The ovarian function and endocrine phenotypes of lactating dairy cows during the estrous cycle were associated with genomic-enhanced predictions of fertility potential

The objectives of this prospective cohort study were to characterize associations among genomic merit for fertility with ovarian and endocrine function and the estrous behavior of dairy cows during an entire, non-hormonally manipulated estrous cycle. Lactating Holstein cows entering their first (n = 82) or second (n = 37) lactation had ear-notch tissue samples collected for genotyping using a commercial genomic test. Based on genomic predicted transmitting ability values for daughter pregnancy rate ( gDPR ) cows were classified into a high ( Hi-Fert ; gDPR > 0.6 n = 36), medium ( Med-Fert ; gDPR −1.3 to 0.6 n = 45), and low fertility ( Lo-Fert ; gDPR < −1.3 n = 38) group. At 33 to 39 DIM, cohorts of cows were enrolled in the Presynch-Ovsynch protocol for synchronization of ovulation and initiation of a new estrous cycle. Thereafter, the ovarian function and endocrine dynamics were monitored daily until the next ovulation by transrectal ultrasonography and concentrations of progesterone ( P4 ), estradiol, and FSH. Estrous behavior was monitored with an ear-attached automated estrus detection system that recorded physical activity and rumination time. Overall, we observed an association between fertility group and the ovarian and hormonal phenotype of dairy cows during the estrous cycle. Cows in the Hi-Fert group had greater circulating concentrations of P4 than cows in the Lo-Fert group from d 4 to 13 after induction of ovulation and from day −3 to −1 before the onset of luteolysis. The frequency of atypical estrous cycles was 3-fold greater for cows in the Lo-Fert than the Hi-Fert group. We also observed other modest associations between genomic merit for fertility with the follicular dynamics and estrous behavior. There were several associations between milk yield and parity with ovarian,


INTRODUCTION
Fertility is a complex phenotype controlled by multiple endocrine and physiological processes that if impaired result in reproductive failure and inefficiency (Garnsworthy et al., 2008;Walsh et al., 2011).For dairy cattle, these processes are influenced by various genetic, environmental, and herd management factors (Butler, 2003;Berry et al., 2019).As a result of this complexity, some of the processes influencing fertility in lactating dairy cows remain poorly understood.There is growing evidence supporting a strong association between genomic merit for fertility and reproductive performance (Lima et al., 2020;Chebel and Veronese, 2020;Sitko et al., 2023a); however, the understanding of the exact mechanisms under genetic control contributing to these differences is limited.Unraveling the relationship between reproductive function and genetic variation for fertility is necessary to fully realize the potential of dairy cows through selection for enhanced fertility and for enabling the development of novel strategies and technologies designed to optimize reproductive The ovarian function and endocrine phenotypes of lactating dairy cows during the estrous cycle were associated with genomic-enhanced predictions of fertility potential performance and dairy herd management.To this end, it is critical to characterize the physiological, endocrine, and behavioral differences between cows divergent in genomic merit for fertility.
Thus far, several studies have provided insight into the associations of genomic merit for fertility with the dynamics of metabolites postpartum, uterine health, resumption of cyclicity, and estrous behavior in lactating dairy cows (Chebel and Veronese, 2020;Cummins et al., 2012b;Moore et al., 2014) and heifers (Veronese et al., 2019a, b).For example, in a recent study, Chebel and Veronese (2020) reported that genomic merit for daughter pregnancy rate (gDPR) was positively associated with IGF-1 and glucose concentrations, and with the hazard and intensity of estrus.In a recent study, we demonstrated that early lactation cows in the highest tertile for gDPR had improved uterine health, greater ovulatory response, and more favorable endocrine conditions throughout Ovsynch compared with cows in the lowest tertile for gDPR (Sitko et al., 2023b).Using a population of Holstein Friesian cows in a pasture-based seasonal calving system, Moore et al. (2014) and Cummins et al. (2012b) reported that cows in the top 20% for genomic merit for calving interval (fert+) had improved uterine health postpartum, earlier resumption of cyclicity, and greater activity levels during estrus.Similarly, Reed et al. (2022) observed that lactating dairy cows with positive genomic merit for fertility traits under grazing conditions had longer estrus and greater physical activity levels during estrus than cows with negative genomic merit for fertility traits.Conversely, less is known about ovarian function and the endocrine dynamics during the estrous cycle in cows divergent in genomic merit for fertility.Estrous cycle abnormalities due to altered ovarian steroid activity and follicle and corpus luteum growth and demise are of particular interest because these alterations have been associated with subfertility in high-producing dairy cows.In this regard, ovarian function alterations such as premature luteolysis, cystic ovarian disease (Lopez-Gatius and Lopez-Bejar, 2002), delayed ovulation (Royal et al., 2000), double-ovulation (Lopez-Gatius et al., 2005) and anovulation (Santos et al., 2016) have been associated with reproductive failure and inefficiency.A better understanding of differences in these key physiological parameters can help elucidate the underlying mechanisms contributing to differences in reproductive performance outcomes for cows of divergent genomic merit for fertility in commercial farms (Lima et al., 2020;Chebel and Veronese, 2020;Sitko et al., 2023a).Moreover, physiological differences for cows of different genomic merit for fertility can be the basis for the development of novel targeted management strategies to enhance dairy herd reproductive performance and management (Zolini et al., 2019;Giordano et al., 2022;Rial et al., 2022).
We hypothesized that cows in discrete fertility groups based on genomic-enhanced predictions of fertility traits would present different ovarian function, endocrine dynamics, and estrous behavior.Specifically, we hypothesized that cows of superior genetic merit for fertility would have more favorable ovarian dynamics, endocrine conditions, and spontaneous estrus features during the estrous cycle than cows of inferior genomic merit for fertility.Specifically, genomic merit for fertility would be associated with differences in follicular dynamics, corpus luteum growth and regression, serum steroid concentrations, estrous behavior, and ovulation outcomes.Therefore, the objectives of this study were to characterize the associations among genomic merit for fertility with ovarian and endocrine function and the estrous behavior of lactating dairy cows classified in fertility groups based on genomic-enhanced predictions of fertility traits.

Animals and Experimental Procedures
This observational prospective cohort study was conducted from January to October 2021 at the Dairy Unit of the Cornell University Ruminant Center (Harford, NY) using lactating Holstein cows entering their first (n = 82) or second (n = 37) lactation.All procedures performed with cows were approved by the Animal Care and Use Committee of Cornell University (Protocol #2020-0069).
Cows were housed in freestall barns with grooved concrete flooring, deep sand-bedded stalls, self-locking headgates, fans above stalls, and fans and sprinklers placed above the feed lane.Cows had ad libitum access to water and were fed a TMR 1 × /d at approximately 0800 h that was formulated to meet or exceed nutrient requirements of lactating dairy cows producing 45 kg of milk/d as determined by the Cornell Net Carbohydrate and Protein System (version 6.5.5, Cornell University; Van Amburgh et al., 2015).Cows were milked 3 × /d at approximately 8 h intervals (0700, 1500, and 2300 h) in a double-16 parallel parlor.Milk weights were collected automatically at each milking and stored on the herd management software (DairyComp305, Valley Ag Software).During the study period, all cows enrolled in the study were in the same pen and commingled with cows not enrolled in the experiment.

Genotyping and Stratification of Cows Based on Genetic Merit
Ear-notch tissue samples were collected and submitted for genotyping by Zoetis Genetics (Clarifide, Zoetis Genetics).Unique animal identification numbers along with pedigree and genotype were submitted to the Council on Dairy Cattle Breeding (CDCB) to obtain genomically enhanced genomic predicted transmitting ability (gPTA) from the CDCB.Low-density genotypes (27,780 markers) were imputed up to 60,671 markers performed by the CDCB.The Zoetis gPTA and associated reliabilities were estimated using the single step evaluation method (Misztal et al., 2009(Misztal et al., , 2014)), as previously described (Vukasinovic et al., 2017;Gonzalez-Peña et al., 2019).
The average predicted gDPR for the study population was −0.4 with a range of −5.3 to 3.6.All cows with genomic test data available were grouped in order from highest to lowest and separated in a high, medium, and a low fertility group (FG) based on tertiles of gDPR.Cows with gDPR >0.6 were classified as high fertility (Hi-Fert; n = 36), cows with gDPR between −1.3 and 0.6 were classified as medium fertility (Med-Fert; n = 45), and cows with gDPR less than −1.3 were classified as low fertility n = 38).

Experimental Procedures
Weekly, cohorts of cows at 33 to 39 DIM eligible for the study were enrolled in the Presynch-Ovsynch protocol (25 mg of PGF 2α , 14 d later 25 mg of PGF 2α , 12 d later 200 μg of GnRH, 7 d later 25 mg of PGF 2α , 24 h later 25 mg of PGF 2α , and 32 h later 200 μg of GnRH) for synchronization of ovulation.Cows (n = 6) that failed to ovulate after synchronization of ovulation with the Presynch-Ovsynch protocol were resynchronized with the Ovsynch portion of the Presynch-Ovsynch protocol as described within 5 d of the failed induction of ovulation.One cow that failed to ovulate after resynchronization with the Ovsynch protocol was removed from the study.Weekly enrollment enabled monitoring cohorts of cows for the subsequent estrous cycle (Figure 1).All PGF 2α treatments were Dinoprost tromethamine (12.5 mg/mL; Lutalyse HighCon, Zoetis), whereas all GnRH treatments were Gonadorelin hydrochloride (50 μg/mL; Factrel, Zoetis).All hormonal treatments were applied by research personnel and compliance with treatments was 100%.Cows (n = 9) detected in estrus did not receive the final GnRH treatment of the synchronization of ovulation protocol because it was assumed that ovulation was triggered by an endogenous surge of LH.

Data Collection from Automated Estrous Detection System and Estrous Behavior Measurements
After calving, cows were fitted with an ear-attached accelerometer-based sensor of an automated system for detection of estrus (AEDS; Smartbow; Zoetis Inc.) previously validated with lactating dairy cows (Schilkowsky et al., 2021).An internal algorithm of the AEDS calculated the probability of estrus based on physical activity and rumination time and triggered an alert when the probability of estrus was ≥40% for at least 2 consecutive hours.The estrus alert ended when the probability of estrus returned to <40% for 2 consecutive hours.Data on estrus alerts (probability of estrus and estrus alert duration), as well as hourly physical activity parameters (high active time, inactive time, active time) and rumination time were stored by the system and later retrieved from the system server.Estrus alert duration, measured and recorded by the AEDS, was the interval between the onset and end of the estrus alert.Intensity of estrus was calculated as the summation of all high active time for the duration of the estrus alert because the AEDS did not record estrus intensity.Peak high active time was defined as the maximum value for minutes of high active time per hour during an estrus alert.

Transrectal Ultrasonography
Transrectal ultrasonography (TUS) of the reproductive tract and ovaries was performed with a 7.5 MHz linear array transducer (Ibex Pro; E.I. Medical Imaging) to measure and record in an ovarian map the location and size of ovarian structures (i.e., corpora lutea and follicles) present.Ultrasonography was performed at the time of each hormonal treatment of the Presynch-Ovsynch protocol (except the fourth PGF 2α ) and then daily from induction of ovulation at the end of the Presynch-Ovsynch protocol (GnRH2) until the next spontaneous estrus, ovulation, or both, and then once at 7 d after ovulation.For cows with extended estrous cycles (i.e., estrous cycles longer than 30 d), TUS was performed daily until d 35 and thereafter every 48 h until 41 d after the previous ovulation.To determine the timing of ovulation for a subset of cows (n = 40), TUS was performed every 8 h for up to 48 h after the onset of the estrus alert or until the disappearance of the ovulatory follicle (OF), whichever occurred first.
The longitudinal (length) and transverse (width) diameters of ovarian structures were measured using internal digital calipers of the ultrasound machine.At each TUS examination, the position and size of every corpus luteum (CL) and the location and size of all follicles >4 mm in diameter were recorded on an ovarian map.The average follicle diameter [(length + width)/2] was calculated for each follicle.The day of follicular wave emergence was defined as the day when the dominant follicle (DF) was first observed to be closest to 6 mm and the day of dominant follicle selection (i.e., deviation) was defined as the day when the DF was first observed to be closest to 8.5 mm.Codominance was defined as concomitant growth of a dominant (F1) and the largest subordinate follicle (F2) ≥ 10 mm during a follicular wave.Corpus luteum volume was calculated using the formula V = 4/3 × π × radius 3 , where V = volume.If present, the volume of a luteal cavity was removed from the total CL volume.Follicle growth was calculated as the difference in size (mm) between 2 specific time points divided by the time in days or hours between time points.Ovulation was defined as the disappearance of at least 1 follicle ≥10 mm between 2 consecutive TUS examinations and confirmed by the presence of a new CL and progesterone (P4) concentrations ≥1 ng/mL 7 d after presumptive ovulation.The day of ovulation was defined as the first day at which the follicle was no longer visualized.

Blood Collection and Laboratory Assays
Blood samples were collected at the time of each hormonal treatment of the Presynch-Ovsynch protocol (except the fourth PGF 2α ) and then daily beginning at GnRH2 of the Presynch-Ovsynch protocol or the day of estrus for cows detected in estrus before GnRH2.Blood samples were collected daily at the time of each TUS session (during the morning) until the next spontaneous estrus, ovulation, or both, and then once at 7 d after ovulation.For cows with no estrus or ovulation observed until 30 d after GnRH2, blood samples were collected every 48 h from 35 d until 41 d after GnRH2.All blood samples were collected by puncture of the coccygeal blood vessels using 10-mL heparinized evacuated tubes (BD Vacutainer; Becton Dickinson and Co.).Blood sampling tubes were immediately placed on a cooler with ice and transported to the laboratory within 3 h of collection.Blood samples were centrifuged for 15 min at 2,000 g at 4°C.Plasma was aliquoted into 2 Eppendorf tubes and stored at −20°C until assayed.
Blood collected at all time points was analyzed for circulating concentrations of P4 in duplicates using a commercial solid-phase, no-extraction radioimmunoassay (ImmuChem Coated Tube, MP Biomedicals).Control samples with known concentrations of P4 (0.7 and 4.0 ng/mL to represent circulating concentrations of P4 during the pro-estrus and diestrus phase of the estrous cycle) were included at the beginning and end of each assay to assess reliability.As a result of assay stock limitations out of the researcher's control, 2 separate batches of RIA kits were required to analyze all samples.Due to variability between the 2 batches, the coefficient of variation (CV) was estimated for the 18 and 14 assays run with the first and second batch, respectively.Intra-and inter-assay CV for the low and high quality control pool samples for batch 1 were 19.9% and 10.9% (low) and 8.4% and 6.9% (high), respectively.Intraand inter-assay CV for the low and high quality control pool samples for batch 2 were 24.1% and 15.9% and 14.7% and 10.9%, respectively.
The day of onset of luteolysis was defined as the day before P4 declined to less than 50% of the average for the 3 greatest P4 concentrations in the estrous cycle.The day of completion of luteolysis was defined as the day P4 declined to <0.50 ng/mL.Maximum P4 concentration was defined as the greatest single day value during the estrous cycle.
All blood samples collected daily from GnRH2 to ovulation of the subsequent estrous cycle were analyzed for circulating concentrations of FSH.Concentrations of FSH were determined by RIA as described in Adams et al. (1992).The intra-assay CV, inter-assay CV, and the sensitivity were 3.9%, 2.9%, and 0.04 ng/mL, respectively.
Estradiol (E2) concentrations from day −9 until ovulation (d 0) were determined in duplicate samples using a commercial 125I Ultra-Sensitive E2 RIA DSL-  (Turzillo and Fortune, 1990) and previously described by Kulick et al. (1990) with some modifications.Specifically, the non-specific binding, maximum binding, and serial dilutions for the standard curve were prepared in charcoal-stripped bovine plasma and double-extracted with diethyl ether (E492, Fisher Scientific).Purified 17 β-Estradiol solution was used to create the standards (E-060, Sigma).Samples were then frozen in a methanol dry ice bath and the ether supernatant was decanted into borosilicate glass tubes.The ether portion was evaporated under filtered, forced air at 36°C in a multiblock heater in a ventilated hood until dry.On the second day, samples were resuspended with assay buffer, vortexed, and then primary antibody was added.Samples were incubated for 1 h before addition on the I 125 tracer and incubation for 2 h.The antigen-antibody complex was precipitated by addition of goat anti-rabbit gamma globulin serum in buffer with polyethylene glycol as precipitating reagent, vortexed, and incubated for 20 min at room temperature.Finally, tubes were centrifuged, inverted to drain, and counted for 1 min in a gamma counter.For the 15 assays, the E2 intra-and inter-assay CV was 23.5% and 29.8, respectively and the mean assay sensitivity was 0.22 pg/ml.

Cow Removal and Definitions of Estrous Cycle Types and Features
A subset of cows were removed from data analysis if pedigree conflict hindered generation of genomic results (n = 1), were sold or died (n = 5), or did not respond to the synchronization of ovulation twice (n = 1).Additionally, cows were excluded from specific analyses due to biological or management issues that precluded collecting all data needed for a valid evaluation of outcomes of interest.Cows with a single follicular wave that concluded in ovulation (n = 4) were removed from outcomes evaluating CL dynamics.Cows in either of the following categories were removed from all analyses of ovulatory wave dynamics outcomes: cows without ovulation within the observational period (n = 9), cows with only 1 ovulatory wave (n = 4), cows that died before completion of the observational period (n = 1), or cows with OF mistakenly ruptured by manual pressure during an ultrasonography session (n = 1).A subgroup of cows (n = 5) were removed from outcomes evaluated after d 10 of the estrous cycle due to non-compliance with study protocols.Specifically, these cows were accidentally enrolled in a synchronization of ovulation protocol by the on-farm herd management software.A consort diagram with number of cows excluded and reasons for exclusion is included as Supplemental Fig- ure S1.
Upon completion of the observational period and after all data became available for analysis, cows were retrospectively classified as having either a typical or atypical estrous cycle.Cows with spontaneous estrus and ovulation between 12 to 30 d after induction of ovulation with the GnRH2 of the Presynch-Ovsynch protocol were included in the typical estrous cycle group (n = 91) whereas the remaining cows (n = 18) were included in the atypical estrous cycle group.Cows with atypical estrous cycles were grouped into the following categories: (1) premature luteolysis were cows with luteolysis before 12 d after ovulation and with 1 follicular wave, (2) anovular type I were cows without estrus or ovulation that underwent luteolysis and a DF reached ovulatory size (≥10 mm) but became atretic, (3) anovular type II were cows without estrus or ovulation that underwent luteolysis and a DF reached ≥25 mm and persisted until the end of the observation period, (4) anovular type III were cows without estrus or ovulation due to an extended luteal phase (i.e., no luteal regression within 30 d after induction of ovulation), ( 5) cows with estrus that failed to ovulate, (6) silent ovulations were cows with ovulation and no estrus, (7) cows with extended estrous cycles were cows with an interovulatory interval >30 d but had luteolysis followed by spontaneous estrus and ovulation during the observation period.
The ovarian and endocrine dynamics of example cows with 1, 2, or 3 follicular waves and cows with atypical estrous cycles classified as anovular type I, II, and III are presented in Supplemental Figure S2 and S3.
Further, cows were classified based on their ovarian dynamics during synchronization of ovulation with the Presynch-Ovsynch protocol and during the spontaneous estrous cycle.Cows were considered to have an ideal response to the Ovsynch portion of the Presynch-Ovsynch protocol and ideal ovarian function during the estrous cycle if the following criteria were met: ovulated to the first GnRH, had P4 concentrations ≥1.0 ng/mL at the first PGF 2α , had P4 concentrations ≤0.5 ng/ mL at the second GnRH, had P4 concentrations ≥1.0 ng/mL 7 d after the second GnRH, and had a typical estrous cycle as defined.Cows that did not meet at least 1 of these criteria were considered to not have an ideal response to Ovsynch and ovarian dynamics during the estrous cycle.

Statistical Analysis
All statistical analyses were conducted using SAS software (version 9.4, SAS Institute Inc.).

Sitko et al.: REPRODUCTIVE FUNCTION AND GENETIC POTENTIAL
Statistical analyses for binary outcomes were performed using logistic regression with the GLIMMIX procedure fitting a binomial distribution with logit link.Binary outcomes with limited observations were analyzed with the Fisher's exact test using the FREQ procedure.For continuous outcomes, ANOVA was performed with the MIXED procedure.For models to evaluate continuous data, assumptions of normality and homoscedasticity of variance were evaluated with normal probability plots (normal Q-Q plot) and plots of residuals versus predicted values.Data for repeated measurements were analyzed by ANOVA with repeated measurements using PROC MIXED.Different covariance structures (i.e., compound symmetry, first-order autoregressive, Toeplitz, and unstructured) were fitted.
Based on Akaike's information criterion values, a firstorder autoregressive covariance structure provided the best fit.All models for repeated measurements included cow and enrollment cohort as random effects.
Fertility group was offered as a fixed effect and enrollment cohort was included as a random effect in all models.All models were offered lactation number (first and second), season at initiation of synchronization of ovulation (cool and warm), and milk yield group For all outcomes, manual backward elimination of explanatory variables with P > 0.10 was adopted to select final models.Variables were considered significant if P ≤ 0.05, whereas 0.10 ≥ P > 0.05 were considered a tendency.When appropriate, the least significant difference (LSD) post hoc mean separation test was used to determine differences between least squares means (LSM).Unless otherwise stated, all proportions reported are LSM [95% confidence intervals] generated using the LSMEANS statement of GLIMMIX.Quantitative outcomes were reported as LSM ± SEM generated using the LSMEANS statement of the MIXED procedure.For some binary outcomes, arithmetic nonadjusted means are presented because the direction of the association based on the LSM was opposite to the observed values or of a much lesser magnitude.

Estrous Cycle Characteristics
Including all cows with a typical or atypical estrous cycle there was no association (P = 0.21) between FG for the proportion of cows with ≤2 or ≥3 follicular waves.
Eighteen out of the 109 cows (16.5%) evaluated had an estrous cycle classified as atypical.
Four cows had an estrous cycle with 1 follicular wave due to premature luteolysis.Ovulation failure was observed for 3 cows with an atypical estrous cycle (2 cows did not ovulate the third follicular wave DF and 1 cow did not ovulate the second follicular wave DF) and lack of detected estrus occurred in 2 cows with an atypical estrous cycle that ovulated a second follicular wave DF.Two cows with an atypical estrous cycle had an extended interovulatory interval with 3 follicular waves due to either the initiation of a new follicular wave following luteolysis or due to delayed luteolysis.Seven cows with an atypical estrous cycle were anovular during the observational period with ≥3 follicular waves visualized [anovular type I (n = 2), anovular type II (n = 1), anovular type III (n = 4)].There was no association between FG and the proportion of cows classified as having an atypical estrous cycle (P = 0.26; Table 1); however, more (P = 0.03) second (25%) than first lactation (9%) cows had an atypical estrous cycle.The proportions of cows in each FG in each category of physiological and endocrine outcomes used to classify cows in the atypical estrous cycles group are presented in Table 2 (not compared statistically due to low number of observations).
The proportion of cows that were considered to have an ideal response to synchrony and ovarian function during the estrous cycle, as defined for this study, was not associated with FG [P = 0.26; Hi-(55%), Med-(33%) and Lo-Fert (28%)].Arithmetic means for this comparison were 56%, 39%, and 37% for the Hi-, Med-, and Lo-Fert group, respectively.Conversely, a greater (P = 0.02) proportion of cows were classified as having ideal ovarian function during the warm (58%) than the cool season (22%) and a greater proportion (P < 0.01) of cows in the first (67%) than the second lactation (16%) were classified as ideal.Arithmetic means for the warm and cool season were 68% and 28%, respectively, whereas for first and second lactation cows arithmetic means were 47% and 37%, respectively.Additionally, a greater proportion (P = 0.05) of cows in the Hi-Milk90 (74%) than in the Med-Milk90 (23%) and Lo-Milk90 (22%) groups were classified as having ideal ovarian function ideal.Arithmetic means for this comparison were 49%, 40%, and 43% for the high-, medium-, and low milk yield group, respectively.
No associations were observed between FG (Table 3) and the day of the onset of luteolysis (P = 0.33), number of days from follicle deviation to the onset of luteolysis (P = 0.51) or the number of days from the onset to completion of luteolysis (P = 0.48).The period from follicle deviation to the onset of luteolysis was longer (P < 0.01) for cows with 2 than 3 follicular waves (3.3 ± 0.3 d vs. 0.8 ± 0.8 d) and the day of luteolysis was observed earlier (P < 0.01) for cows with 2 versus 3 follicular waves (16.6 ± 0.2 vs. 20.3 ± 0.6 d).No associations were observed between FG and all other outcomes evaluated (P ≥ 0.12; Table 3).Conversely, cows with 3 follicular waves had longer (P < 0.05) intervals for all outcomes related to timing of luteolysis compared with cows with 2 follicular waves.

First Follicular Wave of the Estrous Cycle
Follicular dynamics during the first follicular wave of the estrous cycle are presented in Table 4. Fertility group was not associated (P ≥ 0.42) with day of F1 emergence, size of F1 at emergence, day of F1 deviation, size of F1 at deviation, or the proportion of first follicular waves with codominance.A greater proportion (P = 0.02) of cows in the second (28%) than the first lactation (8%) had codominance.
Growth of the first-wave F1 for the first 10 d after induction of ovulation for all cows, irrespective of having a single dominant or codominant follicles, was not associated with FG (P = 0.56; Figure 4A).In contrast we observed an association between codominance and follicle growth (P = 0.02), whereby F1 size was greater for the first 10 d after induction of ovulation for cows with a single DF than codominant follicles.There was no association (P ≥ 0.76; Table 4) between FG and maximum F1 size or day at which the F1 reached maximum size.Similarly, growth rates from emergence to maximum follicle size and from deviation to maximum follicle size were similar (P ≥ 0.26) between FG.Circulating concentrations of FSH for the first 10 d after induction of ovulation was not associated with FG (P = 0.53); however, we observed an association (P < 0.01) with codominance and the interaction between codominance and day, such that circulating concentrations of FSH were greater on d 0, 1, and 3 after induction of ovulation for cows with codominant follicular waves compared with cows without codominance.
Due to the effect of codominance during the first follicular wave, follicle growth and FSH concentrations were analyzed separately for cows with and without codominance.Growth of the F1, growth of the F2, and circulating FSH concentrations from −3 to 4 d after expected F1 deviation for cows without codominance are illustrated in Figures 5A and 5B.As expected, fol-licle size varied by day (P < 0.01) but there was no overall association between FG and F1 size (P = 0.71), F2 size (P = 0.78), or an interaction between FG and day (P > 0.10).There was also an association with day (P < 0.01) but no association (P = 0.52) between FG and the FSH concentration dynamics from −3 d to 4 d after deviation.Concentrations of FSH around the time of deviation were greater (P < 0.01) for cows in the second than the first lactation (0.23 ± 0.01 ng/mL vs. 0.21 ± 0.01 ng/mL).
Growth of the F1 and F2 and circulating FSH concentrations around deviation for cows with codominance are presented in Supplemental Figure S4.There was a FG by day interaction (P = 0.05) for F1 size whereby the F1 was larger for cows in the Hi-Fert and Med- Fert groups than for cows in the Lo-Fert group on d 3 and 4 after expected deviation.No associations (P ≥ 0.59) were observed for F2.Circulating concentrations of FSH around the time of deviation was not associated with FG (P ≥ 0.48).

Ovulatory Follicular Wave
The follicular dynamics during the ovulatory follicular wave of the estrous cycle are presented in Table 5.On average, the emergence of the OF occurred 13.3 d and 20.5 d after induction of ovulation for cows with 2 and 3 follicular waves, respectively.Fertility group was not associated with OF size at emergence (P = 0.37) or deviation (P = 0.26).The proportion of ovulatory follicular waves with codominance was similar between FG (P = 0.43).Follicle growth and hormone concentrations were evaluated for the last 10 d relative to ovulation to characterize final follicular growth before ovulation.There was a FG by day interaction (P = 0.04) whereby the F1 was larger for the Lo-Fert than the Med-Fert group 9 d before ovulation (Figure 6A).We observed a tendency for codominance by day interaction whereby the F1 was larger on the day before ovulation (−1 d) for cows with a single dominant follicle compared with cows with codominant follicles.
Fertility group was not associated with circulating concentrations of FSH or E2 for the 10 d before ovulation (P ≥ 0.68; Figure 6B and 6C).
Maximum OF size was similar (P = 0.94) between FG (Table 5).Days from OF emergence to ovulation and days from OF deviation to ovulation were similar between FG (P ≥ 0.22; Table 5).The period from emergence to ovulation (P < 0.01) and the period from deviation to ovulation (P = 0.03) were longer for cows with 2 (10.0 ± 0.2 and 8.2 ± 0.2 d, respectively) versus 3 follicular waves (7.9 ± 0.6 and 6.7 ± 0.6 d, respectively).We did not observe an association (P ≥ 0.20) between FG and growth rates from emergence or deviation to maximum size or from emergence or deviation to ovulation.The proportion of cows with double ovulation was greater (P = 0.04) for cows in the Lo-Fert than the Hi-Fert and Med-Fert groups (Table 5).We did not observe an association between inter-ovulatory interval and FG for cows with a typical estrous cycle only (P = 0.85) and for cows with 2 follicular waves only (P = 0.48; Table 5).The average inter-ovulatory interval for cows with 3 follicular waves was 26.2 d.
Circulating concentrations of E2 increased (P < 0.01) from −3 d to 4 d after expected follicle deviation but there was no association (P = 0.63) with FG (Figure  Effects of additional explanatory variables are described in the text.

7
).There was a tendency (P = 0.06) for concentrations of E2 to be greater for the Lo-MilkE (1.63 ± 0.16 pg/ mL) than the Med-MilkE and Hi-MilkE groups (1.13 ± 0.16 pg/mL and 1.37 ± 0.17 pg/mL, respectively).Follicle growth and FSH concentrations around the day of expected deviation for cows without codominance are illustrated in Figures 8A and 8B.There was no association between F1 size from −3 d to 4 d after follicle deviation and FG or a FG by day interaction (P ≥ 0.87).Conversely, there was a milk yield group by day interaction (P < 0.01) whereby the F1 was larger for cows in the Hi-MilkE and Lo-MilkE groups than for cows in the Med-MilkE group on d 2, 3, and 4 after expected deviation.Circulating concentrations of FSH decreased (P < 0.01) from −3 to 4 d around deviation but were not associated with FG (P = 0.22) there was no association with the FG by day interaction (P = 0.84).
Follicle growth and circulating FSH concentrations around the day of expected deviation for cows with codominance during the ovulatory follicular wave by FG are illustrated in Supplemental Figure S5.There was no association between FG or a FG by day interaction for F1 and F2 size from −3 d to 4 d after deviation (P > 0.10).

Estrus Event Features and Pattern of Physical Activity and Rumination Time
Timing of estrus events and estrus event features (probability and duration of estrus, estrus intensity, and high active time peak) as determined by the AEDS are presented in Table 6.No associations were observed between FG (P ≥ 0.67) and any of the outcomes evaluated.
The patterns of physical activity and rumination time around estrus alerts for cows with alerts recorded during the estrus event of the putative ovulatory follicular wave are presented in Figures 9A, 9B, and 9C.There was no association between FG or a FG by time interaction for high active time (P ≥ 0.96) but there was a tendency for a milk yield by time interaction for high active time (P = 0.07; Supplemental Figure S6).There was a tendency for an interaction between FG and time for the amount of inactive time (P = 0.07).Overall, first lactation cows had more (P < 0.01) inactive time (12.2 ± 0.2 min/h) than second lactation cows (10.7 ± 0.4 min/h).Rumination time was similar between FG, and a FG by time interaction was not observed (P ≥ 0.34).Season was associated with rumination time (P < 0.01) as cows spent more time ruminating during the warm (24.4 ± 0.4 min/h) than the cool season (22.6 ± 0.4 min/h).
Average daily milk yield for the 10 d before the onset of estrus was associated with FG (P < 0.01) because yield was greatest for the Lo-Fert group (47.5 ± 1.1 kg), intermediate for the Med-Fert group (43.3 ± 1.1 kg), and lowest for the Hi-Fert group (39.2 ± 1.1 kg).Milk yield was also greater for cows in the second (48.4 ± 0.7 kg) than the first lactation (38.2 ± 1.1 kg).

DISCUSSION
To investigate the physiological mechanisms that might underlie differences in reproductive performance between lactating Holstein cows divergent in genomic merit for fertility as determined by genomic-enhanced predictions (Lima et al., 2020;Chebel and Veronese, 2020;Sitko et al., 2023a), we conducted a comprehensive observational study to evaluate multiple physiological, endocrine, and behavioral parameters associated with reproductive performance outcomes during a spontaneous, non-hormonally manipulated estrous cycle.Our data support the existence of associations between fertility potential based on genomic predictions and the ovarian and hormonal phenotype of dairy cows during the estrous cycle.The most notable manifestation of differences between cows of superior and inferior genomic merit for fertility were consistent and substantial differences in circulating concentrations of P4 during the luteal phase.Increased circulating concentrations of P4 during the mid-to-late luteal phase has been shown to be associated with greater fertility (Rivera et al., 2011;Bisinotto et al., 2013;Pereira et al., 2017) and therefore, may be contributing to some of the previously observed differences in reproductive performance between cows of superior and inferior genomic merit for fertility.Greater circulating concentrations of P4 could also partially explain the reduced likelihood of double ovulation observed in cows in the high and medium FG because P4 concentrations were shown to affect double ovulation rates (Stevenson et al., 2007;Cerri et al., 2011).Another finding not as strongly supported by statistically significant associations but with valuable impli-  cations for the understanding of associations between genetic merit for fertility and reproductive function was the frequency at which aberrant physiological, endocrine, and behavioral phenotypes were expressed in each group.Although not statistically significant, the overall frequency of atypical estrous cycles observed was 3-fold greater for cows of inferior than for cows of superior genomic merit for fertility.Although the incidence of atypical estrous cycles was closer to that of the Lo-Fert group, the proportion of cows with an atypical estrous cycle for the Med-Fert group was intermediate in line with the hierarchy of FG.These findings must be confirmed in future studies with sufficient statistical power to test specific hypotheses related to the incidence of atypical estrous cycles.
We also observed some less consistent and modest follicular wave dynamics differences between cows in the extremes of genomic merit for fertility.Thus, due to the type and magnitude of the observed differences in follicular function, the potential contribution of differences in the follicular wave dynamics to reproductive performance outcomes is less evident than that of the P4 concentrations dynamics and estrous cycle features.

Estrous Cycle Characteristics
In agreement with previous studies with lactating Holstein cows (Taylor and Rajamahendran 1991;Townson et al., 2002;Bleach et al., 2004), we observed that among cows with a typical estrous cycle, there was a predominance of a 2, rather than a 3, follicular wave phenotype.Moreover, the proportion of cows with 2 follicular waves was similar across FG but the proportion of cows with 3 follicular waves was distinct between FG. Almost 20% of cows in the Hi-Fert group had 3 follicular waves compared with <3% of cows in the Lo-Fert group.The Med-Fert group had an intermediate proportion of cows with 3 follicular waves.The low incidence of typical estrous cycles with 3 follicular waves for the Lo-Fert group was not due to a predominance of 2 over 3 follicular wave cycles but rather due to the expression of aberrant estrous cycle phenotypes in a vast majority of cows that did not have a typical 2 follicular wave cycle.Collectively, our data suggested that cows of inferior genomic merit for fertility either have a typical estrous cycle with 2 follicular waves or an abnormal estrous cycle characterized by premature or delayed luteolysis, anovulation, anestrus, or combinations thereof.
The potential contribution of a different ratio of cows with 2 versus 3 follicular waves to differences in OF 3 size at emergence, mm 5.9 6.2 6.0 0.2 0.37 OF size at deviation, mm 8.5 8.6 8.9 0.2 0.26 Codominance4 , % 6.6 [1.4-24.9]6.9 10.9 [37.9-27.6]reproductive performance is not obvious because data for the association between follicular wave patterns and fertility has been inconclusive.Some studies reported greater fertility for cows that ovulated a third-wave DF rather than a 2-wave DF (Townson et al., 2002) whereas others reported no difference in fertility (Bleach et al., 2004;Celik et al., 2005).The potential increase in pregnancies per AI associated with ovulation of a thirdwave dominant follicle is thought to be mediated by a shorter period of dominance of the OF (Townson et al., 2002;Bleach et al., 2004), which we observed for cows with 3 versus 2 follicular waves.Nevertheless, the potential contribution of reduced OF dominance duration to the greater fertility of Hi-Fert cows is uncertain for 2 reasons.First, the evidence for a benefit of 3 versus 2 follicular waves is inconclusive, and second the ratio of cows with 2 versus 3 follicular waves was greatly altered for the Lo-Fert group.
In this study, a subgroup of cows (n = 9) became anovular for >30 d despite previously being cyclic and ovulating after synchronization of ovulation.None of these cows were in the Hi-Fert group.Six of them were in the Lo-Fert and 3 cows were in the Med-Fert group.Ovulation failure in these cows may be attributed to a lack of an LH surge.The 5 other cows were anovular due to delayed luteolysis.All these types of anovular conditions have been previously reported (Taylor and Rajamahendran, 1991;Sartori et al., 2004) but our cur-  rent data suggested a potentially greater incidence in cows of inferior genetic merit for fertility.
Although direct effects of the ovarian and endocrine dynamics of the Lo-Fert group on reproductive performance outcomes could not be validly tested in this study, we speculate that a greater incidence (~20% more cows) of atypical estrous cycles could explain, at least in part, the observed reduction in reproductive performance of low fertility cows in commercial farms reported in previous research (Lima et al., 2020;Chebel and Veronese, 2020;Sitko et al., 2023a).Ultimately, the combination of multiple different abnormalities during the estrous cycle would be expected to reduce reproductive efficiency.The service rate is negatively affected by delayed luteolysis and anestrus (Ranasinghe et al., 2001;Shrestha et al., 2004;Royal et al., 2000), pregnancies per AI are reduced through premature luteolysis, anovulation, and suboptimal circulating concentrations of P4 (Royal et al., 2000;Bisinotto et al., 2010;Santos et al., 2016), and the risk of pregnancy loss increases in cows with double-ovulation (Andreu-Vazquez et al., 2002;Andreu-Vazquez et al., 2012).
Due to the lack of a statistically significant association in this study, further research is needed to confirm if cows with superior fertility potential typically have a greater incidence of a normal estrous cycle phenotype (i.e., more cows with a typical estrous cycle with 2 or 3 follicular waves) and if this is a factor that contributes to the improved reproductive performance of cows of superior fertility potential.
Due to the differences in milk yield between FG (Sitko et al., 2023b) and the known association between productivity and the incidence of atypical estrous cycles (Sartori et al., 2004;Nyman et al., 2014), effects of milk yield could have confounded our observations despite our attempt to control this effect by including milk yield as a covariate in statistical models.Despite this limitation, the incidence of atypical estrous cycles seemed to have a stronger association with genomic merit for fertility than with milk yield because the dif-  ference in incidence of atypical estrous cycles between the high and low milk yield groups was only 6% (22% vs 16%) compared with 20% for the Hi-and Lo-Fert groups.Our findings emphasize the need for a better understanding of the genetic control of ovarian function in dairy cows.More specifically, the underlying physiology driving atypical estrous cycles in cows with low genomic merit for fertility must be elucidated.

Corpus Luteum Dynamics
We observed an association between gDPR and circulating P4 concentrations from 4 to 13 d after induction of ovulation.Specifically, cows in the Hi-Fert and Med-Fert group had on average 15 to 20% greater circulating concentrations of P4 than cows in the Lo-Fert group.This was consistent with the 16% greater CL volume and 34% greater circulating concentrations of P4 reported in Cummins et al. (2012b) for fert+ than fert-Irish cows under grazing conditions.Our observations could be explained, at least in part, by the 10% greater CL volume (not statistically significant) observed from d 4 to 13 for cows in the Hi-and Med-Fert than the Lo-Fert group.In agreement, we reported in another manuscript (Sitko et al., 2023b) that size of the OF from which the CL of the spontaneous estrous cycle originated was on average 7% larger for cows in the Hi-Fert than the Low-Fert group (16.1 mm vs. 15.0 mm on the day of GnRH) suggesting that differences in CL volume were, to some degree, due to follicle size.However, the circulating P4 concentrations differences observed might not be completely explained by CL size because the Lo-Fert group had greater milk yield, which is known to be associated with increased P4 catabolism through increased dry matter intake and liver blood flow (Sangsritavong et al., 2002;Vasconcelos et al., 2003;Wiltbank 2006).We also observed ~15% to 25% greater circulating concentrations of P4 for the Hi-and Med-Fert group than the Lo-Fert group for the 3 d before the onset of luteolysis.Elevated P4 concentrations during the late luteal phase of the ovulatory wave have been associated with improved oocyte quality (Rivera et al., 2011) and pregnancy rate (Bisinotto et al., 2013;Pereira et al., 2017).Manipulative studies have demonstrated an increased pregnancy risk after timed AI in cows with high versus low P4 concentrations during the period from induction of the new follicular wave to induction of luteolysis in Ovsynch-like protocols (Bisinotto et al., 2010;Bisinotto et al., 2013;Pereira et al., 2017).Thus, these results suggested that differences in fertility between cows of high and low genomic merit for fertility may, in part, be mediated by effects of P4 concentrations during early stages of The day of luteolysis after ovulation, the period between emergence of the ovulatory follicle and luteolysis, and the period from luteolysis onset to completion was not associated with FG.Unlike Cummins et al. (2012b) who observed a significantly longer period between luteolysis onset and ovulation in fert-than fert+ cows (5.3 d vs. 4.3 d), our data did not support this association.Differences for the duration of the period of ovulatory dominance are relevant because a longer period of dominance was previously associated with decreased fertility (Bleach et al., 2004;Cerri et al., 2009).Similarly, Bleach et al. (2004) reported that in cows inseminated at detected estrus, the average period from OF emergence to estrus was 1 d shorter in cows that became pregnant compared with cows that did not become pregnant.As the effect of an extended period of OF dominance of a smaller magnitude (i.e., 12 h) than that reported to affect embryo quality in previous studies (Bleach et al., 2004;Cerri et al., 2009) is unknown, it remains to be elucidated if the differences that we observed for cows divergent in genomic merit for fertility are sufficient to generate differences in embryo quality and fertility.

Follicular Wave Dynamics
Except for a doubling in the proportion of cows with codominance during the first follicular wave (not a statistically significant difference) for cows in the Lo-and Med-than High-Fert group, there was no association between FG and any of the first follicular wave outcomes evaluated.These data suggested that differences observed between FG for the overall estrous cycle phenotypes are unlikely to be caused or directly linked to the follicular wave dynamics during the first wave of the estrous cycle.
In this study, FG was not associated with size and growth of follicles during the ovulatory follicular wave and we did not observe differences in maximum ovulatory follicle size.This is in contrast with the observed positive relationship between maximum ovulatory follicle size and genomic merit for fertility reported in Cummins et al. (2012b).
Compared with the Lo-Fert group, we observed a numeric reduction of 11 percentage points in the incidence of codominance for the Hi-Fert group and a similar but significant reduction in the proportion of cows with double ovulation (10 percentage points).The double ovulation rate was also smaller for the Med-Fert than the Lo-Fert group in line with other outcomes for which cows in the Med-Fert group behave more similarly to the Hi-than the Lo-Fert group.These observations which must be confirmed in future studies with large sample size are interesting because multiple ovulations are the most common cause of undesirable twin pregnancies in dairy cattle (Silva del Rio et al., 2006).As reduced circulating concentrations of P4 have been associated with increased incidence of multiple ovulations (Fricke et al., 1999;Lopez et al., 2005), the lower circulating concentrations of P4 around the approximate time of deviation of the ovulatory follicular wave could help explain the greater incidence of double ovulations in the Lo-Fert group in this study.

Estrus Features
We did not observe significant associations between FG and physical activity, rumination time, and estrus features (probability of estrus, duration, and intensity).In contrast, Veronese et al. (2019a) observed a longer duration of estrus for heifers in the highest quartile for gDPR compared with heifers in the lowest quartile for gDPR for synchronized estrus, but not for spontaneous estrus events.Other studies with US Holstein cows (Chebel and Veronese, 2020) and Holstein Friesian cows in Ireland (Cummins et al., 2012b) and New Zealand (Reed et al., 2022) also reported increased estrus duration and intensity for cows of high genomic merit for fertility compared with cows of low genomic merit for fertility.The discrepancy between our study and 3 other studies may be explained by differences in study design and experimental setting.Of note, is the potential confounding effect of the method, definition, and calculation of estrus duration and intensity and the methods used to group cows for comparisons.Our observations do not rule out an association between genetic merit for fertility and estrous behavior, but rather suggested that differences may be of a magnitude that are not easily identified with certain methods for detection of estrus or with study designs that do not fully control for other factors known to affect estrous behavior, such as milk yield.

Study Limitations
This study had limitations including a small sample size for identifying significant differences for some binary outcomes, lack of external validity from using a single research herd, and the approach for selecting the study population and grouping cows.Some limitations were inevitable due to study design constraints.Most notable was the large number of sampling time points which limited the study to a university research herd and the number of cows enrolled.The approach to select the study population and generate groups of cows divergent in genomic merit for fertility aimed to replicate conditions observed at commercial farms that attempt to compare reproductive outcomes or implement targeted reproductive management strategies for groups of cows with different genomic merit for fertility.Despite these limitations, data from the current study could be used as a platform for the design of larger studies in multiple sites with an optimized study population to test novel hypotheses regarding genetic contributions to dairy cow fertility.These data also contributes to the development of targeted management strategies incorporating genomic data.

CONCLUSIONS
We concluded that there are physiological and endocrine phenotypes during the estrous cycle associated with genomic merit for fertility that might help explain part of the differences observed in reproductive performance between cows of superior and inferior genomic merit for fertility.Cows of inferior genomic merit were more likely to present consistent and substantial differences in circulating concentrations of P4 during the luteal phase and some altered ovarian and hormonal phenotypes associated with expression of atypical estrous cycles.Several less consistent and modest differences for the follicular wave dynamics for cows of varying fertility potential were also evident.Findings from this study provide a foundation for a better understanding of the underlying physiology contributing to the differences in reproductive performance between cows divergent in genomic merit for fertility.

Figure 1 .
Figure 1.Graphical depiction of experimental procedures.At 33 to 39 DIM lactating dairy cows were enrolled in the Presynch-Ovsynch protocol (PGF 2α , 14 d later PGF 2α , 12 d later GnRH, 7 d later PGF 2α , 24 h later PGF 2α , and 32 h later GnRH) for synchronization of ovulation.Cows were then monitored daily for an entire estrous cycle until the next spontaneous estrus, ovulation, or both.TUS = transrectal ultrasonography.BC = blood collection.All DIM on the timeline represent the middle day out of the range at which events occurred.
[milk yield >3,673 kg (Hi-Milk90; n = 38), 3,673 kg ≥ milk yield ≥3,079 kg (Med-Milk90; n = 42), and milk yield <3,079 kg (Lo-Milk90; n = 39)] based on accumulated daily milk yield up to 90 DIM.The number of follicular waves (≤2 and ≥3) was offered to models evaluating follicular dynamics and luteolysis outcomes.For models evaluating ovarian dynamics during the ovulatory follicular wave and spontaneous estrus features, milk yield groups were created based on tertiles of average milk yield for the 10 d before the onset of estrus rather than milk yield accumulated up to 90 DIM [milk yield >43.7 kg (Hi-MilkE; n = 34), 43.7 kg ≥ milk yield ≥37.3 kg (Med-MilkE; n = 36), and milk yield <37.3 kg (Lo-MilkE; n = 36)].Models for evaluating concentrations of P4 were offered assay kit batch (1 vs. 2) as a fixed effect to control for variability between assay batches.

2
Effects of additional explanatory variables are described in the text.No p-value available for inter-ovulation interval because in 1 of the groups all observations were equal.3 OF = ovulatory follicle.Typical = cows classified as typical estrous cycle (n = 91); 2 wave = cows classified as 2 follicular wave estrous cycle (n = 80); 3 wave (n = 11) = cows classified as 3 follicular wave estrous cycle.
Figure 9. (A) High active, (B) inactive, and (C) rumination time during the peri-estrus period for lactating dairy cows classified in different fertility groups (FG): high [Hi-Fert; genomic daughter pregnancy rate (gDPR) > 0.6], medium (Med-Fert; 0.6 ≥ gDPR ≥ −1.3) and low fertility (Lo-Fert; gDPR < −1.3) with an estrus alert recorded during the estrus event of the ovulatory follicular wave during a spontaneous, non-hormonally manipulated estrous cycle.Two cows with an estrus alert recorded but failed to ovulate were included.
Sitko et al.: REPRODUCTIVE FUNCTION AND GENETIC POTENTIAL Sitko et al.: REPRODUCTIVE FUNCTION AND GENETIC POTENTIAL 4800 kit (Beckman Coulter) that has been validated for analysis of bovine samples

Table 3 .
Timing of luteolysis during a spontaneous, non-hormonally manipulated estrous cycle for lactating 1Values are presented as LSM.2