Cows and herd management

Data were collected from a commercial Holstein dairy farm located 20 km NE of Sari, Mazandaran Province, northern Iran, (36‒42.5′N latitude and 53‒10.7′E longitude at an altitude of ‒8 m). Characterized by a generally warm and temperate climate with an annual average temperature of 16.7 °C, Sari receives most of its precipitation as rain in the winter with sporadic rainfall in the summer to yield an annual average precipitation of 690 mm (CDCW, 2013). According to the Köppen-Geiger system (Kottek et al., Reference Kottek, Grieser, Beck, Rudolf and Rubel2006), the climate in Sari is classified as warm temperate with hot dry summers when the average temperature in the warmest month reaches above 22 °C (CSA).

The farm milked around 2500 Holstein cows three times a day (at 06:00, 14:00 and 22:00 h) in a double 40 milking parlour. The cows were fed a total mixed ration without access to pasture. Feeds consisted of about 0.45 forage (maize silage, alfalfa hay and wheat straw) and 0.55 concentrate (barley, maize meal, soybean meal, sunflower meal, canola, cottonseed, wheat bran, sugar beet pulp, supplements and di-calcium phosphate). National research council recommendations were used to balance the rations (NRC, 2001).

For the purposes of the current study, dry cows were kept in a separate group to be transferred to a ‘transition group’, depending on their body condition score and age, 20‒30 days prior to parturition. An early postpartum group was established as ‘fresh cows group’ for postpartum nutrition and controls, and 21-day postpartum primiparous and multiparous lactating cows were transferred to separate groups based on their daily milk production and days in milk (DIM). All animals were tested to ensure they were free of tuberculosis and brucellosis. The vaccination programmes included foot-and-mouth disease, anthrax and blackleg vaccines, regularly administered every 4 months, every 10 months and once a year, respectively. The voluntary waiting period from calving to first AI was 42 days for this herd. Lactating cows were maintained in a facility with shade, fans, sprinklers and a concrete slatted floor. Cows were housed in free stall barns bedded with sand and cleaned by scrapers three times a day. Heat detection was accomplished by visual observation three times a day for 30 min for each group of cows.

Time of insemination was recorded relative to that of milking (pre- or post-milking). Both groups were bred with frozen semen by a technician. Almost all cows were bred in the morning, such that some of the cows were inseminated within 2 h before milking and some within 2 h after milking (depending on the oestrous detection time). Any cows with metabolic and reproductive disorders and those milked ⩾3 h before or after AI were excluded from the study.

Reproductive health management, insemination and pregnancy diagnosis

Uterine discharges of cows were monitored for odours and appearance during the first 21 days of lactation. Postpartum checks (daily control) involved treatment of the following puerperal diseases until resolved or until culling: metabolic diseases such as hypocalcaemia and ketosis (the latter diagnosed during the first or second week postpartum), retained placenta (foetal membranes retained longer than 12 h after parturition) or primary metritis (diagnosed during the first or second week postpartum). The herd was maintained on a weekly reproductive health programme, involving examination of the reproductive tract of each animal by palpation per rectum within 28–30 days postpartum to check for normal uterine involution and ovarian structures. Reproductive disorders diagnosed at this time (including incomplete uterine involution, pyometra or ovarian cysts) were treated until resolved. Involution of the uterus was defined as incomplete when the uterine horns and/or cervix were larger than those in non-pregnant cows. Detectable intrauterine fluid was interpreted as pyometra. An ovarian cyst was diagnosed when a structure 20 mm or larger in diameter in either or both ovaries persisted for at least 7 days in the absence of a palpable corpus luteum. Boluses containing oxytetracycline were always administered into the uterus for cows suffering retained placenta or primary metritis. Pregnancy diagnosis was performed 38–45 days after insemination by palpation per rectum. Pregnancy was reconfirmed after 4 months and pregnant animals were re-examined after 7 months for drying.

Data structure

In the current study, information on AI treatments i.e. pre- and post-milking was gathered after the fact. The number of parturitions from July to January and the number of services from October to April were more than those in any other months of the year. A total number of 1294 inseminations (either successful or failed) were performed in 708 cows from 22 May 2011 to 21 March 2012. All the cows were from parities of one to five. Of the 1294 inseminations, 583 were administered before milking and 711 after milking. The frequency of successful and unsuccessful inseminations at each service attempt is shown in Fig. 1.

Fig. 1. Frequency distribution of successful and unsuccessful inseminations at different services.

The mean and standard deviations for service numbers in the two pre- and post-milking groups were 2 ± 1.3 and 1.6 ± 0.87, respectively. Average daily milk yield during the 30 days up to the insemination ranged between 10 and 67.5 kg, with a mean of 41 ± 9.3 kg. There were 116 missing data for this trait in the post-milking insemination group. Average 30-day milk yield observations were divided into two groups of more and less than the median as low (⩽41.0 kg) and high (>41.0 kg) milk producing cows. DIM at insemination time ranged from 42 to 305 days with the mean and standard deviation of 119 ± 63.6 days, that was ultimately introduced into the final analysis as months in milk (MIM with nine levels DIM < 60 as MIM = 1 to 270 < DIM ⩽ 305 as MIM = 9). Table 1 reports the frequency of cows in the two milk yield groups, the pre- and post-milking insemination times and the means for DIM and milk yield.

Table 1. Frequency of services in the pre- and post-milking insemination groups and the associated milk yield levels

MYL, milk yield level: low (⩽41 kg); high (>41 kg); ADIM, average days in milk; AMY, average milk yield; sd, standard deviation

Climate data such as daily temperature and relative humidity were taken from a meteorological station located 22 km away from the herd. The maximum THI for each day was calculated using the following equation (García-Ispierto et al., Reference García-Ispierto, Lopez-Gatius, Bech-Sabat, Santolaria, Yaniz, Nogareda, De Rensis and Lopez-Bejar2007a):

(1)$$\eqalign{{\rm THI}\, & = \,(1{\cdot}8\, \times \,T_{{\rm max}}\, + \,32)\,- [(0{\cdot}55\,- \,0{\cdot}0055\, \times \,{\rm R}{\rm H}_{{\rm min}})\, \cr & \times \,(1{\cdot}8\, \times \,T_{{\rm max}}\,- \,26)]}$$

where T _max is the maximum temperature and RH_min is the minimum relative humidity. For each insemination record, a THI value was assigned which was the average of the day prior to, the day of and the day after the insemination. The average THI values thus obtained were divided into two sets of below and above 70.

Statistical analysis

In the current retrospective study, all statistical analyses were conducted using the SAS package, version 9.2 (SAS Institute Inc., 2011). Descriptive statistics were determined using the UNIVARIATE and FREQ procedures. The GLIMMIX model procedure with a generalized linear mixed model methodology was used to analyse the results of insemination (success = 1 and failure = 0) as a binomial distribution with the logit link function. The model included the fixed effects of insemination time relative to milking (pre- and post-milking), parity (including the two levels of primiparous and multiparous), insemination month (six levels), MIM (nine levels), interaction effect of insemination time and parity, interaction effect of insemination time and milk yield level (less or more than the average; i.e. 41.0 kg), interaction effect of insemination time and THI levels (normal or HS conditions), and the random effect of cow. The results were presented as odds ratios (OR) and 95% confidence intervals (95% CI).