Short communication: scoring system based on five teat morphology traits relates to udder health

ABSTRACT

The aim of this study was to propose and verify a scoring system for the evaluation of udder susceptibility to mastitis, which combined five influential teat traits into one overall score for the udder. The traits included barrel diameter, apex diameter, canal length, the change in wall thickness, and the change in apex diameter. Teat morphology of 38 Holstein cows were measured twice during lactation by ultrasonography. Each teat was scored for the presence of traits with negative implications for the udder health at the start (ScoreLactStart) and the end of lactation (ScoreLactEnd). Results showed that cows with higher score had significantly worse udder health. The number of mastitis warning days was significantly correlated to both scores. In addition, ScoreLactEnd was correlated to somatic cell count (SCC) and lactose content. Each increase of ScoreLactEnd by 1 point represented an increase in SCC by 17,745 cells mL⁻¹, 9.4 more days with mastitis warning, and 4.6 more days with SCC above 400,000 cells mL⁻¹ during observed lactation. Only ScoreLactStart was significantly related to blood content, but relations to SCC were weaker compared to ScoreLactEnd. Ranking udders based on teat morphology traits show potential for detection of mastitis susceptible cows in commercial herds.

KEYWORDS:

• Dairy cow
• Holstein
• mastitis
• somatic cell count
• ultrasonography

Introduction

Mastitis continues to be one of the economically most important diseases in dairy farming (Neijenhuis et al. 2001). Although it is a multifactorial disease, identifying udder and teat conformational risk factors can improve our strategies for on-farm mastitis management by detecting high-risk animals (Miles et al. 2019). Studies over the last four decades found a negative influence of various teat morphology traits on the udder health. Dimensions of these traits and their reaction to milking can make the udder less or more susceptible to mastitis (Seykora and McDaniel 1985).

The functionality of the teat canal is essential for mastitis prevention, and studies found multiple teat canal traits with negative implications for the udder health (Seykora and McDaniel 1985; Martin et al. 2018). However, morphological dimensions of the teat canal are subtle, and scans of excellent image quality, together with appropriate software, are required to ensure high accuracy and reliability (Martin et al. 2018). Teat canal length seems to be the most useful for practical use, as it is the easiest to measure out of teat canal traits. Short and long canals were observed to negatively influence susceptibility to mastitis (Lacy-Hulbert and Hillerton 1995; Klein et al. 2005).

Interaction between the milking machine and the teat is critical for effective milking (Mein et al. 2012). Thick teats have trouble fitting into the liner and the chance of liner slip is increased. In addition, cows with thicker teats were more susceptible to clinical mastitis (Slettbak et al 1995) and had higher somatic cell count - SCC (Seykora and McDaniel 1986). Guarín and Ruegg (2016) observed an increased chance for mastitis by 20% with each increase in the pre-milking diameter of teat apex by 1 mm, and the authors attributed it to wider teat canals and larger orifices in wider teats.

Milking may induce changes in teat tissue, such as oedema (Stádník et al. 2010) and hyperkeratosis (Neijenhuis et al. 2001), which may compromise teat defence mechanisms and increase the risk of infection (Mein 2012). Zwertvaegher et al. (2013) observed that the teats, which thickened during milking had higher SCC compared to the teats that decreased in barrel diameter. Moreover, the changes in teat apex diameter by more than +5% increased teat canal colonization by pathogens (Zecconi et al. 1992), and and a similar threshold of ±5%was suggested by Hamman and Mein (1996) to point out ineffective milking.

As the introduction suggests, multiple teat traits influence udder health and its susceptibility to mastitis. We can still see research efforts to identify influential teat morphology traits, as in the study of Martin et al. (2018) or Miles et al. (2019). Teat morphology traits are underused in practice and breeding, despite their strong relation to the udder health. It might be due to difficulty in getting objective measurements, uncertainty in the importance of individual traits, and lack of a scoring system. The aim of this study was to verify the combined effect of influential teat morphology traits on the udder health, and to suggest a scoring system for evaluation of udder susceptibility to mastitis.

Material and methods

The study was conducted in a production environment of a commercial dairy farm with Holstein cows in the Central Bohemian Region of the Czech Republic. Cows were housed in one stable with free stall housing and recycled manure solids as bedding. Cows were milked twice a day in a herringbone parlour. The critical milk flow for the automatic detachment system was set to 0.5 kg min⁻¹. Pulsation was set to a 60:40 ratio with 55 pulses per minute. The vacuum level was set to 42 kPa. Teat liners had a three-sided concave design with a 22.5 mm orifice diameter (Milkrite ImpulseAir IP10U ventilated; Johnson Creek; Wisconsin; USA).

Study design

Teat morphology traits of 38 Holstein cows (1. lactation n = 12; 2. lactation n = 12; 3. lactation n = 7; 4. lactation n = 7) were measured at the beginning (3-17 days in milk - DIM) and the end of lactation (293-314 DIM) by ultrasonography (warm water bath method, Aloka SSD500 with 7.5 MHz linear probe; Hitachi Aloka Medical; Tokyo; Japan). All cows that calved on this farm during July and August 2017, and were not culled during their lactation, were included into the experiment (n = 38). Ultrasonography took place immediately before and after evening milking by a trained researcher. Teat barrel diameter (10 mm from the Furstenberg’s rosette), teat apex diameter (at the Furstenberg’s rosette), and teat canal length (from the Furstenberg’s rosette to teat end) were measured by another trained researcher in programme NIS-Elements AR 3.2 (Nikon Corp., Japan) in the standard locations (Martin et al. 2018). Milking-induced change of teat wall thickness (10 mm from the Furstenberg’s rosette) and teat apex width was calculated as [(post-milking value – pre-milking values)/pre-milking value] * 100.

All teats of tested cows were scored based on the criteria in Table 1 at the beginning (ScoreLactStart; teats n = 152, cows n = 38) and at the end of the lactation (ScoreLactEnd; teats n = 152, cows n = 38). Evaluated teats got +0 points for traits of desirable dimensions and got +1 point for each trait, which dimensions were found to be undesirable in relation to the udder health by previous studies (used studies are referenced in Table 1). Therefore, the teat could achieve 0–5 points (teat level), while the cow could achieve 0–20 points (udder level). Teat wall thickness change was evaluated instead of teat barrel width change, as suggested by Zwertvaegher et al. (2013), because it might better represent the underlying mechanisms behind teat thickening during milking. Used studies (Table 1) measured teat structures with various techniques. We uniformly measured them by ultrasonography, which might cause some discrepancies. However, it was the most viable approach from a practical standpoint, and the findings of these studies should be applicable to ultrasonography measurements as well.

Table 1. The criteria for the scoring system based on five teat morphology traits. Examples of studies, where the trait was observed as undesirable in relation to udder health are provided for each selected trait together with the studied breed.

Scored teat traits	Observed to have a negative influence when	Scoring criteria for the tested herd (n = 38 cows, 152 teats)
		Undesirable dimensions within tested herd = + 1 point	Dimensions for ScoreLactStart		Dimensions for ScoreLactEnd
			+0 points when	+1 point when	+0 points when	+1 point when
Teat barrel width (mm)	too wide^1^, ^2	The widest 25%	17.99–27.73	27.74–34.34	20.02–27.31	27.38–33.83
Teat wall thickness change (%)	too high^3	The highest 25%	−37.33–54.13	55.69–134.47	−40.62–40.17	40.33–88.67
Teat apex width (mm)	too wide^4	The widest 25%	17.18–23.09	23.11–29.28	18.19–22.94	22.95–25.99
Teat apex width change (%)	too low or too high^5^, ^6	Less than −5% and more than +5%	−4.89–4.97	−18.51 to −5.05; 5.24–26.31	−4.75–4.16	−13.98 to −5.02; 5.06–11.97
Teat canal length (mm)	too short and too long^7^, ^8^, ^9	The shortest 12.5% and the longest 12.5%	9.62–14.68	6.85–9.6; 14.7–18.84	10.38–15.93	7.23–10.3; 15.99–20.83

ScoreLactStart – scoring of teat traits at the beginning of lactation; ScoreLactEnd – scoring of teat traits at the end of lactation

¹ Seykora and McDaniel 1986 – Holstein; ²Slettbakk et al. 1995 – Norwegian Cattle; ³Zwertvaegher et al. 2013 - Hosltein; ⁴Guarín and Ruegg 2016 - Holstein; ⁵Zecconi et al. 1992 – not specified, probably Holstein based on high milk yield; ⁶Hamann and Mein 1996 - Holstein; ⁷Lacy-Hulbert and Hillerton 1995 - Holstein; ⁸Klein et al. 2005 – Holstein, Simmental, Brown Swiss; ⁹Celik et al. 2008 – Brown Swiss

Data collection

Data about milk yield (kg), milk conductivity (mS m⁻¹), somatic cell bracket (SCC_AFI; 1 = 0–199,000 cells mL⁻¹; 2 = 200,000–399,000 cells mL⁻¹; 3 = 400,000–799,000 cells mL⁻¹; 4 = 800,000 and more cells mL⁻¹), lactose content (%) and blood content (%) were obtained from ‘in-line real-time’ milk analysers (Afilab with software Afifarm 4.1; Afifarm; Afikim; Israel) for each milking. Milk analysers were calibrated before the start of the experiment. SCC_AFI, lactose, and blood content represented the cow’s lactation average (1–285 DIM). The number of days above 200,000 cells mL⁻¹ (SCC200) and days above 400,000 cells mL⁻¹ (SCC400) were counted for each cow from 1 to 285 DIM. Mastitis warning days (MasWD) were counted for each cow during the monitored period and were based on: 1. Decrease in lactose content (dropped below 4.8% or decrease by more than 0.15% compared to individual physiological optimum), 2. Increase in milk conductivity (>0.5 mS m⁻¹ compared to individual physiological optimum), 3. Increase in SCC (>200,000 cells mL⁻¹), 4. Drop in milk yield (>10% compared to lactation curve) and 5. Presence of blood. Three of these five conditions had to be true for the period to be counted as MasWD. The physiological optimum for lactose and conductivity varies among healthy cows. For the purpose of this study, we calculated it as an average from lactating days in which SCC was below 200,000 cells mL⁻¹. The period of MasWD ended when lactose (<0.05% compared to physiological optimum), milk yield (lactation curve), and SCC (<200,000 cells mL⁻¹) returned to optimal values for three days in a row. The period of increase SCC (PISCC) represented an increase of SCC above 200,000 cells mL⁻¹ for at least five consecutive days. The period ended after three days of SCC below 200,000 cells mL⁻¹. Increased SCC is generally observed after calving even in uninfected quarters (Ferronatto et al. 2018), therefore first 20 DIM were not counted for PISCC. SCC results from monthly performance control (SCC_PC; Fossomatic 7, FOSS, Denmark) were recalculated to the lactation average and evaluated.

Statistical analysis

Programme SAS 9.3 (SAS / STAT® 9.3, 2011) was used for statistical evaluation. The UNIVARIATE procedure was used for basic statistics. For the main evaluation, we calculated partial correlations and regressions between the score for teat morphology traits (ScoreLactStartteat level; ScoreLactStart udder level; ScoreLactEnd teat level; ScoreLactEnd udder level) and parameters of udder health (MasWD, SCC_AFI, SCC_PC, SCC200, SCC400, PISCC, lactose and blood content). Partial correlations were calculated by CORR procedure and were adjusted for lactation number and DIM. Partial regressions were calculated using the GLM procedure that included the fixed effect of lactation number (1, 2, 3, 4) with DIM and score (ScoreLactStart or ScoreLactEnd) as linear regressions. The experimental units for the evaluation at the teat level were individual teats, while the experimental unit for the evaluation at the udder level was the cow (with four evaluated teats). The best model for evaluation was selected in accordance with the values of the Akaike Information Criterion (AIC). Significance levels P < 0.001, P < 0.01, and P < 0.05 were used to evaluate statistical significance.

Results and discussion

The model equation was statistically significant for all monitored parameters at the teat level (P < 0.01), but became insignificant at the udder level for SCC_PC, PISCC; and blood content (Score A), and for SCC_PC, lactose and blood content (ScoreLactEnd). The model equation for partial regression explained variability from:

• 11.1% (SCC_PC; P < 0.01) to 30.8% (MasWD; P < 0.001) for ScoreLactStart at the teat level;

• 12.6% (blood content; P < 0.001) to 30.7% (SCC200; P < 0.001) for ScoreLactEnd at the teat level;

• 27.9% (lactose content; P < 0.05) to 39.4% (MasWD; P < 0.01) for ScoreLactStart at the udder level;

• 29.7% (SCC400; P < 0.05) to 40.7% (SCC200; P < 0.01) for ScoreLactEnd at the udder level.

The average daily milk yield during the tested period (1-285 DIM) was 32.3 kg with mean lactose content of 4.98%, fat of 3.97%, and protein of 3.58%. The average SCC_PC was slightly under the Czech national average with 184,000 cells mL⁻¹. Cows spent on average 58 days with SCC200 and 22.4 days with SCC400. Teat characteristics of the tested population could be deduced from Table 1, where we presented minimal and maximal values for monitored teat structures, as well as criteria for the scoring system. The presence of undesirable teat traits was evaluated and cows scored on average 5.82 points for ScoreLactStart and 5.26 for ScoreLactEnd, with a minimum of 0 and a maximum of 17 undesirable traits within one udder. As we suggested in the introduction and Table 1, individual teat morphology defects negatively influence udder health. After combining these individual traits into one score, our correlation analysis showed that the cows with higher score had significantly worse udder health (Table 2). The occurrence of MasWD was significantly correlated to obtained score, while the strongest correlation was observed for ScoreLactEnd at the udder level (r = 0.494; P < 0.01). Each increase in ScoreLactEnd at the udder level by 1 point represented 9.4 more MasWD during observed lactation (P < 0.01; Table 2).

Table 2. Partial correlations and regressions between the score obtained for teat traits and selected parameters of the udder health.

	Teat level (n = 152)				Udder level (n = 38)
	ScoreLactStart		ScoreLactEnd		ScoreLactStart		ScoreLactEnd
	Corr	Reg	Corr	Reg	Corr	Reg	Corr	Reg
MasWD	***0.273	***19.753	***0.340	***17.647	*0.364	**9.010	**0.494	**9.397
SCCAFI	**0.229	***0.125	***0.289	***0.113	NS0.306	*0.057	**0.425	*0.061
SCCPC	NS0.100	NS12.573	**0.244	**33.806	NS0.134	NS5.735	*0.347	*17.745
SCC200	*0.192	***17.082	***0.335	***20.088	NS0.256	*7.792	**0.488	**10.717
SCC400	**0.234	***11.871	**0.230	**8.415	NS0.311	*5.415	*0.341	*4.574
PISCC	NS0.136	*0.254	***0.298	***0.380	NS0.181	NS0.116	**0.424	**0.198
Lactose content	*−0.162	**−0.023	**−0.227	*−0.020	NS−0.216	NS−0.010	*−0.326	NS−0.011
Blood content	***0.270	***0.014	^NS0.104	^NS0.004	*0.359	*0.006	^NS0.147	^NS0.002

Used levels of statistical significance P < 0.05 = *; P < 0.01 = **; P < 0.001 = ***; no statistical significance = NS

ScoreLactStart – scoring of teat traits at the beginning of lactation; ScoreLactEnd – scoring of teat traits at the end of lactation; Corr - correlation coefficient; Reg - regression coefficient; MasWD – mastitis warning days, SCC_AFI – somatic cell score from Afilab milk analysers; SCC_PC – somatic cell count from monthly performance control; SCC200 - number of days above 200,000 cells mL⁻¹; SCC400 - number of days above 400,000 cells mL⁻¹; PISCC – period of increased somatic cell count.

In addition to MasWD, our correlation analysis showed that the cows with higher score had significantly higher SCC (Table 2). The increase of SCC during intramammary infections is well documented (Bezman et al. 2015), and we closely monitored it with five different parameters (SCC_AFI, SCC_PC, SCC200, SCC400, and PISCC). Analysis showed that ScoreLactEnd was significantly correlated to all SCC parameters, while ScoreLactStart was significant only for SCC_AFI, SCC200, and SCC400 (P < 0.05; teal level; Table 2). Parameters of SCC_AFI and SCC400 at least showed a tendency to ScoreLactStart at the udder level, although the results for SCC_PC remained insignificant. Even though the method used for counting SCC_PC is more precise, SCC_AFI or similar systems might be expedient for research purposes thanks to its high frequency of control, ease to use, and reliable calibration possibilities (Hanuš et al. 2016). Ferronatto et al. (2018) concluded that various methods of assessing milk cellularity have the potential to be used to diagnose bovine mastitis. Even though the amount of SCC varies in the healthy udder, SCC cut-off values for diagnosing mastitis have been established at between 100,000 and 272,000 cells mL⁻¹ (Ferrenatto et al. 2018), while the most used cut-off value would be 200,000 cells mL⁻¹. In our study, SCC200 was mostly strongly correlated to obtained score, while regressions were always significant. The strongest correlation was observed for ScoreLactEnd at the udder level (r = 0.488; P < 0.01). Increases in SCC above 400,000 cells mL⁻¹ represent further worsening of udder health and were achieved much less frequently than SCC200. Correlations and regressions of SCC400 to obtained score were similar as for SCC200 (Table 2). Each increase of ScoreLactEnd at the udder level by one point represented 4.6 more days with SCC400. In addition, increased occurrence of PISCC for cows with high score underlined weaker defence mechanisms of teats with undesirable morphology.

Furthermore, intramammary infections can cause a decrease in lactose content (Bezman et al. 2015), which we also observed in our study. Cows with higher ScoreLactEnd had less lactose in their milk (r = −0.326; P < 0.05; Table 2), which could represent udder health problems during lactation. At last, the presence of blood in milk was also monitored as a parameter of udder health, because it can be a consequence of intramammary infection or heavily damaged teat. Interestingly, blood was the only monitored parameter that was not significantly correlated with ScoreLactEnd. On the other hand, we observed significant correlations between blood content and ScoreLactStart.

Both ScoreLactStart and ScoreLactEnd aimed to verify the combined effect of influential teat traits on udder health. In addition, ScoreLactStart tested the viability of the system as an early diagnostics tool, while ScoreLactEnd showcased how these relationships developed during lactation. One of the reasons for better results for ScoreLactEnd could be resolving of physiologic udder oedema, which commonly appears 2–4 weeks after calving (Divers and Peek 2007), which might influence measurements of the teat dimension. Finding highly susceptible cows would be more useful at the beginning of lactation, so moving teat measurements to 30–60 DIM after the example of linear type evaluation might correct this distortion. Early lactation showed higher variability in reaction to milking compared to a more uniform response in later stages (Gašparík et al. 2019). In addition, ScoreLactEnd also reflected teat morphology changes induced by continual milking as well as other damage suffered by teat throughout lactation. Therefore, it corresponded better with udder health for given lactation. Perhaps in the future, milking robots could be equipped with ultrasound, and artificial intelligence could measure teat structures in real-time for each milking to reliably identify susceptible dairy cows.

Ranking udders based on teat morphology traits showed potential for detection of mastitis susceptible cows in commercial herds. Furthermore, the addition of teat scoring into the linear type evaluation of dairy cattle would provide the measurements to calculate breeding values for teat structures, which could enhance breeding for udder conformation and ultimately improve udder health. Finding new tools to protect udder health might be essential for the future, as the European Union plans to reduce sales of antibiotics for farm animals by 50% by 2030 (EC 2020). The methodological approach used in this study could be further improved by adding morphology traits with proven connections to mastitis incidence. Furthermore, the addition of traits that describe important morphological changes induced by milking could prove expedient for the estimation of teats’ ability to physically defend against invading microorganisms (Martin et al. 2018). At last, the model could be further improved by including factors with a strong association to intramammary infections and teat dimensions like age (Celik et al. 2008), teat position (Guarín and Ruegg 2016), or breed (Stádník et al. 2010).

Ethical statement

This experiment was carried out in accordance with Czech legislation for protection of the animals against abuse (No. 246/1992) and with directive 2010/63/EU on the protection of animals used for scientific purposes.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Ministry of Education, Youth and Sports of the Czech Republic under ‘S’ Grant and by the National Agency for Agricultural Research of the Czech Republic under grant number QK21010123.