Accurate and practical evaluation of the fertilizing capacity of frozen-thawed bull semen used for artificial insemination (AI) is of considerable importance to the cattle industry. Currently, the most accurate method of estimating the fertility of a bull and its processed semen is the retrospective analysis of the pregnancy rates achieved after AI. This is an expensive and time-consuming approach. Many semen tests and test combinations have been trialed with varying correlation with field fertility. None has proven to be consistently and accurately predictive. This thesis assesses the frozen-thawed semen of Australian beef and dairy bulls, using conventional and more novel semen tests, and compares laboratory semen observations with field fertility estimates following AI. The approach described in this thesis has not previously been conducted in Australia using Australian bulls under Australian AI conditions.
The first study described in the thesis is a postal survey of the major livestock semen processing (LSP) centres in Australia. It sought to determine which semen parameters are currently being examined and how, what threshold values are considered acceptable and which tests are considered to be predictive of fertility. The results suggested that the selection of tests, the analytical procedures and the threshold limits used were inconsistent in the Australian semen processing industry. It appears that a different semen assessment protocol is used within each individual laboratory. The percentage of progressively motile (%PM) spermatozoa and the percentage of morphologically normal (%norm) spermatozoa were the most commonly used assessments (100% and 94% of respondents, respectively) but quite a range in threshold levels existed between respondents for these tests. There is a clear need for standardization of semen assessment methods in the Australian AB industry and for training of those involved in assessing semen.
The second study was conducted to identify the degree of variation among semen batches and among same-batch semen straws to validate comparisons between bulls and straw replicates in subsequent experiments. Three to eight straws were assessed from each batch of semen from six different bulls. No significant variation was evident among straws within-batch for ten of the fifteen spermatozoan parameters (66%) examined. However, spermatozoan concentration, progressive motility, rate of motility and membrane integrity at thaw and rate of motility after two hours incubation did differ significantly. Five to seven consecutive semen batches from each of four different bulls were tested and no significant difference was apparent for any spermatozoan parameter among batches, except for membrane integrity at thaw. Those parameters later found to be predictive of field fertility (normal spermatozoa at thaw, number and percentage of normal spermatozoa after the swim-up spermatozoa filtration procedure, and the rate of in vitro zygote cleavage) were all found to be consistent among straws and batches and all differed significantly among bulls, except for the number of morphologically normal spermatozoa after swim-up (nSuNorm).
The third study was conducted as a pilot project to provide a template for subsequent larger studies with regard to technique, equipment and the appropriateness of proposed tests, analysis and statistical modeling. This study also identified potential predictive parameters and models. Frozen-thawed semen from eight different bulls was examined using thirty different semen parameters. The degree of correlation between each of these parameters and with each bull's corrected conception rate (CCR) was determined. The percentage of morphologically normal spermatozoa after swim-up (SuNorm) and the rate of spermatozoal movement at thaw (zeroRate) were the only significant correlants with CCR. Finally, a step-forward multiple regression analysis developed a predictive model that included the percentage of morphologically normal spermatozoa at thaw (zeroNorm) and SuNorm and that predicted respective CCR values quite accurately:
CCR = 78.2 - 0.3 SuNorm + 0.04zeroNorm (adjusted R2=086; P<0.01)
Limitations identified in this study included the need to strengthen the dependent variable with higher bull numbers, more inseminations per bull (minimum of 100) and for those bulls to cover a wider CCR range.
Study 4 was based on the analysis of semen from eleven high-use Australian AI dairy bulls for 29 different spermatozoal parameters, and on the provision of corrected non-return rates (cNRR) and CCRs from those same bulls calculated from inseminations using semen collected during the same period. ZeroNorm was the only parameter significantly correlated with both CCR and cNRR. No other significant relationships were established with CRR, but cNRR also correlated strongly with six other spermatozoal parameters. They were all related to the spermatozoan population after swim-up and included nSuNorm, spermatozoa concentration at swim-up (SuConc), percentage of normal motile spermatozoa at swim-up (SuNM), the number of progressively motile spermatozoa after swim-up (nSuPM), the yield of motile spermatozoa after swim-up (motSrvSu) and the percentage of morphologically normal spermatozoa after swim-up (SuNorm). The step-forward multiple regression analysis produced two significant predictive equations that were each able to accurately estimate the actual CCR and the cNRR of that semen:
cNRR = 45 + 0.2 zeroNorm - 0.11 twoMem (adjusted R2=0.78; P<0.01)
CCR - 4.24 + 0.5zeroNorm - 2.9 nSuNorm + 0.24 Clv (adjusted R2=0.70; P<0.05)
The proportion of membrane intact spermatozoa after two hours of incubation (twoMem) and the rate of in vitro zygote cleavage (Clv) were the two equation parameters not previously mentioned. Both predictive equations estimated the bull fertility figures and rankings closely although most bulls lay within a range too narrow to confidently differentiate within. Importantly however, the outliers were readily identified. It is clearly only those very good bulls and more importantly, those very poor bulls that will be confidently differentiated.
The final study was identical to experiment four, but examined the frozen-thawed semen from Australian beef bulls and corrected calving rates (CalvRate) derived from three successive AI breeding programmes in central Queensland, Australia. This study used a much wider range in field fertility (18%-60.7%), more bulls, far fewer (40.2) inseminations per bull, and calving rates. Similarly to the dairy experiments, the proportion of morphologically normal spermatozoa (in this case SuNorm; P<0.1) after swim-up correlated with CalvRate and was also included in the predictive equation along with rate of spermatozoan motility at thaw.
CalvRate =134-0.56SuNorm -14.9zeroRate (adjustedR2=0.30; P<0.1)
However, this predictive equation only accounted for 30% of variance, and this was reflected in the relatively poor prediction of calving rate for individual bulls and of ranking of the bulls overall. Relative to the dairy studies, the greater number of bulls assessed and the wider field fertility range should have strengthened the predictive accuracy of the regression; the very low insemination number per bull may have weakened this capacity.
In summary, this thesis was successful in identifying bull spermatozoal parameters that strongly correlated with both CCR and cNRR, and in developing predictive equations for field fertility as measured by CCR and cNRR. A significant, but much weaker, predictive model was produced for CalvRate. It is notable that all the measures of field fertility tested for both beef and dairy cattle held significant correlative relationships with normal spermatozoan morphology. Similarly, all regression analyses had enlisted normal spermatozoal morphology (at thaw or after swim-up) as leading components of the respective equations. The results encourage a final prospective study to test the predictive capacity of these models. This would entail analysis of semen currently in use, use of these equations to calculate CCR, cNRR or CalvRate and comparison of the results using actual CCRs and NRRs as the insemination results are analysed and become available. If the predictive models prove to be accurate, they could find use in situations such as progeny testing, presale and preseason testing and in fertility assessments post-injury or post-illness.