Data-driven technologies and management practices for improving the sustainability of reproductive management

Reproductive programs and dairy sustainability

There are numerous connections between dairy cattle reproductive management and performance with dairy farm sustainability. Reproductive management plays a pivotal role in sustainable dairy production due to the role of pregnancy and lactation on the productive lifecycle of dairy cattle. Optimal days to pregnancy for both non-lactating heifers and cows increase the lifetime productivity, health, and well-being of dairy cattle which collectively increases farm profitability and reduces environmental impact (De Vries 2011; Liang and Cabrera 2015; Giordano 2019). Reproductive programs for heifers and cows that achieve optimal timing of pregnancy during rearing and lactation optimize milk production efficiency per unit of time and space at the dairy (Stangaferro et al. 2018a; Masello et al. 2021). Heifers that conceive early after becoming eligible for pregnancy generate more milk revenue and have reduced rearing costs per day of life than heifers with delayed pregnancy (Masello et al. 2021). Cows that conceive between the second and fourth month of lactation have a lactation length that maximizes income over feed costs during their productive life (De Vries 2006; Stangaferro et al. 2018b). Generating pregnancies efficiently also contributes to increasing the lifespan of dairy cattle. Timely pregnancy reduces the risk of herd exit (i.e. culling) directly because a pregnant cow can have a subsequent lactation and indirectly because of improved milk production efficiency and health. An effective reproductive program also provides the heifer replacement pipeline needed to achieve a financially optimal replacement dynamic. Farms that experience major improvements in reproductive performance are also able to benefit from a growing number of offspring marketing opportunities. Dairy farms with excellent reproductive performance and heifer rearing programs can increase revenue and reduce costs by diversifying the type of offspring born. These dairies can generate enough dairy offspring to meet the adult herd replacement needs and produce surplus dairy and crossbred or purebred beef calves through AI or embryo transfer (ET). Surplus dairy calves can be marketed as high valued dairy replacements whereas crossbred or purebred beef calves are sold at a premium in the beef market. Finally, minimizing the number of interventions needed for establishing and monitoring pregnancy until calving improves farm sustainability through a reduction in production costs. Thus, the ideal reproductive program for a dairy herd is the one that minimizes the number of interventions needed to achieve the maximum number of pregnancies at a time during the rearing period of heifers and lactation in cows that optimizes the lifetime profitability and well-being of cows. Moreover, an effective reproductive program also generates enough pregnancies to yield the replacements needed to maintain an optimal cow replacement dynamic while generating surplus calves for marketing dairy replacements and beef.

Behavioral, physiological, and performance data from precision livestock farming technologies

Dairy farms are rapidly adopting multiple PLF technologies for automating reproductive monitoring and other herd management tasks. Both wearable and non-wearable sensor-based systems offer unprecedented opportunities to record automatically and in real time a vast array of behavioral, physiological, and performance parameters directly or indirectly associated with cow reproductive biology and performance. Sensor technologies enable data collection from more cows and their environment, in real time, at more frequent intervals, without cow manipulation, and at lower cost. Other sensor and non-sensor technologies used to monitor and record dairy herd management practices and environmental conditions are also readily available on commercial farms.

The growing availability of data from automated estrus monitoring systems that record physical activity and other behaviors, such as rumination, eating, and body temperature, enables the study of associations between expression of estrus in early lactation and subsequent reproductive performance. Several recent observational studies have extensively documented that cows with one or more estrus events recorded by automated estrus detection systems (AEDS) are more likely to express estrus again after the end of the voluntary waiting period (VWP), have greater pregnancies per AI (P/AI) to first and subsequent AI services, and are more likely to become pregnant throughout lactation. For example, in two recent studies from our group conducted on commercial farms in the USA and Mexico we reported similar results for cows fitted with either ear-attached or neck-attached sensors for automated detection of estrus (Rial et al. 2022; Laplacette et al. 2024). Cows with ≥1 automated estrus alerts (AEA) during a 50-day VWP (53% of cows) were more likely to be inseminated at detected estrus for first service when managed with programs that prioritized insemination in estrus (Table 1). Cows with AEA also had greater first service P/AI and more cows were pregnant by 150 days in milk (DIM) compared with cows with no AEA during the VWP (Table 1; Rial et al. 2022; Laplacette et al. 2024). Interestingly, the associations with P/AI and cows pregnant by mid-lactation were observed for cows managed with programs designed to maximize insemination in estrus or programs that used all-timed AI (TAI) for first service. In agreement with our findings, other research groups in North America and Europe have reported similar results (Borchardt et al. 2021; Bretzinger et al. 2023; Gonzalez et al. 2023). Moreover, others reported that features of automated estrus alerts during the VWP such as intensity of estrus, are associated with positive reproductive outcomes (Gonzalez et al. 2023). Collectively, a large body of evidence for a strong association between AEA during early lactation and subsequent reproductive performance suggested an opportunity to develop and implement TRM strategies on cows grouped based on having or not having a record of AEA during the VWP.

Data generated by PLF technologies could also be used to characterize behavioral, physiological, and performance drivers of variation in cow fertility that individually or collectively could be used to predict reproductive outcomes, such as the likelihood of pregnancy to an insemination. Thus, we recently focused on characterizing associations between cow, herd, and environmental data collected by PLF technologies with the outcome of individual inseminations. In one study conducted at a commercial farm where cows and facilities were fitted with several wearable and non-wearable sensors, we evaluated the pattern of several behavioral, physiological, and performance parameters recorded from late gestation to the time immediately before insemination in cows that became pregnant or not to the first service postpartum (Granados 2022). We also explored associations between the first service outcome with early lactation events, previous lactation performance, and environmental conditions. For primiparous cows, the most notable differences observed were greater milk yield, milk components yield, and fewer lying bouts per day for pregnant than non-pregnant cows. On the other hand, numerous and more diverse associations of different magnitude were observed for multiparous cows. In this group, cows that failed to become pregnant produced more milk and milk fat, had greater body temperature, more activity, more resting time, and had larger body weight changes after calving compared with pregnant cows. We also observed several associations of varying direction and magnitude between pregnancy outcome to first service with previous gestation, and previous and current lactation features, events, and performance for both primiparous and multiparous cows. Evidence of multiple associations between these several cow, herd management, and environmental parameters with the likelihood of pregnancy to first insemination suggested a potential path towards the development of pregnancy prediction models that could help group cows for TRM. Moreover, the large variation between parity groups for the direction and magnitude of the associations observed confirmed that cow parity is a major driver of variability and suggested that models for predicting pregnancy based on PLF data might have to be parity specific.

Although more research is needed to fully characterize the numerous and diverse potential associations between PLF data and reproductive outcomes, existing evidence provided the foundation for the exploration of some TRM strategies based on PLF technologies data.

Ovarian physiological status data from reproductive diagnostic tools

Another technology readily available for reproductive diagnostics in dairy farms is B-mode or Doppler ultrasonography. Coupled with recent advances in the understanding of ovarian function and pregnancy biology (Wijma et al. 2016; Ricci et al. 2017; Domingues et al. 2024), ultrasonography can be used to approximate the ovarian physiological status of cows (Wijma et al. 2017), enabling the development and implementation of targeted synchronization of ovulation protocols. As one of the most common uses of ultrasonography in dairy farms is non-pregnancy diagnosis (NPD), much of the efforts to develop TRM strategies including ultrasonography data have focused on previously inseminated cows. Recent studies that characterized in detail the follicular wave and corpus luteum (CL) growth and regression dynamics in cows with and without pregnancy loss helped identify subgroups of cows with contrasting ovarian status at the typical time when ultrasonography is used for NPD in many dairy herds (Wijma et al. 2016, 2017; Ricci et al. 2017). At 29 to 35 days after a previous AI, approximately 70–75% of non-pregnant cows had a functional CL responsive to one or two prostaglandin (PGF)_2α treatments and at least one follicle in the growing or static phase of development (Wijma et al. 2017, 2018; Pérez et al. 2020). In most of these cows the follicle was also large enough (i.e. ≥10 mm) to growth to preovulatory size and ovulate after gonadotrophin-releasing hormone (GnRH) within 3–4 days of induction of luteolysis at NPD. On the other hand, the remaining 25–30% of cows either did not have a functional CL responsive to PGF or had a follicle that was too small to growth to preovulatory size or was at a stage of development at which it would be unresponsive to induction of ovulation within 3–4 days of induction of luteolysis at NPD. Thus, these two distinct groups of cows were considered good candidates to receive protocols for synchronization of estrus or ovulation tailored to their specific ovarian status at NPD. This improved understanding of the ovarian function of previously inseminated cows was the gateway for the design of programs that enable achieving key reproductive performance and management goals, including maximizing fertility, reducing the interbreeding interval, minimizing the number of unnecessary management interventions, and avoiding disruption of the normal pattern of return to estrus after a previous insemination.

Availability of novel methods for pregnancy testing and determination of the ovarian physiological status of cows such as Doppler ultrasonography, cow-side, in-line milk, or bench top on-farm assays for detection and measurement of markers of pregnancy and CL function can lead to further gains in the understanding of ovarian function status and potential before or at NPD. A greater ability to predict ovarian function can enable the development and refinement of novel programs to reduce the interbreeding interval and optimize the fertility of previously inseminated cows beyond what is possible with existing programs. For example, the ability to determine CL functionality on a single observation using Doppler ultrasonography could improve the classification of cows into subgroups with and without a functional CL at NPD. As cow-side or on-farm assay methods to determine progesterone levels become available or more widespread, these could also be used to identify cows in CL functionality groups. Another area of opportunity to improve the performance of re-insemination programs based on the ovarian status of cows at NPD, is improving the prediction of follicle function because follicle size is a poor predictor of the stage of follicle development. Although more difficult because of the less understood association between blood flow and follicular function, and the complexities of defining follicle blood flow on-farm, predicting the stage of follicle development based on Doppler US could help optimize tailored resynchronization of ovulation and estrus protocols.

Data for cow features, events, and performance records

A myriad of associations between cow features, events, and performance records of individual cows and herds with reproductive outcomes have been extensively documented. Therefore, any cow characteristic, application of a herd management practice, or the occurrence of events that are consistently associated with reproductive outcomes could be used as a predictor in TRM. For herds that genomically tests cows, fertility traits could be used to group cows because of the well-documented associations between traits such as genomic daughter pregnancy rate (gDPR) with several outcomes including the likelihood of insemination in estrus, P/AI, and pregnancy rate throughout lactation (Chebel and Veronese 2020; Lima et al. 2020; Sitko et al. 2023). Cow parity, which has been associated with the probability of estrus expression and fertility in multiple studies (Madureira et al. 2015; Pinedo et al. 2020), or records of health disorder events, which were associated with fewer P/AI, reduced pregnancy rates, and increased pregnancy losses (Carvalho et al. 2019) could also be used to classify cows in groups of different reproductive potential. Milk yield level within herds or for cows managed under the same conditions has also been linked to reproductive performance and could be used to group cows for TRM (Stangaferro et al. 2018b). Ultimately, each one of these factors and data associated with reproductive outcomes could be used as predictors for creating subgroups of cows for TRM. In contrast to PLF technologies used for reproductive management, data for several of these predictors are routinely collected or easily recorded in the herd management software at no extra cost as these are typically used for other herd management practices or decision-making.

Combination of multiple predictors

The drivers and factors of biological variation and cow performance that can be used for prediction can act independently and synergistically. Moreover, individual cows can present one or more of these features or be affected by one or more of the factors that influence performance. Therefore, a potential strategy to create subgroups of cows with the largest possible differences in predicted reproductive potential is to combine data from multiple predictors. This was recently explored in a study from our group in which we documented the effect on reproductive outcomes of grouping cows based on the combination cow features, performance, and early lactation events data with AEA during the VWP (Rial and Giordano 2024). Cows were first grouped based on the presence or absence of factors that were associated with poor reproductive outcomes. These included a record of health disorders during the early postpartum period, ranking in the low tertile for gDPR, and ranking in top tertile of milk yield within the herd. Cows were also grouped based on the presence or absence of a record of AEA during the VWP. Finally, cows were grouped based on the combination of risk factors for poor reproductive performance and AEA during the VWP. In support of our hypothesis, grouping cows based on the combination of the risk factors and records of AEA resulted in the creation of groups of cows with larger and more consistent differences in reproductive outcomes than when a single predictor was used to group cows. For example, cows that had at least one AEA during the VWP and did not present any of the risk factors for poor reproductive performance had the greatest likelihood of expressing estrus, had the most P/AI at first service, and the most cows pregnant by 150 DIM compared with the rest of the cows in the herd. On the other hand, cows with no AEA and at least one of the risk factors for poor reproductive performance had the worst reproductive outcomes. These encouraging results provided foundational evidence to explore the combination of multiple predictors for creating groups of cows with stark differences in reproductive performance potential for TRM.

In summary, a large body of evidence currently supports several consistent associations between predictors of reproductive potential and reproductive and management outcomes of interest. The differences observed for cows grouped based on these predictors were of sufficient magnitude to justify exploration and implementation of TRM strategies. Combining multiple predictors might be beneficial for creating subgroups of cows for TRM because no single factor can explain all the variation within and across herds and the multiple predictors available are likely to interact with each other.

Automated detection of estrus during the VWP

Based on the known association between occurrence of AEA during the VWP and reproductive performance (Borchardt et al. 2021; Rial et al. 2022; Gonzalez et al. 2023), first service management programs aimed at optimizing use of insemination in estrus and TAI have been designed for cows with or without AEA during the VWP.

In a recent experiment, Rial et al. (2022) compared the reproductive performance of cows managed with a TRM program designed to prioritize insemination in estrus (TP-AIE) and optimize timing of first service by grouping cows based on AEA recorded during the VWP (Fig. 2). The TP-AIE program aimed to maximize AI to estrus for cows with AEA during the VWP by providing these cows 31 days for detection of estrus. If not inseminated in estrus, cows were enrolled in the P4-Ovsynch protocol (GnRH plus controlled internal drug release (CIDR) insertion-7 days later PGF_2α and CIDR removal-1 day later PGF_2α-32 h later GnRH-16–18 h later AI) for TAI. Conversely, cows with no AEA during the VWP were given only 17 days for detection of estrus and thereafter were enrolled in the P4-Ovsynch protocol. The idea behind this strategy was to minimize use of TAI and maximize use of AI in estrus in cows most likely to express estrus and have good fertility if inseminated to estrus. On the other hand, the more aggressive use of TAI for cows with no AEA during the VWP was expected to help these cows which are less likely to express estrus and have poor fertility if inseminated in estrus. The performance and herd management benefits and drawbacks of the TP-AIE program were compared with a program that used TAI after a Double-Ovsynch protocol and had a longer VWP (ALL-TAI), and a non-TRM program that also prioritized AI in estrus (P-AIE) but did not use AEA during the VWP to group cows. The latter enabled evaluating the effects of managing cows based on AEA during the VWP. The ALL-TAI treatment served as positive control because the program used is known to maximize first service fertility and contribute to an excellent pregnancy rate.

Fig. 2.

Lactating Holstein cows were assigned to a program that used TAI after a Double-Ovsynch protocol with two PGF_2α and an extended VWP for all cows (ALL-TAI), a program that prioritized AIE and used TAI for cows not detected in estrus (P-AIE), or a TRM program that prioritized AIE and provided different periods for detection of estrus to cows that had or did not have AEA during the VWP (TP-AIE). AEA, automated estrus alerts; AIE, AI at detected estrus based on AEA; P4-Ovsynch, Ovsynch protocol with two PGF2α and progesterone supplementation via a CIDR device. Adapted from Rial et al. (2022).

Interestingly, the TP-AIE program resulted in a greater pregnancy rate after calving (HR = 1.2; 95% CI 1.1–1.4) but the same proportion of cows pregnant by 150 DIM than the ALL-TAI program (Fig. 3). Moreover, TP-AIE generated pregnancies faster after calving despite having reduced P/AI to first service (~8 percentage points (p.p.)). As a result, ~50% of the cows were pregnant by the time that cows in the ALL-TAI treatment received first service. This was due to a combination of earlier DIM at first service and early re-insemination of non-pregnant cows. In addition, overall P/AI to first and second services was relatively high (~40%) which contributed to generate many early pregnancies. Although the TP-AIE treatment was effective at increasing the pregnancy rate early in lactation, this gain was diluted as the lactation progressed because the ALL-TAI treatment compensated for the delayed timing of first service through greater P/AI. Thus, from a practical perspective, the TP-AIE program can help maximize insemination of cows in estrus while targeting TAI to cows that otherwise would have a delayed first service. This approach reduces reliance on reproductive hormones and provides a better justification for the use of synchronization of ovulation for TAI. For example, in the experiment of Rial et al. (2022), only 15 and 53% of the cows (~30% overall) with and without AEA during the VWP received TAI as compared with 100% of the cows in the ALL-TAI treatment. In contrast, other than providing an optimized period for AI in estrus and a data-driven justification for the use of TAI (i.e. only cows that need synchronized are synchronized), no major benefits on performance were observed for the TP-AIE than the P-AIE program. Overall, the proportion of cows AI in estrus (P-AIE = 72.3%, TP-AIE = 68.2%), P/AI (P-AIE = 40.2% and TP-AIE = 39.5%), and the pregnancy rate (HR = 1.2; 95% CI 1.1–1.4) did not differ for these two programs. Although the experiment reported by Rial et al. (2022) was conducted at a single herd in the USA, the results were replicated almost exactly in another more recent experiment from our group in which used a similar experimental design but with lactating dairy cows at a dairy in Mexico and used a neck-attached instead of an ear-attached sensor based automated estrus detection system (Laplacette et al. 2024). Similar experiments from other groups which have implemented TRM management based on AEA during the VWP to optimize AI in estrus, and TAI have also reported similar results and benefits to herd management than our experiments (Fricke et al. 2014; Gonzalez et al. 2023).

Fig. 3.

Survival curves for time to pregnancy up to 150 days in milk (DIM). The hazard ratio (HR) and 95% confidence interval (CI) for time to pregnancy was greater (P = 0.008) for the targeted predominant AI in estrus program (TP-AIE; HR = 1.2; 95% CI 1.1–1.4) and predominant AI in estrus program (P-AIE; HR = 1.2; 95% CI 1.1–1.4) than for the all-timed AI (ALL-TAI) treatment. The was no difference between the P-AIE and TP-AIE treatments (HR = 1.0; 95% CI 0.9–1.2). Median time to pregnancy was 102, 89, and 89 days for the ALL-TAI, P-AIE, and TP-AIE treatments, respectively. Adapted from Rial et al. (2022).

Taken together, current evidence suggests that TRM strategies that use AEA during the VWP to prioritize insemination of cows in estrus and optimize timing of insemination might be a reasonable strategy to use when the pregnancy rate after calving and the percentage of cows pregnant by mid-lactation are the most relevant outcomes of interest. Conversely, TRM programs that prioritize insemination in estrus might not be ideal for herds that prefer to or need to maximize first service P/AI. Critical to the success of this TRM based on AEA during the VWP is the implementation of efficient and accurate estrus detection programs combined with systematic and timely use of synchronization of ovulation. This can be easily accomplished in farms that use AEDS during the VWP because estrus alerts can be easily transferred to the dairy herd management software to assign cows to tailored management programs.

Ovarian status at non-pregnancy diagnosis

Recently, TRM strategies designed to tailor hormonal therapy for cows with different ovarian physiological status at NPD were developed and evaluated. These strategies aim to increase AI in estrus and fertility to TAI while reducing the interbreeding interval without disrupting spontaneous expression of estrus. From a management perspective, reinseminating more cows in estrus is appealing to some farms because fewer interventions and cow manipulation are needed for application of hormonal treatments and implementation of pregnancy tests.

A program specifically developed for optimizing reproductive performance and management through more AI in estrus, a short interbreeding interval, and targeted use of TAI for cows that truly need synchronization and hormonal supplementation is Short-Resynch plus P4-Ovsynch. Based on evidence that most cows with a CL ≥ 15 mm and a follicle >10 mm (herein CL cows) at NPD 32 ± 3 days after AI can be synchronized and have acceptable P/AI without the first GnRH of Ovsynch (Wijma et al. 2017), the Short-Resynch (PGF_2α-24 h later-PGF_2α-32 h later-GnRH-16–18 h later TAI; Fig. 4) protocol is applied on these cows. The TAI interbreeding interval for CL cows, which represent 70–75% of the non-pregnant cows, is 35 ± 3 days. This is a full week shorter than if these cows were enrolled in the Ovsynch protocol at NPD. On the other hand, for the 25–30% of cows that have no luteal tissue, a CL < 15 mm, or a follicle >10 mm at NPD (herein NoCL cows), the P4-Ovsynch protocol (Fig. 4) is used. These cows benefit from the targeted use of P4 supplementation and two PGF_2α treatments. The extra P4 optimizes the endocrine environment during follicle growth. The additional PGF treatment ensures complete luteal regression in a group of cows that due to the high ovulatory response to GnRH (Giordano et al. 2016) is likely to have a single 7-day-old CL somewhat refractory to a single PGF treatment at induction of luteolysis (Carvalho et al. 2018). Ultimately, the overall goal of the Short-Resynch plus P4-Ovsynch program is to achieve a combination of benefits. These include maximizing AI in estrus before NPD, reducing the interbreeding interval for CL cows, and optimizing the fertility of NoCL cows. Experiments at commercial farms confirmed the expected positive effects of these programs on herd management outcomes and demonstrated improved herd performance compared with the commonly used D32-Resynch protocol (GnRH-7 days later-PGF_2α-56 h later GnRH-16–18 h later TAI). The Short-Resynch plus P4-Ovsynch program increased the hazard of pregnancy by ~20%, reduced days to pregnancy by 11 days, and reduced the proportion of non-pregnant cows at the end of lactation by ~7 p.p. (Wijma et al. 2018). Notably, these positive effects were attained without disrupting insemination in estrus before NPD, which not only contributes to achieving excellent pregnancy rates but also reduces the number of hormonal treatments (i.e. only cows that need TAI receive hormonal treatments) and pregnancy tests needed.

Fig. 4.

Schematic representation of the Short-Resynch (PGF_2α-24 h later-PGF_2α-32 h later-GnRH-16–18 h later TAI) plus P4-Ovsynch (GnRH plus CIDR insertion-7 days later PGF_2α and CIDR removal-1 day later PGF_2α-32 h later GnRH-16–18 h later AI) program. This management strategy is aimed at maximizing the proportion of inseminations at detected estrus before non-pregnancy diagnosis, reduce the interbreeding interval for most cows (i.e. cows with CL), and increase the fertility of cows without a CL at non-pregnancy diagnosis. AIE, insemination at detected estrus; NPD, non-pregnancy diagnosis; TAI, timed AI.

Some of the herd management benefits of the Short-Resynch plus P4-Ovsynch protocol are also obvious when compared with the alternative D25-Resynch (25 ± 3 days after AI GnRH-7 days later NPD and PGF_2α-24 h later PGF_2α-32 h later GnRH-16–18 h later TAI) for CL cows and P4-Ovsynch program for NoCL cows. In this program, all previously inseminated cows receive a GnRH treatment to initiate a new follicular wave 7 days before NPD. As expected, including the GnRH treatment increases the percentage of CL cows at NPD (~10 p.p.) and their P/AI (~6–9 p.p.). Nevertheless, fewer cows are re-inseminated in estrus before NPD (10 p.p.), more cows require NPD and synchronization, and the interbreeding interval for CL cows is longer than with Short-Resynch because these cows are forced to receive TAI. Notably, the more P/AI for CL cows that receive GnRH did not reduce days to pregnancy or increase the percentage of cows pregnant by the end of lactation because inseminating more cows in estrus and the overall shorter interbreeding interval obtained with the Short-Resynch plus P4-Ovsych program fully compensated for the difference in P/AI (Pérez et al. 2020). From a management perspective, a major caveat of including the GnRH treatment before NPD is that all previously inseminated cows must receive the injection. This causes unnecessary application of one extra dose of GnRH to all pregnant cows, adding costs and disrupting cows and farm management routines. Altogether, evidence from multiple experiments (Wijma et al. 2017, 2018; Pérez et al. 2020) and anecdotal evidence from commercial farms, suggests that the Short-Resynch plus P4-Ovsynch program might be a valuable strategy not only for achieving optimal reproductive performance but also for meeting herd management goals including AI of more cows in estrus, reducing hormonal interventions and reliance on TAI, reducing pregnancy testing needs, and reducing cow manipulation and disruption.

Another type of program that aims to maximize re-insemination of cows at detected estrus for reducing interbreeding intervals and reliance on hormonal interventions for TAI consists of treatment with PGF_2α on CL cows and P4-Ovsynch for NoCL cows at NPD 32 ± 3 days after AI (Fig. 5) (Giordano et al. 2015; Masello et al. 2020a). The PGF_2α treatment promotes expression of estrus which allows for immediate re-insemination after NPD of a substantial proportion of cows. This program is especially appealing to farms that have high estrus detection efficiency with traditional estrus detection methods or the growing number of farms that utilize AEDS. Treatment of cows based on ovarian structures increases efficiency and fertility of subgroups of cows because PGF_2α is only applied to cows likely to express estrus in response to the treatment, whereas the P4-Ovsynch protocol is applied to cows not expected to respond to PGF_2α and known to have increased fertility with this protocol (Wijma et al. 2018). Experiments that evaluated the re-insemination dynamic of lactating dairy cows managed with this type of program (Giordano et al. 2015; Masello et al. 2020a) demonstrated similar performance when compared with enrollment of all cows in Ovsynch-like protocols when at least ≥40% of the open cows with a CL receive AI in estrus after NPD. Because neither overall P/AI nor time to pregnancy improve with these programs compared with alternative resynchronization protocols, their greatest value is for increasing re-insemination of cows at detected estrus rather than by TAI. Ultimately, farms can expect a reduction in the number of hormonal interventions for TAI (Giordano et al. 2015; Masello et al. 2020a).

Fig. 5.

Schematic representation of a reproductive management strategy aimed at maximizing the proportion of inseminations at detected estrus through synchronization of estrus for cows with a corpus luteum and synchronization of ovulation for cows without a corpus luteum at the time of non-pregnancy diagnosis. AIE, insemination at detected estrus; NPD, non-pregnancy diagnosis; TAI, timed AI; P4-Ovsynch, GnRH plus CIDR insertion-7 days later PGF_2α and CIDR removal-1 day later PGF_2α-32 h later GnRH-16–18 h later AI.

As novel data-driven tools become more readily available to more accurately classify cows in the CL and NoCL group, determine the ability of follicles to grow to preovulatory size after induction of luteolysis, predict the likelihood of expression of estrus after AI, and enable earlier NPD, programs that rely on determination of the ovarian status of cows at NPD for tailoring treatments will become more effective and will help achieve more stringent herd management goals relevant to dairy sustainability.

Use of genomic predictions for targeted reproductive management

Based on the known association between genetic merit for fertility and reproductive performance (Chebel and Veronese 2020; Lima et al. 2020; Sitko et al. 2023), genomic predictions have been used to create subgroups of cows for TRM. In an experiment using primiparous lactating Holstein cows from six commercial farms in NY (Sitko et al. 2023), we tested the hypothesis that cows of different genetic merit for fertility would have different responses to a reproductive management program that prioritized AI in estrus (P-AIE) versus one that prioritized TAI (P-TAI) for all AI services (Fig. 6). More specifically, we hypothesized that cows of superior genetic merit for fertility would have similar reproductive performance when managed with the P-AIE or P-TAI program, whereas cows with inferior fertility potential would have improved reproductive performance when managed with the P-TAI program. The idea behind this hypothesis was that high-fertility cows would have good performance regardless of type of management received, whereas the low-fertility cows would benefit from aggressive use of TAI. To test this hypothesis, cows classified in a high (Hi-Fert), medium (Med-Fert), and low (Lo-Fert) fertility group based on a Reproduction Index (RI) value calculated from several gPTA values provided by the Clarifide test (Zoetis Inc.) were assigned to the P-AIE or P-TAI treatment.

Fig. 6.

Cows in the predominant timed AI (P-TAI) treatment received timed AI (TAI) at 84 ± 3 DIM after Double-Ovsynch. Thereafter, cows received AI at detected estrus (AIE) any time or TAI if not AIE. Non-pregnant cows with a corpus luteum ≥15 mm and an ovarian follicle ≥10 mm received a D25-Resynch protocol with two PGF_2α treatments. Cows that did not meet these criteria received a modified Ovsynch-56 protocol with two PGF_2α treatments and progesterone supplementation (P4-Ovsynch; GnRH plus CIDR insertion-7 days later PGF_2α and CIDR removal-1 day later PGF_2α-32 h later GnRH-16–18 h later AI). Cows in the P-AIE treatment were eligible for AIE after a PGF_2α treatment at 53 ± 3 days in milk (DIM). Cows not AIE by 74 ± 3 DIM received timed AI after P4-Ovsynch. Thereafter, cows were AIE at any time. Non-pregnant cows at the pregnancy diagnosis at 32 ± 3 days after AI received P4-Ovsynch for TAI at 42 ± 3 days after AI. AIE*, cows AIE during synchronization of ovulation. Adapted from Sitko et al. (2023).

Data partially supported our hypothesis because cows in different fertility groups had a different response to treatments but only for some of the outcomes evaluated. Moreover, some responses were not in line with our expectations. For the first service, the P-TAI treatment increased P/AI for cows in all fertility groups (Table 2). This agreed with the idea that the Lo-Fert group would benefit by the combination of a longer VWP and use of a fertility protocol (Double-Ovsynch). Thus, these two strategies combined might be used to increase the fertility of the Lo-Fert group when the main goal is to maximize first service P/AI. Alternatively, data also showed that the P-AIE program, which might be simpler, cheaper to implement, and minimizes use of TAI, would most likely reduce P/AI for low-fertility cows (Table 2). Interestingly, cows in the Hi-Fert and Med-Fert group in the P-TAI treatment had increased P/AI, and the increase was of a similar magnitude than for cows in the Lo-Fert group. Thus, our hypothesis that cows of superior genetic merit for fertility would have a similar response to the P-AIE and P-TAI programs was not supported. Data suggested that either the mechanisms by which genetics improves fertility are different but synergistic with those enhanced by the fertility protocol and longer VWP, or the latter enhance the physiological mechanisms under genetic control that cause differences in fertility.

Regardless of the reasons for these observations, our data suggested that certain herd management strategies are possible for cows of superior and intermediate fertility potential. If the goal is to maximize fertility to first service and using a complex protocol for synchronization plus an extended VWP is not a problem, this strategy could be used to maximize P/AI of Hi- and Med-Fert cows. These cows might be good candidates for TRM such as AI with sexed-semen, costly semen from high genetic merit bulls, longer VWP, or other strategies that require the highest possible fertility to first service. Alternatively, if the main goal is to simplify management or reduce reliance on synchronization of ovulation, cows of superior genetic merit for fertility are expected to have good fertility if submitted to AI with a program that prioritizes AI in estrus. Thus, for farms in which management constraints are more important than maximizing first service fertility, prioritizing AI in estrus might be a good alternative for cows of superior fertility potential. On the other hand, for cows of inferior fertility potential, a program that prioritizes TAI with a fertility protocol and an extended VWP could improve performance.

Like for first service outcomes, data for the pregnancy rate by 200 DIM partially supported the hypothesis that cows of different genetic merit for fertility would respond differently to the P-AIE and P-TAI programs (Fig. 7a, b). We observed differences for some but not all outcomes, and for the most part the differences were the result of the different VWP and the insemination pattern caused by the synchronization protocols. Cows in the Hi-Fert group in the P-AIE treatment (Fig. 7a) had a greater pregnancy rate after calving which reduced time to pregnancy and had more cows pregnant by 100 DIM (66.8 vs 61.2%; P < 0.05). Earlier re-insemination because of the shorter VWP and more cows AI in estrus before NPD compensated for the lower first service P/AI than for the P-TAI group. Despite these early lactation benefits for the P-AIE treatment, the proportion of cows pregnant at 200 DIM was the same for both treatments (91.2 vs 91.3%; P > 0.05). On the other hand, for the Lo-Fert group, there was no difference in pregnancy rate (Fig. 7b) or cows pregnant for the P-AIE and P-TAI treatments by 100 (53.7 vs 47.8%; P > 0.05) or 200 DIM (85.6 vs 85.9%; P > 0.05). This was likely the result of a smaller percentage of cows AI in estrus (during the warm season only) and P/AI to first AIE services for the P-AIE program. Reduced fertility for AI in estrus meant that more cows needed re-insemination which offset the benefits of a shorter VWP. Collectively, our results indicated that the Hi- and Med-Fert groups had larger differences in pregnancy rate after calving than the Lo-Fert group when managed with the P-AIE or P-TAI treatment. Nevertheless, effects were only obvious in early lactation with either marginal or no differences after 100 DIM. Thus, if the choice of reproductive management program for primiparous lactating Holstein cows is based purely on reducing time to pregnancy after calving, cows of superior genetic merit for fertility (based on RI or alike) might benefit from programs like the P-AIE in the experiment of Sitko et al. (2023). Cows of inferior genetic merit for fertility could be managed with programs that prioritize either AI in estrus or TAI as no major differences in performance should be expected.

Fig. 7.

(a) Effect of treatments on time to pregnancy for the high-fertility group. The hazard of pregnancy was greater (P < 0.01) for the predominant AI in estrus (P-AIE) than the predominant timed AI (P-TAI) treatment (HR = 1.4, 95% CI 1.2–1.6). Mean and median days to pregnancy were 94 and 81 for the P-AIE treatment and 107 and 86 for the P-TAI treatment, respectively. (b) Effect of treatments on time to pregnancy for the low-fertility group. No difference was observed for the hazard ratio (HR) for time to pregnancy between the P-AIE and P-TAI treatment (HR = 1.1, 95% CI 1.0–1.3). Mean and median days to pregnancy were 106 and 92 for the P-AIE treatment and 114 and 100 for the P-TAI treatment, respectively. Adapted from Sitko et al. (2023).

Thus, current evidence suggests that the effect of TRM based on genetic merit for fertility depends on the individual or sets of reproductive outcomes under consideration. The potential advantages of tailoring management based on genetic merit for fertility depend mostly on the value of inseminating more cows in estrus versus TAI, maximizing fertility, reducing, or increasing days to pregnancy, or a combination of all. Likewise, the value of TRM based on genetic merit for fertility for some farms might be more obvious for improving herd management practices rather than performance. For example, farms could simplify reproductive program implementation or reduce reliance on hormonal interventions while not compromising reproductive performance if cows of different genetic merit are targeted with programs that prioritize AIE or TAI. Ultimately, the choice of outcome or set of outcomes to optimize is farm dependent.