Factors associated with the reproductive performance of ewes inseminated laparoscopically in England and Wales: A longitudinal study

Abstract

Background

Laparoscopic intrauterine artificial insemination (Lap‐AI) is used in sheep genetic improvement programmes. However, the lambing rate (LR) (i.e., the proportion of inseminated ewes that lamb) following Lap‐AI is variable. Previously, one small UK study investigated factors associated with LR, but it did not adjust for dependencies or confounding in the data.

Methods

Data were collected from 11,493 Lap‐AIs carried out commercially in Southwest England and Wales between 1995 and 2005. Multilevel multivariable analyses were then used to identify factors associated with LR.

Results

The mean unadjusted average LR was 71%. However, ewes inseminated with fresh semen on days when the maximum ambient temperature (MAT°C) was lower than 25°C had a higher LR (adjusted LR ∼78%; OR 1.51, 95% CI 1.30‐1.77) than ewes inseminated with frozen semen (adjusted LR ~70%). LRs were lower on days when MAT°C was ≥25°C (adjusted LRs: fresh ~64%; frozen ~65%). LR decreased as the time between sponge removal and insemination increased (OR: 0.08, 95% CI: 0.01‒0.87) and varied between months and years.

Limitations

Due to the design of the study, individual ewe‐ and ram‐level factors could not be investigated. Furthermore, some breeds were less well represented, which hindered breed‐level analysis.

Conclusion

Keeping ewes and rams cool on the day of insemination and inseminating ewes earlier after sponge removal may improve LR.

INTRODUCTION

Development of laparoscopic intrauterine artificial insemination (Lap‐AI) in ewes ¹ has facilitated the widespread use of fresh and frozen ram semen in genetic improvement programmes in major sheep producing countries around the world including Australia, ² , ³ , ⁴ New Zealand, ⁵ Spain, ⁶ , ⁷ and the UK. ⁸ , ⁹ Lap‐AI increased rates of change in genetic improvement programmes, such as Sire Reference Schemes, through increased availability of semen from elite sires to more flocks and ewes, which in turn enabled more rigorous and rapid selection of breeding rams. ² , ⁸ , ⁹ Previously, genetic improvement programmes had relied on natural mating, ¹⁰ , ¹¹ requiring movements of rams between flocks, or cervical insemination using fresh (∼30°C) or chilled (4°C or 15°C) semen, which has a short fertile lifespan (<10 hours at 15°C and ≤24 hours at 4°C). ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ Both practices limit the geographical distance over which ram semen can be transported. ¹⁴ , ¹⁵ , ¹⁷ Frozen semen can be stored indefinitely and transported nationally and internationally, enabling movement of high‐quality genetics while minimising disease transmission, because rams are health screened and only screened semen is introduced to ewe flocks. ¹⁶

The reproductive performance of ewes inseminated laparoscopically has been estimated as proportions of ewes inseminated that do not return to oestrus, ⁵ proportions pregnant at mid‐pregnancy ultrasonography ³ , ⁷ and proportions that lamb at full term. ³ , ⁴ , ⁶ Previous studies have indicated that reproductive performance is satisfactory but widely variable between Lap‐AI programmes, rams, farms, seasons and years. ³ , ⁴ , ⁶ , ¹⁸ Concerns around this unexplained variability in success are thought to prevent wider adoption of Lap‐AI by breeders and consequently reduce the rate of genetic progress in the sheep industry. ² , ¹⁹ , ²⁰ , ²¹ , ²²

A small number of large observational studies outside the UK have investigated factors associated with the reproductive performance of ewes inseminated by Lap‐AI in commercial settings. ³ , ⁶ , ⁷ , ¹⁸ , ²⁰ , ²³ , ²⁴ However, the results are variable, and there are few consistent associations between factors and lambing rates (LRs). LRs varied by operator ⁶ ; year ⁶ , ⁷ ; season ⁶ , ⁷ ; farm ³ , ⁶ , ⁷ ; breed ³ , ²⁴ ; semen type (fresh or frozen) ³ ; ram ³ , ⁶ ; in vitro measures of semen quality ¹⁸ , ²⁰ ; frozen semen storage type (pellets or straws) ³ ; ewe age, ⁶ body condition, ²⁴ internal fat ²⁰ and uterine tone ²³ ; method of synchronisation ³ ; and climate, in particular rainfall and lower than median maximum temperatures, which are associated with better LRs in summer in a country with a Mediterranean climate. ⁷ The timing of insemination relative to synchronisation treatment was not associated with LR. ²³

To date, no large observational studies have been carried out in the UK. The one previously published study was small, with 646 ewes from 13 farms in one season (autumn 2006), and only reported univariable statistics not adjusted for dependencies or confounding in the data. ²⁵ In 2011, more UK‐based research was called for ²⁶ , ²⁷ to refine and extend the findings of the previous UK study and the studies highlighted above.

This longitudinal study aimed to investigate the reproductive performance of purebred ewes inseminated laparoscopically in commercial settings in Southwest England and Wales and identify factors associated with variation in performance.

METHODS

Study design and setting

The study analysed data on 11,493 Lap‐AIs carried out commercially in ewes in Southwest England and Wales by one artificial breeding company (South West Sheep Breeding Services [SWSBS]). Data were collected for eleven consecutive years from 1995 to 2005. A total of 97 sheep breeders participated.

At the end of each breeding season, all breeders were sent a self‐administered questionnaire (Appendix A) and accompanying letter (Appendix B). The letter explained that the aims of the study were to use the data to improve the quality of the service provided by SWSBS and improve LRs. A prize (free‐of‐charge Lap‐AI of five ewes, or equipment of equal value) was offered to encourage completion and return of questionnaires within 1 month. Breeders who did not respond within 1 month received a follow‐up telephone call as a reminder to complete and return questionnaires.

Ewes were prepared for insemination by synchronising oestrus and ovulation using intravaginal progestagen (30 mg flugestone acetate [FGA]) sponges (CHRONOGEST; Intervet UK) inserted for 12‒14 days and equine chorionic gonadotrophin (eCG; 240–600 iu) (PMSG [pregnant mare serum gonadotrophin] 5000 IU; Intervet UK) injected intramuscularly into ewes at the time of sponge removal. ²⁸ Some breeders introduced vasectomised rams on the day of sponge removal to identify ewes in oestrus and possibly to improve synchronisation. ²⁹ , ³⁰ , ³¹ Details of synchronisation treatments and artificial insemination (AI) protocols were sent to breeders and prescribing veterinary surgeons (Appendix C).

Ewes were housed 18–24 hours before insemination ³² , ³³ on minimal bedding without feed, and water was removed unless the maximum ambient temperature (MAT°C) was forecasted to be 25°C or above (the upper critical ambient temperature for sheep ³⁴ ), in which case ewes continued to have free access to drinking water. Ewes were inseminated on their home farm or after transport to another farm or breeding centre (SWSBS).

On the day of insemination, ewes were sorted into sire groups (SG, defined as a group of ewes from one flock that were to receive semen from one ram). Ewes were inseminated with fresh or frozen‒thawed semen (details of sources and processing of semen are in Appendix D) by an experienced Lap‐AI operator (M.J.G.) supported by a technician from SWSBS. The Lap‐AI technique used has been described elsewhere. ¹ , ¹⁶ , ³³ For each SG inseminated, the following data were recorded by SWSBS: identity of breeder, date, identity of ram semen, semen type (fresh or frozen), frozen semen storage type (pellets or straws), ewe identity, breed and AI start time (time when the first ewe in each SG was inseminated). Breeders were advised to introduce a fertile ram to the SG 10 days after insemination to mate ewes that returned to oestrus. Breeders were responsible for the care of ewes at all times other than during the insemination procedure itself and were asked to record whether the Lap‐AI was successful based on ewes lambing 140‒155 days after insemination. ⁴ , ⁶

Variables and sources of data

The variables investigated in this study are summarised in Table 1.

TABLE 1.

Explanatory variables investigated and source of data.

Explanatory variable and source of data	Variable type	Description
1. Data sourced from participants and cross‐checked against SWSBS records
#Farm name	Categorical	97 farm categories
Breed  ‘Main’ breed categories	Categorical Categorical	17 breed categories Grouped; six ‘main’ breed categoriesa: CHAR, HAMP, ROUGE, SUFF, TEXEL, ‘OTHER’ (combined)
SG size (n)	Discrete	Range: 1‒97; median: 10; IQR: 4‒16
Month of insemination  Year of insemination  Vasectomised ram used	Categorical Ordinal Binary	11 categories: January to November/December (combined) 11 consecutive years: 1995‒2005 Vasectomised ram introduced or not at sponge removal
Transported for insemination	Binary	Ewes transported or not to other premises
#Ram identity	Categorical	434 ram categories
Semen type used	Binary	Fresh or frozen‒thawed (frozen) semen
2. Data sourced from SWSBS records
eCG dose injected at sponge removal	Ordinal	Grouped; four categories: 240/300 iu (combined); 400/440/460 iu (combined); 500/600 iu (combined); and unknown dose
AI start time	Continuous	Time interval in hours from sponge removal and insemination of the first ewe in each SG; all semen range: 48– 63 hours, fresh semen: 48–61 hours, frozen semen: 48–63 hours
Frozen semen storage type used	Binary	Semen frozen in pellets or straws
No. of AI programmes per day	Discrete	Range: 1–3 AI programmes
No. of AI programmes per day	Binary	Grouped; two categories: 1 or >1 programmes carried out per day
3. Data downloaded from the Met Office Hadley Centre for Climate Science and Services 35
MAT°C	Continuous	MAT°C for the 5 km × 5 km grid that included the location on the day of insemination; range: 5.5°C–32.1°C
MAT°C	Binary	Grouped; two categories: MAT <25°C or ≥25°C. The cut‐point was set at the upper critical temperature (25°C) for sheep 34

Note: SG denotes a group of ewes (n) from one farm inseminated with semen from one ram on 1 day. ^#Farm and ^#Ram were analysed as random effects; other variables were analysed as fixed effects.

Abbreviations: AI, artificial insemination; eCG, equine chorionic gonadotrophin; IQR, interquartile range; MAT°C, maximum ambient (air) temperature; SG, sire group; SWSBS, South West Sheep Breeding Services.

^a‘Main’ breed categories: CHAR = Charollais, HAMP = Hampshire, ROUGE = Rouge de l'Ouest, SUFF = Suffolk, TEXEL, ‘OTHER’ = all 11 ‘minor’ breeds (combined), each representing less than 2% of all ewes inseminated (see Table S2 for further details of minor breeds in the dataset).

The outcome variable, LR, was calculated from the number of ewes that lambed in a SG (r) divided by the number of ewes in the SG (n) and expressed as a percentage (see Equation 1):

$$\mathrm{LR}\left(\%\right)=\frac{r}{n}\times 100$$ (1)

Sample size—estimates of detectable effect sizes

The study analysed existing data, so sample size estimates were not calculated a priori. Estimates of effect sizes likely to be detected were calculated and are shown in Appendix E.

Statistical methods

The data were entered into the database Epi Info 6.04d ³⁶ and exported into Microsoft Excel (version: 14.0.7249.5000) for data cleaning and preliminary analysis. SGs were included in the analyses if farm, date, ram identity and semen type (fresh or frozen) were known and LR could be calculated. The overall mean and median, range and interquartile range of LR by categories of explanatory variables were calculated.

Statistical analysis was conducted in R. ³⁷ Binomial mixed effects models were created using the lme4 package ³⁸ to investigate associations between LRs and explanatory variables. The outcome of the models was the number of ewes lambed as a proportion of the total number of ewes in the SG, and SG size was included as a weight to account for differing SG sizes, as described in ref. 39. Ram and farm were included as cross‐classified random effects and adjusted intraclass correlation coefficients (ICCs) were calculated from null models with only the random effects included using the performance package. ⁴⁰ The fixed‐effect explanatory variables tested in the models are described in Table 1. Continuous variables were rescaled and included as polynomial terms if this resulted in better model fit, as measured by the Akaike information criterion (AIC); coefficients and standard errors were converted to the raw data scale. For the variable month of Lap‐AI, observations from November and December were combined because of low numbers.

Model selection and model averaging were performed using the MuMiN package. ⁴¹ This approach accounts for the fact that there is often no single best model to fit to the data; therefore, a subset of potential models is considered, and the outputs of these models are averaged together to provide more robust estimates. ⁴² The process used was as follows: second‐order AIC (AICc) values of all possible additive models containing any or all of the fixed‐effect variables (n = 2560) were determined. All models within two AICc units of the best fitting model (lowest AICc value) were retained in a ‘best models’ subset, and average coefficients and associated 95% confidence intervals (CIs) across this ‘best models’ subset were calculated for each fixed‐effect variable. Odds ratios (ORs) were calculated as the exponential of the model‐averaged coefficient. Variables for which the 95% CI for the OR did not include 1 were considered to have a significant association with LR.

RESULTS

Descriptive statistics

Response rates

Between 1995 and 2005, 82.9% (97/117) of breeders returned at least one completed questionnaire with usable data from 78.2% (333/426) of AI programmes. The response rate was highest in 2004 and lowest in 2003, when 92.5% (37/40) and 44.8% (13/29) of breeders responded, respectively. Overall, the dataset included data on 80.4% (11,493/14,299) of ewes inseminated in 82.1% (897/1093) of SG with semen from 88.2% (434/492) of rams. Thirty‐eight breeders (39.2%, 38/97) contributed data for 1 year, 32 breeders (33.0%, 32/97) for 2‒4 years, 25 breeders (25.8%, 27/97) for 5‒10 years, and two breeders (2.1%, 2/97) contributed data for 11 consecutive years. Overall, breeders contributed data for a median of 2 years and a mean of 3.3 years.

Maximum ambient temperature

The MAT°C at the location on the day of insemination ranged from 5.5°C to 32.1°C, with a mean of 20.1°C. The annual mean, minimum and maximum MAT°C values on the day of insemination are shown in Figure 1. Overall, 7.4% of all ewes in 8.1% of all SGs were inseminated when local MAT°C was 25°C or above (Table S2).

FIGURE 1.

Annual mean maximum ambient temperature (MAT°C) ± 1 standard deviation (SD) on the day of artificial insemination (green line), annual maximum (red line) and annual minimum (blue line).

Unadjusted LRs

The overall unadjusted mean and median LRs were 71.1% and 72.1%, respectively, for inseminations with ‘all semen’; 77.2% and 78.2%, respectively, for inseminations with fresh semen; and 67.9% and 70.0%, respectively, for inseminations with frozen semen (Table 2). Between farms, the mean LR ranged from 25% to 100% (Figure S1), and between rams and SGs, the mean LR ranged from 0% to 100% (data not shown). Overall, 50% or more of ewes lambed at full term in ∼86% (773/897) of SGs inseminated (data not shown), on ∼93% (90/97) of farms (Figure S1), and following use of semen from ∼86% (373/434) of rams (data not shown).

TABLE 2.

Unadjusted average lambing rates (LRs) by semen type (1995–2005).

Semen type	Farms, N a	Rams, N a	Ewes, n	Sire groups, N	Mean LR (%)	Median LR (%)	IQR LR (%)
All semen	97	434	11,493 (100%)	897 (100%)	71.1	72.1	50.0‒85.0
Fresh semen	39	105	4005 (34.8%)	170 (19.0%)	77.2	78.2	68.4‒85.9
Frozen semen	91	361	7488 (65.2%)	727 (81.0%)	67.9	70.0	50.0‒83.3

Abbreviation: IQR, interquartile range.

^a Columns do not summate.

The annual unadjusted mean LRs for inseminations with ‘all semen’ ranged from 67.5% in 2005 to 77.4% in 1995 (Table S3). Between 1995 and 1998, unadjusted annual mean LRs were similar for inseminations with fresh or frozen semen; however, between 1999 and 2005, they were substantially lower for inseminations with frozen semen than fresh semen (Figure 2).

FIGURE 2.

Unadjusted annual mean lambing rates (LRs), ±1 standard deviation (SD), by semen type.

Unadjusted LR tabulated by semen type and categories of explanatory variables, including for SGs with missing values for some variables, are presented in Table S3.

Logistic regression analysis of factors influencing LRs

Ram and farm identity explained a relatively low proportion of the total variance in LR between SGs (adjusted ICC = 0.12 for the null logistic regression model with no fixed‐effect variables).

At the 95% confidence level, both the semen type used (fresh or frozen) and MAT°C on the day of insemination had significant effects on LR, and there was a significant interaction between semen type and MAT°C (Table 3). Figure 3 illustrates this interaction and highlights that LR was highest in SGs inseminated with fresh semen (adjusted LR ∼78%) on days when MAT°C was lower than 25°C, and lower in ewes inseminated with frozen semen (adjusted LR ∼70%) and in ewes inseminated with either type of semen when MAT°C was 25°C or above (adjusted LRs: fresh, ∼64%; frozen, ∼65%).

TABLE 3.

Multilevel multivariable logistic regression analysis of factors associated with the lambing rate of 11,493 ewes in 897 sire groups inseminated between 1995 and 2005.

Explanatory variable	N (%)	Beta (SE)	OR	LCI	UCI
Semen type: MAT°C
Ref. frozen, <25	671 (74.8)
Fresh, <25	153 (17)	0.41 (0.08)	1.51	1.30	1.77
Frozen, ≥25	56 (6.2)	−0.24 (0.14)	0.79	0.60	1.05
Fresh, ≥25	17 (1.9)	−0.46 (0.22)	0.63	0.41	0.98
AI start time (hours)	897 (100)	−2.55 (1.23)	0.08	0.01	0.87
Year: Ref. 1995	56 (6.2)
1996	81 (9)	−0.06 (0.14)	0.94	0.71	1.25
1997	89 (9.9)	−0.35 (0.14)	0.71	0.53	0.94
1998	63 (7)	−0.35 (0.15)	0.71	0.53	0.94
1999	89 (9.9)	−0.37 (0.15)	0.69	0.52	0.93
2000	94 (10.5)	−0.26 (0.15)	0.77	0.58	1.03
2001	72 (8)	−0.31 (0.16)	0.73	0.54	1.00
2002	92 (10.3)	−0.39 (0.15)	0.68	0.50	0.92
2003	44 (4.9)	−0.32 (0.18)	0.73	0.51	1.02
2004	111 (12.4)	−0.58 (0.15)	0.56	0.42	0.75
2005	106 (11.8)	−0.38 (0.16)	0.68	0.50	0.93
Month: Ref. August	510 (56.9)
April	3 (0.3)	0.95 (0.64)	2.59	0.73	9.16
July	128 (14.3)	0.42 (0.15)	1.52	1.12	2.04
September	182 (20.3)	−0.24 (0.12)	0.79	0.62	0.99
October	62 (6.9)	−0.19 (0.14)	0.83	0.62	1.10
November/December	12 (1.3)	−0.57 (0.27)	0.57	0.34	0.95
Programmes per day: Ref. 1	669 (74.6)
>1	228 (25.4)	−0.01 (0.04)	0.99	0.93	1.07
eCG dose: Ref. 500/600 iu	494 (55.1)
240/300 iu	77 (8.6)	−0.06 (0.15)	0.94	0.70	1.28
400/440/460 iu	299 (33.3)	−0.05 (0.10)	0.95	0.78	1.16
Unknown	27 (3)	0.02 (0.10)	1.02	0.84	1.24

Note: OR > 1 indicates a positive association, and OR < 1 indicates a negative association. The bold text indicates that the association with the explanatory variable was significant at the 95% confidence level.

Abbreviations: AI, artificial insemination; Beta, regression coefficient; LCI, lower 95% confidence interval; MAT°C, maximum ambient temperature; N, number of sire groups; OR, odds ratio; SE, standard error of coefficient; UCI, upper 95% confidence interval.

FIGURE 3.

Predicted lambing rates (LRs) by semen type and maximum ambient temperature on the day of insemination. LRs are predicted from the model averaged coefficients presented in Table 3 across the range of all fixed‐effect variables retained in the model.

There was also a slight linear decrease in LR with increasing time between sponge removal and insemination of the first ewe in an SG (OR: 0.08, 95% CI: 0.01‒0.87), but there was no significant interaction at the 95% confidence level between AI start time and semen type. There were significant variations in LR between years and months compared to the reference year of 1995 and the reference month of August (Table 3).

There were no significant associations at the 95% confidence level between LR and the size of SGs inseminated, between LR and the number of AI programmes carried out per day, or between LR and the dose of eCG injected at the time of sponge removal (Table 3; SG size was not retained in any of the models in the ‘best models’ subset, so it is not shown in the table).

DISCUSSION

This is the first UK observational study to report a negative association between LR and the exposure of breeding sheep to an ambient temperature of 25°C or above on the day of insemination. A MAT°C of 25°C is above the upper critical temperature for sheep (24°C–26°C), ³⁴ and since the association was significant at the 95% confidence level for inseminations with fresh but not frozen semen, it suggests that there was a greater effect on ram reproductive function than on ewe reproductive function. Two Spanish field studies reported negative associations between LRs and a MAT°C of more than 30°C in the 2 days before cervical insemination with fresh semen ⁴³ and on the day of Lap‐AI in summer with frozen semen. ⁷ In contrast to our findings, those of Palacios and Abecia ⁷ involved inseminations with frozen semen, suggesting that ewe reproductive function was significantly affected; however, the MATs in that study were higher (>30°C) than the cut‐off point of 25°C used in our study, which might explain the difference in results.

Heat stress in sheep arises when the ambient temperature exceeds 24°C–26°C. ³⁴ Most evidence of negative impacts of heat stress on reproductive performance comes from experimental studies in which sheep were exposed to whole‐body heating at a MAT°C above 32°C and in rams exposed to local heating of the testes and epididymides by scrotal insulation. These studies have shown that the negative effects of heat stress on ewe reproduction are greatest after exposure from 5 days before to 5 days after oestrus and that on ram reproduction after exposure to high ambient temperature 14–50 days before semen is collected. ⁴⁴ , ⁴⁵ However, embryo mortality is higher in embryos sired by rams exposed to mild scrotal heat stress in the previous 4 days than in those sired by unexposed control rams. ⁴⁶ Overall, the experimental and observational findings suggest that heat stress can occur in rams in England and Wales when ambient temperatures are above 25°C on the day of breeding and this impairs sperm quality. Should temperatures be above 30°C, especially for longer periods around insemination days, there might also be reduced reproductive performance in ewes, ⁷ , ⁴⁴ , ⁴⁵ as has been well documented in heat‐stressed lactating dairy cows inseminated with frozen semen. ⁴⁷ , ⁴⁸ , ⁴⁹ Given the consistency of our results with those of studies reported previously, it is likely that the reduction in LRs with high ambient temperatures is generalisable to the wider sheep industry.

As the frequencies of hot summers and heatwaves are increasing, ⁵⁰ , ⁵¹ it would be sensible for stakeholders to consider practical short‐term measures to mitigate the negative impacts of high ambient temperature when hot weather is forecast ³⁴ , ⁵² , ⁵³ : shearing animals well before insemination; providing animals with shade or housing with electric fan ventilation and/or wetting the skin of sheep with water sprays/misters, and ensuring a continuous supply of fresh drinking water; and if transporting sheep, using well‐ventilated trailers. Longer term measures will be needed as high ambient temperatures become more frequent: breeding sheep able to maintain good reproductive performance despite exposure to high ambient temperatures ⁴⁹ , ⁵² , ⁵⁴ , ⁵⁵ , ⁵⁶ ; and possibly changing the seasonal timing of insemination to avoid breeding in summer when heat stress is likely. ⁵⁷

In our study, on days when MAT°C was below 25°C, LR was higher in ewes inseminated with fresh than frozen semen. This is as also reported by others. ³ , ⁵⁸ It is widely accepted that frozen ram semen is less fertile than fresh ⁵⁹ , ⁶⁰ due to reduced viability of sperm and adverse effects of freeze‒thawing on the structure and function of a proportion of sperm surviving these processes. ⁶¹ , ⁶² , ⁶³ In our study, there was no association between LRs and whether frozen semen was stored as pellets or in straws. Others have investigated this, ³ , ⁶⁴ , ⁶⁵ and only one small study ⁶⁵ reported a difference in pregnancy rates by frozen semen storage type.

To our knowledge, the current observational study is the first in the UK, and the second worldwide, ²³ to investigate associations between LR and time of Lap‐AI in ewes in commercial settings, where timing is typically less well controlled than in experimental studies. We found a slight linear decrease in LR with increasing time between sponge removal and insemination of the first ewe in an SG (Table 3), irrespective of the semen type (fresh or frozen) used. Model predictions suggest that starting insemination of an SG ∼48 hours after sponge removal is likely to lead to an increase in LR of ∼9% overall when compared with starting insemination ∼63 hours after sponge removal, which is the current industry practice. The time of insemination for the last ewes in an SG would be longer than for those in smaller SGs and would have reduced the precision and increased the variability of the effect we detected.

A previous large observational study found no association between LR and time of insemination for individual ewes between 43 and 60 hours after the removal of controlled intravaginal drug release devices (CIDRs) and the injection of eCG in Australian Merino ewes. ²³ The optimal timing of Lap‐AI in synchronised ewes has also been investigated experimentally in Australia ⁶⁶ , ⁶⁷ , ⁶⁸ , ⁶⁹ and the UK ⁵⁸ , ⁷⁰ with equivocal results: higher LRs were reported in some studies when ewes were inseminated with frozen semen ∼60–72 hours after sponge (FGA) removal, that is, after the expected time of ovulation, ⁶⁶ , ⁶⁷ whereas in others, earlier insemination with frozen semen ∼48–55 hours after sponge removal, that is, before the expected time of ovulation, was equally successful. ⁵⁸ , ⁶⁸ , ⁶⁹ , ⁷⁰

The wide variations in LRs between years, farms, rams and SGs of ewes in our study were comparable to those reported previously in the UK ²⁵ , ⁵⁸ and in large observational studies elsewhere. ³ , ⁶ , ¹⁸ Given the study design, we were unable to investigate associations between LRs and individual ram‐ or ewe‐level factors, including the timing of insemination in individual ewes.

We found no association between LR and breed after adjusting for confounding and dependencies in our data. Overall, data on 17 breeds were analysed: five ‘main’ breeds (Charollais, Hampshire, Rouge, Suffolk and Texel), largely specialised meat (or terminal sire) breeds breeding rams commonly used commercially to produce lambs for meat ⁸ , ⁹ , ⁷¹ ; and 12 ‘minor’ less‐well represented breeds, a mix of meat and maternal breeds each contributing less than 2% of all inseminations in the dataset (Table S2), which were analysed as a combined ‘OTHER’ ‘main’ breed category. Two previous observational studies have reported significant breed effects on LR following Lap‐AI: Hill et al. ³ found higher LR in fine‐wool Merino strains than in strong‐wool strains in Australia, and Aké‐Villanueva et al. ²⁴ found higher LR in Blackbelly and Pelibuey breeds than in the Dorper breed in Mexico. Differences between study findings may reflect true variability in the relative fertility of breeds investigated in the UK and other studies and/or relative differences in their management, exposures and reproductive responses to risk factors such as heat stress.

The optimal timing of Lap‐AI is likely to vary with the type of intravaginal device and progestagen used to synchronise oestrus. ³ , ⁷² All ewes in our study were synchronised with intravaginal sponges impregnated with 30 mg FGA; however, intravaginal CIDRs—silicone elastomers containing natural progesterone (0.35 g P4)—are now also available in the UK and used by some breeders as an alternative to the currently available 20 mg FGA sponges. The optimal timing of intrauterine insemination in ewes synchronised with CIDRs has not been investigated as extensively as with FGA sponges ¹⁹ ; but what is known is ewes synchronised with CIDRs typically ovulate earlier after device removal on average than with FGA sponges ⁷³ and so may need to be inseminated earlier.

LR was not significantly associated with the eCG dose injected at sponge removal at the 95% confidence level. This broadly agrees with Hill et al., ³ who only found a reduction in the LR when eCG doses of less than 250 iu were used. In our study, only two SGs (0.2%; data not shown) were injected with less than 250 iu, and 77 SGs were injected with 300 iu or less (8.6%; Table S2); thus, it is unlikely that we would have detected an association between LR and eCG dose.

We found no associations between LR and SG size, number of AI programmes carried out per day, or introduction of a vasectomised ram at sponge removal. Lucidi et al. ³⁰ experimentally investigated the effect of introducing a vasectomised ram 24 hours after sponge removal to synchronised ewes in pro‐oestrus/oestrus and reported a significant positive association with LR following insemination with frozen semen. There remains insufficient clarity on the optimal timing of introduction of vasectomised rams to synchronised ewes, at sponge removal or later, and the effectiveness of their introduction on the synchronisation of ovulation and LR following insemination. ¹⁹ , ⁷⁴

In our study, the transport of ewes for insemination was not significantly associated with LR at the 95% confidence level, although Dobson et al. ⁷⁵ reported that experimentally transporting ewes can reduce fertility because of effects on follicular function and oocyte viability. It is possible that ewes transported in our study were from small pedigree flocks, used to being transported and less stressed than those in the experimental study; thus, their fertility was not negatively affected (unpublished observation cited by Dobson and Smith ⁷⁶ ). It is also possible that eCG treatment at sponge removal overrode negative effects of transport stress that reduces the pre‐ovulatory LH surge in ewes because it is an external source of LH.

CONCLUSIONS

This is the first large‐scale longitudinal study to describe LRs in ewes inseminated laparoscopically in commercial settings in the UK. We report, for the first time in the UK, that MAT of 25°C or more on the day of insemination reduce LRs in ewes inseminated with fresh semen compared to lower temperatures and a time interval of 48 hours between FGA sponge removal and insemination of the first ewes in SGs increased LR when compared with an interval of 63 hours. These findings can be used to help improve the success rate of commercial Lap‐AI programmes in the UK.

AUTHOR CONTRIBUTIONS

Mike J. Glover was responsible for collecting and collating the data and for preliminary analysis. Laura E. Green and Rachel Clifton carried out further analysis of the data, including multivariable multilevel modelling. All the authors contributed to the manuscript and approved the final version submitted.

CONFLICT OF INTEREST STATEMENT

The authors declare they have no conflicts of interest.

ETHICS STATEMENT

Breeders gave informed consent for their data to be analysed for the purposes of improving the quality of Lap‐AI services delivered by SWSBS and informing efforts to improve success rates. An anonymised dataset was shared with co‐authors for further analysis, and anonymised results were shared informally with participating breeders and stakeholders. Ethical approval was provided retrospectively by the University of Birmingham ethics committee on 19 September 2025.

Supporting information

Supporting information

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available upon reasonable request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.