4.1. Farm and Producer Features
Worldwide, extensive management systems are the most common type of sheep production [
29], and this is not an exception for Colombian systems, where sheep farms are located in steep mountain areas in cloud forest ecosystems, where other agricultural and livestock activities are performed as well [
30]. The lack of technology, basic traditional management without appropriate records, and thus, inappropriate decision-making, as well as low incomes, are the common denominators of Colombian sheep production systems [
14,
26]. These same conditions were found in the sheep farms evaluated, where it was observed that poor decision-making was carried regarding the selection and reproduction of sheep, whereby the farmer focuses on selecting animals that are adapted to the local environment and capable of surviving and growing with little intervention and care by the farmer and the veterinarian. Under these conditions, the lack of care for newborn lambs has caused a reduction in flock size, especially due to predation by wild dogs (
Canis familiaris) and cougars (
Puma concolor) [
28].
Farm conditions in this study were similar to those seen in Africa and Asia [
31,
32], where the main characteristics are small flocks, extensive grazing systems with a lot of land available, the presence of predators, and lack of vigilance and monitoring by the farmers. Predation by carnivores was reported anecdotally by all farmers in the present study, but the lack of records about the attacks made it impossible to include it in the analysis. All farmers were men, although small ruminant production systems in developing countries like India and Latin American countries have been actively incentivized for women due to their easy management [
33]. Producer education level was low, but no significant correlation was found in this study between formal education and sheep welfare, as was the case with training.
4.2. Animal Welfare Assessment
Overall, most sheep farms in this study had a “good” welfare score for the environment domain, with access to shade and shelter, fleece cleanliness, and panting indicators, along with the behavior domain with stereotypes, excessive itching, social withdrawal, and flight. While the nutrition domain, with the prevalent BCS ≤ 2, no water availability and high lamb mortality reported due to predation and the health domain with fecal soiling, FAMACHA© scores ≥ 4, and ocular discharge had lower average scores. In this study, no significant differences were found for animal welfare classification between the two large farms and the eleven small farms. This observation may be related to the socio-political conditions of the armed conflict, which kept these areas geographically isolated, without state presence, and subjected them to forced displacement. For more than fifty years, affected rural producers did not have access to technology and knowledge sharing related to management practices and veterinary assistance [
10,
34], which affected producers in the area, as evidenced in the study by the absence of basic production records (records for birth, mortality, etc.), traditional management practices, and a lack of professional assistance.
A study performed in Mexico [
35] suggests that small flock sheep farmers due to low density, easiness of handling, and family managed systems have an advantage in achieving good animal welfare over larger production systems. However, in these small and extensive systems, the domains of Behavior and Environment tend to be favored due to the lack of space constrains and a free roaming environment. Nonetheless, the domains of Nutrition and Health can be negatively impacted due to the lack of monitoring, so overall welfare may be affected by this unbalanced relation [
36]. It has also been reported that cultural differences among producers have an impact over production parameters in small farmers due to management and the availability of resources [
37].
4.3. Iceberg Indicators
This study found that BCS, FAMACHA© score, fleece cleanliness, and flight distance had a significant impact on the overall on-farm welfare score, and QBA represented the mind state of sheep (mental domain). These indicators have been reported to be used all around the world for extensive sheep systems and have been proven as reliable, feasible, and valid [
38]. Low BCS measured through palpation at the lumbar section was the best indicator identified in this study to evaluate the Nutrition welfare domain, being BCS ≤ 2 the most prevalent value found in all farms. This could be explained due to the fact that the nutritional value of pastures on extensive hill systems are normally not adequate and the nourishment of animals is sub-optimal [
39,
40].
This study found fleece cleanliness to be a significant indicator for the environment welfare domain, as it helps to determine when small rural extensive farms with sheep in all year outdoor conditions have no protection against extreme weather, which can affect thermoregulation and energy consumption [
37]. It has also been reported that its measurement presents high inter-observer agreement in extensive farming systems [
20], which makes it useful to evaluate changes throughout all production stages on farm.
Regarding the Health welfare domain, we identified that the FAMACHA© score (FS) ≥ 4 (a.k.a. anemic sheep) had an impact on the overall welfare score. Generally, FS is related to gastrointestinal parasite infestation. Since it has been reported that gastrointestinal parasite prevalence is common in sheep productions worldwide [
41,
42] it is important to detect them prematurely in order to avoid economic and productive losses; however, in this study, the correlation between FEC and FS was low (
r = 0.21). This is because the relation is highly dependent on the presence of
Haemonchus contortus, and FS is not sensitive to other parasitic burdens [
18]. Therefore, ideally, FA has to be measured along with BCS or weight to improve its effectiveness. In addition, FAMACHA© is a subjective indicator that depends strictly on the appreciation of the mucous color by the observer, so in order to use this properly and as a reliable indicator, proper training is required [
43,
44].
With regard to the evaluation of the animals’ mental states, the QBA is considered a qualitative measurement scale used and validated in ovine [
45], bovine [
25], porcine [
46], and equine species [
47]; however, to avoid measurement biases, adequate instruction and training in the use of the visual analog scale and the interpretation of the descriptors is required [
48]. In the present study, this factor was controlled by the training of a postgraduate veterinary doctor with training in animal welfare. There are quantitative methods for behavior assessment, such as the open field test or the new object test, that have been used in experimental conditions and have proven to be useful for the assessment of mental states and temperament in animals [
49]; however, they need to be validated for sheep species under field conditions.
For evaluated farms, on average long flight distances (average 8.7 ± 4 m) were found, this is a common situation in sheep production systems where painful procedures, such as tail docking and castration, are performed without anesthetics or pain relief methods [
50]. Only one of the farms in this study had no flight distance but positive approximation to the observer; this farm did not do any castration or tail docking procedures, and the farmer kept close contact with his sheep. In addition, handling procedures such as deworming, transport and identification use a variety of physical and sound signals, which increase the sheep’s reactivity and affect human–animal interactions [
51]. Shearing has also been reported to be a major cause for increased reactivity and stress in wool sheep, due not only to the handling procedure itself but also for the energy consumption necessary for thermoregulation after shearing [
52], especially in high lands with low temperatures such as the ones in this study.
There were relationships between reactivity, energy level, sheep handling training and job satisfaction and flight distance in this study. Regarding the amount of training received by producers (in the last 2 years), we found that as the amount of handling training increased, average flight distance decreased. This result is similar to findings in other studies that have demonstrated that livestock handling training is directly correlated with behavioral indicators and easiness of handling in abattoirs and farms [
53]. This suggests that proper training of farmers may lead to an improvement in reducing the stress produced in situations such as shearing. Knowledge about animal welfare itself does not avoid bad handling practices or bad human-animal interaction experiences, since even when people have knowledge about animal welfare, their attitude towards animals can be negatively influenced by their own internal perceptions of the world [
47]. Whatever the mechanism by which job satisfaction in farmers significantly reduced the flight distance in their sheep, one possible explanation could be that there is a known relationship among farmers’ or handlers’ behavior toward animals that is influenced by the attitude and personality of the farmers [
54]. Farmers in post-conflict areas suffered from violence that affected their lives in many respects. Besides displacement, the physiological trauma and emotional distress [
55] negatively affected farmers’ mental health in a scenario where almost ten percent of adults in post conflict areas suffer from a mental health issue such as depression, post-traumatic stress disorder, and anxiety [
56]. Therefore, since farmers and stock people have the greatest influence on animal welfare [
51], training that involves all these dimensions is required to positively affect animal welfare.
Our findings present some key aspects for small flock rural farmers that could have an important effect on animal welfare problems on farms, and some iceberg indicators to serve farmers as a tool to study and compare animal welfare over time. Since high hill, extensive grazing lands tend to lack good pastures and have low average BCS, as long as farmers have proper training, this could be a powerful tool for quickly assessing the nutritional state of animals as an aid to determine when to implement solutions to ensure that sheep obtain all nutritional requirements and improve both production and welfare. Meanwhile, extensive rural systems tend to lack shelter, whether natural or man-made, for protecting sheep against extreme weather conditions. Furthermore, average fleece cleanliness allows farmers to identify protection from environmental conditions during any handling procedure. We suggest that farmers could use silvopastoral systems to provide natural shelter for sheep kept outdoors that will help to diminish this problem while improving animals’ nutrition [
57,
58].
Low contact with sheep in these extensive systems could lead to health problems to go unnoticed, until they affect production and welfare parameters, for instance gastrointestinal infestation (i.e., Haemonchus contortus) as one of the prevalent conditions for sheep production systems. FAMACHA© allows farmers to detect anemic sheep (Score ≥ 4) so they can be selected and dewormed during handling procedures; however, FAMACHA© requires proper training to use on farms, since it relies on subjective color appreciation of mucosa.
Flight distance, as well as QBA, reflected the mind state of sheep, which is affected mainly by the farmer’s attitude, training and knowledge about animal welfare, and the production procedures such as tail docking, identification, or castration, all of which are affected by the farmer’s environment and traditions. The interactions among farmers and livestock should not be studied isolated from their cultural and social background. In fact, good understanding of culture and background will help to create specialized training for farmers that target their key characteristics to reduce flight distance and ease handling to improve productivity [
59].
According to this, the next step in this program and further studies to impact small sheep farmers in extensive production systems are: first, to promote and evaluate different methods for production record management that allows for constant monitoring of extensive production systems. Second, to perform and evaluate the impact of training sessions for farmers to use iceberg animal welfare indicators consistently and accurately to allow them to identify early welfare problems in their farms; and last but not least, design and evaluate training strategies for sheep handlers taking into consideration their key socio-demographic characteristics to find the best approach to improve Human-animal interactions.
Finally, we found BCS, fleece Cleanliness, flight distance and FAMACHA© to be iceberg indicators that explain 85% of variance of the overall welfare score, and QBA to be an important tool for inspecting and assessing the mind state of sheep on-farm. The 22 indicators for the AWIN sheep protocol were reduced to five; this reduction allows for a practical and quick animal welfare assessment by small farmers on extensive sheep production systems. In addition, we found interactions with respect to sheep handling training and job satisfaction to be key aspects for interventions to improve human animal interactions and sheep handling for extensive rural sheep farmers.