Extensive farming is generally perceived by society and consumers as a more sustainable and animal-friendly method of animal production. However, this remains to be validated by science. First, the welfare of animals kept, permanently or temporarily, in these systems must be scientifically and transparently studied ^[1][2]. Multiple approaches have been used to assess sheep and goat welfare in intensive systems ^[3], but ways for adapting and integrating this knowledge into protocols for small ruminants in extensive systems is still open to discussion.

The first welfare assessment protocols using animal-based indicators were developed for intensive production systems, especially for dairy/beef cattle, poultry, and pigs (for example, Welfare Quality), due to the overall high animal density per unit area associated with such systems. On the other hand, welfare assessment protocols for small ruminants, developed under the Animal Welfare Indicators (AWIN) project ^[4][5], were tested in different rearing conditions: dairy goats kept in intensive systems and sheep in extensive and semi-intensive farms ^[4]. Although these protocols may be useful for the assessment of welfare in all systems, it is necessary to ensure that appropriate and evidence-based changes are inserted and integrated. For example, some original measures may not be applicable or may show a very low prevalence and should thus be withdrawn, while new indicators will undoubtedly need to be added. Additionally, the assessors may need to record variable indicators dependable on various factors (for example, terrain, distance, access, time, and weather constraints). Recently, it ^[6] tested the feasibility and reliability of the AWIN protocol for the welfare assessment of dairy goats in semi-extensive farming conditions. It showed that some indicators from the AWIN sheep protocol could be successfully integrated into the original goat protocol, but some validation ones were nonetheless still needed.

It was discuss some of the main limits of these protocols when applied to fully extensive systems, as an opportunity for the use of novel technologies that are presented in the following sections.

Neither small-ruminant AWIN protocol includes males in the evaluation. Although this is not an issue for dairy cattle or pigs in intensive farms, it may be an important flaw for extensive and pasture farming systems ^[7][8][9]. Most published ones only assess the welfare and behavior of males in relation to the use of reproductive technologies ^[10] or in castration ^[11]. When applying an assessment protocol, it should be remembered that behavior is inherently different between females and males and that the presence of males may influence the behavior of the females (and vice-versa); for example, ewes are usually more active and vocal than rams ^[12][13]. Consequently, the welfare of males in the flock has often been overlooked. Therefore, developing or adapting protocols to groups that include males should consider such particularities.

Some proposals have been made to adapt existing protocols to less intensive farming systems ^{[6][14][15][16][17]} although this should be regarded as a complex and challenging task considering the settings’ diversity ^[3][16]. Small ruminants bred in extensive farming systems are more exposed to weather conditions, which implies a need for considering temperature, humidity, and even wind exposure in welfare assessments. Extreme conditions, as well as sudden variations in weather, may significantly change the expression or intensity of some indicators and, thus, the welfare of the animals. Heat stress must be considered in these evaluations as well as each breed’s capacity to adapt and cope with its environment ^[18]. Breeds selected to live in extensive or pasture farming systems are usually more resilient and well adapted to the climatic conditions ^[19][20]. However, when exposed to temperatures outside their thermal comfort zone, more sturdy animals may mask poor welfare signs, making assessment very difficult ^[20]. The volatility of these circumstances poses an enormous and complex challenge for welfare assessment in small ruminant species kept in harsher environmental conditions. Nevertheless, indicators related to thermal comfort should be seen as an essential component of protocols to be used in extensive farming systems. For example, weather and environment monitoring stations may be important assets as they will allow for accurate correlations between welfare indicators and weather conditions. Other ways novel technologies may help in overcoming some of these issues will be further addressed in subsequent sections.

One of the principles suggested by the Animal Welfare Indicators (AWIN)  protocol is good housing. Although in extensive systems actual housing is very limited or even absent, the provision of shade or shelter from wind, rain, or snow should be considered as the lack may result in very poor welfare, particularly for young animals. Providing some sort of shelter, whether for feeding or for protection from climate extremes, will result in higher welfare levels ^[21][22]. However, the extension and type of landscape in which these animals are generally kept may preclude the building of such shelters ^[3]. Thus, including the presence of shade/shelter and its accessibility, should be considered as an important indicator of animal comfort.

Extensively kept small ruminants may be more prone to lameness problems due to constant exposure to wet soils and infection agents or due to the need to walk long distances along rough paths. Additionally, they are usually less frequently checked for signs of pain or discomfort and very rarely treated early, compared to animals in intensive farms ^[14]. Although conducting a thorough hoof examination poses great challenges in extensive systems, lameness assessments should include traditional gait scoring but also the careful examination of hooves if the prevalence of severely lame animals is detected. Moreover, the time of the year, climate, and terrain conditions should also be considered in these protocols. Technologies such as thermography, image scanning, or weight pads may be an expeditious and reliable way of detecting lame animals going through a race.

Several other problems deserve special attention in extensively kept flocks due to potential differences in prevalence and severity when compared with intensive systems ^[23]. Body condition score, diarrhea, and fleece condition are just a few examples of measures that may need substantiation and critical analysis before being approved ^[16][24]. These may vary according to weather conditions, season, and type of feeding (e.g., fresh grass), and the welfare impact of these changes must be verified. In extensive systems, in which the handling of individuals may be difficult, sorting gates and electronic weighing scales combined with e-ID may be used for the automatic collection of weight data.

Well-designed ones on the behavior of small ruminants in extensive conditions are required to ensure the validity of the many animal-based measures, including the group mental state assessment trough the Quality Behavior Assessment (QBA) used in the existing AWIN protocols ^[25]. Despite the merit of on-farm welfare assessment schemes like the AWIN protocols, there is a need to introduce indicators that address positive aspects of animal welfare ^[26]. In this sense, other methods, including the five-domain model, were developed ^[27] and applied to sheep ^[14][28]. This model incorporates three survival-related domains (nutrition, environment, and health), a behavior-related domain, and a fifth domain that results from a comprehensive assessment of how the other domains impact the animal’s affective experience ^[29][30]. Besides, it is an ever-evolving model ^[31], which also allows for new interpretations and adaptations to extensively farmed species. In this way, identifying valid, reliable, and viable animal-based indicators related to the positive aspects of animal welfare will improve the quality of life of animals and strengthen communication about animal welfare to stakeholders ^[31]. Likewise, measuring human–animal interactions in extensive systems might differ from the intensive systems norm. Therefore, it is relevant to combine the knowledge acquired in intensive systems, but also to understand how the human–animal interaction is in extensive systems, and to not infer by mistake that the applicable behavioral parameters might be the same.

An equally important issue in applying protocols to extensive systems is feasibility. For example, the time needed to apply the protocol in very large settings, or the difficulties associated with the exposure to open-field constraints, all have to be considered to ensure validity and feasibility. In this context, on-farm welfare assessment feasibility can be increased by adopting a strategy supported on a few valid and reliable animal-based indicators ^[14] complemented by the introduction of new technologies, such as automatic-recording devices or drones ^[32].

Finally, a word on an often-demoted issue—the need for specifically trained evaluators ^[6][33] so that the repeatability and credibility of the protocols are guaranteed. Experienced and competent auditors in intensive system assessments may not be prepared needed in extensive settings.

In summary, welfare assessment of extensively kept small ruminants should be seen as a very specific subject, and not just an extension or simple adaptation of the protocols validated for intensive systems. Due to particular features and limitations, full-field assessment may be too difficult to manage through traditional farm-level personal observation. This provides excellent opportunities for new sensor technologies, as will be discussed in the following sections. The main constraints to the use of new technologies in small ruminants in extensive settings, such as drones, intra-ruminal sensors or ear-tags containing accelerometers, are cost and the difficulty of getting accurate, real-time readings.

2. The Use of Precision Livestock Farming Applied to Sheep and Goat Extensive Production

Technological developments that have been applied to sheep and goat extensive production systems, although very diverse, can be framed within the concept of precision livestock farming (PLF). PLF has been described as the use of real-time monitoring technologies to manage the smallest manageable production unit ^[34]. PLF uses equipment, data, and software that allow individual animal information to guide decisions and inputs more precisely in an animal production system ^[35][36]. As mentioned, PLF approaches critically depend on identifying the animals individually, and electronic identification (EID) allows the achievement of this goal. EID has undergone significant developments since the early 1980s and is typically linked to the use of tags or boluses ^[35]. In 2004, the European Union made EID mandatory for all sheep and goats ^[37], and it currently represents an opportunity to further increase the scope of PLF technologies into extensive management systems ^[36]. EID is linked to technologies such as global positioning systems; behavior–activity loggers; virtual fencing; stationary management systems, such as walk-over-weighing systems; and automatic drafters ^[35][36][38]. These technologies allow for the precise management of sheep and goats, individually, in small groups, and as a flock ^[34][38]. Individual animal performance provides support for better decision making, which could benefit animal performance, economic performance, labor ^[37], and animal health and welfare ^[39]. As these management systems develop, vast amounts of data can be collected from thousands of farms, further assisting and directing agricultural policies on sheep and goat production, global warming mitigation, and antibiotic resistance ^[36]. Furthermore, such precision data can be used and integrated to find solutions to disease, welfare, productivity, and environmental issues and improve farming outputs ^[39][40][41]. Also, positive economic results have been observed in different reports ^[37][42]. As extensive systems are very diverse, there are many circumstances in which PLF is not suitable or even feasible. Despite promising results, most of the technologies have not yet reached an applicability level similar to those introduced in intensive systems ^[35]. It should also be considered that cultural dynamics, financial stability aspects, confidence in new technologies, and the openness of farmers to new ideas do not always encourage wider adoption of innovative technologies in sheep and goat extensive systems ^[36][43]. In this sense, and considering that PLF is a collection of relatively novel technologies, the effects on animal welfare in extensive systems are not yet apparent ^[40]. However, it is expected that PLF solutions will play a key role in assessing welfare in extensive systems and will be driven by a greater capacity of technologies to recognize welfare and, more significantly, whether the welfare of farm animals is improved by the application of technologies ^[40]. Nevertheless, available solutions that include PLF approaches have been used to assess various issues related to sheep and goats’ health, behavior, and welfare in extensive systems.

3. The Use of Omics Applied to Sheep and Goat Extensive Production

Omics refers to the use of novel molecular biological approaches that allow for the profiling of a particular organism, tissue, or cell concerning its genes (genomics), mRNA transcripts (transcriptomics), proteins (proteomics), and metabolites (metabolomics) at a particular point in time ^[44]. Post-genomic platforms, namely proteomics, metabolomics, and transcriptomics, are gaining importance in the context of animal production, and more recently, the integration of these different platforms with food and nutrition science have been demonstrated to be a very interesting asset to obtain an in-depth analysis on animal physiology, production, and other related fields of animal science ^[45]. Despite many are concerning animal welfare in the behavior and ethology fields, the establishment of biomarkers can be a great complement to improving the welfare assessment ^[46] and the knowledge of animals’ physiological processes and regulatory mechanisms of adaptation to harsh conditions ^[47]. Overall, it can be considered that the different omics are a valuable tool for addressing several key aspects of small-ruminant science, particularly in the framework of production and welfare in extensive sheep and goat farming. However, there is an aspect that is particularly associated with small-ruminant-production systems. It is related to the year-round fluctuations in the rain pattern that in turn cause important changes in the availability and quality of pasture and fodder for ruminants, particularly those in the extensive systems. Indeed, the occurrence of a dry season that can last several months leads to the unavailability and lignification (decreased nutritional value) of pasture during such months. In turn, this leads to seasonal weight loss (SWL), a problem to which several small ruminant breeds have adapted over the selection process. This issue is particularly pertinent in the framework. SWL is one of the most pressing issues in extensive animal production in tropical and Mediterranean regions. There are two solutions to address this problem. First, supplementation with additional feed is often problematic, if not impossible, to implement in the large areas that characterize these extensive systems. Second, the most cost-effective approach is using breeds adapted to feed scarcity ^[48]. To select such breeds, novel technologies are available, allowing for the identification of biomarkers and molecular patterns related to SWL resilience. Using omics have been conducted over recent years ^[49]. Here, it will be focus on two examples where omics were used to know SWL: meat-producing sheep in Australia and dairy-goat production in the Canary Islands. Sheep production in Australia is mainly based on extensive systems, primarily designed for wool production using the Australian Merino (AM) breed. In recent years, these production systems have been increasingly exposed to droughts that compromise animal welfare and the economic viability of farms due to undernutrition ^[50]. Moreover, AM sheep are highly susceptible to myiasis (caused by blowfly strikes), which compromises their health and welfare ^[51]. To deal with this issue, farmers routinely remove the hind-quarters skin folds which are susceptible to blowfly strikes in a surgical procedure called mulesing ^[52]. In addition to decreasing wool prices and the consequently reduced profitability of AM flocks, these welfare concerns have motivated a shift in these production systems. Indeed, producers have been steering towards meat production, particularly destined for live animal exports bound for the Middle East and Asia. Because the AM sheep is primarily bred for wool production and is also highly susceptible to SWL and external parasites, it is less appealing for meat production compared to South-African breeds such as the Dorper. The latter is a composite breed conceptualized for meat production, originating in the breeding of Persian Blackhead and Dorset Horn. In addition to this breed, using fat-tailed breeds (for example, Damara, another South-African breed) poses another alternative, taking advantage of their superior fat depots to endure SWL ^[53]. To evaluate the response of the AM, Dorper, and Damara to SWL, a live-animal trial was carried out to induce weight loss experimentally ^[54]. Since then, several different analytical approaches have been carried out to assess the physiological response of these breeds. Briefly, the restricted groups of Damara and Dorper lost a smaller percentage of their initial live weight (LW) than the AM group. Unrestricted animals increased by 7%, 13%, and 10% of their initial LW, respectively ^[54]. The differences between breeds extended to carcass and meat characteristics, with both South-African breeds having heavier carcasses, higher fat deposition, and darker meat compared to the AM breed ^[55]. The different muscle development inherent to each breed was reflected in the muscle proteome ^[56][57]. The muscle structure of the Dorper breed is particularly affected when restricted, lowering the abundance of contractile apparatus proteins, such as myosin and tubulin. In addition, a higher number of cellular functions were impacted in the AM breed as a consequence of SWL, such as ATP and actin binding ^[56]. This was corroborated by a metabolomics ^[58] and amino acid ^[59] profiling analysis of the muscle tissue, which identified lower levels of amino acids (for example, tyrosine, glycine, and taurine) in the muscle of the AM breed, suggesting lower muscle growth and increased endogenous protein mobilization compared to the other two breeds. Interestingly, the Damara breed was seen to increase the abundance of structural proteins such as desmin because of SWL ^[57]. This highlights an increased resilience of Damara sheep under SWL, where they counter-balance muscle amino acid mobilization ^[59] by attempting to maintain structural integrity. The liver proteome of these sheep has been learned, including the mitochondrial proteome ^[60][61]. These revealed that the Damara breed under SWL mobilizes more lipids than the AM breed through the higher abundance of lipid transport proteins, such as apolipoprotein E, and lipid metabolism proteins, such as annexin. In turn, the unrestricted group has a metabolism oriented for the synthesis of fatty acids, particularly branch-chain fatty acids, which accumulate in the tail ^[62]. Indeed, the mobilization of tail fat under SWL is the distinct mechanism of the resilience of the Damara breed against SWL. Its mobilization has caused the increase of fat tissue mineral concentrations ^[63] since the presence of fat has a diluting effect on the tissue mineral profiles ^[64]. The quality of wool from the AM breed was also negatively influenced by SWL, which caused a reduction in fiber diameter and an increase of the high-sulfur protein KAP13-1 and the glycine–tyrosine-rich KAP6 family of proteins ^[65]. The data mentioned above demonstrate the physiological mechanisms behind the improved adaptation of SWL of the Dorper and Damara breeds compared to the AM breed. Moreover, several different biomarkers have emerged from these ones that can be used to choose hardy breeds whose welfare is not so negatively affected by current conditions. Dairy-goat production in the Canary Islands is another example where the reared animals are subjected to SWL, particularly in the easternmost islands, which are very dry compared to the western islands with a more temperate climate. The different rainfall in La Palma (a humid island) and Fuerteventura (a dry island) has an impact on the available pasture, and consequently, on animal production. The Majorera goat from the latter island has been adapted to weight loss, whereas the Palmera goat from the former is more susceptible to feed restriction, which threatens welfare in dairy production systems. Similar to the sheep example described above, a trial was conducted to compare the response of both of these breeds to SWL. Restricted groups lost 13% of their initial live weight and 87% of their initial milk yield ^[66]. This had repercussions on the FA composition of milk and the mammary gland, particularly for the Palmera breed, where restriction increased oleic acid and reduced palmitic acid in the secretory tissue, whereas the Majorera had no differences ^[67]. Despite this, feed restriction caused the mobilization of endogenous FA in both breeds, as indicated by higher levels of circulating non-esterified FA ^[68]. However, omics approaches have revealed that the response was different in the two breeds, with more resistance features in the Majorera. Indeed, a transcriptomics approach identified a wide set of genes with differential expression in the mammary gland caused by SWL. The restricted Majorera increased the expression of genes related to amino acid, lipid, carbohydrate, and nucleoside transport, indicating reduced metabolic activity. Contrarily, the restricted Palmera goats upregulated genes involved in suppressing cell differentiation and related to the response to DNA damage, demonstrating the effects of mammary gland involution. Comparing both restricted groups identified two genes associated with unregulated tissue development in Palmera goats (CPM and ASB11) ^[69]. This is confirmed through two different proteomic approaches that identified a high abundance of apoptotic proteins in the restricted Palmeras and suggested cadherin-13, collagen alpha-1, and clusterin as another set of putative biomarkers to SWL tolerance in the Majorera goat breed ^[70][71]. The detrimental effect of feed restriction extended to the metabolome of the mammary gland and milk ^[72]. Restricted groups had lower AMP, ADP, ATP, and IMP, all energy-related molecules characteristic of low metabolic rates in both restricted breeds. In addition, feed restriction influenced the rumen metabolism, which seems to have contributed to the lower levels of Krebs-cycle intermediates (citrate, fumarate, succinate) in the milk of the restricted goats. So, as SWL can represent a problem in the dairy sector, it is essential to establish biomarkers to ensure the health-status monitoring, apply new breeding systems, and essentially guarantee animal welfare ^[47][73]. Similar to the Australian sheep, these goat have yielded several putative biomarkers of SWL-resistance that could be used to select animals for enhanced response to SWL. This is particularly important given that the climate is rapidly changing, and susceptible breeds could soon be subjected to harsh droughts that threaten not only animal welfare related to undernutrition, but also local food security and economies. The information obtained from these omics approaches provides a detailed look at the impact of SWL on a molecular level. This allows for a deeper understanding of the metabolic response to weight loss differentiation among adapted and susceptible breeds in two distinguished contexts: dairy and meat production. The identified differences are supported by classical approaches, including mineral, amino acid, and fatty acid profiling. Identifying biomarkers for SWL-resistance enables the improvement of breeding programs for the selection of hardy breeds towards the economic viability and welfare of animals in the extensive production systems in tropical and Mediterranean regions.