FSCR and PL proportion considering SCC before first AI, SCC after first AI, 305-d milk yield, DFS, parity, and season are presented in Table 1. Regarding both FSCR and PL, no significant interactions between factors were found in this study.

Table 1. First service conception rate and pregnancy loss proportion considering several variables. Blank variables in the last 3 columns were not included in the final model of the logistic regression. The binary logistic regression model for first service conception rate (FSCR) includes the results for animals with a somatic cell count (SCC) record within 30 days after artificial insemination (AI).

DFS: days to first service. PL: pregnancy loss. AOR: adjusted odds ratio. CI: confidence interval. ^[a] Hosmer & Lemeshow test: p=0.863. ^[b] Cox & Snell R² =0.026

First service conception rate

Overall, the average FSCR for all the cows involved in this study was 28.9%. FSCR did not differ between non-SCM and SCM groups considering SCC before first AI (29.6% and 25.8%, respectively; p=0.134). After first AI, the SCM group had lower FSCR than the non-SCM group (25.1% and 31.2%, respectively; p=0.016). In relation to 305-d milk yield, FSCR was lower for cows in the >8780 L group than for cows in the ≤8780 L group (22.7% and 35.1%, respectively; p=0.00). The FSCR for primiparous cows was higher than for multiparous cows (35.5% and 26.5%, respectively; p=0.00). Finally, the FSCR was 30.1% and 27.6% for cows in <71 DFS and >70 DFS groups, respectively (p=0.210), and was not influenced by season (30.8%, 26% 28.5%, and 30.3% for winter, spring, summer, and autumn, respectively, p=0.401).

The results of the logistic regression for FSCR, including animals with a SCC record within 30 days after AI, are depicted in Table 1. Regarding SCC after first AI, cows in the non-SCM group were more likely to conceive pregnancy than cows in the SCM group (OR=1.285; 95% CI: 1.000-1.653; p=0.050). Moreover, cows with milk productions ≤8780 L/305-d were 1.820 (95% CI: 1.469-2.257) times more likely to become pregnant than cows that produced >8780 L/305-d (p=0.00). Concerning cows with a SCC record within 30 days before AI, our results show that SCC factor was not included in the final model. In addition, primiparous cows, animals inseminated during spring or summer and <70 d after parturition were more likely to become pregnant (OR=1.332, 95% CI: 1.052-1.688, p=0.011; OR=1.407, 95% CI: 1.026-1.929, p=0.034; OR=1.351, 95% CI: 1.003-1.819, p=0.044; OR=1.250, 95% CI: 1.011-1.546, p=0.033, respectively). Furthermore, cows with milk productions ≤8780 L/305-d had a higher probability of conceiving pregnancy (OR=1.803, 95% CI: 1.448-2.245, p=0.00).

According to our analysis, an increased SCC level within 30 days after first AI, higher than 200×103 cell/ mL, significantly reduced FSCR. This result is in accordance with previous studies suggesting that an elevation in SCC within the first month after AI was associated with a decrease in pregnancy rate (Pinedo et al., 2009; Lavon et al., 2011; Hudson et al., 2012). Considering that mastitis after AI may interfere with corpus luteum formation and regression, progesterone secretion, endometrial functions, and embryonic development (Gilbert et al., 1990; Mann & Lamming, 2001; Spencer et al., 2004), the negative effect of a sudden increase in the SCC level after first AI on FSCR in our study could have resulted from those disruptive effects. Hence, it could be suggested that, to achieve the best CR and diminish economic loss, farmers need to recognize the consequences that subclinical mastitis can have on FSCR and focus on preventing this disease within 30 days after first AI, which is one of the critical periods concerning the success of pregnancy.

It should be noted that, in our study, an increased SCC before first AI had no influence on FSCR. According to the literature, the effect of SCC before AI on FSCR is controversial. Two studies stated that an increase in SCC 30 days before AI (Pinedo et al., 2009) and 10 days before AI (Lavon et al., 2011) diminished the odds of conception. Hudson et al. (2012) also showed that clinical and subclinical intramammary infections (IMI) occurring before or after AI reduce reproductive performance. Furthermore, Schrick et al. (2001) indicated that clinical and subclinical mastitis before AI increase days open and services per conception, irrespective of the pathogen type. Additionally, McDougall et al. (2016) observed a slower conception in cows diagnosed with mastitis before rather than after the first AI. Besides, Miller et al. (2001) have reported that high SCC before AI has a minimal effect on the non-return rate. These results may be explained by the production of pro-inflammatory cytokines triggered by mastitis (SCC >400,000 cell/mL), which reduce the concentrations of steroid and gonadotrophins, leading to an impairment of follicular development and lower oocyte quality (Santos et al., 2018). In contrast, there are some studies that support our findings that SCC before AI has no effect on conception. In this regard, Klaas et al. (2004) claimed that there is no effect on CR or DO. Moreover, Fernandes et al. (2021) reported that SCM before first AI did not affect the likelihood of pregnancy. Reasons for the differences among studies could result from differences in cut-off values of SCC between healthy and infected cows, or timing of IMI relative to AI (Wolfenson et al., 2019).

We found that 305-d milk yields >8780 L and parity negatively affected CR. This is in accordance with several previous studies. Buckley et al. (2003) reported that the hastily increased milk production diminishes FSCR due to the physiological stress. Other researchers claimed that genetics plays a key role regarding this issue (Pryce & Veerkamp, 2001; Grimard et al., 2006), in combination with the disruptive effects of negative energy balance in high producing cows (Grimard et al., 2006). In addition, it is noteworthy that not only does 305-d milk yield increase with increasing parity, but also does the occurrence of periparturient disorders such as retained placenta, endometritis, and metabolic disorders (Lee & Kim, 2006; Pinedo et al., 2020). Consequently, several authors reported a significant effect of parity on pregnancy rate and success at first AI (Tenhagen et al., 2003; Windig et al., 2005; Balendran et al., 2008; Inchaisri et al., 2010; Pinedo et al., 2020). In spite of this, Lee & Kim (2006) claimed that it is difficult to determine the relationship between parity and fertility due to the confusing effect of culling under farm conditions. In a similar way, Lucy (2001), Rocha et al. (2001), Melendez & Pinedo (2007) and Yehia et al. (2020) suggested that primiparous cows show lesser conception rates at first insemination, which may be due to a more severe negative energy balance caused by a greater energy requirement for growth, the impact of the first lactation, and the stress of calving.

As far as season and calving-first AI interval are concerned, different results of the logistic regressions could be due to the exclusion of the SCC factor of the final model when filtering by 30 days before AI, therefore including these factors that were not significant when selecting cases by 30 days after AI. Our results disagree with those obtained by other researchers, who stated that heat stress during summer, either by reducing feed intake or negatively affecting the activity of the hipotalamus-pituitary-ovary axis, has a negative influence on reproductive performance (De Rensis et al., 2017). One possible explanation could be that the area where this study was carried out is not characterised by remarkable high temperatures neither during summer nor throughout the year. In fact, Lopez-Gatius (2003) declared that a cool environment could reduce the risk of reproductive disorders. Regarding calving-first AI interval, our results may be explained by the fact that, given the VWP established in all farms, animals in which the first AI was performed >70 d after calving were probably undergoing certain issues that not only led to a delayed first AI, but may also interfere with fertility. Nevertheless, Kim & Jeong (2019) stated that this factor was not associated with FSCR; however, it should be noted that the VWPs differ between our study and the one performed by these authors (50-60 vs 45 d).

Pregnancy loss

The prevalence of PL during the second month of gestation was 7.6%, and it was not influenced by SCC before first AI, SCC after first AI, DFS, and season (p=0.585, p=0.616, p=0.572 and p=0.232, respectively, Table1). However, cows belonging to the >8780 L group had a doubled proportion of PL than cows with lower milk production (12.0% and 5.6%, p=0.006). In addition, multiparous cows were more likely to tend towards PL than primiparous cows (9.6% and 4.9%, p=0.055). According to the logistic regressions, cows with milk productions >8780 L/305-d were more likely to undergo PL (OR=2.074, 95% CI: 1.063-4.047, p=0.032; OR=2.027, 95% CI: 1.057-3.889, p=0.033 in cows with SCC within 30 days before and after AI, respectively).

The results of this study showed that a sudden increase in SCC neither before nor after AI affects the prevalence of PL between ~28-60 days after AI. On the contrary, Pinedo et al. (2009) reported that cows experiencing a linear somatic cell counts (LNSCC) ≥4.5 (≥300,000 cell/ mL) during the first 90 days of gestation had an increased risk of abortion. Similarly, Moore et al. (2005) concluded that cows with a LNSCC >4.5 before breeding were twice as likely to lose their embryo by 35 to 41 days compared with cows with a score <4.5. We analysed the effect of SCC (higher than 200×103 cell/mL) within 30 days before and after first AI on PL by day between 28-60 of gestation, whereas other studies analysed the effect of LNSCC on PL during the first 90 days of gestation (Pinedo et al., 2009) and before breeding (Moore et al., 2005). Altogether, a different result from our study and other studies may be explained by differences in our design and other designs above-mentioned, as well as different environmental conditions and herd managements among studies. Due to the divergence of results and to the fact that we defined PL based on the second pregnancy diagnosis, there is a possibility that SCM may have a short rather than a long-term effect on pregnancy loss. According to our method, we could only detect pregnancy losses that occurred between ~28 to 60 days of pregnancy (pivotal period 3, according to Wiltbank et al., 2016). However, if SCM had a short-term negative effect on pregnancy loss, that is, by causing either early embryonic loss (EEL, before maternal recognition of pregnancy) or embryonic loss up to 28 days of gestation (pivotal periods 1 and 2; Wiltbank et al., 2016), we would not be able to report it. As it was previously mentioned, mastitis may play a disruptive role on corpus luteum formation and progesterone secretion, which are essential to achieve a successful maintenance of pregnancy (Mann & Lamming, 2001). Moreover, Roth et al. (2013) observed that the proportion of blastocyst obtained 7-8 d after fertilization was significantly less in medium (>200×10³ to <600×10³ cell/mL) and high (>600×103 cell/mL) SCC groups, and they related this impairment to the oocyte quality.

Herein, it was observed that the season variable did not have any influence on PL. This result may be explained by the cool ambient temperature in Galicia region. Besides, it has been claimed that the embryo is susceptible to heat stress mainly during the peri-implantation period, as high temperatures especially affect pre-attachment stage embryos and the magnitude of this effect decreases as embryos develop (De Rensis & Scaramuzzi, 2003; De Rensis et al., 2017). Consequently, this variable may have an influence on EEL, a period that was beyond the scope of this study. Moreover, PL did not differ between <71 and >70 DFS groups. This result is in agreement with the finding of a previous study indicating that calving to insemination interval had no significant influence on embryonic loss (Silke et al., 2002). However, a threshold of 70 days was set in our study, which may be considered not early enough to detect any possible effect of DFS on PL.

On the other hand, among all the variables analysed, our results show that cows in the group with a higher milk yield were more prone to suffer PL, as previously indicated by other researchers (Michel et al., 2003; Grimard et al., 2006). Despite this, Silke et al. (2002) reported that there was no significant relationship between total lactation yield and embryonic loss. Nevertheless, differences between studies may be due to the divergence in milk production, as the mean milk yield used in our study noticeably differed from the one used by Silke et al. (2002) (8780 L and ~7035 L, respectively). It should be noted that not a significant effect, but a tendency, was observed between parity and PL in the current study. This subject is still controversial, as there are researchers who claimed that parity was positively linked to PL (Humblot, 2001; Lee & Kim, 2007; Fernandez-Novo et al., 2020), and those who have reported no effect regarding these two variables (Labernia et al., 1996; Moore et al., 2005). One explanation for a possible relationship between both variables is that the high milk production at third parity in comparison to previous parities would mean a major mobilisation of body fat and a severe loss of body condition (Lee & Kim, 2006). Finally, other factors that were not considered in this study, such as fixed-time-AI and number of AI, were associated with PL (Fernandez-Novo et al., 2020).

It can be concluded that an episode of SCM within 30 days after AI has a disruptive effect on FSCR. Consequently, the adoption of preventive measures should be considered to tackle this issue and diminish the economic loss. Moreover, it may be possible to use the reports of the GDCP as a source of information to predict fertility based on the SCC levels. In addition, although we did not observe any influence of SCM on PL between ~28-60 days after AI, further research should be carried out regarding SCM and both EEL and PL before the first pregnancy diagnosis.