The 150th anniversary of the publication of The Origin of Species… is a good time to consider how selection can affect welfare — the quality of life. Darwin (1859) quoted Youatt's description of selective breeding: “…the magician's wand, by means of which he may summon into life whatever form and mould he pleases”. Evolution has fairly recently included us humans in its toolbox, alongside its older instruments, such as climate and disease, as significant agents of selection. We have taken to this work vigorously and have summoned into life an extraordinary array of creatures. It is only much more recently, with the development of interest in animal welfare science, that the welfare consequences of this have begun to be critically reviewed. There are two ways that selection can affect welfare: (i) by resulting in changes that make aversive feelings more likely, eg by predisposing to disease or by altering behaviour such as to increase risk of disease or injury, and (ii) by altering sensitivity of the affect systems such that animals feel, for example, more (or less) pain or fear in response to a stimulus than their ancestors would have. Comparing natural and human selection — that is, the simultaneous scrutiny of all aspects of biology as opposed to our selection for one or two features that appeal to us — Darwin (1859) wrote: “Can we wonder, then, that nature's productions should be far ‘truer’ in character than man's productions; that they should be infinitely better adapted to the most complex conditions of life, and should plainly bear the stamp of far higher workmanship”. The aims of this meeting were to discuss how selection can affect welfare and how we can improve our workmanship in the interests of animal welfare.

Increased production has been the major goal of animal breeding for many decades, and the correlated side-effects have grown to become a major issue in animal welfare. In this paper, the main genetic mechanisms in which such side-effects may occur are reviewed with examples from our own research in chickens. Pleiotropy, linkage and regulatory pathways are the most important means by which a number of traits may be affected simultaneously by the same selection pressure. Pleiotropy can be exemplified by the gene PMEL17 which causes a lack of black pigmentation in chickens and, simultaneously, predisposes them to become the victims of feather pecking. Linkage is a probable reason why a limited region on chicken chromosome 1 affects many different traits, such as growth, reproduction and fear-related behaviour. Gene regulation is affected by stress, and may cause modifications in behaviour and phenotype which are transferred from parents to offspring by means of epigenetic modifications. Insights into phenomena, such as these, may increase our understanding not only of how artificial selection works, but also evolution at large.

By definition, the companion-animal niche demands merely that animals must provide companionship. At first glance, this may seem easy enough, but the forces that contribute to success in this niche are complex. Indeed, success as a companion is rarely measured in terms of biological fitness, and empirical measures of the breeding value of stock remain elusive. The challenges in the niche are manifold and reflect the need for companion animals to show behavioural flexibility, an attribute variously labelled compliance, tolerance, and even forgiveness. The borders of the niche are blurred and there is often negligible communication between buyers and suppliers of companion animals. In addition, demand for a given phenotype is subject to considerable flux. Paradoxically, companion animals may be victims of their own success. We value the social feedback they provide and yet often leave them alone for lengthy periods. There is an inherent tension between the desire to share the company of these animals and the reality that some humans find an animal's need for social contact, and indeed many species-specific behaviours, unacceptable. Also, the animal-sense of owners may be declining, reflecting reduced community exposure to animals in non-companion contexts, such as on farms and as modes of transport. Often, in the case of dogs, the companion-animal niche is occupied by a breed that was developed to work in a specific role that required endless energy and high reactivity. We select for conformation and movement in what were once working animals and yet many owners reject animals for behavioural traits that were subject to scarcely any primary selection. Since neutering of companion animals is, for many excellent reasons, now so common, the genes of outstandingly suitable pets are routinely lost to the gene pool. Companion animals may be living longer and yet, as they age, the dog-human relationship can shift diametrically. Senior dogs often become less appealing to and yet more dependent on, and needful of attention from, their owners. In Australia, urban companion-animal ownership per capita is declining in tandem with falls in living space. Despite this reduced demand, the pet industry uses positive imagery and targeted research to promote pet acquisition, helping to maintain a situation in which supply generally exceeds demand. This results in the annual euthanasia of thousands of excess animals in shelters and pounds. The pet industry also motivates owners to be consumers so it is unsurprising that expenditure on pets in Australia is rising. Sometimes food is promoted as a means of demonstrating affection. In many developed nations, unfortunately, pet owners have the resources to respond to marketing (among other forces) by overfeeding animals, often to the point of obesity. Obesity is considered to be a significant welfare problem for companion dogs. In summary, it seems that these shifts and growing paradoxes are making the companion-animal niche more challenging than ever. Perhaps science will help make the niche more predictable, but this alone will not guarantee the welfare of the animals that occupy it.

In farm animal breeding, behavioural traits are rarely included in selection programmes despite their potential to improve animal production and welfare. Breeding goals have been broadened beyond production traits in most farm animal species to include health and functional traits, and opportunities exist to increase the inclusion of behaviour in breeding indices. On a technical level, breeding for behaviour presents a number of particular challenges compared to physical traits. It is much more difficult and time-consuming to directly measure behaviour in a consistent and reliable manner in order to evaluate the large numbers of animals necessary for a breeding programme. For this reason, the development and validation of proxy measures of key behavioural traits is often required. Despite these difficulties, behavioural traits have been introduced by certain breeders. For example, ease of handling is now included in some beef cattle breeding programmes. While breeding for behaviour is potentially beneficial, ethical concerns have been raised. Since animals are adapted to the environment rather than the other way around, there may be a loss of ‘naturalness’ and/or animal integrity. Some examples, such as breeding for good maternal behaviour, could enhance welfare, production and naturalness, although dilemmas emerge where improved welfare could result from breeding away from natural behaviour. Selection against certain behaviours may carry a risk of creating animals which are generally unreactive (‘zombies’), although such broad effects could be measured and controlled. Finally, breeding against behavioural measures of welfare could inadvertently result in resilient animals (‘stoics’) that do not show behavioural signs of low welfare yet may still be suffering. To prevent this, other measures of the underlying problem should be used, although cases where this is not possible remain troubling.

Farming and laboratory industries face questions about whether to breed animals with altered capacities for pleasure and pain. This paper addresses this issue from different approaches to animal welfare based on experiences, fitness and naturalness. This can illuminate both the breeding-related issues and the different approaches themselves. These differences have practical implications for decisions about animal breeding. All three approaches will agree that pleasure that is adaptive in natural environments has positive value and that maladaptive pain has negative value. However, where animals’ environments will not be natural, experiences-based approaches may support breeding animals that experience more pleasure and less pain or insentient animals; whereas, in some cases, fitness-based and naturalness-based approaches might favour the breeding of animals that experience more pain and less pleasure.

Milk yield per cow has more than doubled in the previous 40 years and many cows now produce more than 20,000 kg of milk per lactation. The increase in production should be viewed with concern because: i) the increase in milk yield has been accompanied by declining fertility, increasing leg and metabolic problems and declining longevity; ii) there are unfavourable genetic correlations between milk yield and fertility, mastitis and other production diseases, indicating that deterioration in fertility and health is largely a consequence of selection for increased milk yield; and iii) high disease incidence, reduced fertility, decreased longevity and modification of normal behaviour are indicative of substantial decline in cow welfare. Improving welfare is important as good welfare is regarded by the public as indicative of sustainable systems and good product quality and may also be economically beneficial. Expansion of the Profitable Lifetime Index used in the UK to include mastitis resistance and fertility could increase economic response to selection by up to 80%, compared with selection for milk production alone. In the last 10 years, several breeding organisations in Europe and North America followed the example of Nordic Countries and have included improving fertility and reducing incidence of mastitis in their breeding objectives, but these efforts are still timid. A multi-trait selection programme in which improving health, fertility and other welfare traits are included in the breeding objective, and appropriately weighted relative to production traits, should be adopted by all breeding organisations motivated in their goal of improving welfare.

Farm assurance schemes can set standards to assure compliance with specific requirements relating to animal welfare. As such, standards can be set to address genetic-related welfare problems in farm animals, such as those associated with fast growth rates in meat chickens (broilers) (Gallus gallus domesticus). Based on discussions with broiler breeding companies, broiler producers and in line with published research, the RSPCA placed a maximum limit on the genetic growth rate potential of broilers that could be used within its own farm assurance scheme — Freedom Food. Despite the introduction of this requirement, the number of birds reared on the scheme increased from 25 million to 55 million per year over a three-year period, with all of these birds meeting the genetically slower growing requirement. In addition, the two largest global broiler breeding companies responded to this change in the market by each developing a genetically slower growing bird. This demonstrates that assurance schemes can have a pivotal role in tackling genetic welfare problems in farm animals, such as those associated with fast growth in broilers.

Primates are bred in captivity for a number of purposes, from zoo-based captive breeding programmes for conservation to breeding for biomedical research. In each case, breeding animals that are fit for purpose, either as viable candidates for reintroduction or as valid research models, has presented challenges and resulted in steep learning curves. The breeding of animals for biomedical research has become increasingly focused on the production of animals that are less stressed by captive (specifically laboratory) environments. This is because elevated, particularly chronic, stress responses can result in altered physiological, neurological and behavioural states that have the potential to compromise the validity of scientific results. Selective breeding in captivity to, for example, maximise production, select for docile temperament or specific genotypes for biomedical research, is likely to be counter to natural selective pressures for evolutionary fitness. Given that many natural selective pressures active in the wild are absent in captivity, this paper reviews the selective breeding of primates (especially Old World monkeys) in captivity, its potential negative effects, and options that exist for ameliorating these negative effects.

The effect that breed standards and selective breeding practices have on the welfare of pedigree dogs has recently come under scrutiny from both the general public and scientific community. Recent research has suggested that breeding for particular aesthetic traits, such as tightly curled tails, highly domed skulls and short muzzles predisposes dogs with these traits to certain inherited defects, such as spina bifida, syringomyelia and brachycephalic airway obstruction syndrome, respectively. Further to this, there is a very large number of inherited diseases that are not related to breed standards, which are thought to be prevalent, partly as a consequence of inbreeding and restricted breeding pools. Inherited diseases, whether linked to conformation or not, have varying impact on the individuals affected by them, and affect varying proportions of the pedigree dog population. Some diseases affect few breeds but are highly prevalent in predisposed breeds. Other diseases affect many breeds, but have low prevalence within each breed. In this paper, we discuss the use of risk analysis and severity diagrams as means of mapping the overall problem of inherited disorders in pedigree dogs and, more specifically, the welfare impact of specific diseases in particular breeds.

Social interactions between individuals, such as co-operation and competition, are key factors in evolution by natural selection. As a consequence, evolutionary biologists have developed extensive theories to understand the consequences of social interactions for response to natural selection. Current genetic improvement programmes in animal husbandry, in contrast, largely ignore the implications of social interactions for the design of breeding programmes. Recently, we have developed theoretical and empirical tools to quantify the magnitude of heritable social effects, ie the heritable effects that animals have on their group mates’ traits, in livestock populations, and to utilise those effects in genetic improvement programmes. Results in commercial populations of pigs and laying hens indicate large heritable social effects, and the potential to substantially increase responses to selection in traits affected by social interactions. In pigs, including social effects into the breeding programme affected aggressive behaviour, both at mixing and in stable groups, indicating changes in the way dominance relationships are established and in aggressiveness. In laying hens, we applied selection between kin-groups to reduce mortality due to cannibalistic pecking. This resulted in a considerable difference in mortality between the low mortality line and the unselected control line in the first generation (20 vs 30%). Furthermore, changes in behavioural and neurobiological responses to stress were detected in the low mortality line, pointing to reduced fearfulness and stress sensitivity. These first results indicate that including social effects into breeding programmes is a promising way to reduce negative social interactions in farm animals, and possibly to also increase positive social interactions, by breeding animals with better social skills.

Current alterations in the farm environment, such as a reduced number of farm workers, may mean that sheep genotypes that are highly dependent on man for nutritional and reproductive success will experience poorer welfare within that environment. In the past 30 years, average flock size has doubled, and flocks of over 1,000 ewes managed by one stockperson are common. The reduction in the ratio of stockpeople to sheep affects animal welfare, with less time for tasks such as healthcare and inspection. It has also led to increased interest in the development of new genotypes that are better able to look after themselves. Selection and management of sheep to promote behaviours associated with survival, and selection of robust animals that require less human intervention for good welfare, are important breeding goals. As these animals will receive less inspection at close quarters, selection for resistance to disease will have significant animal welfare benefits. In addition, the development of sheep lines that require little or no intervention at lambing will be important. In areas where wool is not valuable, the use of wool-shedding breeds to avoid the stress associated with shearing, and to reduce the incidence of flystrike are already proving to be beneficial. Importantly, this selection should not be interpreted as providing no care to these animals, and careful management during the production of these genotypes is needed to avoid at least transient welfare problems where genotypes and environment (eg lower shepherding) are mismatched.

One option for improving the welfare of purebred dog breeds is to implement health breeding programmes, which allow selection to be directed against known diseases while controlling the rate of inbreeding to a minimal level in order to maintain the long-term health of the breed. The aim of this study is to evaluate the predicted impact of selection against disease in two breeds: the Cavalier King Charles spaniel (CKCS) and the Labrador Retriever. Heritabilities for mitral valve disease, syringomyelia in the CKCS and hip dysplasia in the Labrador were estimated to be 0.64 (± 0.07), 0.32 (± 0.125) and 0.35 (± 0.016), respectively, which suggest encouraging selection responses are feasible based upon the estimation of breeding values (EBVs) if monitoring schemes are maintained for these breeds. Although using data from disease databases can introduce problems due to bias, as a result of individuals and families with disease usually being over-represented, the data presented is a step forward in providing information on risk. EBVs will allow breeders to distinguish between potential parents of high and low risk, after removing the influence of life history events. Analysis of current population structure, including numbers of dogs used for breeding, average kinship and average inbreeding provides a basis from which to compare breeding strategies. Predictions can then be made about the number of generations it will take to eradicate disease, the number of affected individuals that will be born during the course of selective breeding and the benefits that can be obtained by using optimisation to constrain inbreeding to a pre-defined sustainable rate.

This study investigates the genetic basis of lamb vigour (defined as neonatal lamb activity and sucking ability) and lambing difficulty as potential traits to be included in selection programmes to improve ewe and lamb welfare. Scores for lamb birth difficulty, vigour and sucking ability were collected shortly after birth on 1,520 lambs born in 2006 in 19 different flocks that were members of the UK Suffolk Sire Referencing Scheme. Scores evaluated each trait on a scale of 1 to 4; 1 being no assistance given either during birth or to suck, or excellent vigour, through to 4 where a large degree of assistance was required, or poor vigour. Genetic parameters (heritabilities, genetic correlations) were estimated by fitting an individual animal model using ASREML. Variance components obtained from univariate and bivariate analyses were averaged to provide genetic parameter estimates. Heritabilities for birth difficulty and vigour were moderate but heritability for sucking ability was not significant. The genetic correlation between vigour and sucking ability was positive and high, that between vigour and birth difficulty moderately negative, and that between birth difficulty and sucking ability not significant. Birth difficulty and vigour could be included in Suffolk breeding programmes to help reduce health and welfare problems associated with these traits in Suffolk sheep, and in flocks producing crossbred lambs sired by Suffolk rams. Further work is required to evaluate correlations between these traits and performance traits and to comprehensively validate the scoring system once more data become available.

In humans, empathy has emotional and cognitive components, both of which are linked to caring and nurturant behaviour. Variations in each of these facets of empathy were likely to have been accessible to natural selection during the evolution of Homo, although the likely details of their respective adaptive values has so far only been considered in the context of intraspecific (human-human) behaviour. We propose that evolutionary psychology may provide a useful additional framework for examining why humans feel empathy for certain animals but not others. Phobias towards noxious animals, such as snakes and spiders, have been explained in terms of gene-culture co-evolution, but the possibility of an analogous ‘biophilia’ directed towards other animals has received less attention. The redirection of primarily intraspecific nurturant behaviour towards the young of non-human species may be a general human trait since it is practiced in a wide variety of cultures, including hunter-gatherers, and may arise from the merging of natural history and social intelligences that the archaeologist Steven Mithen suggests evolved ~100,000 years before present (YBP). The visual stimuli that evoke such nurturant behaviour, Lorenz's ‘Kindschenschema’, or ‘cuteness’, have been compared with the super-stimuli whereby parasitic cuckoos induce care-giving from their hosts, but recent evidence suggests that human females of childbearing age are especially sensitised to respond most strongly to characteristics of human infants, and may correspondingly become less attracted towards ‘cute’ animals. It is also possible that during human prehistory, the ability to care for young animals was selected for, in adolescent females, as an honest indicator of future quality as a mother. An ability to empathise with animals may also have given certain individuals and/or groups of kin an evolutionary advantage in hunting, and subsequently herding and domesticating, animals. Concern for animal welfare may therefore stem from an evolved human trait, even though its degree and extent of expression are undoubtedly strongly influenced by culture.

The aim of this paper is to discuss the potential consequences of modern dairy cattle breeding for the welfare of dairy cows. The paper focuses on so-called genomic selection, which deploys thousands of genetic markers to estimate breeding values. The discussion should help to structure the thoughts of breeders and other stakeholders on how to best make use of genomic breeding in the future. Intensive breeding has played a major role in securing dramatic increases in milk yield since the Second World War. Until recently, the main focus in dairy cattle breeding was on production traits, but during the past couple of decades more emphasis has been placed on a few rough, but useful, measures of traits relevant to cow welfare, including calving ease score and ‘clinical disease or not’; the aim being to counteract the unfavourable genetic association with production traits. However, unfavourable genetic trends for metabolic, reproductive, claw and leg diseases indicate that these attempts have been insufficient. Today, novel genome-wide sequencing techniques are revolutionising dairy cattle breeding; these enable genetic changes to occur at least twice as rapidly as previously. While these new genomic tools are especially useful for traits relating to animal welfare that are difficult to improve using traditional breeding tools, they may also facilitate breeding schemes with reduced generation intervals carrying a higher risk of unwanted side-effects on animal welfare. In this paper, a number of potential risks are discussed, including detrimental genetic trends for non-measured welfare traits, the increased chance of spreading unfavourable mutations, reduced sharing of information arising from concerns over patents, and an increased monopoly within dairy cattle breeding that may make it less accountable to the concern of private farmers for the welfare of their animals. It is argued that there is a need to mobilise a wide range of stakeholders to monitor developments and maintain pressure on breeding companies so that they are aware of the need to take precautionary measures to avoid negative effects on animal welfare and to invest in breeding for increased animal welfare. Researchers are encouraged to further investigate the long-term effects of various breeding schemes that rely on genomic breeding values.

The routine mixing of pigs causes aggression that cannot be greatly reduced by low-cost environmental changes. The variability and heritability of aggressiveness are discussed and both appear adequate to make selection against aggressiveness worthwhile in grower-stage pigs. Selection would require rapid phenotyping of many animals for which a validated indicator genetically correlated to aggressive behaviour is required. Three potential indicators are discussed (attack latency, number of skin lesions and relationship to non-social behavioural traits). Attack latency correlates with post-mixing aggressiveness under research conditions but attacks are delayed under commercial conditions reducing the practicability of the trait for selection. Correlations between aggressiveness and responses to non-social challenges, such as the back-test, are not always consistent. Lastly, the counting of skin lesions is rapid, and the number of lesions has a moderate heritability and is genetically correlated with involvement in aggressive behaviour. The wider effects of selection against post-mixing aggressiveness are discussed. Examining the behavioural strategies of unaggressive pigs, especially their response to defeat, would reveal how selection may alter aggressive tactics. Selection against lesions from mixing is also expected to reduce their number in more stable social conditions, but the implications for aggression between sows and that of sows towards their piglets and humans needs to be investigated. Aggressiveness is genetically correlated with response to handling involving components of social isolation, human presence and novelty. Identifying how unaggressive pigs respond to other challenging situations differing in these components may be worthwhile. Selection against aggression using skin lesions appears to be achievable although the full value of this would benefit from estimations of the genetic correlations with the traits outlined above.

In the UK, numerous pedigree dogs of many breeds experience compromised welfare due to the direct and indirect effects of selective breeding. Many breeds are selected to have physical conformations which, although perceived by some to be desirable, have direct negative effects upon their welfare. Dogs are regularly bred whose heads are too large and pelvises too small to birth naturally or whose faces are so flat that they are unable to breathe or exercise normally. There are also many indirect effects of selective breeding for appearance, including significantly elevated prevalence of specific diseases within particular breeds. Current breeding practices can therefore result in unnecessary suffering due to pain, disability, disease and behavioural problems. In this paper, we summarise and review the current scientific evidence for such suffering, and difficulties associated with assessing the impact of current breeding practices. Limited record-keeping, lack of transparency in the breeding and showing world, and the absence of sufficient research, mean that the full extent of the problem is difficult to assess. Furthermore, the collection of data is currently unsystematic, and although there are specific case studies of individual breeds and particular disorders, relatively few have been conducted in the UK. Individual breeds each suffer from their own array of problems, so each breed's survival and improvement (in terms of health and welfare) is likely to require a different specific course of action. With 209 breeds currently registered in the UK, this makes the situation complex. We collate and present a range of suggestions which may help to improve pedigree dog welfare significantly, and prioritise these based on expert opinion.

The breeding of domestic dogs for dog fighting has resulted in numerous genetic alterations in a breed widely acknowledged to be the most successful fighting dog: the American Pit Bull Terrier (APBT). Much of the genetic foundation underlying the motivation and ability for pit fighting can be traced back to the earliest use of dogs for hunting purposes and continued through the selective breeding for use of dogs in wars and bull and bear baiting. In the development of the APBT as a fighting dog, there were two main breeding criteria. The first was, and remains, fighting success. The trait most prized by breeders of fighting dogs and considered most contributory to fighting success is ‘gameness’, which is the perseverance at a task even under extreme adversity, such as injury, pain, or fatigue. The second criterion was the absence of human-directed aggression. Since dogs are handled extensively before, during, and after the fights, dogs that showed aggression toward humans were eliminated from the gene pool. Indeed, anecdotal reports suggest that breeding may have been carried beyond that of simply selecting against human aggressiveness to a degree of enhanced affinity for humans. The result is that today's fight-bred APBT is genetically predisposed — but not predestined — to aggressiveness toward other dogs and a strong social attraction to humans. The human affinity trait is a highly valuable characteristic that ought to be preserved. With the appropriate breeding decisions, the power of genetic selection suggests that this goal, along with normalising the genetic disposition for conspecific aggressive tendencies, should be ultimately achievable.