Twenty-four lactating Holstein dairy cows were selected for the current study from a larger herd (120 lactating dairy cows) housed in the LaMaster Dairy Farm AMS facility at Clemson, SC, USA. Cows were enrolled in the study when they reached day 97 of lactation and continued for 6 consecutive days until day 102 of lactation. Cows were divided into 2 groups (n = 12/group) balanced for the stage in lactation, with each group having access to two AMSs. Group A consisted of 12 primiparous (PR) cows (mean age 3.0 ± 0.2 yr), and group B contained 12 multiparous (MU) cows (mean age 6.1 ± 1.9 yr and parity 4.1 ± 1.9). The cows were selected based on the expectation that they would be lactating for the full duration of the trial, and were allowed 6 milking permissions per day.

The AMS entrance and exit gate design for cow traffic flow was the same between the groups. Both AMS were left-handed milking robots, forcing clockwise cow traffic. The Holding area (HA) had an automatic selection gate that enabled access to the CP of the AMS. Gates only opened to a commitment pen (CP) if cows were granted permission based on their last milking time, and exit gates only opened toward the feed alley if the milking was complete. There were two selection gates, one granting access to the holding area (the area next to the CP) and one for exiting the CP post milking. All gates and alleys were of equal lengths for both groups.

The AMS (Delaval VMS V300), feeding alley, stall layouts, and management practices were identical for both groups. The AMS was operational 24/7 except for the extent of 3 20-minute robot cleaning cycles (0400, 1230, and 2000 h). Feeding of a PMR occurred twice daily (0730 and 1530 h) and diet feed and chemical composition and feeding schedule were identical for both groups. Cows were offered a concentrate pellet during the milking process; the amount of concentrate was dependent upon the individual projected milk yield. Footbaths were present in the CP exit alley containing a 5% copper sulfate solution.

Six Wyze Cam v3 cameras (Wyze labs, Inc, Seattle, USA) recorded behavior for each group 24 h/d for 6 consecutive days. Cameras were focused on the CP in front of the AMS, CP entry gate, the AMS itself, and the AMS exit and entrance gates. Focal cows were identified on video by specific colored paint markings using livestock paint (LA-CO Industries, Inc, IL, USA) across their back. Parity was identified by colored livestock paint down the spine from the shoulder to the tail head. Primiparous cows were marked with red livestock spray paint, and multiparous cows were marked with blue livestock paint. Focal cows were equipped with a Polar H10 heart rate sensor (Polar Electro Europe BV, Fleurier, Switzerland), a data transmission watch, and a triaxial accelerometer (Onset HOBO PendantG acceleration data loggers) to record Heart Rate Variability HRV (Hopster and Blokhuis, 1994) and activity data (Wolfger et al., 2015). The number of daily milking visits, total milk per visit, and daily milk yield (kg) were recorded for each of the focal cows.

All statistical calculations for milk yield, AMS behavior, CP behavior, and cow activity and HRV were preformed using JMP Pro 16.1.0, and P ≤ 0.05 was considered evidence of statistical significance. Evaluation of data with a Shapiro-Wilk’s test (P>0.05) and a visual inspection of histograms revealed that data from all measurements were normally distributed. To describe the influence of parity on the response variables of interest (milk production, AMS behaviors, CP behaviors, HRV, and activity) during the six days of observations, general linear mixed models were developed that related the response variables (i.e., daily milk yield, idle, laying time, etc.), to the fixed effect of parity and the random effects of day of observation and individual cow. If parity was found to have significant effects on the response variables, the nature of the effects was further studied using T-tests (i.e., Fisher’s Protected Least Significant Difference Test). Since more than one observer contributed to data collection, inter-observer reliability with the primary observer was calculated using Cohen’s kappa Agreement coefficient (K), following Landis and Koch (1977). A power analysis was executed, employing CP time as a determinant factor to identify a medium effect size (Cohen’s d = 0.5) with a two-tailed alpha level set at 0.05 and a desired power of 0.80. The analysis revealed that to attain adequate statistical power for the variable of interest in our study, a minimum sample size of 10 cows would be necessary. In anticipation of potential attrition or data loss, we opted to enlist 12 cows for each group, thereby ensuring that our study maintains a robust sample size capable of detecting meaningful effects. Inter-observer reliability was measured during the observer-training period before data collection occurred when trainees simultaneously observed the same videos. Inter-observer agreement was considered good [Kappa = 0.96 (P< 0.001), CI (0.90, 0.99)].

Multiparous cows produced more daily milk yield than PP cows (P = 0.021), while parity showed no effect on daily milking frequency or milk yield per visit ( Table 2 ).

It is well established that cows higher in parity produce more milk than cows of lower parity, but this can vary based on many factors, including but not limited to; facility traffic flow, type of milking system, and feed provided (Beauchemin and Rode, 1994; Deming et al., 2013; Neave et al., 2017; Siewert et al., 2019). In the current study, parity influenced daily milk yield, as MP cows produced more milk than PP cows during the observation period. Adrien et al. (2012) reported that total milk production at 60 DIM was greater in multiparous cows than primiparous cows. Siewert et al. (2019) also found that multiparous cows produced greater daily milk yield than primiparous cows within a similar DIM range (89-118 DIM) to the current study.

The current study showed that parity significantly impacted time spent inside the CP, AMS approach attempts, tail swishing, idle time, and displacement behavior. PP cows spent greater durations inside the CP prior to milking, and PP also stood idle significantly longer than MP cows in the CP. This is consistent with Solano et al. (2022), who reported that PP cows spent longer waiting times than MP cows before entering the milking stalls in early and mid-lactation and attributed these findings to the mixed grouping of PP and MP cows inside the CP, competing for AMS entry.

Similarly, Dijkstra et al. (2012) stated that submissive cows, like first lactation cows, are forced to spend greater time waiting to milk due to the dominant structure dairy cows have been known to express. Another study reported that lower-ranking cows generally had to wait longer in the CP than dominant cows before accessing the AMS (Halachmi, 2009). Primiparous cows are lower in social ranking than MP cows, which prevents them from entering the milking robot as quickly as other cows inside the CP. This is likely a stressful situation for PP cows as they are unable to feed, drink, or lie down in this area (Dijkstra et al., 2012; Norring et al., 2012). The lack of locomotion activity permitted in the restricted space of CPs accounts for our findings of longer idle times in PP cows inside the CP than MP cows. However, no previous research has investigated the influence of parity on cow idle time, specifically inside the CP. However, some studies have observed an effect of parity on idle time within other areas of the barn. Gomez and Cook (2010) reported reduced idle time in PP cows within free stalls, possibly related to social competition and increased feeding time. Thus, the influence of parity on idle time cows inside the CP waiting to milk should be further investigated in the future as it can influence daily feeding and resting activity and negatively impact milk production.

Parity also significantly influenced the number of AMS approach attempts, tail swishing, and displacement behavior in MP more than PP cows in the CP. The number of times the cows approached the AMS milking robot was recorded while confined in the CP, waiting to access the milking robot. This behavior was evaluated to assess the cow’s motivation to enter the milking robot. MP cows made more AMS approach attempts and entered through the gates faster than PP cows in this study, suggesting MP cows were more motivated to milk. If cows are highly motivated to approach the AMS, waiting times inside the CP could be reduced, thus preventing prolonged stress (Dijkstra et al., 2012; Norring et al., 2012) and potentially increasing milk production. However, various social interactions can also interfere with cow motivation to approach AMS milking stalls, like head-butting and displacement agonistic behaviors.

Time spent inside the CP can be influenced by factors like competition to enter milking stalls, leading to agonistic behaviors. Halachmi (2009) reported dominant cows would not wait in the pre-established milking queue, instead would displace lower-ranking cows until they reached the head of the line to milk. However, most research investigating parity influence on agonistic behavior has focused on displacement in feed bunk or free stall areas. Val-Laillet et al. (2009) reported restricted feed space resulted in significantly greater DP-A behavior in MP cows. Whereas Westin et al. (2016) discussed how social competition could also impact lying behavior, with lower-ranking cows receiving more displacements from lying stalls. Primiparous cows are typically socially inferior to MP dairy cows due to being smaller in size, younger, and unfamiliar with preestablished social groups in the facility (Nogues et al., 2020; Fadul-Pacheco et al., 2021). Thus, these reports support the current findings of MP cows displaying greater DP-A behavior in the AMS CP. The area outside AMS robots can cause substantial social competition as cows are competing to enter the milking stalls. Similarly, in the current study, we hypothesized to see MP cows displace lower-ranking cows, like PP cows, more frequently in the CP. This behavior was attributed to the higher dominance status of MP cows and their better familiarity with the milking process in the AMS robots.

Tail swishing is a natural cow behavior performed in situations that irritate the cows (Weary et al., 2011). In the current study, we observed tail swishing inside the CP as a sign of annoyance or irritation to other cows inside the CP while waiting to enter the milking stalls. However, previous research generally evaluated tail swishing as an annoyance to flies (Wehrend et al., 2006; Weary et al., 2011; Perttu et al., 2020). We found MP cows tail-swished more frequently than PP cows inside the CP, thus, suggesting MP cows were more irritated while waiting or competing to enter milking stalls. Although there is no previous research with similar findings as the current study, other studies have found increased tail-swishing could be a sign of cow agitation or a response to fear, stress, or painful stimuli (Eicher and Dailey, 2002; Willson et al., 2021).

MP cows were observed stepping more frequently and spending greater time milking inside AMS milking stalls than PP cows in the current study. We also found PP cows took significantly longer to exit the milking stalls post-milking than MP cows. Castro et al. (2012) reported the average time spent in the AMS was approximately 7 min per cow. Similar to current findings, Sitkowska et al. (2016) found MP cows spent greater time (383 s) than PP cows (361 s) in the milking robot. They also reported the average milking time accounted for 70% of MP cows and 66% PP cows (Sitkowska et al., 2016). Considering MP cows typically produce higher milk yield per visit than PP cows (Johnson et al., 2003), we would expect longer AMS times in MP cows (Kliś et al., 2021). We also observed greater stepping frequency in MP cows in the present study, which agrees with Gygax et al. (2008), who reported cows stepping in the AMS >95% of all milking events, and that the stepping rate increased in cows of higher parity. Although the milking process can be stressful, stepping behavior is considered a minor stress response in cows (Lupoli et al., 2001; Gygax et al., 2008), indicating that MP cows were more stressed than PP while milking in AMS. In the present study, PP cows took significantly longer to exit the milking robot than MP cows once the milking process was complete. This could be due to a multitude of reasons, including crowding at AMS exit gate, desire to receive more feed concentrate, or lack of understanding when milking in the novel AMS environment. While little research has been conducted investigating the effect of parity on AMS exit duration, Jacobs and Siegford (2012) reported cows were more likely to hesitate when other cows were near the exit gate. Conversely, Jacobs et al. (2012) found parity did not influence exit hesitation in their study. However, investigating possible influences of factors such as parity on AMS exit duration would be beneficial as occupying AMS for longer durations creates longer waiting times for other cows and can hinder maximum milk production, system efficiency, and cow welfare (Rodenburg, 2013).

The findings of the present study suggest that PP cows were more stressed while waiting to be milked in the CP. PP cows exhibited lower RMSSD and SDRR values and higher LF/HF ratios than MP cows. RMSSD accounts for short-term, high-frequency components of HRV, strongly reflects the vagal tone, and is highly correlated to LF/HF ratio (Kleiger et al., 1992; Hagen et al., 2005; Gygax et al., 2008). Lower RMSSD values indicate increased cow stress levels (Hagen et al., 2005; Von Borell et al., 2007; Buck et al., 2013). Although previous research has not investigated HRV parameters while cows were explicitly waiting in the CP, other studies have reported similar parity effects on HRV values. Kovács et al. (2013) reported that MP cows showed lower levels of stress than PP cows. Another study reported that SDRR tends to be lower in PP cows than in MP cows in all behavioral patterns (Buck et al., 2013). Such a lower HRV in PP cows suggests that these cows might not have enough training time or sufficient experience milking in AMS, resulting in greater stress responses while waiting in AMS CPs (Szentléleki et al., 2015). Although previous studies reported significant differences in HRV parameters when cows wait to milk and inside the AMS, for instance, Jurkovich et al. (2017) reported a higher sympathetic tone during periods before milking in AMS, suggesting cows were more stressed.

Parity significantly impacted multiple cow HRV parameters while milking in AMS. The current study found PP cows had lower RMSSD and SDRR values and higher LF/HF ratios, which suggests that PP cows were more stressed than MP cows inside the AMS. This is consistent with other research findings of PP cows being more susceptible to AMS technology than MP cows. For instance, Kovács et al. (2013) reported MP cows had higher SDRR values than PP cows, thus suggesting PP cows were more stressed while milking in AMS. Whereas Hagen and colleagues (2005) found the only factor that influenced HRV parameters while milking was cow body weight, thus, suggesting PP and MP cows would likely differ in stress during the milking process solely due to MP cows being larger in size. Moreover, mechanisms that cause such physiological stress should be further examined to control for cow-level effects of stress when milking in AMS (Frei et al., 2022).

Increased cow daily lying time is often associated with parity and stage in lactation (Vasseur et al., 2012; Westin et al., 2016). Westin et al. (2016) reported higher parity cows spent 0.5 h/d more lying time than PP cows. Norring and colleagues (2012) also found that daily lying time increased with cow age and parity. However, other researchers reported a weak correlation between parity and daily lying time (Charlton et al., 2016). Our results partially agree with the previous research findings, as MP cows spent more time lying down in total, particularly during nighttime hours (1831 – 0629h), than PP cows. A possible explanation could be that cows in AMS farms spend considerable amounts of time waiting in the CP to milk, thus, limiting the available time to lay down, especially for PP or lower-ranking cows (Melin et al., 2006; Halachmi, 2009; Westin et al., 2016). Moreover, MP cows in the current study spent less time waiting to enter the milking stalls, which may allow them more time to rest longer than PP cows.

In the present study, PP cows performed higher daily and daytime lying frequencies than MP cows. Charlton et al. (2016) reported moderate correlations between parity and lying bout frequency, and Westin et al. (2016) found lying frequency was influenced by parity with higher frequencies in younger cows. Other researchers also observed a complex interaction between parity and lying bouts, where PP cows performed higher lying bout frequencies than older cows (Vasseur et al., 2012). It is plausible PP cows laid down more frequently than MP cows because of a lack of understanding of AMS procedures, as PP cows are fetched to milk more regularly than MP cows (Gednalske, 2021).

MP cows in the current study showed longer daily lying bouts than PP cows, which agrees with previous research. Charlton and colleagues (2016) also found a moderate correlation between parity and mean lying bout duration. Similarly, Vasseur et al. (2012) reported cow parity being moderately positively correlated with mean lying bout duration. At the same time, other researchers observed no effect of parity on daily lying bouts (Hart et al., 2013). Nonetheless, shorter and more frequent lying bouts may indicate increased cow discomfort (Bell et al., 2009; Dippel et al., 2009; Charlton et al., 2016).

Inactive standing or idle time was considered when cows were resting or standing without engaging in any other activity or interaction (Hagen et al., 2005). The majority of previous research investigating inactive standing was conducted in terms of feeding space/activity or stall flooring type effect on cow idle time. For instance, DeVries and Von Keyserlingk (2006) found feeding competition had no influence on PP cow standing time, and Rushen et al. (2007) reported cows standing idle longer in facilities with concrete flooring. No research has found a parity effect on daily idle time. However, in the present study, we found a parity influence on daily and daytime inactive standing, as PP cows spent greater time standing idle than MP cows. The elevated idle times reported in PP cows might suggest overcrowding or social competition in the AMS CP, as primiparous cows are typically lower ranking than MP cows (Olofsson, 1999; DeVries et al., 2004; Huzzey et al., 2006; Proudfoot et al., 2009).

While acceleration activity has yet to be extensively evaluated in dairy cows in AMS farms, one study found acceleration (g) activity varied greatly among individual cows (Yin et al., 2013). The accelerometers used in this study recorded three-dimensional movement used to monitor cow behavior within the AMS facility remotely. In the current study, we found PP cows showed greater daily and daytime acceleration (g) activity than MP cows. No previous research has investigated the impact of parity on cow acceleration. However, activity data are consistent with the reported lying and idling durations from the present study, indicating that PP cows were more active and performed more locomotory behaviors than MU. These findings could be attributed to the fact that primiparous cows are smaller in size and produce less milk than MP cows (Beauchemin and Rode, 1994; Neave et al., 2017), and spend less time milking (Castro et al., 2012), lying (Lobeck-Luchterhand et al., 2014), and standing (Gomez and Cook, 2010), enabling them to perform more activity throughout the day.