Factors influencing the success of an artificial insemination program in Florida goats

An artificial insemination (AI) program using cooled semen was evaluated over a 7-year period in Florida goats. The effect of the following variables on pregnancy rates was analysed: production system, year and season of AI, synchronisation treatment, bucks, AI technicians, semen deposition site, days in milk at AI, milk yield and parity. Animals were reared under field conditions on commercial farms in southern Spain. Semen was collected from nine bucks and cooled at 4°C until use. A total of 3,941 goats were synchronised using intravaginal progesterone sponges and inseminated 46.0 ± 0.5 h. after sponge removal. Pregnancy was diagnosed by ultrasonography 42-46 days after AI, obtaining an average pregnancy rate of 48.7%. Logistic regression showed that production system, AI year and season, bucks and semen deposition site had a significant effect (p < 0.05) on pregnancy rate, while the other variables analysed were removed from the model. The final statistical model accounted for 59.7% of the cases analysed, suggesting that other factors not studied here may influence pregnancy rates in Florida goats. Additional key words: buck; fertility; fresh semen; goat; pregnancy.


Introduction
The Florida goat is a dairy breed reared in southern Spain, with an average milk yield of 650 kg per goat per year and an average lactation period of 10 months (Sánchez, 2007).This breed was created in the early 20 th century by crossing Pyrenean and Nubian goats, in a process similar to that which gave rise to several current breeds, such as Anglonubian goat.In 1997, it was included as threatened Spanish breed in the Official Catalogue of Spanish Livestock Breeds (BOE, 2009).An artificial insemination (AI) program was implemented in 2005 in Florida goats in order to optimise selection schemes and control kidding dates.This technique is an essential tool in goat breeding programs, since it increases the efficiency of sire genetic evaluation and the extension of genetic improvements; at the same time, it enables control of parturition dates with a view to meeting market demands.Insemination is usually carried out transcervically, and semen is deposited either at the entry of the cervix or, more rarely, in the uterus (Nunes & Salgueiro, 2011).However, AI is not widely used in goats.Cooled semen provides better results than frozen-thawed semen.Hormone treatments are required for fixed-time artificial insemination (FTAI) in order to avoid oestrus detection and synchronise oestrus in a group of goats.This enables more goats to be inseminated over a shorter period, as well as allowing insemination and births to be carefully scheduled; it also permits an increase in the number of kids obtained by AI at the start of the kidding season, thus enabling a more efficient use of human resources (Nunes & Salgueiro, 2011).
A number of factors are reported to affect the success of AI, including nutrition, breeding season, environmental conditions, parity, breed, farm, depth of semen deposition, extender composition and hormone treatment (Mellado et al., 2004 and2006;Salvador et al., 2005;Nunes & Salgueiro, 2011).The present study sought to clarify the effect of different factors (production system, year of AI, season, synchronisation treatment, bucks, AI technicians, semen deposition site, days in milk (DIM) at AI, milk yield and parity) on pregnancy rates after AI with cooled semen in Florida goats reared under commercial farm conditions.

Animals, management and housing
From 2005 to 2011, a total of 3941 inseminations were carried out in goats registered in the Florida breed stud-book, as a part of a breeding and selection program.Goats were reared on commercial farms and, considering the size of the herd, the grazing surface and the concentrate consumption (Nahed et al., 2006), three different production systems were evaluated: intensive, semi-intensive and semi-extensive.Milk yield was officially recorded.All females had at least one kidding and were evaluated by ultrasonography prior to AI in order to exclude individuals with reproductive disorders such as pseudo-pregnancy.Does with previous abortions or reproductive problems were not included in the AI program.
Nine Florida bucks were housed in individual stalls at the Andalusian Government's IFAPA Research Centre (Hinojosa del Duque, Cordoba, Spain; 38.30° N, 5.09° W) in barns with a concrete floor and large roof windows with access to external paddocks, allowing some interaction with the environment.Bucks were daily fed with a commercial concentrate (0.5 kg) and given ad libitum access to alfalfa hay, water, and mineral supplementation blocks.They were kept in a light-controlled environment (two months of long days alternating with two months of short days).To this end, artificial light was switched on at 8:00 a.m. and off at 16:00 p.m. or 00:00 during the short and long photoperiod treatments, respectively.The supplementary light source provided a light intensity of roughly 200 lx.

Semen collection, assessment and processing
Semen was collected twice a week by artificial vagina, and volume and sperm concentration (Accucell, IMV, France) were assessed.Semen samples were diluted in a skimmed milk-based extender to a final concentration of 800 × 10 6 spermatozoa mL -1 , and then cooled at 4°C.Briefly, one sample was warmed to asses sperm motility (S.C.A., Microptic SL, Spain) and membrane integrity (hypo-osmotic swelling test) in order to determine semen quality before preparing doses.Semen was used for AI only when it complied with the following criteria: membrane integrity up to 40%, over 70% progressive motility and over 60% mass motility.Inseminating doses were packed into 0.25 ml straws with 200 × 10 6 spermatozoa and transported under refrigerated conditions to the farm, where insemination was performed by qualified technicians.The whole procedure was carried out early on the same day.

Synchronisation and insemination
Oestrus synchronisation was carried out using intravaginal progestagen-impregnated sponges with varying doses of fluorogestone acetate (FGA, 20 and 45 mg, Chronogest, Intervet International, Boxmeer Holland) inserted for 11 ± 1.0 days.Forty-eight hours before sponge withdrawal, goats received an intramuscular injection of 50 mg cloprostenol (Estrumate, Schering-Plough Animal Health, Upper Hutt, New Zealand) and 400 UI of equine chorionic gonadotropin (eCG) (Foligon, Intervet, Spain).Only one hormonal synchronisation treatment per goat and per year was used to reduce the effects of repeated use of eCG, specifically the production of anti-eCG antibodies which reduce the fertility of artificially-inseminated females.Fixed-time artificial insemination with cooled semen was performed 46.0 ± 0.5 h after sponge withdrawal, using a lighted speculum and an ovinecaprine AI catheter.The aim was to deposit the sperm dose in the uterine cervix; information on the semen deposition site (cervical vs. uterine) was recorded at each insemination.
Pregnancy was diagnosed around day 42-46 after AI using an ultrasound scanner (Pie-Medical, Netherland) equipped with a 5.0 MHz linear-array transducer.Animals were restrained against railing in the standing position and scanning was performed in the inguinal region.

Statistical analysis
Records were routinely kept on farms and classified for data analysis purposes.Logistic regression analysis was performed on data for each insemination, using pregnancy diagnosis as the dependent variable (0 or 1).The following independent variables were tested: production system (intensive, semi-intensive, semi-extensive); year (2005 to 2011); AI season (cold: autumnwinter, hot: spring-summer); synchronisation treatment (20 mg or 45 mg FGA); buck (9 inseminating males); depth of semen deposition (cervical or uterine); AI technician (1, 2, 3, 4, 5); parity (1, 2, 3 and ≥ 4); days in milk at AI (DIM) (≤ 154 days, 155-181 days, ≥ 182 days) and milk yield (≤ 1.80 kg, 1.81-2.69kg, ≥ 2.70 kg).All categorical independent variables were first screened for univariate association to pregnancy rate using the crosstabs procedure.Only variables displaying significant values were included in the logistic regression analysis, which was run using the "Forward Wald" stepwise selection method to specify how independent variables are entered into the analysis.The level of statistical significance for variables included in the model was set at p < 0.05.This procedure enabled calculation of the odds ratio (OR) and the 95% confidence interval as an indicator of risk or probability of pregnancy in an AI program using cooled semen.The class whose value was nearest to the mean pregnancy rate was selected in order to establish categorical variable reference values.
The Hosmer-Lemeshow method (Hosmer & Lemeshow, 1989) was used to test the goodness of fit of logistic regression models.All statistical analyses were performed using the SPSS 15.0 statistical software package.
Regression analysis of the OR for pregnancy rate on milk yield categorised in 0.5 kg increments was evaluated to determine the fitted line plot that best described this relation.

Results and discussion
A mean pregnancy rate of 48.7% was recorded over the 7-year period.The mean number of DIM and milk yield at AI were 168.66 ± 58.49 days (mean ± SD; range: 27-489 days) and 2.34 ± 1.0 kg (range: 0.18-7.04kg), respectively.Parity ranged from 1 to 11 births.
Preliminary crosstabs procedure showed that AI technician, parity and DIM had no significant effect on fertility.Otherwise, production system, AI year, AI season, synchronization treatment, buck, depth of semen deposition and milk yield at insemination were included in the initial logistic model.After run the logistic procedure, synchronization treatment and milk yield at insemination were removed from the final model as they did not affected significantly the dependent variable.The odds ratios are shown in Table 1.
Results showed that the probability of pregnancy using cooled semen was significantly higher under the intensive production system than under the semi-intensive system.The probability of pregnancy varied significantly as a function of AI year, dropping to 35.3% in 2007 (OR = 0.56) and rising to 54.8% in 2008 (OR = 1.32).Goats inseminated in cold season (autumn and winter) recorded lower pregnancy rates (OR = 0.74).The probability of pregnancy was greater when semen from buck n.2 (OR = 1.41) and n.5 (OR = 1.42) was used.Cervical semen deposition was Artificial insemination in Florida goats associated with a lower probability of pregnancy than uterine insemination (OR = 1.82).The final model correctly classified 59.7% of the cases analysed, with a sensitivity of 57.8% and a specificity of 61.8%.The Cox & Snell R Square value showed that only 4.3% of the variance in fertility was attributable to the variables included in the model for the studied population.
Data obtained from 2005 to 2011 were used to analyse the effect of several factors on fertility after AI using cooled semen in Florida goats.Results showed that the pregnancy rate was significantly affected by the production system, AI year, AI season, buck, and semen deposition site.The mean pregnancy rate (48.7%) was similar to that reported by Siqueira et al. (2009), but lower than described under controlled conditions (Roca et al., 1997;Romano et al., 2000).
The production system (intensive, semi-extensive or semi-intensive) strongly influenced the success of AI.Pregnancy rates were significantly higher under intensive systems (Table 1).This may be due in part to a greater degree of animal management in intensive production, leading to lower stress during goat-human interactions.
The marked year-on-year variations in fertility, with a sharp fall in 2007 and a marked rise in 2008, may reflect the interaction of a range of factors, including weather conditions, animal welfare, nutrition and animal breeding.
Goat reproduction is affected by changes in day length over the year.Photoperiodic control of reproductive patterns is mediated through circadian rhythmic secretions of melatonin by the pineal gland during dark- ness, which influences gonadotropin-releasing hormone pulse generation and the hypothalamic-pituitary-gonadal feedback loop (Fatet et al., 2011).Reported goat conception rates range from 50% to 70%, depending on the season of insemination (Amoah & Gelaye, 1997).In the present study, fertility was significantly influenced by AI season.The probability of pregnancy was reduced when insemination took place during the colder seasons (autumn and winter).In contrast, the fertility of Mediterranean breeds inseminated with cooled semen is reported to be high throughout the year (Roca et al., 1997).
The dairy-goat market demands insemination in spring, when fertility declines due to seasonal reproductive rhythms.For this reason, oestrus synchronisation has become essential to optimise spring pregnancy rates in order to enhance milk yield in autumn.
Hormone treatments based on progestagens, eCG and/or prostaglandins, established over the last forty years, enable the synchronisation of oestrous and ovulation during both the breeding and non-breeding seasons (Fatet et al., 2011).These treatments are essential for any AI program in goats cycling naturally and also in order to induce oestrus during the non-breeding season with fixed-time artificial insemination (FTAI) (Leboeuf et al., 2000).Lower pregnancy rates have been reported when AI is performed in synchronisedoestrus goats than with natural oestrus (Hafez & Hafez, 2000), and also during the non-breeding with respect to the breeding season (Karatzas et al., 1997), although these claims remain controversial (Summermatter, 1987;Roca et al., 1997).Goats synchronised with lower FGA sponges displayed a slightly higher probability of pregnancy than those synchronised with the 45 mg FGA treatment (p > 0.05).These results suggest that FGA doses could be lowered in the treatment of Florida goats, thus reducing hormone residues in milk whilst ensuring good fertility results.
Although a number of authors report no difference in pregnancy rates as a function of cervical or vaginal AI (Roca et al., 1997;Paulenz et al., 2005), others have noted a direct correlation between depth of semen deposition and fertility (Ritar & Salamon, 1983;Salvador et al., 2005;Barbosa et al., 2009), uterine insemination being associated -as observed here-with increased fertility.The present study found that uterine insemination took place on only a few occasions (5.1%); similar findings are reported by Meza & Ross (2000), though higher percentages of 18% and 30% were recorded by Salvador et al. (2005) and Ritar & Salamon (1983), respectively.The results obtained here suggest that uterine or post-cervical deposition is to be recommended whenever it can be done easily and quickly as part of routine AI practice, since it is associated with increased pregnancy rates.Nevertheless, uterine deposition should not become a handicap for the inseminator, since it might cause a delay in insemination and/or cervical injuries that could limit the success of the AI program.For the present program, the interval elapsing between sponge withdrawal and insemination was meticulously recorded.Artificial insemination was considered to be properly performed if completed within 46.0 h ± 0.5 after sponge withdrawal.Only 10.6% of inseminations were performed incorrectly (over 47.0 h after sponge withdrawal), and this was reflected in a drop in the pregnancy rate (31.7%).This confirms the need to avoid any delay in the AI routine in order to improve pregnancy rates.
The transportation of liquid semen over long distances has been reported as a limiting factor in goat AI programs (Paulenz et al., 2005).The influence of the distance between the AI centre and the farm on fertility was considered in the initial statistical model, but no significance was observed, perhaps due to good semen quality and/or the adequate transport system used.Recently, it has been suggested that liquid semen does not have to be cooled to 4°C, nor stored and transported under cooled conditions when used within a day after collection (Peterson et al., 2007).
Inseminating bucks affected fertility rates, a finding also reported in other studies (Paulenz et al., 2005;Salvador et al., 2005).Although seminal parameters were carefully evaluated and only good ejaculates were used, variations were noted in pregnancy rates as a function of buck used.Additional sperm assessment might reveal differences between bucks which could account for this effect.
Few reports have addressed the effects of milk yield on fertility in goats (Leboeuf et al., 1998).However, numerous investigations in dairy cattle have shown that milk yield at AI has negligible effects on conception rates in high-producing dairy cattle (Chebel et al., 2004;Lopez-Gatius et al., 2006;Garcia-Ispierto et al., 2007).In the present study, the initial univariate analysis showed that OR for fertility was higher as milk production increased (Fig. 1).However, after controlling for several factors, results obtained by multivariate logistic regression model indicate that milk yield is not significantly associated with pregnancy rate, and suggest that other factors modify the relation between milk yield and pregnancy, as reported in cows (Chebel et al., Artificial insemination in Florida goats 2004).In addition, in the population evaluated, the probability of pregnancy did not vary significantly as a function of DIM.
As conclusions, numerous factors are involved in the success of a cooled-semen AI program in Florida goats, although here only 4.3% of variability was accounted for by the final model.Production system, AI year, AI season, buck and semen deposition site considerably influenced pregnancy rates, while the other factors analysed had no significant effect.These results show how difficult it is to monitor critical aspects in AI programs.All potentially influential factors need to be controlled, and the AI technician must be aware of any variations that could reduce the probability of pregnancy.The good fertility results obtained during the non-breeding season highlight the value of AI for improving the competitiveness of Florida dairy herds.

Figure 1 .
Figure 1.Relationship between odds ratio for fertility and milk yield.Odds = 0.760 + 0.061 * milk; R 2 = 0.52.Effect of milk yield on pregnancy according to the multivariate logistic regression model: p > 0.05.