Characteristics of organic dairy major farm types in seven European countries

This study aimed to identify organic dairy major farm types (MFTs) in seven European countries, describe these MFTs in an open research database and assess central characteristics of the MFTs. This was conducted in a three-step procedure including (1) Identification of organic MFTs in seven European countries: Austria, Switzerland, Germany, Denmark, Lithuania, Poland and Sweden, based on existing data from dairy databases and consultations with experts within the respective fields of knowledge; (2) Collection of data on farm characteristics, management procedures, production level and herd health from at least 10 farms per MFT and country and (3) Creating an open research database on MFT characteristics, description of essential characteristics of MFTs and assessment of similarities and differences between farms within and across MFTs. The results indicate variations in herd characteristics such as production level, herd size, farm size, housing system, milking system and cow health status between organic dairy farms in these seven European countries. It also indicates variations in management strategies such as feeding, animal health management and recruitment strategies across the organic dairy sector in Europe. These variations seem to be associated with differences between regions and countries in the conditions for organic dairy production, such as topography, land availability and regulations.


Introduction
Consumer demand for organic products is increasing (European Commission 2014;Eurostat 2016) and this, together with the premium price for organic products, has prompted an increase in organic production (Rosati and Aumaitre 2004;Escribano et al. 2015). Moreover, organic production contributes public goods such as biodiversity and human and animal health and welfare (Jespersen et al. 2017). In Europe, the organic dairy sector has seen a steady increase, reaching a market share of between 2 and 11% in 2016 depending on country, with the main markets being in Western Europe (Sørensen et al. 2006). Consequently, organic dairy production in the EU has almost doubled since 2007, and by 2016 constituted more than 2.8% (4.1 million metric tons) of total EU milk production (Willer and Lernoud 2018). Organic dairy production is characterised by high-roughage diets based on home-grown feedstuffs, outdoor access on pasture and restrictive use of antibiotics (EC 2007). In organic dairy production, animal health and welfare is promoted by preventative actions such as using suitable breeds, feed, management and housing system. Accordingly, compared with conventional farms, organic dairy cattle farms have a higher amount of pasture-based feeding systems and a lower amount of concentrate feeding (Kristensen and Kristensen 1998;Valle et al. 2007;Thomassen et al. 2008) and on average a lower use of medicines (Richert et al. 2013;Van Wagenberg et al. 2016). Some of the major challenges within European organic dairy production are preventative handling of mastitis and metabolic diseases (Bennedsgaard et al. 2010;Ivemeyer et al. 2012), improvement of production efficiency based on local feed resources and development of suitable breeding strategies (Röös et al. 2018).
The organic dairy cattle sector must adapt to differences in topography, infrastructure, national and regional organic regulations, traditions and typical management procedures. Consequently, organic dairy farm types vary across Europe. These differences arise within individual countries (e.g. Ivemeyer et al. 2017), but can be expected to be of even greater magnitude between European countries. Thus, there is not one, but many, organic dairy production systems in Europe. However, no systematic mapping of organic dairy production environments in Europe is available at present. Thus, an essential step in the development process in order to enhance knowledge exchange within the European organic dairy sector, and also to provide evidencebased information for advisors, policymakers and regulation developers, is to identify and map major organic dairy farm types in Europe.

Aim
The aim of this study was to identify organic dairy major farm types (MFTs) in seven European countries, describe these MFTs in an open research database and assess central characteristics of the MFTs.

Material and methods
This study was conducted in a three-step procedure: (1) Identification of organic MFTs in seven European countries: Austria (AT), Switzerland (CH), Germany (DE), Denmark (DK), Lithuania (LT), Poland (PL) and Sweden (SE), based on existing data from dairy databases and consultations with experts (production advisors (including veterinarians) and researchers) within the respective fields of knowledge.
(2) Collection of data on farm characteristics (herd size and structure, housing and milking system), management procedures (related to feed, health and breeding), production level and herd health from at least 10 farms per MFT (Table 2). This three-step procedure was chosen in order to identify and describe some major variations and patterns in the European organic dairy sector, and was not intended to give a representative quantification of characteristics of European organic dairy farms.

Identification of MFTs in seven European countries
The MFTs in each country were identified according to a standardised protocol that covered four criteria: geographical location, herd size, housing type and production level. The protocol was completed using data from different national data sources and expert panel interviews, although methods for MFT identification varied to some extent between countries (Table 1). The relevance of the criteria considered also varied between countries, e.g. there is greater variation in geographical location and milk production level in Switzerland and Austria than in Denmark. Thus, the main criteria included in MFT identification also varied between countries (Table 1).
Collecting data on farm characteristics for each MFT The characteristics of each farm type were investigated in an in-depth farm survey that included detailed questions on approximately 90 items, such as farm location, housing and milking system, herd size and structure, production level, herd health status and preventive health management, feeding including ratio formulation, pasture management, fodder production and breeding strategies, including breeds kept on the farm and reproduction techniques used. For details of the survey, see OrganicDairyHealth database (2018). The survey protocol was developed jointly by all partners, translated into native languages and used in all countries. Data collection according to the survey protocol was performed as interviews, questionnaires, data collection from databases or a combination of these methods. Methods for identification of farms for inclusion in data collection and methods for data collection varied between countries (Table 2). Protocol information was Herd size, production level (nested within milking equipment type) 1 Poland: MFTs were defined using data from 463 organic dairy farms obtained through the national cattle database. Using PROC FACTOR method = prin in SAS 9.4® software (SAS Institute Inc. 2017), principal component analysis (PCA) was used to identify variables that explained the largest degree of variance. Input variables were the continuous variables 'herd size' and 'milk yield', and the binominal variables 'location mountain', 'location lowland', 'loose housing' and 'tie-stall'. Number of components to be retained was decided using Kaiser's criterion. On the basis of the PCA, location mountain, location lowland, loose housing and tie-stall were chosen as defining variables for MFTs. However, this created a farm type with too few observations (location mountain, loose housing; n = 6). The two mountain farm types were therefore merged into one, and three final MFTs were defined (lowland, loose housing; lowland, tie-stall; mountain, loose housing and tie-stall) collected from at least 10 farms per farm type. A survey was considered complete when critical information on housing and milking system, herd size and structure, production level and herd health status was recorded. In total, 715 completed surveys were obtained. All data collected referred to production year 2014, except for Denmark, for which information for production year 2015 was recorded.
Building a database on organic dairy farm characteristics Data obtained in the data collection process were compiled into a common database with information on farm level to be used as input to other work packages in the OrganicDairyHealth project (OrganicDairyHealth database 2018). Database information summarised on MFT level is publicly available and can be accessed at http://projects.au.dk/coreorganicplus/researchprojects/organicdairyhealth/database-summary/. It is intended for use in future research projects. The database is extensive and includes more than 250 variables describing detailed characteristics of the 715 farms surveyed. In this paper, we do not aim to provide a complete description of all recorded characteristics, but focus on describing critical farm, herd and management characteristics.

Identified MFTs
The number of MFTs identified varied between countries. Definitions of the MFTs identified in each partner country are presented in Table 3.
Farm size and structure Farm size, in terms of number of cows and area of arable land, varied between MFTs (median 11 to 143 lactating cows per herd, median 28 to 385 ha per farm). Size showed skewed distributions within MFTs, as indicated by the interquartile ranges (IQRs) and the relationships between medians and IQRs (Table 4). The availability of arable land for pasture and access to semi-natural pasture varied between regions, and thus between MFTs (Table 4). The age distribution of the cows in the herd varied between a MFT median of 16.7 and 35.8% for first-parity cows in the herd. Age distribution also varied between countries, with a higher proportion of first-and second-parity cows in MFTs in Austria, Denmark and Sweden and a larger proportion of cows in parity 3 or older in MFTs in Germany, Lithuania and Switzerland (Table 4). Variation was also observed in average age at  Table 4 Farm size and structure, housing and milking system, feeding, production level, herd health and use of breeds for the major farm types (MFTs) in different countries. For MFT definitions, see Table 3. IQR, interquartile range     Code structure for MFT names for each country is given in Table 3 first calving, which was higher and showed greater variation in MFTs in Austria, Switzerland and Germany compared with MFTs in Denmark, Lithuania and Sweden (Table 4).

Housing and milking system
Milking and housing systems were interrelated. A milking parlour was the most common milking system in loose housing systems and a mobile single milking machine was the most common in tie-stall housing systems ( Table 4). The majority of all farms had cubicles with straw in the lying area (Table 4). Farms with straw yards or sloped floors were identified as relevant housing systems within MFTs only in Germany, Sweden and Poland.

Feeding strategy
The proportion of roughage in the winter ration varied between a median of 56 and 85% on dry matter (DM) basis for the MFTs surveyed. A feeding strategy including only hay as roughage was most common among German, Swiss and Austrian MFTs (Table 4).

Production level and breeds
Median production level varied from 3000 to 9505 kg energy-corrected milk (ECM) per year, with the lowest production levels on Polish MFTs and the highest on MFTs in Denmark and Sweden. Milk protein and fat content varied between MFTs, but with no clear patterns.
A majority of all farms had one major breed in the herd, but crossbreeds in the herd were common. The proportion of farms with only one breed was lower in MFTs in Sweden, compared with the other countries (Table 4).

Herd health and treatment routines
Average herd somatic cell count (SCC) over the year varied between a median of 135 and 858 × 10 3 cells/mL, with lower levels in some Austrian and Swiss MFTs and higher levels in Polish MFTs (Table 4). Most MFTs had an average mastitis prevalence of between 10 and 20% of lactating cows. Average culling rate varied between 8 and 34% of lactating cows (Table 4). A majority of farmers on all MFTs except one (a German MFT) used antibiotics routinely for treatment of clinical mastitis. Use of the practice of drying off teat quarters and use of homoeopathic treatments varied more between MFTs, both between and within countries. A higher proportion of farmers in Lithuania and Sweden used drying off quarters as a mastitis treatment and a higher proportion of Lithuanian and Swiss farms used homoeopathic treatment (Fig. 1).
Most farms used veterinary treatments to treat metabolic disease. Homoeopathic treatment was more commonly used by farmers in Lithuania and Switzerland and small German farms (Fig. 2).

Discussion
The objectives of this study were to identify major farm types (MFT) in seven European countries, describe these MFTs and assess their critical characteristics. We identified 22 MFTs in the countries studied and entered descriptions of these MFTs into a public research database (http://projects.au.dk/coreorganicplus/researchprojects/organicdairyhealth/database-summary/), where the variation in farm characteristics is illustrated. The data collected is not, and was not intended to be, a representative sample of the organic herds in each country, as we did not intend to quantify organic dairy production characteristics. This study, and the methodology used, aimed to capture some major variations and patterns in European MFTs and the characteristics of these. The same basic protocol for data collection was used in all seven countries, but the method of data collection varied between countries, from internet-based questionnaires and compiling data from databases to telephone or in-person interviews. Differences in data collection methods, in combination with differences in management and interpretation of questions between countries and farms, may have influenced the accuracy of the answers, which must be taken into account when interpreting the results.
The results from the survey indicated that the highest milk production level and largest herd sizes are found in northern regions of Europe, i.e. Denmark, Sweden and northern Germany. The smallest herd sizes and lowest milk producution levels occur on Polish, Austrian and Swiss organic dairy MFTs, especially in mountain regions, which also displayed relatively larger variation in these characteristics. Moreover, the Lithuanian data on herd size showed great reported variation, varying from relatively small to the largest of all countries included in the survey. These results are in line with Sørensen et al. (2006), who described dairy cattle production in the industrialised world as highly specialised and with high milk production levels. However, that study also reported large differences, with dairy farms in the Netherlands, Germany, Denmark and Sweden typically being characterised as highly intensive. In the present study, we found that the countries reporting the largest variation in MFTs were Austria and Germany. These countries are situated in a region of Europe that has both lowland and mountain dairy systems. Nevertheless, in Germany, the most extreme differences between MFTs were constituted not only by topography (lowlands in the north and mountainous in the south), but more regarding production intensity (partly within the same regions; Ivemeyer et al. 2017). According to Stuaro et al. (2009), in recent decades, the number of traditional extensive farming units has decreased in favour of highly mechanised and intensive production practices and this decrease has been particularly dramatic for the traditional extensive livestock farms of the Alpine region. Apart from herd size, the amount of farm land also varied between MFTs, indicating differences in availability of arable land for fodder production, pasture and access to semi-natural pasture. Thus, conditions for dairy production differ between regions, explaining part of the variation in dairy production systems observed in the countries and regions included in this study.
Among the farms surveyed in this study, the majority of cows were kept in loose housing systems, but there are still some tie-stall housing systems, typically in Alpine regions (Austria and Switzerland), Sweden and central Europe (Poland and Lithuania). This is in line with previous findings (e.g. Krieger et al. 2017), indicating that the farms and MFTs included in this study reflect the situation in the European dairy sector. These differences are partly explained by different conditions due to topography in mountain and lowland regions, partly by cultural, traditional and socio-economic differences and partly by differences in regulations in different countries and regions. Since 2010, EU regulations on organic production demand that cows be loose-housed, with the exception of small farms (EC 2007). However, it is not permitted to build new tie-stalls even for small farms. The threshold for 'small dairy herds' was not defined consistently in the previous version of EUs regulation (before 2018) and is instead defined at national level (Swensson 2008;Barkema et al. 2015). Countries and federal states within countries have defined these cut-offs individually, e.g. 20-35 cows, depending on federal state, in Germany, 35 livestock units in Austria, and 45 cows in Sweden (Swensson 2008). Altogether, this indicates that a considerable proportion of the cows on organic farms in Europe are kept in tie-stalls with access to pasture during the vegetation period and regular exercise during the remainder of the year, but this proportion can be expected to decrease in the future. Milking and housing system were found to be interrelated. A milking parlour was the most common milking system in loose-house systems and mobile single milking machine was the most common in tie-stall housing systems. Milking robot systems were used in some MFTs and seemed to be more common in regions with higher milk yields, i.e. northern Germany, Denmark and Sweden. This may partly be explained by high investment costs for automated milking systems and the need for high yields for good returns on this investment. However, it may also be attributable to differences in culture and experiences between regions, e.g. regions where automatic milking is common also in conventional dairy production (Svennersten-Sjaunja and Pettersson 2008).
According to the EU regulations on organic agriculture, at least 60% (50% at the beginning of lactation) of the dry feed matter has to be roughage (EC 2007). The results of our survey indicate a wide range in feeding intensity on organic dairy farms. The highest values are more than 90% roughage in the dry matter ration on Swiss MFTs and typically between 80 and 90% on MFTs in Austria, Germany and Lithuania. The lowest shares of roughage in the ration (below 70%) are used especially on all MFTs in high-yielding regions, such as in Sweden and Denmark. This variation reflects differences in national and regional regulations. For example, according to 'BioSuisse' standards of organic farming in Switzerland, the proportion of concentrates in the yearly ration for cattle-defined as cereals and grain legumes such as soybeans-may not exceed 10% of dry matter (BioSuisse, 2017). Krieger et al. (2017) observed differences in concentrate feeding levels for organic dairy farms in four European countries, with an average of 616, 1200, 1500 and 2373 kg concentrates per cow and year in France, Germany, Spain and Sweden, respectively. These results support our findings of a wide range between countries and of high average concentrate amounts fed in Sweden of more than 30% in the dry matter ration. Organic dairy farms in the UK are reported to use on average 1800 kg concentrates per year (Langford et al. 2009). Thus, our results support und complement previous findings and reflect typical variations in feeding management within the European organic dairy sector, caused by differences in available resources and differences in regulations between regions.
According to organic principles and standards, disease prevention must be a high priority in organic dairy production (IFOAM 2014). Disease prevention should primarily be ensured by implementing site-adapted breeding, management, nutrition and housing appropriate to the animal species (EC 2007). However, the regulations clearly state that diseased animals should be treated with appropriate methods to prevent animals suffer (e.g. antibiotic treatments in cases of serious infections). Regarding herd health, in the present study, we focused on mastitis and metabolic diseases as two important health impairments in dairy cows in general (reviewed by Martin et al. 2018 andSundrum 2015) but also in organic dairy cows (Ivemeyer et al. 2012). Udder health was measured as arithmetic mean of somatic cell count (SCC) over 1 year from test day data and, for both udder and metabolic health, treatment incidence was used to assess disease incidence. When interpreting the results, it has to be taken into account that differences in the use of antibiotics in particular, but also other veterinary medicines, may be due not only to infection levels but also to farmers' attitudes to treatment choice (Vaarst et al. 2006;Bennedsgaard et al. 2010).
The results show a wide range of udder health conditions in the European organic sector, with the lowest SCC in Alpine and northern European regions and the highest in central European regions. While the results have to be carefully interpreted regarding absolute SCC, they are in line with other findings of similar patterns between regions. They indicate that SCC in organic herds in several European countries is on such a high level that improvements are necessary (Ivemeyer et al. 2012;Krieger et al. 2017). Mastitis treatment incidences found in the present survey support findings by Ivemeyer et al. (2012) that mastitis incidences range widely and that mastitis is a major reason for antibiotic treatment of organic dairy cows. Moreover, all treatment incidences on MFTs (average ranging from 7 to 40 treatments due to udder problems per 100 cow-years) were within the range reported in other studies (e.g. Fall and Emanuelson 2009;Bennedsgaard et al. 2010). Besides production method, management intensity has been found to have an impact on medicine use. Intensive production systems are reported to be closely associated with frequent veterinary treatment (Richert et al. 2013). However, the results from our survey did not indicate interrelations between production intensity (e.g. milk yield and feeding ratio) and treatment incidence for mastitis or metabolic diseases within organic production systems.
Interestingly, the results of the survey indicated that non-antibiotic (non-allopathic) treatment regimes varied more between regions and countries than between MFTs. While e.g. drying off single mastitic udder quarters during lactation seems to be a well-known measure in Sweden and Lithuania, and is also used by some organic farms in Switzerland and Denmark, it seems to be uncommon in Germany and Austria. The use of homoeopathic treatment also varies, being quite common in countries like Switzerland and Lithuania and less used in Denmark and Sweden. The results related to treatment strategies indicate differences between regions and countries caused by differences in regulations and culture. This emphasises the importance of cross-country and region knowledge transfer on farmer and advisor level.
The age distribution of the cows in the herds varied between MFTs, with higher proportions of first-and second-parity cows on Austrian, Danish and Swedish MFTs and higher proportions of cows in parity 3 or older on German and Lithuanian MFTs. Moreover, there was variation in average age at first calving, which was reported to be higher and to have a larger variation on MFTs in Austria, Switzerland and Germany compared with MFTs in Denmark, Lithuania and Sweden. This might partly be explained by breed differences where dual-purpose breeds, which usually have a higher age at first calving, are more common in Alpine areas. Accordingly, culling rates varied widely between MFTs. MFTs with a long productive lifetime and a low culling rate were found particularly in Lithuania. A higher culling rate than a median of 25% was found on the highyielding MFTs in Sweden and Germany. Taken together, these results could indicate differences in dairy cow recruitment strategies between regions, but are also clearly linked to production intensity, where a high intensity is connected to reduced longevity and fertility.
More than one-third of the farms surveyed had one major breed in the herd, but also kept some crossbred cows in the herd, indicating that breeding strategies including crossing of dairy breeds are widespread in the European organic dairy sector. The majority of these crosses were between high-producing dairy breeds, but crosses with native dairy breeds also occurred. The IFOAM norms for organic agriculture state that the breeds used in organic production should be adapted to local conditions, and the use of native breeds is promoted (IFOAM 2014). The widespread use of crossbreeds may be an indication that organic dairy farmers across Europe try to take advantage of heterosis and to benefit from both high-producing modern breeds and the functional traits of locally adapted native breeds.
One of the aims of the study was to describe characteristics of the MFTs in an open research database. The primary aim was to provide input to other work packages of the OrganicDairyHealth project, which has been achieved. A secondary aim was to make the information in the database openly available for future research projects interested in farm characteristics of organic dairy farms. Thus, information from the database summarised on MFT level has been made publicly available for anyone interested to use it for research purposes. However, as the project is finalised, there are no plans for further expansion or development of the database.

Conclusions
This survey revealed variations in herd characteristics such as production level, herd size, farm size, housing system, milking system and cow health status between organic dairy farms in seven European countries. There were also variations in management strategies such as feeding, animal health management and recruitment strategies across the organic dairy sector in Europe. These variations seem to be caused by differences between regions and countries in the conditions for organic dairy production, such as topography, land availability and regulations. Future development of organic dairy production in Europe would benefit from exchange of knowledge and experience, on farm and advisory level, across regions and countries.