An individual reproduction model sensitive to milk yield and body condition in Holstein dairy cows

Changes in body condition score from calving to first insemination and milk yield, pregnancy per AI, and pregnancy loss in lactating dairy cows: A meta-analysis

We determined the association of body condition score (BCS) at calving, at first postpartum artificial insemination (AI), and change in BCS between calving and first AI on pregnancy per AI (P/AI) at 30-45 d, pregnancy loss to 60-85 d, and milk yield in lactating dairy cows. Outcome data were included from 15 studies and 47 herd-year combinations. Additional variables included season of AI, herd, days in milk at first AI, parity, and of mean daily milk yield within 2 wk of first AI. The BCS scale employed was a standard 1-5 scale (1 = severe under conditioning or emaciated and 5 = severe over conditioning) with 0.25 cut points. Presynchronization treatments that included PGF_2α and GnRH increased (P < 0.05) the proportion of cows with luteal function before AI compared with PGF_2α alone. Compared with no presynchronization treatment those that included PGF_2α or PGF_2α and GnRH increased (P < 0.05) first P/AI. Cows having BCS ≥2.75 at AI had greater (P < 0.01) first P/AI than cows with BCS <2.75. As BCS at first AI increased, P/AI increased in a linear (P = 0.04) fashion and was greater in cows expressing estrus when BCS at AI was <2.50. Presynchronization had no association with P/AI for cows with BCS at calving <3.00 compared with those with BCS ≥3.00. In contrast, multiparous cows tended (P = 0.06) to have greater P/AI when they calved with BCS ≥3.00 compared with <3.00. Increasing BCS at AI was associated with decreased (P = 0.01) pregnancy loss. Pregnancy per AI did not differ among cows according to the magnitude of prebreeding BCS loss, but more multiparous cows losing more than 0.5 units of BCS tended to have greater pregnancy losses in second-parity cows (P = 0.09) and in cows of third or greater (P < 0.001) parity. Daily milk yields at first AI differed among parities as expected, but a parity by BCS at calving interaction was detected (P = 0.008). Daily milk yield at first AI decreased (P < 0.001) linearly as BCS at AI increased, with an exacerbated greater negative effect during summer. More prebreeding loss in BCS was associated with more (P < 0.05) milk yield in first- and second-parity cows. We concluded that greater BCS at first AI was associated with improved P/AI, but magnitude of prebreeding BCS loss was not associated with P/AI. In contrast, more pregnancy loss was associated with more prebreeding BCS loss in multiparous cows. Cow having lesser BCS at AI and greater prebreeding loss in BCS produced more milk than their herd mates of greater BCS and lesser prebreeding loss in BCS, respectively.

Coupling a reproductive function model to a productive function model to simulate lifetime performance in dairy cows

Reproductive success is a key component of lifetime performance in dairy cows but is difficult to predict due to interactions with productive function. Accordingly, this study introduces a dynamic model to simulate the productive and reproductive performance of a cow during her lifetime. The cow model consists of an existing productive function model (GARUNS) which is coupled to a new reproductive function model (RFM). The GARUNS model simulates the individual productive performance of a dairy cow throughout her lifespan. It provides, with a daily time step, changes in BW and composition, fetal growth, milk yield and composition and food intake. Genetic-scaling parameters are incorporated to scale individual performance and simulate differences within and between breeds. GARUNS responds to the discrete event signals 'conception' and 'death' (of embryo or fetus) generated by RFM. In turn, RFM responds to the GARUNS outputs concerning the cow's energetic status: the daily total processed metabolizable energy per kg BW (TPEW) and the net energy balance (EB). Reproductive function model models the reproductive system as a compartmental system transitioning between nine competence stages: prepubertal (PRPB), anestrous (ANST), anovulatory (ANOV), pre-ovulating (PREO), ovulating (OVUL), post-ovulating (PSTO), luteinizing (LUTZ), luteal (LUTL) and gestating (GEST). The transition from PRPB to ANST represents the start of reproductive activity at puberty. The cyclic path through ANST, PREO, OVUL, PSTO, LUTZ and LUTL forms the regime of ovulatory cycles, whereas ANOV and GEST are transient stages that interrupt this regime. Anovulatory refers explicitly to a stage in which ovulation cannot occur (i.e. interrupted cyclicity), whereas ANST is a pivotal stage within ovulatory cycles. Reproductive function model generates estradiol and progesterone hormonal profiles consistent with reference profiles derived from literature. Cyclicity is impacted by the GARUNS output EB and clearance of estradiol is impacted by TPEW. A farming system model was designed to describe different farm protocols of heat detection, insemination, feeding (amount and energy density), drying-off and culling. Results of model simulation (10 000 simulations of individual cows over 5000 days lifetime period, with randomly drawn genetic-scaling parameters and standard diet) are consistent with literature for reproductive performance. This model allows simulation of deviations in reproductive trajectories along physiological stages of the cow reproductive cycle. It thus provides the basis for evaluation of the relative importance of different factors affecting fertility at individual cow and herd levels across different breeds and management environments.

Evaluating the ability of a lifetime nutrient-partitioning model for simulating the performance of Australian Holstein dairy cows

The present study determined the ability of a lifetime nutrient-partitioning model to simulate individual genetic potentials of Australian Holstein cows. The model was initially developed in France and has been shown to be able to accurately simulate performance of individual cows from various breeds. Generally, it assumes that the curves of cow performance differ only in terms of scaling, but the dynamic shape is universal. In other words, simulations of genetic variability in performance between cow genotypes can be performed using scaling parameters to simply scale the performance curves up or down. Validation of the model used performance data from 63 lactations of Australian Holstein cows offered lucerne cubes plus grain-based supplement. Individual cow records were used to derive genetic scaling parameters for each animal by calibrating the model to minimise root mean-square errors between observed and fitted values, cow by cow. The model was able to accurately fit the curves of bodyweight, milk fat concentration, milk protein concentration and milk lactose concentration with a high degree of accuracy (relative prediction errors <5%). Daily milk yield and weekly body condition score were satisfactorily predicted, although slight under-predictions of milk yield were identified during the last stage of lactation (relative prediction errors ≈11.1–15.6%). The prediction of feed intake was promising, with the value of relative prediction error of 18.1%. The results also suggest that the current recommendation of energy required for maintenance of pasture-based cows might be under-estimated. In conclusion, this model can be used to simulate genetic variability in the production potential of Australian cows. Thus, it can be used for simulation of consequences of future genetic-selection strategies on lifetime performance and efficiency of individual cows.