Influences on the assessment of resource- and animal-based welfare indicators in unweaned dairy calves for usage by farmers

Abstract

Consumers, industrial stakeholders, and the legislature demand a stronger focus on animal welfare of all livestock at the farm level by using suitable indicators in self-assessments. In order to deduce farms’ animal welfare status reliably, factors that influence indicators’ results need to be identified. Hence, this study aimed to apply possible animal welfare indicators for unweaned dairy calves on conventional dairy farms with early cow–calf separation and evaluate influencing factors such as age and sex of calves or climatic conditions on the applied indicators’ results. An animal welfare assessment using 7 resource-based and 14 animal-based indicators was conducted at 42 typical Western German dairy farms (844 calves) in 2018 and 2019 by two observers. The effect of influencing factors was calculated by binary and ordinal logistic regressions and expressed as odds ratios. Although every unweaned calf was assessed during the farm visits, most farms had relatively few unweaned calves (average number of calves ± standard deviation = 20.1 ± 6.7 calves), with six farms having not more than 10 calves. The small sample sizes question the usage of those indicators to compare between farms and to set thresholds at the farm level. Only one assessed indicator (cleanliness core body) was not statistically affected by the evaluated influencing factors. Calf age was identified as the most decisive factor, as it affected 16 of 21 evaluated indicators, and calf age distribution on-farm varied greatly. Climatic conditions (ambient temperature and rainfall) influenced resource-based indicators such as access to concentrate and water or the cleanliness of feeding implements and bedding as well as animal-based cleanliness indicators and the occurrence of health-related impairments such as coughing and diarrhea. The authors found differences between calves on farms assessed by the different observers not only in resource-based hygiene indicators but also in animal-based indicators such as hyperthermia or hypothermia, highlighting the need for further evaluation of quality criteria in dairy calf welfare assessments. Nevertheless, animal welfare assessments by farmers themselves could be useful tools to sensitize farmers to animal welfare and thereby improve calves’ welfare.

Introduction

Consumers’ demand for animal-friendly production systems has risen over the past decades, forcing the dairy industry to improve its production systems’ welfare (Weary and von Keyserlingk, 2017). Animal welfare assessment protocols have been developed to evaluate animal welfare at the farm level to ensure high animal welfare standards (Krueger et al., 2020). The Welfare Quality protocol for dairy cows, developed in the EU-funded Welfare Quality project, incorporates social, scientific, and industrial demands focusing on the animal itself by using primarily animal-based indicators (Blokhuis et al., 2010). However, dairy calves are not included in this program, despite their unquestionable importance (Krueger et al., 2020) and inherent presence on dairy farms.

The focus on animal welfare indicators has increased since then as more assessment protocols were developed by scientists for different species or husbandry systems (Can et al., 2017; Barry et al., 2019a; Berteselli et al., 2019). In 2014, the assessment of animal welfare indicators by farmers themselves was firstly included in animal welfare regulations as the German Animal Welfare Act was revised. The revised § 11 (8) of the German Animal Welfare Act demands collecting and assessing suitable animal-based animal welfare indicators by farmers for every livestock species. However, neither suitable indicators nor the form of the assessment or the assessment interval is specified by law. To solve this, the German Association for Technology and Structures in Agriculture (KTBL) developed several sets of animal welfare indicators for the different livestock species, including one for reared calves consisting of eight indicators (Brinkmann et al., 2016). Currently, no results of an application of this self-assessment protocol have been published yet. Also, important aspects of dairy calves’ welfare, such as individual health status, are missing, and national legal requirements that were not convertible into animal-based indicators (e.g., access to fresh feed and water) were left aside. Although Germany is currently the only country requiring farmers to conduct welfare self-assessments, it could be seen as a role model for other countries as social and political awareness for animal welfare is already high (Vogeler, 2019).

Using animal welfare indicators for comparison between farms and in third-party verifications of quality programs with fixed thresholds is only possible when indicators show sufficient reliability (Waiblinger et al., 2001; Pfeifer et al., 2020). Recently, projects aimed to assess intra- and inter-observer reliability and retest reliability for established indicator programs of the Welfare Quality project (Kirchner et al., 2014; Czycholl et al., 2016) or the KTBL protocol for pigs (Pfeifer et al., 2020). Another vital quality criterion is consistency over time, which describes that the assessments’ results represent the long-term welfare state of farms (Kirchner et al., 2014). Despite limitations in reliability or consistency over time, animal welfare indicators are increasingly used in practice for benchmarking, for third-party verifications in quality programs, or for self-assessments (Pfeifer et al., 2019; Krueger et al., 2020).

Animal welfare assessment protocols including animal-based welfare indicators for unweaned calves were only recently proposed (Barry et al., 2019a), and no validation has been made yet. Poor reliability and consistency over time of animal welfare indicators can be associated with the indicator’s dependence on animals’ characteristics (e.g., sex or age) or conditions on-site during the visit (Mullan et al., 2009). Calves in general, and especially unweaned calves, are fast-growing animals, undergoing considerable changes in the gastrointestinal tract, and are exposed to different health risks depending on their age (Svensson et al., 2003; Windeyer et al., 2014; Wickramasinghe et al., 2019). Also, and contrary to piglets, calves are mostly housed in outdoor or semi-outdoor environments and, therefore, influenced by ambient climatic conditions, which might influence calves’ welfare (Roland et al., 2016). Influencing factors such as calf age or the environmental conditions at the time of an assessment affected the reliability or the consistency over time of animal welfare assessments on the farm level in previous studies (Mullan et al., 2009; Kirchner et al., 2014; Can et al., 2017; Berteselli et al., 2019) and need to be evaluated to control for them.

The present study aimed to test possible resource- and animal-based animal welfare indicators to be used in assessments for unweaned dairy calves under practical conditions on Western German dairy farms. Secondly, we aimed to identify influencing factors on those indicators’ results by analyzing the effect of group size, calf sex, age, climatic conditions before and during the visit, and the observer.

Materials and Methods

This study was conducted in accordance with federal and institutional animal use guidelines (Az. 84—02.05.40.16.038), the data privacy agreement (University of Bonn, 38/2018), and ethical standards.

Resource- and animal-based indicators

Firstly, an assessment protocol including management-, resource-, and animal-based indicators was designed based on international and national welfare assessment systems (e.g., Welfare Quality, 2009; Vasseur et al., 2012; Brinkmann et al., 2016) and publications on the welfare of calves (Tables 1 and 2; Supplementary Material S1). Resource-based indicators were mainly adapted from Vasseur et al. (2012). The indicators such as diarrhea, coughing, hypothermia, and hyperthermia were included as a simplified and shorter version of the health scoring method from McGuirk, University of Wisconsin (https://fyi.extension.wisc.edu/heifermgmt/files/2015/02/calf_health_scoring_chart.pdf). Animal-based hygiene indicators can provide valid information on bedding quality, which is important for disease prevention, thermoregulation, and resting behavior of calves. As no uniform hygiene assessment exists, three different scorings (core body, carpal joints, and claws) were integrated into this study. Beside the unquestionable value of behavioral indicators, at the time of the study, an integration of behavioral indicators for calves on-farm was hampered by their ongoing evaluation status but should be addressed in future studies. Assessment and results of management-based indicators are described in detail in a related publication (Hayer et al., 2021). Only resource- and animal-based indicators were assessed, which farmers could use without extensive training or additional material.

Table 1.

Evaluated resource-based animal welfare indicators (n = 9) used to assess dairy calf housing conditions on 42 Western German dairy farms

Indicator	Definition	Score levels
Access to roughage	Calves were provided with additional roughage (straw bedding excluded)	0 = No access to additional roughage 1 = Access to additional roughage
Access to concentrate	Calf having access to concentrate	0 = No access to concentrate 1 = Access to concentrate
Access to water	Calf having access to water	0 = No access to water 1 = Access to water
Cleanliness of milk bucket	Appearance and condition of each calf’s milk bucket	0 = No soiling visible 1 = Minor milk residues visible 2 = Coarse soiling
Cleanliness of feeding trough	Appearance and condition of each calf’s feeding trough	0 = No soiling visible 1 = Minor soiling 2= Coarse soiling and spoiled feed left
Cleanliness of water trough	Appearance and condition of each calf’s water trough	0 = Clean drinking water 1 = Low turbidity, feed residues 2 = High turbidity, feed residues in water, visible biofilm development
Cleanliness bedding	Appearance and condition of each calf’s bedding	0 = Fresh, clean, dry bedding 1 = Minor soiling, slightly damp 2 = Coarse soiling, wet spots

Table 2.

Evaluated animal-based animal welfare indicators (n = 14) used to assess unweaned dairy calves’ animal welfare on 42 Western German dairy farms

Indicator	Definition	Score levels
Cleanliness core body1	Cleanliness of core body, one site was chosen randomly	0 = Less than 25% of the surface is dirty 1 = More than 25% of the surface is dirty
Cleanliness carpal joints	Cleanliness of surface around the carpal joints	0 = Clean carpal joints 1 = Minor soiled carpal joints 2 = Coarse soiling, wet carpal joints
Cleanliness claws	Cleanliness of all four claws	0 = Clean claws 1 = Slight soiling around the claws 2 = Thick crust of dirt around claws
Nesting score1	Evaluation of amount and quality of bedding	0 = Limbs not visible in lying calves 1 = Limbs partly visible in lying calves 2 = Limbs fully visible in lying calves
Underdevelopment/runt1	Evaluation of general appearance (muscles, visibility of ribs, and coat)	0 = Lack of muscles, ribs visible, dull coat 1 = Good muscles, ribs not visible, shiny coat
Hypothermia	Rectal temperature below 38.5 °C	0 = Rectal temperature > 38. 5 °C 1 = Rectal temperature < 38.5 °C
Hyperthermia	Rectal temperature above 39.5 °C	0 = Rectal temperature < 39.5 °C 1 = Rectal temperature > 39.5 °C
Ear tag wounds	Evidence of wound healing disorder around the ear tag	0 = No clinical sign of wound healing disorder 1 = Discharge of pus or blood, deformation of tissue
Horn bud inflammation1	Evidence of wound infection around the removed horn bud	0 = No clinical sign of inflammation 1 = Reddening, swelling, or pus discharge around the removed horn bud
Navel inflammation	Evidence of navel infection	0 = Normal, pain-free to handle 1 = Swelling, inflammation of navel area
Diarrhea	Evidence of diarrhea	0 = Solid or paste-like feces 1 = Watery fluid feces, pungent smell
Coughing	Presence of a cough	0 = Normal breathing 1 = Spontaneous or continuous coughing
Visible skin injuries	Assessment of abrasions or skin damage	0 = Absence of lesions or wounds 1 = Visible lesions or wounds
Sucked teats	Evidence of sucked teats	0 = Normal tissue around teats 1 = Swollen tissue around teats

¹Indicator based on Brinkmann et al. (2016).

Farm visits

A total of 21 resource- and animal-based indicators (7 resource-based indicators and 14 animal-based indicators) were applied in a single animal welfare assessment of a total of 844 unweaned dairy calves held at 42 dairy farms in the regions of Rhineland-Palatinate (n = 22) and North Rhine-Westphalia (n = 20) in two time periods during hibernal climatic conditions (First: November 2018 to April 2019; Second: October 2019 to December 2019). This study was limited to farms of the two regions in Western Germany as characteristics of German dairy farms differ by region in breeds used, farm size, and structure. Approximately 1,800 and 5,600 dairy farms are located in Rhineland-Palatinate and North Rhine-Westphalia, respectively. The average herd size in these regions is relatively similar, with 63 and 74 cows, respectively (BMEL, 2019). Farm selection and farm visit procedure are described in detail in the study of Hayer et al. (2021). In summary, farms, which participated in previous studies administered by the University of Bonn, were contacted, asked for participation in a study about calf welfare, and a visit was scheduled in consultation with the farmer. Upon arrival, stockpersons were interviewed (Hayer et al., 2021), followed by an assessment of the husbandry system and the recording of animal welfare indicators without the presence of the stockperson. Assessments were performed by one of two researchers, who were used to working with livestock. The assessment of housing facilities included the housing system (e.g., single, pair, or group housing), the housing position, the feeding implements in both single and group housing, and the level of social contact between adjacent calves. The two observers trained the developed animal welfare assessment protocol on two farms with 60 calves. During and after the training assessments, observers discussed the interpretation and classification of the used indicators to assure a common understanding. Like other recent studies, only unweaned dairy calves were assessed as unweaned and weaned calves differ greatly in aspects of physiology, feeding, behavior, and husbandry systems (Barry et al., 2019a). All unweaned calves on each farm were examined, except for three farms (n > 40 calves), on which the number was limited to two-third of the assessable calves due to limitation in time and resources. In this case, calves from each age segment were chosen by rolling a dice and excluding all calves getting a one or a four.

Influencing factors

This study focused on the influencing effect of the observer, group size, calf sex, age, and the ambient climatic conditions before and during the assessment. Ambient climatic conditions at the assessed farms were extracted from the open-access data bank of the German Meteorological Service for 2 d before the visit and on the day of the visit. The linear distance of farms to weather stations ranged from approximately 1 to 20 km. From these data, the average of these 3 d was calculated for minimal temperature (°C), maximum temperature (°C), average temperature (°C), rainfall (L/m²), number of daily hours of sunshine (h), and maximum wind velocity (km/h). Evaluated characteristics of calves were age, which was extracted from farmers’ records, and sex, which the observer assessed during the protocol. Furthermore, the housing form (individual, pair, or group housing) and the number of calves housed together (continuous variable) were recorded.

Data analysis

Descriptive results were illustrated as percentages of scores or farms for qualitative variables, while quantitative data were described using means with standard deviation. The number of observations (respectively, the number of calves assessed) differed due to each indicator’s limitations. For example, the nesting score from Brinkmann et al. (2016) can only be assessed for lying animals, or the wound healing of disbudding or ear tagging can only be evaluated for an animal that underwent these procedures. Each indicator was assessed for each calf whenever possible and was differentiated by a scheme with either two- or three-level scoring (0, 1 and 0, 1, 2, respectively). A linear model was created to detect and adjust to multicollinearity between the evaluated influencing factors using the assessed influencing factors as independent variables and the animal welfare indicator as the dependent variable for each animal welfare indicator. For each model and variable, the variance inflation factor was calculated using the “vif”-function of the “car” package (Version 3.0-10) in R (Version 4.0.3). All influencing factors with a variance inflation indicator greater than five, indicating critical multicollinearity levels, were removed from the analysis.

Influencing variables left after this process and included in the following analysis were: “group size,” “calf sex,” “calf age,” “average ambient temperature,” “average rainfall,” and “observer.” The influencing factor “observer” was included in the statistics to account for the possible effect of the two different investigators. However, due to the assessment of each farm by only one observer, farm effects cannot be excluded from the observers effect. For analyzing the effect of influencing factors on the assessed animal welfare indicators, logistic regressions were calculated. Ordinal indicators (cleanliness of milk bucket, feeding trough, water trough and bedding, cleanliness of carpal joints and claws, and the nesting score) were evaluated using R’s “polr” function of the “MASS” package (Version 7.3-53) to perform an ordinal logistic regression as described by UCLA (2020). For binary indicators (access to concentrate, roughage and water, cleanliness core body, runt, hyperthermia, hypothermia, ear tag wounds, horn bud inflammation, diarrhea, coughing, visible skin injuries, and sucked teats), a binary logistic regression was applied using R’s “glm” function of the “MASS” package based on Rawat (2017). Calculated logistic odds ratios (ORs) and confidence intervals were exponentiated to get ORs and confidence intervals. It is important to account for the different scales of influencing factors. In the case of nominal influencing factors (e.g., observer 1 vs. observer 2 or male vs female), the calculated OR describes one level of the influencing factor’s effect compared with the other level. For example, an OR of 2.0 for the factor “observer” can be interpreted so that the risk for a higher outcome level is two times higher for observer 2 than observer 1. However, in the case of continuous influencing factors (e.g., age or average temperature), ORs describe the increase or decrease of probabilities for a higher outcome level with every unit of the influencing factor—e.g., in the case of calf age with every day. ORs were considered statistically significant at P ≤ 0.05, with P ≤ 0.01 indicating statistically highly significant odd ratios and 0.05 < P < 0.1 indicating a statistical tendency.

Results

Descriptive results

Detailed characteristics and calf management of enrolled farms have been described previously (Hayer et al., 2021). Of the 42 assessed farms, 24 were assessed by the first observer and the other 18 by the second observer. The ambient climatic conditions and the distribution of numbers of calves assessed per farm and their sex are presented in Table 3. An average of 20.1 calves was assessed per farm, which was predominantly female. Six of the 42 farms reared bull calves, whereas the other 36 farms sold their bull calves at approximately 14 d of age. The average age of male calves was 25.2 d (range 1 to 107) and 47.3 d (range 1 to 128) for female calves. Reared calves were weaned with an average age of 10.8 ± 1.6 wk. The age distribution of unweaned calves on the farm differed greatly between the 42 farms (Figure 1). All calves were first housed in individual (n = 40) or pair housings (n = 2) after birth (time of separation between 1 h and > 1 d postpartum) and then in groups of 7.2 ± 2.9 calves. The average duration of individual and pair housing was 16.9 ± 5.4 d. Calf hutches were the only housing type used by 38.1% of the farms visited, calf pens were the only housing type used by 38.1%, and 23.8% used both most common was housing the groups in deep straw stables inside a semi-closed barn (50%) followed by deep straw stables in a barn with an open side or group hutches without a roof to cover the enclosure (23.1% each). Group hutches under a roof and group hutches inside a barn were less commonly found at the farms (14.3% and 4.8%, respectively).

Table 3.

Distribution of evaluated influencing factors on animal welfare assessments in unweaned dairy calves on 42 Western German dairy farms

Variable (unit, number of farms)	Mean ± SD	Minimum	Median	Maximum
Average temperature1, °C (n = 40)	6.1 ± 3.3	0.0	5.0	19.9
Rainfall1, L/m2 (n = 42)	2.2 ± 1.8	0.0	1.5	8.7
Amount of daily sunshine1, h (n = 38)	2.0 ± 1.4	0.0	1.4	9.7
Maximum velocity1, km/h (n = 36)	5.4 ± 1.1	3.0	5.3	7.3
Number of calves assessed (n = 42)	20.1 ± 6.7	5.0	19.0	59.0
Percentage of female calves, % (n = 42)	83.0 ± 11.4	52.0	85	100.0
Percentage of male calves, % (n = 42)	17.0 ± 11.4	0.0	15	48.0
Average age of calves assessed, d (n = 42)	42.1 ± 9.2	14.6	42.6	66.3

¹Average ambient environmental conditions for the day of the visit and 2 d before the visit were extracted from official weather stations, which differed in assessed parameters.

Figure 1.

Distribution of calf age on 42 Western German dairy farms ranked according to the median calf age. The mean calf age for each farm is visualized by an “x” in each boxplot. The black line inside each box represents the media; lower and upper hinges represent the 25th and 75th percentiles, respectively. The whiskers end at the lowest and highest values that are no outliers (marked as dots). The number of observations (n) for each farm is displayed below the farm ranks.

In total, 21 animal welfare indicators were assessed in 844 unweaned dairy calves. The distribution of the results of resource-based indicators is shown in Figure 2. High proportions of calves had no access to water, roughage, or concentrate (28.5%, 34.1%, and 41.0%, respectively). Regarding evaluating the housing environment’s cleanliness, milk buckets were identified as the main concern, with over 75.8% of milk buckets having minor or coarse soiling, while feed troughs and bedding were most often scored as clean.

Figure 2.

Percentage distribution of the sores (0, 1, and 2) of seven resource-based indicators to assess the housing environment of unweaned dairy calves (n = 844) on 42 Western German dairy farms. Marked indicators (*) were defined only by a two-level scoring system (0 and 1).

Results of animal-based indicators are shown in Figure 3 as a proportion of given scores. The calves’ core body was scored as clean in the case of 93.2% of calves, whereas carpal joints and claws were more often soiled (45.1% clean and 49.7% clean, respectively). Clinical signs of impaired animal health were below 15% (diarrhea: 14.4%, hypothermia: 12.0%, skin injuries: 8.1%, coughing: 7.9%, hyperthermia: 7.3%, navel inflammation: 7.2%, sucked teats: 3.3%, and runt: 1.9%), except for ear tag wound inflammation and horn bud inflammation, which was recorded for 53.1% and 18.2% of all calves assessed, respectively.

Figure 3.

Percentage distribution of the sores (0, 1, and 2) of 14 animal-based indicators to assess the welfare of unweaned dairy calves (n = 844) on 42 Western German dairy farms. Marked indicators (*) were defined only by a two-level scoring system (0 and 1).

Effect of influencing factors

The calculated effect of recorded influencing factors is shown as ORs in Figure 4. Due to multicollinearity reduction, only six influencing factors were left in the final model and were evaluated. A detailed description of the calculated ORs is provided as additional material (Supplementary Table S1).

Figure 4.

Calculated odds ratios (ORs) for the effect of group size (animals per group), calf sex (male vs. female), calf age (d), average rainfall of the 2 d before the visit and the day of the visit (L/m²), the ambient temperature (°C) during this period, and the observer (observer 1 vs. observer 2) on the results of 20 animal welfare indicators, with asterisks indicating statistical differences (*P < 0.05, **P < 0.01, and ***P < 0.001). ORs of continuous factors (group size, calf age, rainfall, and temperature) represent the risk per unit increase, whereas the ORs of categorical factors (calf sex and observer) show the relationship between the factors’ levels. Not assessable data are shown as “na.” Dots are marking the OR, and the whiskers end at the lower and upper end of the 95% confidence interval.

ORs for denied access to concentrate, roughage, and water decreased with the number of animals per group (OR: 0.56 to 0.83). In contrast to the cleanliness of feed troughs, which had decreased odds to be rated as soiled in decreasing group sizes (OR: 1.09), milk buckets and bedding had lower odds to be rated as soiled in smaller groups (OR: 0.79 and 0.92, respectively). Calves housed in larger groups were at higher risk of higher scores in claws’ cleanliness (OR: 1.19). Except for skin lesions, for which the odds increased with the number of animals per group (OR: 1.22), no effect of group size on animal health indicators was found.

Compared with female calves, male calves had lower odds for soiled milk buckets, feed troughs, and bedding (OR: 0.35 to 0.56). In contrast, being female decreased the odds of being categorized as runt (OR: 0.31) and having a navel inflammation (OR: 0.25). No effect of calf sex on horn bud inflammation and sucked teats was calculatable, as no male calves were dehorned.

The age of calves assessed influenced the evaluated indicators’ results greatly as 16 of 21 assessed indicators were influenced by calf age. An increase in calf age decreased the odds of having a denied access to concentrate (OR: 0.97), roughage (OR: 0.96), and water (OR: 0.99). On the other hand, an increase in calf age led to increased odds of all other environmental and animal-based hygiene indicators (except for the core body’s cleanliness), indicating heavier soiling. Also, an increased calf age entailed increased odds for high nesting scores (OR: 0.99). ORs for hypothermia, ear tag wound inflammation, navel inflammation, and diarrhea decreased with calf age (OR: 0.99, 0.97, 0.97, and 0.97, respectively), while the odds for sucked teats and coughing increased (OR: 1.06 and 1.01, respectively).

An increase in ambient temperature was associated with lower odds for a denied access to concentrate, soiled feed trough and soiled water, and higher odds of having no access to water and soiled milk buckets. Also, increasing ambient temperatures resulted in lower odds for soiled carpal joints (OR: 0.93) but higher nesting scores (OR: 1.13). Animal health indicators affected by the ambient temperatures were hypothermia, hyperthermia, visible injuries, coughing, and diarrhea (Figure 4). Increasing rainfall resulted in higher odds of being denied access to water (OR: 1.12). Milk buckets were rated as cleaner with an increase in rainfall (OR: 0.87), whereas feed trough and calf claws had higher odds to be rated as soiled with increasing rainfall (OR: 1.70 and 0.87). No statistically significant effect of rainfall on animal health indicators was found.

Calves on farms that were assessed by observer 1 had lower odds for high hygiene scorings (increased soiling) of environmental conditions (milk buckets, water, feed trough, and bedding; OR: 0.28 to 0.53) and the cleanliness of carpal joints and claws (OR: 0.14 and 0.16, respectively) (Figure 4). Hypothermia and hyper-thermia were recorded more frequently in calves on farms assessed by observer 2 (OR: 2.04 and 2.06, respectively). However, calves on farms evaluated by observer 1 had more often lower scores of visible skin injuries, ear tag wounds, and horn bud inflammations (OR: 0.16, 0.40, and 0.16, respectively).

Discussion

Animal welfare assessment in unweaned calves

Although the increasing interest of consumers in animal welfare and legal regulations requires assessing calves’ welfare, to our knowledge, no standardized, evaluated recording protocol for calves or data exists. Promisingly, Barry et al. (2019a, 2019b) developed an animal welfare protocol to assess unweaned calves by trained observers. Comparable to the approach of Barry et al. (2019a), we aimed to evaluate every calf on the farm present during the visit, while Brinkmann et al. (2016) proposed an evaluation of all calves for sample sizes up to 30 calves and a representative, stepwise selection for larger calf herds. Nevertheless, this proposed threshold of 30 calves was only exceeded on seven farms. A limitation regarding the comparability is more likely posed by small farms with small dairy calf herds, as 28 of the assessed farms held not more than 20 unweaned calves and 6 not more than 10 calves. For example, on farms with less than 10 calves, an occurrence of an adverse animal welfare result in one calf already represents a proportion of more than 10%. Hampton et al. (2019) evaluated the role of sample sizes in animal welfare assessments statistically and demonstrated that far larger sample sizes are needed to determine the probability of adverse animal welfare results reliably. For an adverse result with a true probability of 0.15 (see low prevalence in most of the evaluated health indicators), an acceptable upper limit of 0.1 and a confidence interval of 0.9, a sample size of 37 animals would be required (Hampton et al., 2019). For adverse results with lower probabilities (i.e., indicator runt in this study), far larger sample sizes would be needed (Hampton et al., 2019). Pfeifer et al. (2020) evaluated five different sampling strategies (100 evaluated animals each) for the KTBL protocol for pigs and compared the calculated prevalence of 13 indicators with the true prevalence on the farm (636 evaluated animals). They concluded that all five strategies could not predict true prevalence, and the authors proposed an extension of the sample size to 150 pigs—a sample size unreachable in unweaned calves on most German dairy farms. Thus, the usage of thresholds for the evaluated indicators for unweaned calves for on-farm animal welfare monitoring or the usage of assessments in calves for benchmarking between farmers must be seen critically.

The descriptive results of this study show that more than a quarter of all calves assessed had no access to water, additional roughage, or concentrate, although it has been shown that water and solid feed is crucial for the development of calves (Khan et al., 2016; Wickramasinghe et al., 2019). Regarding hygiene assessments, milk buckets received higher soiling scores than water, feed troughs, or bedding. Heinemann et al. (2021) evaluated the sanitation of individually housed dairy calves and identified feeding implements such as teat buckets as hygienical weak points, enhancing pathogens’ spread. Most animal health impairments were relatively rare in this study. However, more than half of the assessed calves suffered from inflammations around the ear tag, an animal health issue that has not been included in the existing animal welfare assessment so far. Literature on ear tagging is relatively rare, but studies showed that ear tagging was associated with behavior that is specifically indicative of pain (Turner et al., 2020). Johnston and Edwards (1996) examined the ears of calves after slaughter and found damages to ears in 46.4% of calves tagged with metal tags and 1.1% of calves tagged with polyurethane tags lower than the prevalence found in our study. Further studies on animal welfare issues related to ear tagging are needed, and their inclusion in animal welfare protocols seems beneficial.

Effect of influencing factors

The evaluation of influencing factors showed that the results of all indicators, except for the cleanliness of the core body, were affected by at least one of the six defined factors. Our analysis showed that the odds of having access to concentrate, roughage, and water increased with the group size. Of the farms surveyed, 57.1% stated to provide only milk to individually housed calves (Hayer et al., 2021), which is similar to the results of other studies where water was not provided in single housings on 10% of farms (Johnsen et al., 2021). Other studies linking the access of concentrate or roughage to the housing form are, to our knowledge, currently missing. However, a survey from Austria reported that concentrate and roughage are provided at an age older than 4 wk on 39.5% and 15.1% of dairy farms surveyed (Klein-Jöbstl et al., 2015). Increasing group size increased the risk of skin injuries, which was frequently associated with recorded symptoms of Trichophyton verrucosum infections. Lesions caused by T. verrucosum were associated with close contact among animals (Moretti et al., 1998), which aligns with our findings for increased risk factors of skin injuries in larger groups. It is important to note that we only recorded the number of calves per group but not the stocking density (available space per calf). Jorgensen et al. (2017) found a stronger relationship between calf health and stocking density than with group size, a factor worth assessing in future research.

Sex of calves as an essential factor has gained additional interest as male calves have higher risks of dystocia, have relatively high mortalities, and are treated differently on dairy farms (Mee et al., 2011; Raboisson et al., 2013; Hayer et al., 2021). Our analysis showed that male calves had lower risks of soiled milk buckets, feed troughs, or bedding compared with female calves, which could be associated with the relatively young age of male calves (average of 25.2 ± 25.7 d) and the circumstance that most farms sold their surplus calves after 2 wk (Hayer et al., 2021). Soiling scores reflect more long-term conditions that might explain male calves’ different evaluations as they were predominantly kept for 2 wk after birth (Hayer et al., 2021). Being male increased the odds of being categorized as a runt, which might be caused by different milk feeding practices for male and female calves on-farm (Hayer et al., 2021). Furthermore, a recent study among dairy producers indicated that male calves are prioritized lower than female calves due to low economic values (Wilson et al., 2021), which might impact health, growth, and development. Another possible reason could be that male calves are more likely to need assistance and have a greater likelihood of dystocia during birth, which affects further development of these calves negatively (Mee et al., 2011).

Calf age was the most significant confounder of the 21 evaluated indicators; 16 out of 21 indicators were influenced by calf age. Many of the effects reported for calf age mirror standard nutritional management changes prior to weaning. However, the calf age deviation analysis revealed that the age of unweaned calves assessed on the farm during the visit differed significantly between farms (median age of 7 d to median age of 77 d). Therefore, it seems necessary to assess calf age and interpret animal welfare assessment in calves at the farm level. Possible reasons for the differences in calf age between farms were the broad weaning age range between farms (7 to 16 wk) and the additional rearing of male calves (6 of 42 farms) instead of selling bull calves with 14 d (Hayer et al., 2021). Assessed farms provided additional feed and water often relatively late (14 d postpartum) (Hayer et al., 2021), which caused the lower odds of calves having a denied access with increasing age.

Furthermore, the odds of heavier soilage increased with calf age for every resource- and animal-based hygienic indicators. Neonatal calves are especially vulnerable to infectious diseases and at risk of death during the first weeks of life (Gulliksen et al., 2009; Windeyer et al., 2014), which is also shown in the increased odds for diarrhea in younger calves by our analysis, a frequent cause for calf losses during the first weeks postpartum (Svensson et al., 2006). Also, farms surveyed reported most calf losses occurring in the first 14 d postpartum (median of 5.0%; Hayer et al., 2021). As a result, farms might focus their workforce and attention on young calves, as older calves are more resilient, explaining the variance of hygienic indicators found for calf age and sex. Nevertheless, the odds of coughing increased with age, which has also been reported by Windeyer et al. (2014).

Ambient temperature and rainfall showed to influence the provision of concentrate and water and the cleanliness of milk buckets, water, and feed troughs. Differences caused by variance in ambient conditions might be a result of changes in dairy calf management. Barry et al. (2019b) reported a change in calf management practices at the beginning and the end of calving season in Ireland (6 to 12 wk difference). Although calving seasons are uncommon on Western German dairy farms, it is imaginable that calf management on the farms surveyed changes depending on the season or climatic conditions, although it has not been reported yet. Especially in outdoor-housed calves, stockpersons are exposed to climatic and sometimes uncomfortable conditions (Jorgenson et al., 1970), which might influence calf management. Also, providing calves with water during wintertime has been reported as challenging as provided water is at risk of freezing in case of low temperatures (Jorgenson et al., 1970). Most of the calves assessed were housed outdoor, and therefore animals and stockperson were at least partly exposed to climatic influences, which require different management adaptations during the different seasons (Jorgenson et al., 1970; Jorgensen et al., 2017).

On the one hand, this exposition to environmental conditions might be positive as rainfall might enhance a visual cleaning effect of calf claws due to wet surfaces’ soaking effect, explaining the decreased risk for soiled claws with increasing rainfall. Whereas increased rainfall could increase the spoilage rate of feed, it could explain the effect on the cleanliness rating of feed troughs in feed troughs. Lago et al. (2006) showed an increased ability to nest, estimated with a nesting score, lowering the risk of respiratory diseases in calves in winter. Higher ambient temperatures were associated with higher nesting scores in our study, indicating that farmers provide more bedding material in case of lower temperatures. Although all farm visits were performed at the end of autumn, winter, and the beginning of spring, increased ambient temperature increased the risk of hyperthermia and decreased hypothermia risk, according to Hill et al. (2016) and Jorgensen et al. (2017). Although studies on skin injuries in calves and the association to temperature are rare, the increased risk of skin injuries with higher temperature in our study was also found in dogs’ animal welfare assessments (Berteselli et al., 2019). Respiratory disease incidence and diarrhea are more prevalent during winter than in summer (Svensson et al., 2003; Windeyer et al., 2014; Medrano-Galarza et al., 2018), which supports our findings of increased odds with higher ambient temperatures for both indicators. The effects of warmer ambient temperature are likely greater, as reported for summer by Windeyer et al. (2014) or Jorgensen et al. (2017).

The observer’s effect was evaluated as an influencing factor to adjust for differences in assessments due to the different observers. However, due to our study’s limitations, no simultaneous assessment of the same farms by the two observers was feasible. Nevertheless, extreme odds for the observer’s effect on the results of indicators might indicate a difference caused by the different observers. For example, the odds for high scorings of cleanliness indicators were 1.9 to 7.1 times higher for assessments of observer 2 compared with observer 1, although both trained their common understanding of the indicator scores on two farms and used the same evaluation sheet. This deviation is unlikely explainable by the variance of the different farms assessed. Furthermore, the odds for hypothermia and hyperthermia were higher for assessments by observer 2, indicating that the measurement of rectal temperature is influenced by the person making the measurement. Burfeind et al. (2010) evaluated the variability of rectal temperature measurements in dairy cows and found high inter-observer reliability (r = 0.98) but highlighted possible effects of the procedure (up to 0.5 °C), type of thermometer (up to 0.3 °C), and the penetration depth (up to 0.4 °C). Our findings support their recommendation for standardization of measurement procedures for rectal temperature measurement.

Reflection of the study design

The chosen study design and selection criteria have the advantage that it is likely to represent practical conditions in dairy calf rearing in Western Germany and showed the feasibility of data collection of the used indicators under these conditions (Hayer et al., 2021). The farms assessed were used to cooperate with researchers and, therefore, did most likely not adjust their management and housing to avoid the evaluated indicators’ negative results. However, assessments of practical conditions are limited in the ability to control influencing factors and the conclusions reached are most likely not as specific as in controlled studies, in which only single variables are changed while others were kept the same. Also, due to one observer’s limitation to one-time visits by one observer, true inter-observer reliability was not calculatable. However, differences were found between calves on farms assessed by the different observers (e.g., cleanliness scorings or rectal temperature measurements), highlighting the need for further studies. All assessments were conducted in late autumn, winter, and early spring, and thereby only limited conclusions can be drawn from the effect of climate as an influencing factor. In particular, the effect of summer conditions on the results of animal welfare assessments would be an interesting addition to our research, as heat stress has a major impact on animal welfare (Roland et al., 2016; Jorgensen et al., 2017) and calf management on-farm might be different in summer (Barry et al., 2019b). Although macroclimate strongly influences microclimate in barns, pens, or hutches to which calves are exposed, changes in microclimate will be most likely smaller, and further studies are needed to evaluate its effect. Furthermore, this study only evaluated the effect of ambient temperature, which differs from microclimate in barns, pens, or hutches to which calves are exposed. Especially considering climate change, a stronger focus on climatic effects on calves welfare seems necessary (Roland et al., 2016).

Implications of the evaluation of influencing factors

Livestock farmers are obliged to assess livestock’s animal welfare, including dairy calves on dairy farms, by using animal welfare indicators. This study aimed to determine the variance of possible resource- and animal-based indicators, evaluate influencing factors on these indicators’ results, and identify associated challenges. In relation to the number of dairy cows kept, the number of evaluable dairy calves was small, limiting reliability and comparability between farms. On small farms, individual calves account for much of the variance in the overall result. This relationship is even more critical given that the herd size of this study was relatively large (150 ± 16.6 cows) compared with Germany in general (65.3 cows) (BMEL, 2019) or other European countries. Further research on how to adjust for small calf herds could help overcome insufficient sample sizes to implement self-assessment of dairy calves. Based on the evaluation of influencing factors in this study, the general applicability of resource- and animal-based indicators to evaluate farm animal welfare for controlling or benchmarking between farms should be carefully reconsidered. It seems to be important to adjust the results of animal welfare indicators for influencing factors such as the climatic condition during the assessment or the age and sex structure of the assessed calf herd. Furthermore, inter-observer reliability of single indicators has been critically discussed (Czycholl et al., 2016; Vieira et al., 2018; Pfeifer et al., 2019) and may limit their usage in welfare assessments. One solution for the issues of small sample sizes and influencing factors could be repeated assessments during different seasons and the calculation of rolling averages as suggested by Kirchner et al. (2014). However, this would require additional effort and time by the farmers. In line with Pfeifer et al. (2019), we assume that the most significant benefit of self-assessed animal welfare indicators in calves is farmers’ sensitization to animal welfare during the continuously repeated process of assessing and evaluating animal welfare indicators in their animals. Studies have shown that stockpersons’ ability to empathize with calves contributes to better handling, management, and productivity of calves (Lensink et al., 2000; Hayer et al., 2021). Anneberg et al. (2013) conducted an interview study of Danish animal welfare inspectors, in which they highlighted the importance of farmers’ motivation to improve animal welfare. The assessment and evaluation of animal welfare indicators by farmers could enhance this ability and improve animal welfare. Although farmers monitor their animals daily, additional self-assessments can sensitize farmers to aspects that are not routinely monitored (e.g., ear tag and horn bud wounds or specific cleanliness indicators). However, studies indicate that some farmers might see the obligation of animal welfare self-assessments as an insult and interference into their stockmanship (van Dijk et al., 2018), which might hinder positive effects. Informing farmers about the benefits of self-assessments might help to overcome this barrier.

In conclusion, the assessment of animal welfare by resource- and animal-based indicators in unweaned dairy calves showed to be hampered by the relatively small calf herds on the dairy farms visited and their variation regarding calf age and sex. Group size, calf sex, age, and climatic conditions at the time of the farm visit, as well as the role of the observer, were identified as influencing factors on the result of animal welfare indicators. Although the usage of animal welfare indicators from self-assessment for controlling or benchmarking animal welfare of calves on a farm level seems critical, resource- and animal-based welfare indicators might be useful tools to sensitize farmers to the welfare of calves via the assessment itself and thereby improving animal welfare.

Glossary

Abbreviation

OR

odds ratio

Conflict of interest statement

J.J.H., D.N., C.H., C.D.L., and J.S.-W. declare no conflict of interest.