that the scoring system reflects an operational framework rather than a definitive representation of an animal’s welfare state (Hampton et al. Reference Hampton, Hemsworth, Hemsworth, Hyndman and Sandøe2023). In conclusion, the YFP-WAP tool represents a critical advancement in cetacean welfare assessment, combining scientific rigour with practical applicability. While challenges remain, the tool’s emphasis on transparency, specificity, and prioritisation provides a solid foundation for improving the welfare of YFP. Ongoing refinement and validation efforts will be essential to further enhance its effectiveness and ensure it continues to serve as a valuable resource for animal welfare practitioners and decision-makers.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/awf.2025.10040.
Acknowledgements
We would like to thank the training staff of the Yangtze Cetacean Breeding and Research Centre (YCBRC) for their support during the development of the framework. This project was funded by the 武汉市知识创新专项项目 (Wuhan Science and Technology Bureau Special Project), under project number 2022022101015012.
Competing interests
None.
Open access
