Killing Them ‘Softly’ (!): Exploring Work Experiences in Care-Based Animal Dirty Work

Working with animals is a daily occurrence for millions of people who often complete tasks which are tainted, in spite of the work being seen as essential in modern society. Animal shelter-work is such an occupation. This article contributes to a deeper understanding of the caring–killing paradox (a dissonance that workers face when killing animals they are also caring for), through an insider ethnographic study. We find that care-based animal dirty work consists of unique ambiguities and tensions related to powerlessness, deception and secrecy in the work based on a ‘processing-plant’ framework which informs how workers deal with unwanted animals. We find competing ideologies of care and control to be foundational in this work.


Introduction
Engaging with animals at work is a daily occurrence for millions of human workers across a range of occupations and organisational contexts (Hamilton and Taylor, 2013), yet our understanding of the field of human-animal work (HAW) is very much in its infancy (Hannah and Robertson, 2017). Traditionally, animals have been excluded from management and organisation studies (Connolly and Cullen, 2018;Labatut et al., 2016;O'Doherty, 2016), with limited focus on how animals in the workplace affect workers and organisations. Across society, work with animals is considered 'dirty work', but there is limited understanding of how dirty work which includes nonhumans differs from pure human contexts.
A unique aspect to animal-related dirty work is that in some situations animals are killed and, for many workers, killing is a central work-task 'sanitised' through organisational processes and societal cultural norms. In a recent quantitative study on slaughterhouse work, Baran et al. (2016) found the intentional, routinised killing of animals negatively affects worker wellbeing and health, beyond levels of prestige and 'dirtiness' of the occupation, suggesting that animal killing is unique within dirty work. However, the slaughterhouse context differs from some other human-animal interactions, as animal killing is a main work-task, while in other occupations this is one of many other tasks and as such may be hidden. One such occupation is animal shelter-work. Studies (e.g. Baran et al., 2012;Reeve et al., 2006;Rogelberg et al., 2007) show that killing is problematic for shelter-workers in spite of this not being the main work-task. Here, there is an emotional toll of first caring and then killing some of the animals: a caring-killing paradox. Previous research suggests this is a 'moderate' and 'manageable' tension for experienced workers, while being more intense for newcomers (Arluke, 1994). In our research, we present a more nuanced case study that suggests killing animals is highly problematic for most individuals. This paradox is largely hidden to others (not involved in the killing), and rather than being a 'moderate' experience, workers who undertake this task of supporting organisational processes and societal 'processing' of animals experience significant dissonance.
Ethnographic methods have long been used as a method to conduct animal shelter studies (e.g. Arluke, 1994;Hamilton and Taylor, 2013;Irvine, 2003), but these have not been based on a full insider position in a kill-shelter. As Chavez (2008: 479) notes, advantages to insider status include the ability to understand nuances of the field, access to activities and data interpretation/representation including insights 'into the linguistic, cognitive, emotional, sensory and psychological principles of participants', as well as detection of participants' hidden behaviours and perceptions. As killing is a highly tainted task, its effects on wellbeing could easily be downplayed to outsiders, leading to an incomplete understanding of the work and implications for workers as well as the animals. Our purpose in this article is to address this limited understanding of HAW by questioning: what is the experience of the caring-killing paradox and what are the reasons behind it? We offer deeper insights into this phenomenon based on our unique insider access to this experience. We find that the killing experience and the process involved implies a moral ambiguity in HAW, reinforced by a largely unexamined and undisputed 'socially-sanctioned killing' (Göring, 2014) happening in these largely hidden workspaces some frame as 'caring'-organisations. We suggest the caring-killing paradox is a result of ideologies of care being at odds with rationalised systems of control.
The article is structured as follows: first, we discuss animal dirty work exploring instrumental and care-based framings; second, we introduce our shelter context and ethnographic methods; third, we illustrate the lived experience of the caring-killing paradox with elements of dissonance; and lastly, we discuss these findings for a deeper understanding of animal work and its governing 'processing-plant' paradigm.

Animal dirty work
HAW is a largely 'understudied domain of human activity' and can be defined as 'human work that is substantially focused on live non-human animals' (Hannah and Robertson, 2017: 116). These scholars estimate that hundreds of millions of humans work within HAW and the economic impact of animal-related industries is larger than many nations' GDP. 1 However, the impact of these industries extends beyond their massive economic impact to involve significant employee health and wellbeing implications (e.g. Baran et al., 2016;Schabram and Maitlis, 2017). There is an incomplete understanding of this as animals have traditionally been neglected in work studies, although this anthropocentric limitation is slowly shifting as organisational scholars (e.g. Connolly and Cullen, 2018;Labatut et al., 2016;O'Doherty, 2016) increasingly acknowledge animals and their impact in the world of work.
Understanding has further remained incomplete most likely because much animal work is socially constructed as stigmatised 'dirty work' with layers of taint affecting identities. Research indicates that employees engage in dirty work for both intrinsic (e.g. personally rewarding) and extrinsic (e.g. financial) reasons (Ashforth and Kreiner, 2002;Sanders, 2010). While work with animals can be highly tainted, many choose animal care-work due to an affection for animals. In fact, research in zoo-keeping (Bunderson and Thompson, 2009) and animal shelter-work (Schabram and Maitlis, 2017) reveals intriguing insights of work callings, psychological contracts and why individuals choose the psychologically and physically challenging work. Working with animals is highly emotionalised and slaughterhouse studies have pointed to aspects of detachment and compartmentalising emotions (Hamilton and McCabe, 2016), as well as repressing and denying emotions to uphold masculine 'animals as food' categorisations (McLoughlin, 2019: 332). Hence, previous studies offer initial insights into emotions of animal work, yet as they are based on outsider researcher positionings, the nuances of dissonance and paradox in the work may be limited.
A central tenet to many HAW occupations is that this work is often hidden from 'sanitised' society, considered 'dirty' in some way. Dirty work includes tasks that have a physical, social, moral (Ashforth and Kreiner, 1999;Hughes, 1958) and/or emotional taint (McMurray and Ward, 2014;Rivera, 2015). This contributes to workers being stigmatised by wider society but can also include the workers identifying themselves as 'dirty' (Ashforth and Kreiner, 2002;Baran et al., 2012). Dirty work has been extensively examined with empirical findings advancing understandings of identity, socialisation, coping and emotions in the workplace. However, we suggest animal dirty work is a unique work experience differing from pure human dirty work as work with animals involves a different set of moral norms (including, at times, killing) -treating animals in ways that would not be sanctioned for a human population. Hence, anthropocentric perspectives have impacted understanding of such work, neglecting to account for the true impacts on workers of objectifying and killing animals. Many studies have focused on slaughterhouse work, but this gives only a partial understanding of emotions in animal dirty work due to the high focus on killing and clear instrumental framing of animals in the work. In contrast, care-based animal dirty work seems more ambiguous. Hence, we briefly present these two aspects of animal dirty work to theoretically contextualise our study.

Instrumental animal dirty work
Slaughterhouses are animal dirty work contexts that rely on an instrumental construction of live animals at work. Organisation and sociology scholars have long used slaughterhouses to uncover aspects related to work identity, coping and emotions (e.g. Ackroyd and Crowdy, 1990;Baran et al., 2016;Hamilton and McCabe, 2016). In slaughterhouses, animals are constructed as 'commodities', 'input' and 'raw materials' to be processed, similar to manufactured goods in 'efficient' ways (Ackroyd and Crowdy, 1990). The identity and occupational culture of the 'highly routinised' and 'extremely dirty work' slaughterhouse workers perform is shaped both by dominant surrounding societal values as well as by internal organisational values (Ackroyd and Crowdy, 1990: 4). Historically, this routinisation of mass animal killing had significant influences on work processes beyond this specific context. For example, Patterson (2002) notes that Henry Ford's car assembly-line process was inspired by the Chicago meatpacking industry's 'dis-assembly line' of turning live animals into dead meat for consumption. Market-driven values of profit, efficiency and routinisation feed the instrumental construction of animals in such industries, but often with societal suppression of acknowledging animal death as part of the production process.
Throughout most animal dirty work, researchers have suggested that emotional management, coping and ambiguity are key in the work. Dependent on speciesist food culture and companion animal habits, certain animals are seen as 'food' with a greater degree of instrumental value, while others as companion animal with greater intrinsic value. Such cultural rationalisations and categorisations are used by, for example, slaughterhouse meat inspectors to justify killing chickens, but not horses, through emotional subjectivity (Hamilton and McCabe, 2016). Research, however, reveals that killing animals negatively affects workers' health in spite of such rationalisations (Baran et al., 2016). This suggests these typical rationalisations may not be fully internalised by workers.

Care-based animal dirty work
Unlike the slaughterhouse context, animal shelters are usually seen as ideological and caring, focused on intrinsic valuations of animals rather than profit-driven meat work. Shelter-work includes a duty of care of animals with workers reporting strong animal affections, often seeing the work as a calling (Schabram and Maitlis, 2017). Animal shelter-work relies foremost on ideologies of 'saving' animals, yet much of the work relies on organisational processes based on control and efficiency. Such processes mean that many animals are managed through 'euthanasia' (in kill-shelters) with the selection criteria of identifying those deemed less attractive and hence not adoptable. However, 'euthanasia' implies terminating life with consent based on compassionate reasons, such as ill health. With animals there is no consent given. Furthermore, many shelter euthanasia decisions are based on limited resources (financial and physical), rather than the animals' health, making this word inaccurate in spite of its frequent use.
Care-based animal dirty work demands a great deal from the individual worker who feels compelled to care yet is often emotionally challenged due to organisational resource limitations (Arluke, 1994). Workers face generalised dirty work stigma from society at large for doing work others find difficult or unattractive (Ashforth and Kreiner, 1999), but much of the moral taint in animal shelter-work is largely hidden from public scrutiny through marketing positive stories of adoptions, rather than focusing on those killed or the processes involved. This sanitised image of animal shelter-work may also impact employee expectations, as altruistic values often drive work motives, but these frequently do not match experiences, thus leading to high turnover. Tallberg et al. (2014) suggest animal shelter-work includes different, but often layered, coping-identities to make sense of the work, suggesting within-group differences to manage negative emotions. Furthermore, Schabram and Maitlis (2017) suggest being 'animal-motivated' creates negative worker wellbeing (an 'unhealthy' calling leading to burnout), while a focus on work-tasks and career development (rather than the animals) creates 'healthy' calling paths. As most animal shelter-workers identify as 'animal lovers', refocusing on individual and rationalised benefits, rather than the animals, may not be authentic nor necessarily benefiting those in 'care' -the animals.
Veterinary work is often seen as another care-based animal occupation that involves ambiguities, such as 'convenience euthanasia', creating moral stress (Rollin, 2011) or perpetration-induced traumatic stress (PITS) (Bennett and Rohlf, 2005). Convenience euthanasia resembles kill-shelters' use of 'euthanasia' yet consent here is only given by 'owners' rather than the animals themselves. Sanders (2010) notes veterinary technicians normalise 'euthanasia' by refocusing on animals saved over those killed, while Hamilton (2007: 495) suggests within-group occupational differences where farm veterinarians use physical dirt ('muck and blood') as 'badge[s] of courage for physical masculine heroism', referring to small animal veterinarians as 'dandies'. Similarly, Clarke and Knights (2019: 269) show 'anthropocentric masculinities' dominating veterinary work, requiring veterinarians to strictly control animal-bodies (and their own) for professional gains to satisfy commercial demands of human 'owners'. They conclude veterinary work appears zoocentric (animal-focused), but actually enforce human domination over animals through anthropocentric, masculine occupational narratives from their training and work identity. Such a rationalised focus may suggest a buffer in dissonance, but the high prevalence of moral stress in veterinary work suggests that internally this may be more complex. Ethnographic veterinary research is limited (Clarke and Knights, 2018), while full internal research positionings, such as ours, is unheard of in veterinary studies (as well as in shelter-work).

Context and methods
Ethnographic materials were collected as part of the first author's Doctoral research working as a paid frontline animal shelter-worker for 10 months. She worked on average five days a week in the case-shelter and our study includes participant observations (e.g. diary entries, reflective notes, pictures) and secondary sources (e.g. organisational staff briefings, training materials) and 20 semi-structured shelter-employee interviews. The first author initially worked as a volunteer, then completed an organisational internship shadowing employees across all departments, thus gaining insights across the organisation. However, these yielded an incomplete understanding of the embodied emotions of the 'dirtiest' tasks shielded from those not performing them. Therefore, she was positioned as a full insider, obtaining a paid role as a frontline animal shelter-worker, hence establishing rapport with study participants (Ross, 2017) and gaining access to the personal experience of the work, including activities shielded from outsiders. Insider experiences are valuable in generating deeper knowledge (Chavez, 2008) and participatory designs key in conducting ethical research (Heron and Reason, 1997), an aspect we are mindful of when studying hidden phenomena. According to Flyvbjerg (2011), paradigmatic case studies (such as ours), are useful for learning deeper about phenomena with the purpose of creating metaphors to illustrate general societal characteristics as well as developing a more 'nuanced view of reality' and better understanding of human behaviour (p. 303). Thus, our case suggests a 'processing-plant' metaphor to understand how unwanted animals and frontline workers are organised to meet organisational (and societal) objectives.
Our case-shelter was run by a large animal welfare organisation taking in 20,000 animals annually, predominantly those considered 'companion animals', but also farm animals and native wildlife. There was high turnover among the frontline shelter-team consistent with other studies (Rogelberg et al., 2007) but, on average, there were 10 paid shelter-workers working on a roster to manage the animals and volunteers. The first author fitted the demographics of the workforce, which, in line with other animal shelter studies (Baran et al., 2012;Schabram and Maitlis, 2017), consisted mainly of younger women working on the frontline while management consisted predominantly of older men (Table 1).
Our research design was approved by a university ethical committee and the caseorganisation who gave written consent for this study. Interview participants were recruited through an email on a voluntary basis and each interview started by explaining the research, confirming anonymity (we use pseudonyms) and freedom of withdrawing from the study. The interviews lasted between 30 minutes and two hours, were audiorecorded and transcribed.
For both the interviews and the diary entries, a manual thematic analysis was used for data analysis (Glaser and Strauss, 1967), looking for patterns and recurring themes across all data. Key themes emerged such as power(lessness), secrecy, conflict, negative/difficult emotions, care/killing. Based on the participatory observational data, self-reflexive autoethnographic 'evocative' narratives were constructed to invite others into the subjective world (Ellis and Bochner, 2000) and lived experience of shelter-work. The narratives were constructed from text in the first author's field-diary (which had undergone thematic analysis) and retroactively pieced together through a self-reflexive lens to offer the evocative, embodied experience of the work (Tracy, 2013). To further enhance qualitative credibility and multivocality, the narratives are weaved alongside interviews to illustrate the shared lived experience (Tracy, 2013).

Findings: The lived experience of the caring-killing paradox
Animal shelter-work is characterised by low pay, poor working conditions, long hours and often dangerous work from dealing with the unwanted and often traumatised animals. The main work-tasks include animal husbandry duties (i.e. feeding, cleaning and medicating animals) and 'processing' animals to manage the shelter population and keep the charity running 'smoothly'. New shelter animals arrived as private surrenders (by owners who no longer wanted them), as strays (homeless animals), or by official inspector seizures (due to owner neglect or abuse). The latter group remained in protective shelter-custody awaiting legal prosecutions -an often lengthy and expensive process. A controversial aspect of shelter-work (in kill-shelters such as our case) is killing healthy animals based on organisational spatial-fiscal constraints and preferences. Here, it is the animal shelter-workers who predominantly assess and kill the animals in addition to shelter veterinarians. We start by illustrating the lived experience of killing shelter animals through an autoethnographic narrative: The dog stiffened as we approached the kill-room. The narrative illustrates the conflicting emotions of the requirements in the job (killing animals) being at odds with wanting to help (and save) animals. Noticing animal distress and fear was a normal work experience leading to dissonance, which Baran et al. (2012) refer to as an emotional toll. But where does dissonance come from? Is this just a reluctance to kill? Our data reveal a more complex situation.

Secrecy, deception and normalisation of killing contributing to dissonance
Killing was the most down-played work-task to those outside the animal shelter. Within the organisation, those not directly involved in the killing, reported that they 'didn't want to know' and 'trusted this was done according to values of the organisation' (Lucy, administrator). As the narrative above reveals, this was not always the case. Secrecy of the process extended outside the shelter to the wider society, in this way attempting to protect against the 'stickiness' (Bergman and Chalkley, 2007) of killing which would increase the work's moral taint if exposed: Interviewees reported actively contributing to this secrecy, as the 'public didn't need to know [details]' (Matt, animal shelter-worker). Rationalisations included keeping volunteers and financial donations, which might decrease if realities were revealed. In this way, workers justified concealment, thus supporting and maintaining processes by not publicly questioning organisational methods. Deception and silencing surrounding killing were further evidenced through administrative tactics whereby some owners surrendering their animals would be persuaded to 'euthanise', rather than put the animals through the shelter system for a chance to be rehomed. Animals killed based on an owner's decision did not go on the official shelter euthanasia statistics, as the animal was not yet legally shelter 'property'. Such deception created further worker dissonance, leading to them questioning their work, as Jill explains: [A senior manager] asked customer services to book more animals in as 'TBE' [To Be Euthanised] rather than 'give them a go' because it looks better for our figures . . . that's disgusting and not why I work [here]. (Jill, animal shelter-worker) Such rationalised organisational methods supported secrecy of what happens to unwanted animals, while workers faced the emotional strain from seeing and actively enforcing such processes, whereby negative emotions were suppressed as a 'private problem' to get the job done. Consistent with general dirty work findings (Ashforth and Kreiner, 2002), socialisation and normalisation of shelter processes either conform the employee (to stay and adapt) or cause them to leave the shelter. A break to this 'norm' was in 'unsocialised' new staff who were often openly emotional, questioning processes, confirming previous studies (Arluke, 1994) that negative emotions are most intense for shelter newcomers. Seasoned workers recognised emotional regulation as central to the work: 'if we show emotions, we'll be deemed weak, that we can't handle it and shouldn't be working here' (Jill, animal shelter-worker). Yet, even those who adapted and normalised killing as something that was beyond their control, this still remained the most controversial and tainted work-task across our data. This suggests that those who stayed adapted by regulating their emotional displays to fit the social environment and organisational culture, not necessarily that their emotions were less intense, as in Jill's comment above. Surface acting, such as this, often leads to emotional labour (Hochschild, 1983), another tension impacting dissonance and worker wellbeing, but these aspects may be difficult to recognise externally. Furthermore, as adoptions were handled by customer services rather than frontline animal shelter-workers, the ideological framings buffering some taint for veterinary staff by highlighting those saved above those killed (Sanders, 2010) was not available to these workers. Organisational processes for distribution of tasks meant there was limited access to positive tasks within the work to allow for buffering.
To cope with these predominant negative tasks, our study revealed excessive drinking and substance abuse to be normal. Similar negative coping mechanisms are found in other animal dirty work, such as slaughterhouse-work (Baran et al., 2016), but it is worth noting that shelter-work includes an additional emotional toll of a positive rationale for doing the work (to help animals). To manage such dissonance, some shelter-workers constructed killing through a care framework, thus buffering some taint: 'at least the animals have someone who cares at the end' (Ruth, animal shelter-worker). Such rationalisations were encouraged to normalise the 'socially undesirable emotions' and reframe perceptions of killing, making the 'extraordinary' seem 'ordinary' (Ashforth and Kreiner, 2002: 217). This care-based sensemaking technique might limit some objections, but equally served to support the 'machine'-like processing framework by shifting focus from the act of killing to one around care and responsibility. Thus, difficult emotions were somewhat diffused and killing normalised.

Powerlessness contributing to dissonance in the animal 'processing-plant'
Although it was the animal shelter-workers who made many of the life and death decisions, these were controlled by organisational spatial-fiscal considerations and management. The shelter had an 'open-door' policy, taking in all animals in need, in contrast to 'closed-door' policies of some 'no-kill' shelters. This meant large seasonal intake differences (with more surrendered animals around holidays), or before puppy-mill raids, and as spaces became limited those already in the shelter faced 'culling' to make room for new occupants. There were no explicit organisational guidelines of how many or who to kill, although behavioural assessments along with breed, size, age, health and even colour were used to control and limit those available for adoption (which had limited spaces). Behavioural tests have been critiqued as flawed with limited scientific reliability (Mornement et al., 2010) and the first author (who was involved in conducting such assessments) witnessed subjective tester biases in addition to the challenges individual animals faced at being behaviourally rated while surrounded by loud sounds, strong smells and the sight of other distressed animals. As a result, conflict was common among the workers regarding killing-decisions, creating a context high in interpersonal tension, and workers often campaigned to save specific animals. The following illustrates this normal work experience: In contrast to previous studies (Arluke, 1994), our findings suggest the organisational killing-routines presented even seasoned shelter-workers with emotional turmoil. Lauren, who had 27 years of shelter-work experience, reflected on the process: Animals here, they're numbers, they're nothin' but numbers . . . an eight-year-old German Shepherd that . . . we can't rehome . . . Crazy! He's not done anything wrong. When I put him to sleep, he didn't do anything wrong, he didn't growl, nothing. (Lauren, animal shelter-worker) Lauren buffers against moral taint through downplaying the act of killing, instead referring to putting the dog 'to sleep'. Such perceptual reframing not only buffers from taint, but also regulates emotions by normalising the 'extraordinary' act (Ashforth and Kreiner, 2002). The powerlessness of affecting change to killing processes resulted in constant tensions around decisions. Similarly, Cassie (a seasoned animal shelter-worker, but new to this shelter), noted: 'We [assessed] a puppy that passed but they [her colleagues] said it was untrustworthy . . . It was a five-month-old puppy [cries] and I'm just going "this is crap!".' An animal would have different carers depending on work-rosters and assessments were done on average within three days of the animal's shelter-arrival. Certain workers were more ready to kill than others, creating interpersonal conflicts as emotionality surrounding killing was high. As a result, many shelter-workers who resisted quick assessments left the organisation (such as Cassie soon did), as they felt powerless in executing the societal and organisational 'pet management' processes, which to many simply felt wrong.
Powerlessness was felt both through their inability to help the animals, but also in relation to external societal 'pet' trends. The public were seen as the main offenders, framed as 'ignorant' and 'irresponsible' in their consumption and breeding behaviours. This public hostility was intensely felt at the frontline level seeing 'people [as] the problem, not animals' (Mary, animal shelter-worker), framing themselves as 'true' animal lovers by taking responsibility for societal limitations. These diffusing and reframing techniques are ways to assume control of the emotionally painful situation workers faced (Ashforth and Kreiner, 2002), buffering somewhat against moral taint, with some more adapted to the narrative of giving the animals a 'last gift' (in death). This is typical of 'successfully' coping with the caring-killing paradox (Arluke, 1994: 148): There is a terrible paradox in what you will have to do -you will want to care for the animals, but you will have to kill some of them. It is a painful process of killing animals when you don't want to. It seems so bad, but we'll make it good in your head. You will find yourself in a complex emotional state. Euthanising is not just technical skills. You have to believe it is right to make it matter of fact.
This narrative attempts to assume emotional and cognitive control and reframe taint by making it 'right' and 'good in your head' (similar to Ruth's comments earlier). Controlling workers' paradoxical thoughts and feelings was a way to enforce organisational ideals of 'good' animal shelter-work and constructed as 'line-workers' focused on hierarchical, linear processes rather than the animals: I find the best employees here are the ones that have done some normal work outside, like worked for McDonald's [and] realise, 'Hey, you do as you're told, you get on with it, you follow the procedures, and you don't make up your own mind [about how to do the work and whom to save]'. (Aaron, senior manager) Such perspectives were frequently voiced in our management interviews, who found employee emotionality unprofessional and unconducive to the 'big picture'. Another senior manager explained how workers were free to quit and leave if they could not handle the key task of animal population control. Hence, the charity's rationalised business model suppressed worker emotions while simultaneously relying on the public's emotional responses for donations to ensure organisational functioning. In this way, the caring-killing paradox was approached by many managers as the individual's own responsibility in accepting and continuing with the job. Yet, others acknowledged that workers were ill-equipped to cope with the demands of the work as recruitment was based on their affection and pro-social motivation to help animals (seen as conducive to organisational functioning). For example, many workers stayed over-time without compensation or came to work on days off (as volunteers) due to their commitment to care for the animals. Thus, the workers' intrinsic motivations were used to support the financial constraints of the charity, reflecting the 'double-edged sword' of callings in animal dirty work (Bunderson and Thompson, 2009).
A hierarchical organisational structure also competed with the care-ideology of the charity. Workers felt alone in helping animals, seeing management as controlling their feelings through processes: 'management [is] up there and we're just the shit-pickers, we're nothing' (Ruth, animal shelter-worker). Management saw efficiency and protocol as key in running the organisation, admitting that there was 'a lot of mistrust . . . people complain that it was better before . . . it was more flexible, now there are rules . . . it's just like any business' (Aaron, senior manager). However, frontline workers saw this model conflicting with the organisation's core mission of care: 'we're a charity, not a business' (Jill, animal shelter-worker). Hence, care-based animal dirty work seems to include similar challenges to other animal work, instrumentalising both workers and animals in processes of strict control and efficiency. Thus, the anthropocentric masculinities of veterinary work (Clarke and Knights, 2019) similarly dominate shelter-work as ideologies of control prevail over ideologies of care, both for the human shelter population (the frontline workers) as well as for the nonhuman animal shelter population.

Discussion
Animal shelter-work is an example of care-based animal dirty work ascribing to ideologies of care, but in practice this can involve a 'processing'-mentality which negatively impacts worker wellbeing and intrinsic motivations for doing the work. According to Ashforth et al. (2017), taint can be limited in dirty work contexts through the alignment of worker affinity to the organisational cause. This was not an issue in our study as the workers felt closely aligned to the core organisational mission. As Rothschild and Milofsky (2006) suggest, in establishing not-for-profit sector values, the passions and ethics of its workers are important. In the shelter setting, however, killing and caring are layered in individual and organisational work framings in attempts to make sense of the paradox. Working in a job that aligns with strong personal convictions or passions can be seen as work based on a 'moral duty' (Thompson and Bunderson, 2003), and in our case this was about taking responsibility for unwanted animals. As Irvine (2003: 562) writes on the institutional framing of animal shelter organisations: '[w]e will take in your canine and feline mistakes and inconveniences, and we will shield you from the "dirty work" that takes place here'; hence, protecting 'people more than they do animals'. This anthropocentric responsibility is similar to the zoocentric apparent focus of vet-work as, 'even those who swear an oath to the animal, continue to ensure the human animal [the owner] is served first and foremost' (Clarke and Knights, 2019: 282). Here, animal interests are often placed secondary to those of their human owners -veterinary care being provided (in spite of their oath) to serve human desires first.
In the shelter context, the organisational processes (with input from management at times), assumed these owner privileges with key differences from private owners: (1) utilitarian herd-mentality (i.e. animals received care/were killed in relation to the shelter population); (2) limited time frames for knowing the animal (i.e. killing/adoption decisions made within three days of shelter-arrival with limited opportunities to understand individual animal differences in the stressful environment and often traumatic lifestyle change); (3) those caring for the animal and closest to them (i.e. the frontline workers) were conflicted by killing-decisions and processes, as evidenced throughout our findings. These aspects have not clearly been explored in previous literature but highlight key tensions involved in animal care-based dirty work. 'Care-based' killing or killing 'softly' (sic) was regulated through rationalised processes where emotions were viewed as unprofessional and not valid, despite care often being highly tied to positive emotions, such as empathy and compassion.
The ambiguity in the work was highlighted early on in the employment as the first author went through normal recruitment processes and, although 'euthanasia' was covered, this was downplayed. This lay the tone for killing as something individuals personally coped with, and perhaps why negative coping methods and secrecy prevailed. Secrecy is a core element of stigmatised work, but managerial recognition and support has been suggested as important for shelter-workers' wellbeing (Baran et al., 2012). This aspect was not evident in our managerial or frontline interviews. Rather, a processing-plant paradigm existed for both the low-skilled (seemingly replaceable) workers as well as the animals, upholding rationalised ideologies in practice while advertising ideologies of care.
Our case seemed to predominantly include frontline shelter-workers who reported they were emotionally motivated to complete the work for the animals (foundational to an 'unhealthy' calling path according to Schabram and Maitlis, 2017). Future studies could focus on screening processes which result in perceived 'better suited' new recruits and the impact on their wellbeing, the impact on shelters of reconsidering processes, the impact of offering different coping techniques for workers, or a reconsideration of the 'hidden' framework preferred by most shelters. It was clear that the pro-social motivation of prospective employees was valued, but mainly when it benefited the organisation in terms of cost-savings. Offering quality animal care or attempting to find alternative methods to this type of unwanted animal 'solution' really was seen as challenging the dominant paradigm.
A potential solution is seen in reframing tactics, which suggest changing workers' attitudes about dirty work and creating a sense of self-worth to buffer from taint (Ashforth et al., 1999). This self-worth may be difficult to build if challenged emotionally and morally on a daily basis as the caring-killing paradox is experienced. Although killing was seen as the organisational way to control the shelter population and reflects rationalisations of meeting organisational financial and physical constraints, our study illustrates the emotional toll many workers face when working in this paradigm of care and killing, even when trained to do this 'softly' and being socialised to kill. The caring-killing experiences in this article illustrate and question this perception of 'soft' killing in lieu of the lack of consent to death and suggests that the secrecy, worker emotional suppression and dissonance as well as managerial attitudes point to a fundamental flaw in animal welfare management -to serve people first, then animals. This worker realisation creates moral taint (in the caring-killing paradox), making it qualitatively different from other forms of taint, as it involves questioning the very reasons for undertaking the work in the first place -to help animals. Indeed, further research could focus on how personal values affect the impact of this moral taint and dissonance.
In our study, efficiency, control and life were regulated according to organisational spatialfiscal limitations, but also societal normative views of animal attractiveness. Previous research on animal shelter-work (done from an outsider perspective), concludes that the caring-killing paradox is a challenge predominantly for shelter newcomers. This research also suggests that any negative emotions are replaced with more manageable ones, focusing on objectifying the animals into 'virtual pets' to manage conflictual emotions (Arluke, 1994). These emotional management techniques of distancing and objectifying the shelter animals are seen as key to coping with the moral conflict that otherwise would damage the workers' sense of self (as animal lovers). This appeared not to be the case in our study, as seasoned animal shelter-workers continued to be negatively affected by the paradox, in spite of attempts at normalising killing as ascribed by organisational management. Here, moral and emotional dissonance may be especially complex, as our data suggest management preferred to ignore workers' paradoxical experiences. This may stem in part from not having to perform the killing themselves or may be a coping mechanism on their part to deal with workers in a stressful environment where dissonance is a part of the daily work experience. Further studies could look deeper at understanding these managerial frames in a welfare context. As Baran et al. (2012) explain, there is a reluctance to discuss morally dirty work with others not involved in the tasks (including other organisational actors) and Tallberg et al. (2014) suggest different identity 'storylines' for sensemaking difficult emotions in animal work. Owing to these rationalisations and secrecy, the effects of this work may have remained incomplete as outsider research perspectives may cause workers to report less conflictual emotions than what our insider perspective reveals. Hence, our case illustrates a more nuanced understanding of a complex and 'messy' reality, leading to deeper knowledge (Flyvbjerg, 2011). Our case highlights the paradox animal workers encounter when faced with implementing rationalised objectives of managing unwanted animals, in spite of their desires to save animals. In practice, a deeper understanding of the caring-killing paradox may help organisations and societies rethink current processes that commodify both workers and animals in a processing-plant paradigm, as opposed to ideologies of care and compassion.
Additionally, a comparative study in a no-kill-shelter could offer further insights into the industry's different processes and how these affect dissonances in care-based animal work. Do no-kill-shelters operate according to similar rationalised paradigms? And do workers there experience dissonance? Also of interest is how collective coping mechanisms, such as 'communities of coping' (Korczynski, 2003), apply to animal work, taking into consideration subjective differences in lived experiences.
Our article is limited from a multispecies perspective; thus, re-enforcing anthropocentric binaries of whose voice is considered in organisational research (McLoughlin, 2019). Future animal dirty work research could strive for further nonhuman inclusivity in the research process using multispecies ethnography or other multispecies research methods such as crystallization (Tallberg et al., 2020;Tallberg et al., 2014). In terms of broader implications, our research suggests a deeper understanding of moral paradoxes within care-based work, with similar pressures of moral decision-making that doctors have recently faced in the Covid-19 pandemic regarding whom to save when faced with limited resources. This pragmatic, but highly challenging issue, is similar to those struggling with the caring-killing paradox in animal shelters.

Conclusion
This article illustrates how the current societal solution to unwanted animals creates a caring-killing paradox for workers and commodifies both them and the animals into processing-plants called 'animal shelters'. Shelters operate according to anthropocentric masculine rationalisations, a 'business' model, rather than the branded ideologies of care and compassion which attract workers and donations. As previous studies have been limited to outsider research positionings, we argue that the hidden and morally tainted dirty work involved in this work has largely been unrecognised, leading to false understandings of what current societal animal welfare policies entail and whether these are morally acceptable in today's society. Could anthropocentric treatment be reduced, worker wellbeing enhanced, and could alternative solutions be found if these killing processes and worker challenges become more transparent? We hope so in offering this internal account of the hidden aspects of what happens to those marginalised in society and the pressures workers tasked with fulfilling such work-tasks face.