First evaluation of the practicability of the CLASSYFARM welfare assessment protocol in Italian small-scale mountain dairy farms - a case study

Abstract The challenges facing the health sector and the agri-food industry have become increasingly complex over the years. Animal welfare, drug consumption, the use of the slaughterhouse as an epidemiological observatory and biosecurity, as well as the health risks associated with it, are increasingly interconnected, making an integrated approach necessary. Hence the European Commission developed regulations to ensure a certain standard for animal welfare in livestock husbandry, which led the Animal Health and Veterinary Drugs Directorate of the Italian Ministry of Health to design an integrated system called Classyfarm, aimed at categorising farming according to risk. The aim of our study was to consider the protocol of the Classyfarm system for evaluating the present situation in South Tyrolean dairy farming (1) and discuss the practical implementation of this new system in small-scale mountain dairy farming (2) which largely differs from large dairy enterprises in the lowlands. The results provide important insights about farm structure and management. However, in terms of animal welfare the questionnaire of Classyfarm is not able to properly differentiate and therefore results need to be considered with caution. Similarly, the section about biosafety is hardly applicable for small-scale mountain farms due to limited herd size and building capacity. Therefore, authors underline the necessity of developing a further differentiated questionnaire for the Classyfarm system, which considers the peculiarities of small-scale mountain dairy farming and allow proper conclusions regarding sanitary risk for public health unless the long-term aim is that such structure will stop producing which will have significant effects on many other sectors in the region, e.g. tourism. HIGHLIGHTS The section for Biosafety in the present Classyfarm protocol is hardly applicable to small-scale mountain dairy farms. Results of animal-based measures do not offer differentiated insights in the present situation. The present benchmarks for some animal-based measurements like mortality are not applicable for small scale-farms as they provide misleading information.


Introduction
Over the last few decades, the importance of animal welfare has been constantly increasing to become a broadly debated topic on national and international level (Battini et al. 2014;IFC. 2015).As a result, many countries have developed numerous projects regarding the well-being of livestock (Bock and van Leeuwen 2005;Veissier et al. 2008;Egger-Danner et al. 2012).Due to the influence of urbanisation, civil society organisations, media and increased societal education, consumers demand a certain standard for animal welfare in animal production systems (Koknaroglu and Akunal 2013;Battini et al. 2014).Also, there is evidence that promoting animal welfare helps in preventing transmissible diseases and thereby reducing the use of antibiotics, which helps to fight the rising antibiotic resistance in livestock production (Regula et al. 2003;Bengtsson and Greko 2014).Thus, the on-farm assessment of animal welfare has become an important issue in livestock production and the European Commission has been developing regulations to ensure high standards for animal health in farm animals (Battini et al. 2014;Mariottini et al. 2022).In order to meet these objectives, Italy launched the Classyfarm system in 2018, which is a unified system financed by the Ministry of Health, elaborated by the Istituto Zooprofilattico Sperimentale of Lombardia and Emilia Romagna together with the University of Parma.This new integrated system for assessing the potential risk of farms for sanitary health does not only consider animal welfare assessment (Classyfarm 2023a), but also combines these data with the electronic prescription system for veterinary drugs (AACTING 2021;De Monte et al. 2020), official national registry as well as slaughter animal findings from the slaughterhouses for sanitary risk evaluation.The Classyfarm system operationally rolled out in 2019 for pig husbandry and in 2023 for dairy cow husbandry in Italy.For dairy cows, a questionnaire with requirements in the fields of biosafety, farm management, barn structure, animal-based measures (ABMs) as well as hazards and alarming systems is considered (Bertocchi et al. 2020).Depending on the husbandry system some questions were differentiated for tie stall (TS) and free stall (FS) housing systems.The methods for calculating the coefficients used to determine the level of risk for animal welfare and public health of agricultural holdings are not accessible to the public (Classyfarm 2023a).The participation of famers in Classyfarm is not mandatory in the first place but obligatory for obtaining the animal welfare label Sistema di Qualit� a Nazionale per il Benessere Animale -SQNBA.The aim of SQNBA is not only to improve animal welfare and animal health but also to ameliorate processes in production and supply chain in terms of economic, ecological and social sustainability thanks to accreditation of farms and other stakeholders within the production chain (Classyfarm 2023a;Governo Italiano 2020).
As Classyfarm will become mandatory for dairy farms for receiving the national animal welfare label SQNBA from 2023 onwards, the South Tyrolean dairy association in collaboration with the Free University of Bolzano initiated the project Tierwohl S€ udtirol in January 2022 to prepare and brief the local dairy farmers for the upcoming program.Several auditors visited South Tyrolean dairy farms to collect data according to the protocol for welfare assessment of Classyfarm.The objectives of this study were (1) to evaluate the results of the respective field of the questionnaire in order to highlight strengths and weaknesses of South Tyrol dairy farming, and (2) to discuss the practical implementation of the Classyfarm system especially with regard to small-scale mountain dairy farms, which largely differ in farm structure, management and organisation from large dairy enterprises in the lowlands.

Study area
The province of South Tyrol is the very northern part of Italy with the border to Switzerland and Austria.It is characterised by a mountainous landscape with steep meadows and pastures where livestock farming is practiced on small-scale, family run farms (K€ uhl et al. 2020;Zanon et al. 2021).In the region 61,332 hectares of grassland are used to produce feed and about 127,216 hectares of mountain pastures are being grazed (Agrar-und Forstbericht, 2021).A total of 4354 dairy farms are located in South Tyrol with an average farm size of 15 dairy cows per farm (T€ atigkeitsbericht Sennereiverband, 2021).In 2021, average milk production per lactation and cow was 7,725 litres (Agrar-und Forstbericht 2021).

Data collecting
Between January and December 2022, 27 auditors from the Sennereiverband S€ udtirol (Alpine Dairy Association of South Tyrol), the Free University of Bolzano and BRING (Consulting Ring for Mountain Farming) assessed 3,469 dairy farms, which is about 80% of all dairy farms located in the province of South Tyrol, considering the Classyfarm welfare assessment protocol.The data was collected using a mobile APP called Awenko:360, which was specifically designed for the purpose of this project.The assessors were briefed by the Sennereiverband S€ udtirol and experts form animal science research group of Free University of Bolzano on the use of the mobile APP as well as on each point reported in the checklist and their special characteristics that has to be considered to provide a certain objectivity in the assessment.The checklist was the one considered for the Classyfarm system which includes 99 questions for TS systems (Classyfarm 2019a) and 105 for FS systems (Classyfarm 2019b) split on the five macro-areas Biosecurity, Management and Personnel, Structures, ABMs, Hazards and Risks (Classyfarm 2019a/b).For simplifying the farm audits we have only considered adult animals (lactating cows, dry cows) for assessing ABMs as well as combined some questions regarding management (e.g.cleanliness of the floor) and structure (e.g.laying area, bedding material, drinkers) of the different groups of animals (Tables 6,7,9,10).For scoring ABMs, resource-based measures and management-based measures a two-point as well as a three-point scale were considered, where the highest number (2 or 3) represented the best result and the lowest number (1) the worst result of each indicator.The participation of farmers in this survey was on a voluntary basis.Besides applying the checklist of the Classyfarm system regarding the five macro-areas for assessing small-scale South Tyrolean dairy farms, we considered an additional questionnaire with further questions regarding pasture access and number of days spent on pasture for each group of animals (lactating cows, dry cows, young stock).

Animal based measures according to Classyfarm
Other than the Classyfarm protocol where maximum 30 animals per group (lactating cows, dry cows, heifers, calves) per farm are being assessed, for this project maximum 10 adult cows (lactating cows, dry cows) per farm were assessed individually.If a farm kept fewer than 10 dairy cows, all were considered for this evaluation criteria.The ABMs were performed after the scheme of the Classyfarm manual (Classyfarm 2023b).Body condition score (BCS) was assessed from behind the animal, looking at both sides with special attention on protruding bone points and fat accumulation of the tail base, the lumbar spine, the hip bone, and ischial tuberosity.The cows were scored with 1, when lean (BCS < 2) or fat (BCS > 4,25), and 2 when normal (2 < BCS < 4,25).Avoidance as in the scheme of the Classyfarm protocol distance was evaluated in initially staying in front of the animal with an estimated distance of 2 metres and slowly approaching the animal reaching out the right hand.Avoidance distance was scored with (1) when the animal tried to avoid the hand with a distance more than 1 m from the muzzle, with (2) when the distance was less than 1 m form the muzzle and (3) when the person was able to touch the muzzle.For evaluating the dirtiness, the auditor stayed with a maximum distance of 2 metres from the animal and looked out for dirty spots on udder, upper and lower hindleg.If the dirty spots summed up an area as big as a palm the cows was defined as dirty and scored with 1, if the cow was completely clean it was scored with 2. Skin lesions were distinguished in (1) heavy (10 or more small lesions <4 cm diameter, 1 big lesion >10 cm in diameter, swollen areas, open wounds or big scars), ( 2) slightly (15 small injuries <2 cm in diameter, 2-10 lesions <4 cm in diameter or 1 medium injurie between 4 -10 cm in diameter) or (3) none (up to 14 small lesions < 2 cm in diameter or 1 lesion up to 4 cm in diameter).Lameness was evaluated from behind and both sides of respective animal.Depending on the husbandry system (TS or FS) the animal was assessed while walking and/or standing.During the examination in the stand (especially for TS) it was looked for whether the investigated animal relieved a leg, showed a malposition of a limb, an abnormal pelvic position or unilateral muscle atrophy.Additionally, while the investigated animal was moving it was observed if the animal bent its back, showed head nodding, was limping, or showed other irregularities while walking.The rating was differentiated in (1) lame and (2) not lame.For estimating the claw conditions all four claws were inspected for alterations in sense of length or deformations, as well as lesions or infections and irregularly surface.A distinction was made between (1) not adequate and (2) adequate.The described method was used to evaluate the single animal (Table 8).The average score based on the prevalence of every indicator of the 10 (or less) single assessments was calculated and transferred into the three-point scale after the model of the Classyfarm protocol (Table 9).

Statistical analysis
For the statistical analysis the SAS software v. 9.4 (SAS Institute Inc., Cary, NC) was used.The dataset considered for the analysis comprised 336,448 records of 3,469 dairy farms which corresponds to approximately 80% of all dairy farms located in the area and 23,976 dairy cows.For the present study, descriptive statistic was performed for depicting management and husbandry as well as describing the status quo regarding biosecurity, animal welfare, as well as hazards and risks in South Tyrolean dairy farms (Figure 1).

Characteristics of South Tyrolean dairy farms
On average, investigated farms kept 12.3 lactating cows, 2.5 dry cows as well as 3.5 calves and 6.1 young stock (Table 1).Farms often keep more than one breed within the barn, so we declared the breed which was represented the most in the respective barn as the main breed.The investigated farms were keeping Fleckvieh (42%), Brown Swiss (26%), Holstein Friesian (15%), Alpine Grey (14%) and Crossbreeds, Jersey, Pinzgauer (1%) as their main breed, respectively (Table 2).In terms of husbandry system, 71.6% (2,484) of farms used TS and 28.4% (985) used FS for lactating dairy cows.Further, results revealed breed specific husbandry as the percentage of TS in farms breeding Alpine Grey cattle was 93.4%, while for Holstein Friesian the percentage was lower with 53.4% (Figure 2).For the other breeds the percentages of TS housing varied between 59.3% for Crossbreeds to 73,4% for Fleckvieh (Figure 2).Generally, farms that kept Holstein Friesian cows as the main breed had the largest average herd size of 20.9 cows, while farms that kept Jersey as the main breed had the lowest average herd size of 9.7 cows (Table 2).

Biosafety
The importance of biosecurity gained a lot due to the rising concerns about animal health, the spread of The main breed was considered as the breed that was represented the most within the herd.zoonotic diseases and antibiotic resistance (Gunn et al. 2008;Layton et al. 2017;Baraitareanu and Vidu 2020).Biosecurity refers to all actions done to prevent the introduction and spread of transmissible diseases either inside or between populations (Dutil 1999).According to Baraitareanu and Vidu (2020) biosecurity action plans should mainly be applied on large farms.On large dairy farms, the risk of disease transmission can be higher due to the larger number of animals and the potential for frequent movement of equipment, personnel, and animals, which is one of the major causes of spread (N€ oremark et al. 2013).This can make it more challenging to control and contain diseases and may require more extensive biosecurity measures to be put in place than in small-scale dairy farms.In terms of biosafety requirements according to the Classyfarm protocol, South Tyrolean dairy farms, on average, showed deficits in most of the questions (see Table 5), with exceptions in the fields of unloading and loading animals or cadavers with a minimum distance of 20 metres away from the barn, knowledge and use of antiparasitic care as well as in taking possession of controls of the supplied water (Table 5).The deficits are partly related to the limited building and facility design in small-scale mountain farms as different barn structures, quarantine compartments or hygiene locks are in many cases not realisable due to limited space availability and topographic constraints in mountain area as well as limited financial turnover as a consequence of the limited farm size (Table 1).Consequently, the Classyfarm system does not fully consider the peculiarity of small-scale mountain farms, and therefore, might lead to misleading conclusions from risk analysis.Therefore, the checklist in this section should be adjusted for mountain farms and its special conditions.

Management and personnel
The results of the management-based measures showed that the South Tyrolean farmers have, overall, good skills and knowledge in handling and managing their animals (Table 6).Some farms showed poor results regarding their boxing management.In 13.4% of the cases for dairy cows, 10.2% for dry cows, 16.9% for the young stock and 9.9% for calves, dirty boxes or detrimental lying surface was determined (Table 6).Dairy cows have a strong physiological need for resting with approximately 12 h per day (Cook et al. 2005;Munksgaard et al. 2005;Gomez and Cook 2010).When those needs are not met physiological function, health, productivity, and longevity are compromised (Grant and Miner 2015).Thus, proper boxing management is essential for maintaining cow health and welfare which in turn can improve milk production and farm profitability (Grant and Miner 2015).Further, dirty cows are more susceptible to diseases such as mastitis (Verbeke et al. 2014), hence proper boxing management practices, such as regular cleaning and disinfection of facilities can help prevent the spread of diseases and reduce the need for antibiotics (Gosling 2018).Additionally, 13.3% of the FS showed an absence of calving pens (indicator is not included in the questionnaire for TS).Generally, it can be assumed, that South Tyrolean dairy farmers performed well in management-based measures, which could be down to the reason, that the small herd size gives the farmer the option to have a close observation of every single animal every day and immediately respond to every need that comes up.

Farm structures
The indicators for structures within the barn showed very good results in space requirements per cow and the ratio of number of lying boxes and feeding places to number of animals in all groups (dairy cows, dry cows and young stock) (Table 7).Also, requirements were met in the field of usage of the lying boxes, lighting conditions, temperature, humidity, pollutant gas concentration in the barn as well as the maintenance of the milking equipment and space conditions for the waiting room before milking, with only some exceptions in a small number of farms (Table 7).The majority of investigated farms that had calves at the current time provided a single box for calves younger than two weeks (95%), whereas 32,2% needed some improvements in giving the calves possibilities to make eye or even physical contact with other calves (Table 7).It is mandatory by European law to provide perforated walls for visual and tactile contact between the calves as long as the holding keeps more than 6 calves (Art.3, 2008/119/EC).In this regard, Jensen and Larsen (2014) showed how early social interaction between calves affects their social behaviour towards other calves positively.The main critical points of South Tyrolean dairy farms referred to firm lying surface and/or absence of litter, low water pressure of the drinkers in TS, the presence of a cow trainer in TS and the absence of an infirmary box (Table 7).Several studies have shown the impact of lying surface and the management of the cubicles on skin alterations, hock lesions and locomotion (Hernandez-Mendo et al. 2007;Husfeldt and Endres 2012;Brenninkmeyer et al. 2013;Katzenberger et al. 2020) and thus, directly affects animal welfare.For example, Husfeldt and Endres (2012) compared mattresses and deep-bedding boxes, which both represent soft materials, and showed that the latter performed significantly better on the above-mentioned health indicators.In our study some farms used hard surfaces for lying areas like concrete, tiles, or wood, which is rather responsible for leg injuries than mattresses (Fulwider et al. 2007) and hence even more detrimental.The water flow rate of the drinkers should be adapted to the natural drinking speed of the cow and therefore should be at least 10 l/min (Blokhuis 2009; Ofner-Schr€ ock Schr€ ock 2018).In 47.9% of investigated farms water pressure lower than 10 l/min was measured, which indicates the need to improve water flow rate for ensuring natural drinking speed.The low water flow rate observe in our study might be related to old facilities with old tubing systems or natural springs, where the farmers depend on because of remote locations, that do not build up sufficient water pressure.In addition, cow trainers in TS are not forbidden by law but are criticised in terms of animal welfare.It is proved that the animal experiences physical and psychological stress and shows a significant reduction of natural actions like feeding, grooming, and standing (Pollmann Tschanz 2006).With increasing social awareness for animal welfare (Battini et al. 2014) the cow trainer is not a future-proofed method for keeping the cows clean and should be suggested to be removed in the long term.The absence of an infirmary box by 75.9% (Table 7) of farms could be related to the fact that most barns were built before infirmary boxes were required as well as space and financial limitations by the farmers.

Animal-based measures
For the present study we have considered 10 animals per farm instead of 30 animals per group and farm as prescribed by the Classyfarm protocol because of small herd size with an average of 14.8 adult animals (lactating cows, dry cows) (Table 1) and the time issue which we faced in practice.As approximately 70% of South Tyrolean dairy farms are run as a side-line activity it was not possible to ask a lot of time from the farmers to assess their farms.As seen in Table 8 the mean of each assessed animal welfare indicator on the single animal assessment is very close to the maximum, which represents the best possible evaluation.This is in contrast with results reported in previous studies of Mattiello et al. (2009) and Katzenberger et al. (2020), who showed some deficits regarding animal welfare in small-scale mountain dairy farms, due to the high frequency of TS housing in Alpine area.A possible reason for the overall good outcome of the ABMs might be explained by the simplified scoring scale for respective indicators used in the Classyfarm protocol compared to other welfare protocols like in the study of Katzenberger et al. (2020), who used the same indicators but reported them in a more specific and differentiated way.For instance, in Katzenberger et al. (2020) skin lesions were declared as these already if one hair less patch, lesions or swellings bigger than a 10-cent coin was detected.In contrast, in the Classyfarm protocol a skin lesion was recorded if at least 15 small hairless patches, lesions or swellings smaller than 2 cm, 2 to 10 between 2-4 cm or one of a bigger dimension than 4 cm were observed.Further, the summary of the indicators as foreseen in the Classyfarm protocol of the individual animals could possibly lead to the fact that only animals with extreme rating were considered, since moderate ratings are not considered and thus the information about the animal welfare status of a farm might be biased through this simplified presentation.Therefore, an over-or underestimation might be the logical consequence in the sense of not depicting properly and differently animal welfare in farm animal husbandry.ABMs are direct indicators of animal welfare because they reflect the surrounding influences of a husbandry system (e.g.facility structures, resource and management practices) (Zuliani et al. 2018).The World Organisation for Animal Health (WOAH) and the International Organisation for Standardisation (ISO) both emphasised the importance of using ABMs to evaluate animal welfare (ISO, 2016;Zuliani et al. 2018).In addition, the EFSA (2012) determined that ABMs are the most adequate indicators for evaluating animal welfare, which goes in line with the Classyfarm protocol, where the ABMs are being evaluated strong, with the highest magnitude score for lameness and the lowest for cleanliness (Bertocchi et al. 2018).Moreover, South Tyrolean dairy cows performed very well in the fields regarding udder health (low no. of mastitis), allowed medical interventions (dehorning) and mortality in the past 12 months of young stock and calves.A critical result was the high mortality rate of lactating cows (19%) in the last 12 months (Table 9).The relatively high mortality rate might be related to the small herd sizes on a farm.The worst result for this question is given when 5% or more of the stock died within the past 12 months.Because of limited dilution effect compared to bigger herds the 5% benchmark is mostly fulfilled in small herds (Table 1) as soon as one animal of the herd dies or is getting culled.Hence, adjustment for the threshold of this indicator is highly recommended for small-scale farms to avoid misleading conclusions.Furthermore, an important result was the relatively high percentage cows with the worst score in cleanliness (19.9%,Table 8) which is linked to inappropriate housing/bedding management.The latter was already described for small-scale mountain farms in Katzenberger et al. (2020).

Hazards and alarm systems
The results for the presence of alarm systems are very poor.For instance, 98% (Table 10) of the farms do not own a fire alarm system and the majority does not own a ventilation system, thus also no alarming system.The results for the other requirements regarding this field were very satisfactory.In terms of noise impact and artificial lightning the farms performed the best assessment in 99% (Table 10) of the cases.The requirements for inspections of mechanical devices were in 1.3% not met and 0.3% farms were spotted using illegal substances (Table 10).Only 0.3% were not keeping a stock register and 0.2% farmers were not having a register for all medical treatments (Table 10).Due to the less developed type of farming and the small sizes of the farms some results in this field are not satisfactory.Hence, it should be reconsidered if it makes sense to require alarming systems in those small-scaled farms with an average of 14.8 dairy cows as in our sample (Table 1).

Additional observation: pasture access as a strength of mountain area dairy farming
Overall, TS housing is seen as controversial in terms of animal welfare because the animals do not have the possibility for movement and thus cannot live out their species-specific behaviour (Krohn 1994;EFSA, 2009;Popescu et al. 2013).However, our results reveal that TS are still the dominant housing systems in small-scale mountain dairy farms as small herd size as well as difficult topography limit the possibility in many cases to build cost-intensive FS.Nevertheless, in this regard, it is worthwhile to consider that 63% of South Tyrolean farmers provide access to pasture to their lactating cows with, on average, 96 days per year (Figure 3, Tables 3, 4).Moreover, in almost all farms young stock (animals between 6 months and first calving) access to pasture with, on average, 157 days per year and 53 days for dry cows (Figure 3, Table 3).Furthermore, pasture use was more common in farms with TS (69.3%), than in farms with FS (47.3%).It has been proven by numerous studies, that temporary access to pasture reduces some adverse effects of tethered housing (Gustafson 1993;Krohn and Munksgaard 1993;Regula et al. 2004;Keil et al. 2006;Rutherford et al. 2008) and has positive effects on welfare for dairy cows (Arnott et al. 2017).For instance, there is evidence that the use of pasture and the length of the pasture period reduce the risks for mammary infections (White et al. 2002;Firth et al. 2019;Zanon et al. 2021), the appearance of lameness (Hernandez-Mendo et al. 2007;Armbrecht et al. 2019) as well as skin alterations (Haskell et al. 2006).Pasture gives dairy cows the possibility to live out their species-specific behaviour compared to all year confinement housing (Hemsworth et al. 1995) and thus fulfils one of the five freedoms, namely 'Freedom to Express Normal Behaviour' (FAWC, 2009).Previous studies revealed that cows spent more time lying (Olmos et al. 2009), showed less agonistic behaviour (O'Connell et al. 1989) and more herd synchronisation (Roca-Fern� andez et al. 2013).Moreover, consumers perceive free range and grazing systems as very animal-friendly and positive in terms of animal welfare (Ellis et al. 2009;Miele et al. 2011).The questionnaire of Classyfarm does not neither contain a specific question for the number of pasture days nor the precise circumstances (e.g.hours, space allowance).The effective number of days of pasture is only addressed slightly in some questions in the field of barn structures regarding lying area and feeding places, where the best assessment is given when 60 pasture days are provided.But the high percentage of cows that get outdoor access and the amount of pasture days in the present study can be seen as one of the big strengths as well as could be seen as a  The number behind the item refers to the number of the respective questionnaire (TS or FS) of the Classyfarm protocol.Minimum is the worst and Maximum the best score.TS: tie stall; FS: free stall.
compensation for the high share of TS, which represents the biggest weakness of small-scale dairy farming in mountain area.

General remark regarding the limitations of the Classyfarm protocol for small-scale dairy farming
Beside structural, topographic, and financial limitations of small-scale mountain farms mentioned before, some benchmarks regarding ABMs (e.g.Avoiding distance, BCS, Cleanliness, skin alterations, lameness, claw conformation and mortality (Table 9)) in the present form are not suitable for smaller herds.For instance, since the herd size is 14.8 dairy cows per farm (Table 1), one or two animals with welfare problems already exceed the threshold for certain ABM indicators and therefore the average score of the herd slips into the worst category.
That does not provide reliable information regarding animal welfare deficits derived from indirect animal welfare indicators (resource and management-based measures) and should be reconsidered.Every animal can face poor condition at some points in its life which does not need to be correlated to poor management or structure conditions.The benchmarks for some indicators like mortality should be raised to a higher threshold in order to be able to make well-founded statements for small-scale farms.Furthermore, especially in the mountain region of South Tyrol agritourism is an important source of income and crucial for the continued existence of several farms.However, according to the biosecurity indicators of the Classyfarm protocol, the presence of external persons in the barn or on the farm is not compatible, thus such farms would perform bad in this aspect.Lastly, since there is still a high percentage of TS in South Tyrolean dairy farms, which is seen as a major limitation for animal welfare, South Tyrolean dairy farms might have a disadvantage in the evaluation although the Classyfarm welfare assessment protocol does not score TS worse than FS: there is a specific checklist for TS, as well as for FS, and a farm can get very high scores  regardless of the housing system.Nevertheless, if TS might be generally rated worse also in Italy (e.g.considering the increasing societal criticism and the recent market developments in neighbour regions like Germany and Austria for striving to ban milk from Tie stalls) it would lead to the situation, that small-scale mountain farms have to give up on their business, because they will not be able to be certified with the animal welfare label (SQNBA) which will be mandatory to participate in the market.Furthermore, if farms need to reach a certain score in the results of the biosecurity indicators in order to take part on the subsequent fields and cannot be compensated by other macro areas, a crucial part of small-scale alpine dairy farms would drop out of the market.With the abandonment of those farms, there would not only be the food production lost but also the many ecosystem services like genetic resource conservation, control of water flow, pollination, climate regulation, landscape maintenance, recreation and ecotourism, and preservation of cultural heritage, which is important for rural development and a big contribution to society, especially from an economic perspective in the delicate Alpine environment (MEA 2005;Battaglini et al. 2014).Thus, it should be in the interest of all parties involved to do some adjustments with changing some requirements and benchmarks for some indicators for the peculiarities of small-scale farms in mountain areas while at the same time maintaining a high standard for animal welfare.

Conclusion
The Classyfarm protocol regarding animal welfare was tested in this study for the very first time on smallscale mountain dairy farms.Results provide important insights regarding biosecurity, management, and farm structure, as well as ABMs and risks and hazards South Tyrolean dairy farms.Our results clearly reveal that some sections included in the questionnaire in the present form, especially biosafety, are hardly applicable to the production scenario in small-scale mountain farms as they do not consider the limitations in building and facility design due to topographic and economic constraints in marginal areas.Furthermore, results of ABMs do not offer differentiated insights in the present animal welfare situation of a farm as according to the Classyfarm protocol only extreme cases are reported which to some extend might lead  The number behind the item refers to the number of the respective questionnaire (TS or FS) of the Classyfarm protocol.Minimum is the worst and Maximum the best score.
to misleading conclusions.On the other hand, some thresholds regarding certain ABMs (e.g.mortality) should be specifically adjusted for small herds to not create misleading information.Finally, it should be mentioned that within the framework of the Classyfarm Protocol it should also be possible to compensate for certain deficits which we have observed particularly in animal husbandry and biosecurity in mountain dairy farms through special management measures such as the frequently practiced grazing and transhumance.Consequently, authors recommend a stronger inclusion of the peculiarities of mountain farming and therefore a further differentiation of the Classyfarm protocol for avoiding misleading results, which could have detrimental effects on the survival and development of small-scale mountain livestock farming.The latter are crucial as they are providing multiple ecosystem services as well as are an important pillar for rural economy and development in marginal areas.

Figure 1 .
Figure 1.Distribution of husbandry system in the different groups of animals.

Figure 2 .
Figure 2. Distribution of husbandry system in Correlation with breeds.

Figure 3 .
Figure 3. Pasture access in the different groups of animals.

Table 1 .
Descriptive statistics of herd size in investigated farms.

Table 2 .
Descriptive statistics of herd size of lactating cows regarding main breed of the investigated farms..

Table 4 .
Descriptive statistics between the ratio of pasture access and husbandry system.

Table 3 .
Descriptive statistics of pasture days of lactating cows and young stock.

Table 5 .
Results of the questions regarding biosecurity.The Minimum represents the worst possible answer and the maximum the best possible answer.The number behind the item refers to the number of the respective questionnaire (TS or FS) of the Classyfarm protocol.

Table 6 .
Results of the macro-area management.

Table 7 .
Results of the macro-area structures.The number behind the item refers to the number of the respective questionnaire (TS or FS) of the Classyfarm protocol.Minimum is the worst and Maximum the best score.

Table 8 .
Results of the animal-based measures of the single cows assessment.Minimum is the worst and Maximum the best score.
BCS: body condition score.

Table 9 .
Results of the macro-area ABMs.The number behind the item refers to the number of the respective questionnaire (TS or FS) of the Classyfarm protocol.Minimum is the worst and Maximum the best score.TS: tie stall; FS: free stall.

Table 10 .
Results of the macro-area Hazards and Risks.