In order to achieve maximum success at artificial insemination (AI), capacitated sperm should be present in the female reproductive tract at ovulation time. Then, AI should be carried out close to the ovulation time.

The hormonal fluctuations during the oestrous cycle induce changes on the female reproductive tract at both uterine and ovary level (Carrière et al., 2010; DesCôteaux et al., 2010). The uterine wall presents oedema, and a variable quantity of mucus in the uterine lumen can be observed, which may be greater during oestrus. Besides, uterine blood flow velocity increases during proestrus and oestrus and returns to its lowest values after ovulation (Bollwein et al., 2000). Regarding the ovary, not only an increase in size of the preovulatory follicle can be observed, but also an increase in blood flow around the initiation of the luteinizing hormone (LH) surge that reaches its maximum just before ovulation (Carrière et al., 2010).

In addition, hormonal environment is responsible for the oestrus behaviour, which is characterized by standing immobile while being mounted, restlessness, sniffing the vulva of another cow, resting chin, and mounting (Roelofs et al., 2010). Following these signs, the well-known rule “AM-PM” has been established as a recommendation for AI after oestrus detection (Trimberger, 1948). Overall, it has been stated that the optimal time of insemination is 24 to 12 h before ovulation (Pursley et al., 1998; Roelofs et al., 2006a).

Nowadays, automatic milking machines and monitoring devices are the new tools to perform oestrus detection, and each day more farms are switching to this approach (Roelofs & Van Erp-Van Der Kooij, 2015; Saint-Dizier & Chastant-Maillard, 2018). These technologies can discern between the onset, middle and end of oestrus, and thus determine its duration and even the maximum heat activity.

Therefore, whether using observational oestrus detection or electronic devices, it is necessary to determine which oestrus characteristic features are better indicators of ovulation, aiming to perform AI at the appropriate time. Some studies have been conducted using behavioural signs (Layek et al., 2011), ultrasonographic features of the preovulatory follicle (Siddiqui et al., 2010), and progesterone (P4) concentrations (Roelofs et al., 2006b). Additionally, investigations on the relation between uterine wall changes and hormone concentration and fertility have been reported (Souza et al., 2011; Sugiura et al., 2018). To our knowledge, research on the combined evaluation of uterine features and size and vascularization of the preovulatory follicle to predict ovulation time, with the aid of monitoring devices to perform oestrus detection, has not been published. In addition, another value of this study is that it aims to simulate as closely as possible the conditions veterinarians have to face on farm work in a daily basis, so the results can have a direct application.

Consequently, the objective of this study was to assess the utility of B-mode ultrasonography to predict ovulation time by assessment of uterine and follicular measurements; additionally, we aimed to determine the usefulness of Power Doppler ultrasonography compared to B-mode to predict ovulation, by the assessment of blood flow to the preovulatory follicle.

Of the 33 cows, 8 (24%) did not respond to the treatment. In addition, 2 cows (6%) ovulated before 52 h and were excluded from de analysis. The distribution of ovulation times of the remaining cows (n = 23) is displayed in Fig. 1. Additionally, descriptive statistics are shown in Table 1. No statistically significant differences were observed in this experiment regarding 2 or 3 PGF2α administrations for the variables of interest.

Results for the one-way ANOVA test showed that, for the variable UL, time -6 (1.53 mm) significantly differed from time -42 (5.69 mm, p = 0.016). Concerning DF, significant differences were observed between time -6 (20.48 mm) and time -54 (16.60 mm, p = 0.038). As for P4, significant differences were found between time -6 (0.34 ng/mL) and time -54 (1.03 ng/mL, p = 0.013). No significant differences were found between examination times for END/MYO (p = 0.707). Considering MHA, statistically significant differences were observed for the UL between group AFT (1.60 mm) and groups BEF and DUR (4.90 mm (p = 0.002) and 4.84 mm (p = 0.009), respectively); similar results were obtained for DF, with significant differences between group BEF (18.63 mm) and group AFT (20.90 mm, p = 0.043). As for heat, the UL significantly differed between group AFT (1.57 mm) and groups BEF and DUR (5.09 and 4.25 mm, respectively, p = 0.002).

Results for the Pearson’s χ² test showed that the percentage of cows with Doppler signal in the preovulatory follicle at examination times -54, -42, -30, -18, and -6 h was 0%, 23.1%, 14.3%, 36.4%, and 54.2%, respectively (p = 0.011). For the variable PDCL, the percentage of cows with Doppler signal was 60.0%, 23.1%, 19.0%, 8.7%, and 0% (p < 0.001). Concerning MHA and heat, statistically significant differences were found for both variables (Table 2).

One important finding was the presence of Doppler signal in most preovulatory follicles few hours before ovulation, and the increase in the percentage of follicles with blood flow during and after MHA and heat. Blood flow to the preovulatory follicle is correlated to the increase concentration of oestradiol and the LH surge (Acosta et al., 2003; Pancarci et al., 2012). Moreover, Siddiqui et al. (2010) reported a biphasic increase and decrease in blood flow to the preovulatory follicle, with the first peak occurring 3 h after GnRH treatment and the second peak occurring 8-6 h before ovulation. Probably due to our study design (examinations carried out every 12 h), we were not able to identify the two peaks, but a progressive increase in follicles with detectable peripheral vascularization instead. Our results suggest that the identification of Doppler signal on the follicular wall might be of use as an indicator of ovulation within the following 6 to 18 h. Additionally, this identification took place mostly after MHA and heat, determined by electronic monitoring devices, which offers the possibility of a complementary use of these two approaches.

Regarding the UL, differences were observed among times of examination. During oestrus, the UL may show an accumulation of anechogenic content inside the uterine horns, that will normally disappear during the dioestrus phase (DesCôteaux et al., 2010). However, no significant differences were observed for UL at the examination times closer to ovulation (-18 and -6), in contrast with the PDF. Consequently, UL may be useful as a positive sign of heat, but it may not be an adequate parameter to estimate ovulation time. Special attention should be paid to the characteristics of the uterine content to discern between heat and pathological situations like endometritis (Sheldon et al., 2006).

As for END/MYO, no significant differences were found among examination times. The high level of circulating oestrogens during the perioestrus period is responsible for the greater thickness and oedema of the uterine wall, that will normally decrease between 4 and 5 days after oestrus (DesCôteaux et al., 2010). A possible explanation for the lack of significance could be the fact that the first examination was delayed 52 h after PGF2α administration. Therefore, cows were already at the onset of heat. Nevertheless, no differences were found at the times closer to ovulation, which suggest that END/MYO could be a good parameter to identify cows in heat, but not to predict the time of ovulation. Additionally, it should be noted that endometrial thickness also increases during endometritis due to inflammation. Consequently, a differential diagnosis should be made taking into account other aspects such as the uterine fluid, time of the cycle and the diagnostic tools available like endometrial cytology (Dubuc et al., 2010).

Finally, we used Doppler ultrasonography and P4 serum levels to assess luteal function, since its accuracy has been previously described by numerous researchers (Siqueira et al., 2013; Rocha et al., 2019; Dubuc et al., 2020). In this case, similar results were obtained, as PDCL progressively decreased until heat, and was absent near ovulation.

In conclusion, the use of Power Doppler to predict the time of ovulation needs further research. Additionally, END/MYO and UL could be of use to identify cows in heat; however, their application to determine ovulation time seems to be inaccurate. Therefore, the difficulties to predict the exact moment of ovulation in field conditions have to be considered, and additional studies, including a bigger sample size, are needed.