Skip to main content
European Commission logo print header

Pluridisciplinary study for a RObust and sustainabLe Improvement of Fertility In Cows

Final Report Summary - PROLIFIC (Pluridisciplinary study for a RObust and sustainabLe Improvement of Fertility In Cows)

Executive Summary:
In dairy cattle, intensive genetic selection has resulted in modern cows with very high milk yield but reduced fertility and poor calving performance. PROLIFIC is a large project involving seven research organizations (INRA, AU, TEAGASC, SERIDA, DLO, SLU and SRUC), one industry (LATTEC) and four SMEs (CONAFE, TS, SM and Allice). Its main strategy is to use pluridisciplinary approaches to eliminate the key scientific/methodological blockages and develop innovative solutions for a robust and sustainable improvement of fertility in cows. It is structured in four R&D workpackages (WP1 to WP4), one demonstration (WP5), one outreach (WP6) and one management (WP7). In WP1, different models at the animal and herd levels have been designed to incorporate genetic, endocrine, metabolic and stochastic regulating components of fertility and to perform high speed simulations. An equation that predicts the probability of insemination success based on progesterone (P4) profiles and cow traits was built and tested. Models at the herd level have been also developed to estimate economic results, labor time and environment impacts including the genetic aspects of reproduction management strategies. WP2 identified from in vivo and in vitro models by using transcriptomic, proteomic, metabolomic and epigenetic approaches, molecular markers differentially expressed in relation with diet and/or negative energy balance in fat and reproductive tissues. It also investigated changes induced by well-known pathogens and metabolic disorders in uterine tissue from in vitro models. At the bioinformatics level, a pipeline has been developed to analyze epigenetic data and the platform to gather large amount of data generated by the molecular analyzes from different partners reinforced. WP3 aimed to identify the functional quantitative trait nucleotide for fertility based on progesterone measures, and to estimate genomic breeding values using whole sequence information on individuals, in order to develop optimal breeding strategies. For this, Milk P4 records have been collected and data have been translated into endocrine fertility traits like commencement of luteal activity that is a good trait to select for improved fertility whilst improving milk yield simultaneously. Through genome-wide association studies genetic variants associated with endocrine fertility traits have been searched. Prediction accuracy for fertility can be increased by using bi-variate models for training that includes both endocrine and classical fertility traits. WP4 provided a large amount of phenotypic data related to fertility variables from cows with different genetic backgrounds on different feeding strategies and determined relationships between fertility parameters and milk yield, body score condition profiles, endometrial cytology, and general health status. The effect of pasture availability on metabolic status, body-condition score and postpartum oestrous cycles has also been studied. WP5 developed tools that integrate knowledge created especially in WP1, and that can be used by the farmer for decision support regarding fertility management. The tools built in this WP relied both on real-time on-farm progesterone measurements, and on predictive models. They will help the farmer to determine whether or not an oestrus is worth inseminating and they will use simulations on cow and herd level to predict the effect of a change in fertility management on a given farm. WP6 disseminated the knowledge produced in the project to the relevant stakeholders. A joint Final International Conference with FP7 FECUND has been hold as a satellite workshop of ESDAR 2016.
Project Context and Objectives:
The undesirable environmental and welfare impacts of animal farming systems indicate a need to increase the sustainability of these systems. The sustainability of dairy cattle farming systems relies in large part on the ability of cows to maintain reproductive performance as they cope with the constraints imposed by environmental conditions and livestock practices. Intensive genetic selection has resulted in a modern dairy cow with very high milk yield but reduced fertility as well as poor calving performance. Fertility problems include an increase in postpartum failure to return to oestrus, poor expression of oestrus, defective oogenesis, embryo survival and susceptibility to uterine infections. These problems created a situation where, through decreased reproductive performance, farmers are losing control of a key component of cattle system sustainability. The deterioration of fertility is most pronounced in the Holstein breed, even after reproductive performance was included in both the breeding goals and selection criteria for the breed. Many countries have implemented genetic evaluation of fertility traits in recent years and new traits are gradually being measured and more increasingly sophisticated evaluation methods are being applied. However, the low level of reproductive performance is a real problem for farmers. There is considerable scientific evidence to support the view that infertility is a multi-factorial problem influenced by animal (genotypes, parity, milk production), environmental (nutritional status) and management factors.

Genetic changes can be envisaged that would reduce the risk of reproductive failure at different periods throughout the life of the cow. A part of the solution will come via the implementation of improved genetic selection programmes but this requires new information on component traits of fertility. Reproduction, especially female fertility, is a complex trait influenced by many physiological and disease-related variables. Thus, it is important to better define the important components of fertility and apply these definitions in breeding programmes to achieve a good and targeted selection response.

The rapid development of OMICs-technologies facilitates systematic studies of molecular profiles in reproductive tissues in different conditions of reproductive physiology and fertility problems. The most important point is gaining a deeper understanding of the different steps in reproductive physiology. PROLIFIC will combine the use of models with state-of-the-art OMICs technologies to address problems of oocyte maturation, early embryo development, and implantation that are early steps shown to be the major source of pregnancy loss in cattle. Comprehensive molecular profiles may be used for biomarker discovery and could also be the basis to identify new candidate genes potentially involved in reproductive success. Variation in candidate genes which were identified by expression profiling or reproductive tissues in critical periods of reproductive physiology can be used to identify genetic markers which are positively associated with production and fertility traits.

Nutritional status has been recognized as a modulator of reproductive performance since ancient times. High producing dairy cows meet frequently negative energy balance during the post-partum period and metabolic stressors can negatively impact many stages of reproductive life. Moreover, nutrition is an important environmental factor that regulates fertility performance, primarily through effects on energy balance and body energy reserves. The observations that prolonged periods of anestrous during the postpartum period result in poorer reproductive performance during subsequent breeding emphasizes the importance of exploring the relationships and interactions that exist between nutritional status and resumption of ovulatory ovarian activity. To date, the precise mechanisms by which nutritional status interacts with the reproductive axis have not been delineated.

The strategic aim of PROLIFIC is to unlock the potential for proactive herd management by providing the farmer with improved tools for on-farm reproductive monitoring and management. This can be achieved by a pluridisciplinary approach to eliminate the key scientific/methodological blockages and develop innovative solutions for a robust and sustainable improvement of fertility in cows.

One of the main scientific objectives of PROLIFIC is to provide integrated, quantitative, models to describe interactions between genotype, nutrition and reproductive physiology. More precisely it consists to: (i) develop a generic model of reproductive physiology capable of simulating the dynamic of reproductive performance across a wide range of genotypes and management environments. This objective concerns modelling of the reproductive complex for understanding basic mechanisms of sub fertility. (ii) Develop a model to support on-farm decisions for optimizing insemination timing with respect to the desired calving interval and the individual cow’s physiological state, and thereby increase conception rates. (iii) Develop a herd level diagnostic model that can be used on-farm to test how changing reproductive management (insemination rules, oestrus detection protocols, etc.) can alter reproductive performance at the herd level; and to provide a benchmarking tool using on-farm progesterone monitoring systems to permit diagnosis of which aspects of reproductive and/or nutritional management need attention. PROLIFIC also developed simulation tools that will allow evaluation of the economic and environmental impacts of changes in breeding management at the farm level. More precisely, it developed predictive model of socio-economic consequences of improved reproductive management in different livestock systems (from extensive to intensive systems) and use these to identify optimal cattle reproduction systems according to local system constraints (e.g. breeds, forage availability) and innovative reproductive management approaches.

Another objective of PROLIFIC is to find new phenotypic and genetic markers for fertility as tools to be included in selection schemes. This is very important particularly in the new context of genomic selection to have a more precise phenotyping to improve the quality of the selection basis. A better knowledge of networks of genes involved in fertility will allow to improve the number and quality of the genetic markers used. Identifying groups of genes and pathways involved in the adaptation of the reproductive function to different environmental conditions (especially low input feeding systems) and to specific pathogens affecting the reproductive tract becomes also critical when considering climatic changes and the need to increase lifetime productive efficiency and system sustainability.

At the genetic and genomic levels, PROLIFIC aimed to identify the functional mutations associated with fertility (based on progesterone measurements) and estimate genomic breeding values using whole sequence information on individuals. For this, 8000 lactation records from animals with progesterone phenotypes have been collected and most informative bulls based on the pedigree have been sequenced. Genome-wide association studies (GWAS) has been performed on 85K SNP to identify regions associated with endocrine fertility traits. Sustainable systems of milk production that are profitable for the farmer and that meet industry product quality criteria, maintain or improve animal productivity, health and welfare, and that are compliant with environmental regulations represent the future of the EU dairy industry.

As indicated in the introduction it is well established that nutrition is an important environmental factor that regulates fertility performance, primarily through effects on energy balance and body energy reserves. PROLIFIC generated a large amount of phenotypic data related to fertility variables from cows with different genetic backgrounds on different feeding strategies to determine relationships between fertility parameters and milk yield, body score conditions (BCS) profiles, endometrial cytology, and general health status. PROLIFIC has also identified the effects of different BCS at calving on subsequent BCS dynamics, metabolic status and indicators of reproductive performance and explored nutritional and management strategies to improve BCS in advance of the breeding period in cows with suboptimal postpartum BCS.

The control of reproductive management is a pre-requisite for crafting a cattle production system that maximizes lifetime productive efficiency in a balanced and sustainable manner. Achieving this has not been easy given the recent decline in fertility and reduced expression of estrous behaviour in modern cattle. Fertility tools are available, but they were based mainly on behavioral oestrus, which has been proven to decline in case of pure oestrus behavior. PROLIFIC through a winning combination of disciplines created improved decision support tools for innovative reproductive management strategies. More precisely, PROLIFIC developed, deployed and evaluated on commercial farms decision support tools to optimise the timing of reproductive management decisions, to improve the rate of successful inseminations, and to provide reproductive performance benchmarking. These tools will allow to the farmer to better target reproductive actions at individual cow and herd level, according to the livestock system he/she finds himself in. They have been built by using real-time on-farm progesterone measurements, and on predictive models developed in PROLIFIC.

Finally, the objectives of PROLIFIC have been to make the best possible use of the project results by farmers, to ensure fruitful exchanges with the scientific community, including individual scientist and initiates, to inform to general public about the results and innovations brought by the project and to trigger technology transfer of project results to industrial applications. PROLIFIC combined events, publications, and internet tools to maximise the dissemination of the project results.
Project Results:
WP1 Multilevel integration and modelling of reproductive performances at different scales
One of the main objectives of Work package 1 was to provide farmers with improved tools for on-farm reproductive monitoring and management. This has been addressed by developing models to support on farm decisions at the animal level (D1.2) and at the herd management level (D1.3). A list of scenarios considered to be the most relevant to be studied with models has been proposed by an expert focus group (D1.1). Some of these scenarios were prioritized and tested to quantify socio-economic impacts of changes in reproductive management (D1.4).
The following section reports the main scientific results and implications for WP1. However some of WP1 impacts will also be described in the WP5 part as models developed in WP1 were used to develop decision-support tools in WP5 (IWP and RMTO).

1/ Identification of genes involved in reproductive hormones patterns
A mathematical model has been developed previously consisting of 15 linked differential equations and 60 parameters modelling the processes of cattle reproduction. The output of the model shows the hormonal patterns during the oestrous cycle of the non-pregnant cow correctly with pre-determined values for the parameters. Different values for the parameters indicate different cycling output, possibly related to non-optimal reproduction. However, the biological processes underlying the parameters are poorly known. Knowledge of these biological parameters will enable to measure (i.e. quantify) the biological values of individual cows. This will enable to use the model to evaluate the quantitative effects of the genes / biological mechanisms on the oestrous cycling conditions of individual cows. Therefore, the objective of this research was to determine the potential biological background underlying the parameters in the model. We used two datasets: A QTL dataset with reproduction-related traits measured on a commercial cattle population (WP3), and a dataset consisting of QTL data and transcriptome (endometrium and corpus luteum) data measured in a high / low fertility high productive cattle model (WP1). We showed for each trait, as defined in the WP3 dataset, genes, pathways, networks, and (general) biological mechanisms underlying the different parameters in the mathematical model. Now we can insert these genes, networks, pathways, and biological models in the parameters of the model. These data can be measured in individual cows. This will change the (values calculated by the) mathematical equations, making the results describing the situation in individual cows. From these results we can learn what the biology underlying reproduction problems in cows is, and it will hopefully lead to solutions.

2/ Knowledge integration and systemic representation of reproduction
One of the main objectives of WP1 was to develop a Reproductive Physiology Model (RPM), together with physiologists, researchers in animal sciences and modellers, which could embody the multilevel integration we wanted to achieve in this project. Such an integration has relied on two main pieces: inputs of knowledge from experts and literature and the construction of a theoretical framework able to represent the main biological mechanisms involved in cow fertility. The RPM model is the centerpiece of the multilevel integration achieved in WP1. It provides the most complete representation of reproductive physiology in a quantitative framework to date. It represents the reproductive axis machinery (biological units, hormones dynamics and effects) and simulates the normal functioning (hormones signaling, reproductive status) of a cow over her lifetime. By way of modulations of its parameters, the model can generate deviations around the normal behaviour and simulate phenomenas such as irregular cyclicity, short or prolonged luteal phases, fertilization failure and embryonic mortality. RPM is thus a research tool now available to help us analysing the sensitivity of the reproductive function to combined animal factors (nutritional status, milk yield, and breed). The documentation of this RPM model via a scientific publication is in its final stages (submission of the manuscript is planned for the end of February).

3/ An equation to predict the probability of insemination success at a given estrous (D1.2)
In dairy farms equipped with the Herd Navigator© system, oestrus detection is based on milk progesterone (P4) measurements. Until now, oestrus alerts were given to the farmer according to changes in P4 levels. In the context of progesterone-based oestrus detection with a specificity of almost 100% the option of waiting until the next (better) oestrus to inseminate becomes a viable strategy. This part of the project aimed to provide an equation predicting the probability of insemination success at each oestrus based on progesterone profiles and cow traits. First, a cluster analysis has been performed on 1060 smoothed progesterone profiles coming from the research herd at the Danish Cattle Research Center in Foulum, Denmark, to characterize the variability of progesterone profiles. This analysis showed that progesterone profiles can be classified according to their shapes and that cycles mainly differ on their length and P4 level. A publication was written on this part (Blavy et al, 2016 in Theriogenology, 86, 1061-1071).
This work has then been completed with statistical modelling to provide an equation predicting the success of insemination for a given oestrus based on the features of the previous progesterone profile cycle (cycle length, cycle height) and cow traits (peak milk yield, parity). These results have been reported in a scientific paper, which was submitted to Journal of Dairy Science in December 2016 (Blavy et al., Estimating probability of insemination success using milk progesterone measurements).

4/ A list of reproductive management strategies relevant to go towards more sustainable dairy systems
Before starting the modelling work, a workshop was organized with the aim to ask experts to identify reproductive management strategies they considered relevant and robust to face increasing environmental variability and meet demands for a more sustainable dairy production. The target audiences were organizations that were partners in PROLIFIC or members of the Stakeholder Advisor Board (SAB), but the workshop was also open to other experts on invitation by partners or SAB-members. A detailed description of the method and of the achieved results is available in deliverable 1.1. The outcome of the workshop was two separate lists applicable to future reproduction scenarios in Europe and beyond. One list consists of the European dairy production systems the experts have considered to better meet project demands for a more sustainable production. The second list concerns reproductive management strategies and the factors that appeared to be useful to customize each strategy. In addition, a cross-reference table was created that identifies the most relevant combinations of strategies (i.e. genetically driven mating x sexed semen, improvement methods of heat detection x seasonal calving...). A total of seven dairy production systems and 10 reproductive management strategies were pinpointed and ranked by experts according to their importance. These lists have been considered as a reliable basis to define the scenarios to be tested with herd models.

5/ Simulation tools to test scenarios
WP1 has provided two herd level models to test scenarios.
The first one (RPM_herd) is derived from a simplified version of the cow RPM model, called RPMlite, which has been connected to a Reproductive management model to allow simulation of reproductive performance of a large cohort of individual cows, i.e. a herd of animals under the same reproductive management. The reproductive management model allows different oestrus detection and insemination strategies to be simulated, with the capability to simulate a very wide range of reproductive (e.g. strict seasonal calving through to year-round calving with automated oestrus detection) and feeding (e.g. high, medium or low energy density) management. RPM_herd provides a biologically consistent tracking of reproductive status of the individuals in the herd, including return to cyclicity and conception rates as affected by the animal’s nutritional status and performance potential. RPM herd is thus able to adapt to a great diversity of farming systems (e.g. high and low input) and reproductive strategies. It is now ready for testing scenarios and providing references about the expected deviations in fertility and demographic indicators’ distributions when something is changed in the reproductive strategy (intervals between calving to 1st cycle, between calving to 1st AI, between calving to conception, number of non-detected oestruses, number of 1st lactation cows...). A description of the RPMlite model has been described in a peer-reviewed paper that has just been submitted (Martin et al 2017 submitted in Animal).
The second herd model is derived from the pre-existing SimHerd model. New modules have been added to estimate economic results, labor time and environment impacts at the herd level. Now a prototype model is available and has already been used to estimate such impacts for scenario analyses of alternative reproduction management strategies in high and low input systems. With regard to environment impacts, the results showed that improved reproduction efficiency do not directly improve methane emission. But improved reproduction efficiency allows formulating herd strategies (e.g. use of beef semen) that improves methane emission and economy. Secondly, a new modelling approach including both the SimHerd model and the ADAM model was developed. ADAM is a computer program that models selective breeding schemes for animals and simulates the genetic changes in the population under different selective breeding scenarios The contribution of the ADAM model was to include change in genetic level due to different reproduction management strategies including different use of different semen types (conventional semen, sexed semen and beef semen). In the combined modelling approach the SimHerd model estimate the technical and economic herd effects without the effect of change in genetic level, and the ADAM model was used to quantify the change in genetic level in the scenarios and its corresponding economic values. This has been described in a peer-reviewed paper (Ettema et al, in preparation, see the annex 2 of D1.4 report): Economic opportunities for using sexed semen and semen of beef bulls in dairy herds) where the consequences of seven different scenarios for use of sexed and beef semen were investigated within five different production environments. Within management level, the economically best strategy changed whether the improvement of the genetic level was taken into account or not. These results suggest that it can be important to include the effect of genetic change when evaluating reproduction management strategies at herd level.

WP2 Molecular approached from refined phenotypes
Phenotypic data and results from molecular analyses were generated from two types of models. i) Similar in vivo models were developed in INRA (France) and SLU (Sweden) with the aim to study the impact of metabolic changes on reproductive function. Cows were exposed to different diets inducing changes in energy metabolism and negative energy balance (NEB) during the post-partum period. Fat tissue, plasma, oocytes, follicular cells and follicular fluid, and uterine samples (endometrial biopsies and uterine flushings) were taken to perform a wide range of molecular analyses (epigenetics, transcriptomics, proteomics and metabolomics). ii) In vitro models were also developed to study the impact of metabolic imbalance and pathogens on endometrial function and samples prepared in view of proteomic, transcriptomic and epigenetic analyses. Results from the work performed in the different tasks of this Work Package have been presented in detail in the deliverables and are summarized below.

Results from In Vivo models:

Phenotypic evaluation for fertility and metabolism / results from metabolic analyses and characterization of groups
- Animal models and sampling
The animal models developed in INRA, SLU and Allice on the basis of use of different diets allowed to produce enough biological material for the molecular analyzes planned, exception made of Day 14 embryos. This limitation led to ask for a modification of the DoW.
- Phenotypic characterization (1) general metabolism and blood metabolites
The application of differential diets was followed by variations in dry matter intake, body weight and Body Condition Score, fat tissue thickness, milk yield and composition (case of dairy cows). However, strong individual variations existed and differences between the Holstein breed and Swedish Red Breed (SRB) in response to diets were observed. Effects of the above factors were also observed for blood metabolites (glucose, insulin, Non Esterified Fatty Acids (NRFA’s)). In the heifer model, the regime was not followed by significant changes in mean body weight nor body weight gain. From the in vivo models developed in the dairy cow experiments, extreme cows in terms of Negative Energy Balance were identified both in INRA and SLU and this information has been used when analyzing relationships with other variables (metabolic and reproductive variables) and for the choice of extreme animals from diet groups in view of molecular analyzes.
- Phenotypic characterization (2) reproductive variables and relationships between different types of variables
From all in vivo models no differences were observed in reproductive phenotypes between diet groups ; number of follicles, interval from calving to first AI, number of AI per conception (cow models), number of oocytes and quality (cow and heifer models), number of Day 14 embryos (cow models). However significant effects of diet were found on in vitro embryonic development, LH pulsatility and initiation of ovarian activity. The analyses performed from cow models failed to demonstrate relationships between negative energy balance (NEB) or other metabolic parameters and reproductive variables.

Despite, high individual variations were observed, the differential response of breed to low energy diet is interesting per se (control diet suited better to Holstein whereas low energy suited better to SRB and may lead to developments to further exploit this information in combination with information from Work Package 4.
The corresponding information is presently published (one manuscript based on INRA material, one manuscript based on SLU material).

Molecular Analyses from fat tissue:
Different profiles of markers of fat tissue activity have been found in relation with Negative Energy Balance (INRA/TS).
Analyzes performed with proteomics on fat tissue did show that 22 proteins and 12 proteins were differentially expressed in fat tissue samples collected from INRA and SLU cows respectively. In addition results obtained from INRA cows suggest that pre-partum profiles are predictive of the intensity of fat mobilization during the post-partum period (TS-INRA). However, so far this information has not been confirmed from SLU samples.
From transcriptomics studied by RNA sequencing, strong changes in gene expression in fat tissue occur in relation with post-partum stage, diet and the intensity of negative Energy Balance. Functional analyses are on the way. RNAseq data showed that 977 genes were differentially expressed (DEGs) between 4 weeks before and 1 week after calving and 31 DEGs were found between week 1 and week 16 post-partum in cows in moderate NEB. In cows in severe NEB, a total of 79 DEGs were found between 4 weeks before and 1 week after calving and 9 DEGs were found between week 1 and week 16 post-partum. In addition, 214 DEGs were found between high NEB and moderate NEB cows at 1 week and 510 DEGs at 16 weeks post-partum. Differences between 4 weeks before and 1 week after calving relates with genes involved in cellular growth and proliferation processes in the moderate NEB cows, whereas they were involved in lipid metabolism in the high NEB animals. This work has been completed by the determination of the corresponding methylation patterns from Whole Genome bisulfite sequencing. The full set of results should lead to a functional interpretation of the mechanisms explaining the effect of Negative Energy Balance on cow fat tissue physiology during the post-partum period. A special attention will be given to groups of genes that may interact with (or may be involved) reproductive function.

Molecular Analyses from oocytes and granulosa cells:
Transcriptomic analyses (RNA sequencing) has been produced from oocytes and lipidomic analyses from follicular cells and plasma. From RNA sequencing, changes in gene expression in oocytes (and less changes in granulosa cells) in relation with diet and the intensity of negative Energy Balance were documented. Oocyte transcriptomic profiles were not globally discriminated based on diet (5 animals with high energy diet (HED) vs 4 with low energy diet (LED)), nevertheless, 6 genes were differentially expressed between the two groups. When focusing on 5 animals with contrasted values of energy balance at 8 weeks post-partum (2 with moderate NEB and 3 with severe NEB), 138 differentially expressed genes were revealed.
Cumulus (2 HED vs 4 LED) could be discriminated based on diet, and 868 (268 annotated) differentially expressed genes were revealed. The first integration of data revealed significant differential expression for 8 biological processes, 7 cellular compartments (highlighting differences related to the plasma membrane and extracellular space), and 9 molecular functions (highlighting differences for membrane receptor activity, nucleotide/nucleoside binding, enzyme regulator activity).
New approaches were developed to analyse lipid profiles from different types of ovarian cells (oocytes, cumulus, granulosa cells) and corresponding reference for lipid data were obtained. However, very few changes in relation with diet and NEB have been observed from lipid profiling. Further work is needed to precise if changes in such molecules are just a reflection of the differential diet or they relate to the intensity of NEB.
Methodological approaches have been developed and DNA libraries produced for the study of the epigenome from low numbers of oocytes (groups of 10-20 non-superovulated oocytes) and this preliminary work can be valorised in future studies. Also, more work is needed to interpret the results from RNAseq in a comprehensive way and this will be completed after the end of the project. A special attention will be given to groups of genes that may interact with (or may be involved in) reproductive function.

Molecular Analyses from different types of cells in the endometrium:
New methods were developed for the preparation of tissue and for DNA and RNA extraction that gives new possibilities for use of next generation sequencing analyzes from small amount of well characterized populations of cells from biopsies of the endometrium (INRA-SLU). From RNA sequencing, strong changes in gene expression in the endometrium occur in relation with diet and the intensity of negative Energy Balance. However, the importance of the changes in relation with diet is much more pronounced in the luminal epithelium (more than 2000 differentially expressed genes, DEGs) than in the stroma (more than 500 DEGs) and very few DEGs (less than 10) were found in the glandular epithelium). Globally, less genes are differentially expressed in relation with NEB than in relation with diet and the highest number of DEGs is observed this time in the stroma. The differences observed between the different cell types and also the difference of response in relation with diet and NEB showed the interest of analyzing selected populations of cells and not the entire tissue.
No major changes of the metabolome in relation with diet or NEB have been found from uterine flushings. Only one protein was found with those from proteome in animals with extreme phenotypes in terms of negative energy balance status.

Results from In Vitro models:

Study of interactions between pathogens and bovine endometrial epithelial cells:
A model has been developed with the aim to investigate interactions between components from bacteria and endometrial cells and possible consequences for fertility. Bovine endometrial epithelial cells (bEEC) in culture have been exposed to E coli Lipopolysaccharide (LPS) and their response evaluated from proteomic and transcriptomic analyzes. In both cases a large set of candidates have been found to be strongly deregulated by LPS when compared to controls (around 40 proteins and more than 2000 gene transcripts). Integration of the RNAseq results with those of proteomics, shown that the gene expression changes observed 24 hours after challenge correlate well with the deregulations observed from protein profiles 72 hours after exposure to LPS. Functional pathways are also modified in a similar way; showing that LPS destabilizes cell structure, increases metabolism and glycolysis and associated oxidative stress, stimulates gene transcription as well as immune response and inflammatory processes, while immuno-tolerance is depressed. In addition, results from this model show that almost 90% of the molecules cited as key molecules for successful implantation are deregulated. This model mimics the effects of infection. However, when considering that challenges were performed for a short time and with a single molecule, it can be speculated that alterations of cell functions are much less pronounced than in clinical cases of endometritis. Despite this, the importance of alterations raises questions about the ability of cells (and time taken) to come back to “normal” and show well how dys-regulations persisting in the absence of infection may alter implantation and fertility.

Study of interactions between mediators of NEB and endometrial epithelial cells:
Similar approaches as those described above have been developed to analyze the response of the bEEC to mediators of NEB and their possible impact for the establishment of pregnancy. Exposure of cells to non-esterified fatty acids concentrations in the range of those observed during the post-partum period show at high dosages, detrimental effects of those on cell survival and cell proliferation. The cells also accumulate lipids which is a potential source of oxidative stress. They express abnormal cytokine profiles which may be related to the above and increased apoptosis. More molecular studies will be done in the continuation of Prolific to understand the full range of mechanisms involved. The first studies performed under WP2 show that the uterine epithelium is very sensitive to NEB mediators. This may affect, in addition to possible impact on immune processes not studied here, the capacity of response of the endometrium to infection.

Perspectives and conclusions:
The results from WP2 generated four types of outcome of potential use for the society and research community.
1) New inputs in modelization: General information on mechanisms regulating reproductive function to be included in future simulations to optimize WP1 models has been produced. This includes new insights on the impact of diet on reproductive tissues, and how Negative Energy Balance potentially impairs uterine function which relates to some of the issues under study in other work packages (especially WP1).
2) Functional information to orientate future genetic studies: Both results from in vivo and in vitro models already show genes and network of genes differentially expressed / deregulated under metabolic imbalance or exposure to pathogens. This functional information can be used to reinforce the choice of candidate genes/regions potentially influencing fertility identified through genome wide association studies in WP3. These regions / genes represent interesting targets to be studied more deeply in the future to optimize genomic selection. This functional information and the sets of data already generated will be also of some use to better annotate the bovine genome.
3) Information to design new diagnosis and prognosis tools: Results which may be useful to design future diagnosis and prognosis tools have been generated. This includes for instance a list of genes and proteins modified by bacterial components such as LPS which may be used to develop future tools dedicated to the diagnosis of persistent inflammation in the endometrium or predictors of pregnancy success. These results will be used in ongoing studies aiming to investigate differences between disease cows and healthy cows. This includes also a list of candidate proteins / genes identified in fat tissue from pre-partum cows which could be predictive of fat mobilization/ metabolic disorders in lactating cows. The information from present work should be followed by development activities “post Prolific” that may lead for instance to tools which may be used in research activities like those presently performed under WP4.
4) Basic information for future research on Health and Resilience: Results have been produced from in vitro models showing a strong impact of metabolic imbalance and of pathogens on uterine tissue. This information and corresponding know how will now be used to develop un-invasive models to study how various factors affects health and resilience of uterine tissue in dairy cows. This may help in the future to develop phenotypic tools to be used in genomic selection. Genomic selection may also benefit from future GWAS based on those phenotypic markers.

The present set of results already allows understanding some of the molecular changes occurring in different tissues during the post-partum period in association with the severity of Negative Energy Balance. These results raise also key questions to be addressed in the future by functional studies and as mentioned above, needs to be further integrated with advances made in other work packages. In addition some of the information produced may help to develop in the future new applications such as phenotyping tools and/or diagnostic tools for veterinary research and practice in dairy cattle. The pertinence of approaches for developing applications should be discussed with stakeholders.

WP3 From genomics to selection
The objective of this WP3 was to identify the causal mutation for fertility based on progesterone measures, and estimate genomic breeding values using whole-genome sequence information on individuals, in order to develop optimal breeding strategies.

For WP3 of the PROLIFIC project milk progesterone records have been collected for 5770 lactations from Holsteins at experimental herds but also at commercial herd equipped with the Herd Navigator (Lattec I/S, Denmark) (task 3.1). The raw progesterone measurements have been translated into endocrine fertility traits like commencement of luteal activity (CLA), luteal phase length etc. and genetic parameters were estimated (Tenghe et al., 2015). The heritability for CLA was somewhat higher than the traditional fertility traits (calving interval (CI) and calving to first service) and some of the endocrine traits have a higher repeatability over lactations. The higher repeatability indicates that the endocrine traits reflect more the ability of the cow and are less hampered by environmental factors such as farm management. Noteworthy, CLA had a lower correlation with milk production than CI and is thus a good trait to select for improved fertility whilst improving yield simultaneously.

Most of the cows with progesterone records were also genotyped. A list of bulls to be sequenced was made based on pedigree information of the phenotyped cows, and 14 of them were sequenced through PROLIFIC but a further 25 and 22 Holstein sires from Wageningen UR (including DLO) and TEAGASC, respectively, were also added to surpass the original proposed 50 sequenced bulls (task 3.2). The sequences were incorporated in Run 5 of the international 1000 Bull Genomes Project.

Associations between the endocrine fertility traits and variation on the genomes of the cows have been studied using 80,000 DNA markers (Tenghe et al, 2016a). Through genome-wide association studies we searched for genetic variants associated with endocrine fertility traits. The regions were fine-mapped using imputed whole-genome sequence data. A number of candidate genes were discovered (task 3.3). A list with p-values and SNP effects for significant variants was forwarded to researcher in WP1 & WP2 in order to compare the associated genomic regions with their findings based on other ‘omics’ techniques.

The database with progesterone phenotypes and genotypes was updated in March 2016 and contains 5781 genotyped animals and 9,463 lactations with progesterone records since, hence the milestone of 8000 lactations with progesterone phenotypes was reached after all (task 3.1).

Opportunities for genomic prediction for fertility using endocrine and classical fertility traits in dairy cattle (Tenghe et al., 2016b) have been explored and the results showed that prediction accuracy for fertility can be increased by using bi-variate models for training that includes both endocrine and classical fertility traits (task 3.4). However, this requires progesterone records on a large number of daughters of bulls.

To determine optimal breeding strategies the following scenarios were investigated using input parameters from literature. The number of progeny records contributing to calving interval was increased from 20 to 140 and the proportion of progeny with endocrine phenotypes were maintained at 20% of the progeny with CI records. Results showed that inclusion of endocrine trait(s) as predictor in addition to classical traits for fertility increased the accuracy of selection up to 60% and the response to selection for calving interval from -2.15 days to -3.54 days. The benefit of including genomic predictions for endocrine traits increased accuracy approximately 18%. These scenarios are based on a single selection goal of improved fertility and were diluted when considered as part of a broader breeding goal. These results show that there is the potential to increase accuracy and overall response to selection for fertility in dairy cows using genomics and progesterone data as part of overall breeding programme (task 3.5).

Currently, it is however unlikely that progesterone will be measured on such a large scale as it requires quite an investment. Hence it is more likely that there will be a training population with phenotypes and genotypes to predict SNP-effects that can be used to estimate genomic breeding values of selection candidates. In such scenario, endocrine traits only add to the breeding program if they are the breeding goal trait. If classical traits like CI which are cheaply recorded population wide remain the breeding objective than the limited population with progesterone records does not enhance the response to selection (task 3.5).

In order to get more records on endocrine traits it is more valuable for selection response to record on additional animals than on additional lactations of the same animals. In addition, progesterone records from different countries could be combined to increase the reference population (task 3.5).

WP4 Innovations in farm nutritional management to optimize cow fertility
Tasks 4.1 4.2 and 4.3 (Ireland, France, Sweden)
A major effort was undertaken in WP4 to examine the effects of cow breed and nutritional management on indicators of reproductive efficiency across diverse European feeding systems. Multi-year studies were conducted in Ireland (TEAGASC), France (INRA) and Sweden (SLU), and detailed fertility phenotypes were recorded. These studies compared high and low input feeding systems within both pasture (Ireland and France) and confinement total mixed ration (TMR; Sweden) feeding systems. The same animals and feeding systems were used for Tasks 4.1 4.2 and 4.3.

Identifying the optimum stocking rate (SR) and cow breed for pasture-based dairy systems is essential to maximise output/ha without compromising reproductive performance. The objective of this experiment was to compare the performance of two different cow breeds (Holstein Friesian (HF) and Jersey crossbreds (JEX); n = 69 per breed) managed under three different SR (Low: 2.5 cows/ha; Medium: 2.9 cows/ha; and High: 3.3 cows/ha; n = 46 per treatment) for three consecutive years. Measurements were taken to identify effects on resumption of cyclicity, oestrus intensity, blood indicators of bioenergentic status, uterine health and fertility performance. Neither SR nor breed affected resumption of cyclicity (25.2 ± 1.2 24.5 ± 1.2 25.1± 1.3 d for Low, Med and High SR; 24.7 ± 1.8 vs. 25.2 ± 1.7 d for HF vs. JEX, respectively, P>0.05) or oestrus intensity. Oestrus duration was not affected by treatment but was reduced in Parity ≥3 cows compared to Parity 1 and 2 cows. The different SR treatments and breeds had little or no effect on blood metabolites. Mean plasma glucose (72.9 ± 0.8 71.9 ± 0.8 72.4± 0.9 mg/dl), β-HBA (0.61 ± 0.01 0.61 ± 0.01 0.62 ± 0.01 mmol/l), fatty acids (0.46 ± 0.01 0.48 ± 0.01 0.47 ± 0.01 mmol/l) and IGF1 (93.7 92.9 and 98.9 ± 4.3 mmol/l) concentrations were not affected by SR (Low, Med and High, respectively; all P>0.05). Mean glucose concentrations were significantly greater in JEX cows compared with HF cows (73.5 ± 0.7 vs. 71.3 ± 0.8 mg/dl; P< 0.05 respectively), but breed did not affect (P>0.05) plasma fatty acids or β-HBA concentrations. There was a significant effect of SR and breed on body weight. There was no effect of SR or breed on BCS. Neither SR nor breed affected (P > 0.05) the proportion of cows with endometritis at week 6 postpartum or the overall phenotypic fertility performance during the breeding season. In conclusion, there was no major effect of increasing SR or cow breed on reproductive performance. Providing that nutritional requirements are met and cows are genetically suited to seasonal-calving pasture-based systems, it is possible to increase stocking rate without causing adverse effects on cow metabolic health or reproductive performance.

Strong genetic selection on production traits is considered to be responsible for the declined ability of dairy cows to ensure reproduction. The objective of this experiment was to quantify the effect of genetic characteristics (breeds and genetic merit for production traits) and feeding systems (FS) on the ability of dairy cows to ensure reproduction. An experiment was conducted during 9 years on Normande and Holstein cows assigned to contrasted pasture-based FS. Diets were based on maize silage in winter and grazing plus concentrate in spring in the High FS; and on grass silage in winter and grazing with no concentrate during spring in the low FS. Within breed, cows were classified into two genetic groups with similar estimated breeding values (EBV) for milk solids: cows with high EBV for milk yield were included in a Milk-Group and those with high EBV for fat and protein contents were included in a Content-Group. Holstein produced more milk throughout lactation than Normande cows (+2294 kg in the High FS and +1280 kg in the Low FS, P<0.001) and lost more body condition to nadir (-1.00 point in the High FS and -0.80 kg in the Low FS, P<0.001). They also showed a poorer ability to be inseminated because of both a delayed commencement of luteal activity (CLA) and delayed first service (more days from start of the breeding season to first service, DAI1). Holstein cows had a lower re-calving rate than Normande cows (-19 %). Cows in the Milk-Group produced more milk than cows in the Content-Group, but milk solids production was similar. Cows in the Content-Group had earlier CLA than cows in the Milk-Group (P<0.01). Genetic group neither affected ovulation detection rate nor DAI1. Within breed and FS, cows with high genetic merit for milk yield had later CLA and DAI1. We found no effect of genetic group and FS on fertility of Normande cows. However, according to FS, Holstein cows in the content group exhibited different fertility failure patterns. In the low FS group, Holstein cows in the content group had more non-fertilizations or early embryo mortality (+26 percentage units at first and second services) than Holstein cows in the milk group. In the high FS group, Holstein cows in the content group had a higher proportion of late embryo mortality than in the milk group (+10 percentage units at first and second services). Cows in the High FS produced more milk and lost less condition to nadir than cows in the Low FS. FS did not affect dairy cows' ability to be inseminated nor re-calving rate. However, indicators of energy balance (protein content or body condition score) were positively associated with earlier CLA and DAI1, and with successful conception and pregnancy. In addition, higher milk yield was associated with poorer ovulation detection rate and oestrus intensity (P<0.05). Hence, at similar EBV level for milk solids, selection for increased milk fat and protein content resulted in improved cyclicity and similar oestrous expression and submission rates compared with selection for increased milk yield. However, it was associated with a lower ability of dairy cows to ensure pregnancy because of more non-fertilizations and early or late embryo mortality.

This study examined the effect of two feeding levels during the antepartum and postpartum period on reproductive performance and blood metabolites (glucose, non-esterified fatty acids (NEFA), insulin) in primiparous Holstein and Swedish Red (SRB) cows, in order to identify possible differences in the way these breeds respond to negative energy balance after calving. A total of 44 cows (22 Holstein, 22 SRB) kept in a loose housing system were included in the study. The control group (HE, n = 23) was fed a diet for high-producing cows (target 35 kg/d energy corrected milk, ECM). A lower feeding intensity (LE, n = 21) was achieved by giving -50% concentrate to target 25 kg/d ECM. Diets were implemented 30 days before expected calving and the cows were monitored for 120 days postpartum. Milk yield and composition, dry matter intake (DMI), live body weight and body condition score (BCS) were assessed to calculate the weekly energy balance (residual feed intake). Blood sampling started before diet implementation and was repeated every 2 weeks until Day 60 postpartum and then once monthly until Day 120. Holstein cows had lower mean energy balance than SRB cows (-4.7 ± 1.4 and -0.9 ± 1.4 MJ, respectively; p = 0.05). SRB cows had higher (p<0.001) BCS (3.3 ± 0.1) than Holstein cows (2.7 ± 0.1) and also higher plasma glucose concentrations from Day -30 to Day 120 relative to parturition (4.1 ± 0.1 and 4.2 ± 0.1 log ; mg/100 ml, respectively; p < 0.05). Overall, breed or diet had no effect on NEFA blood plasma concentrations. However, plasma NEFA concentration levels tended to be higher (p = 0.09) in SRB cows than in Holsteins at Day -14 before calving, indicating higher mobilisation of lipid from adipose tissue already before calving. In contrast, Holstein cows had higher NEFA at Day 14 postpartum than SRB cows (p < 0.05). There were no significant effects of diet or breed on reproductive performance (% pregnant at first AI, days open). However, commencement of luteal activity within 21d postpartum was affected (p < 0.05) by the interaction of breed and diet. These results suggest that Holstein cows prioritise milk production to a larger extent than SRB cows, resulting in a less balanced metabolic profile.

Task 4.4 (Ireland)
It is generally accepted that energy balance (energy consumed minus energy required for maintenance and milk) is a key regulator of reproductive status. During early lactation, the energetic cost of milk production can exceed energy consumed, resulting in a prolonged period of negative energy balance (NEB) and consequent mobilization of body tissue reserves. Body condition score (BCS) is an objective assessment of available labile body reserves. A number of scoring systems are available; in Ireland the 1 to 5 scale (1 = emaciated, 5 = obese) has been adopted. The purpose of this study was to use historical data to examine the relationship between BCS at calving with subsequent BCS loss after calving and reproductive performance during the breeding season. Data were sourced from the centralized database at TEAGASC’s Animal and Grassland Research and Innovation Centre, Moorepark, County Cork, Ireland. The animals included in the dataset were from two spring-calving research herds, which were representative of Irish grass-based farming systems with a diverse range of cow genetics. The records consisted of 6,008 AI services for 1,076 lactating cows (3,047 lactations) from 2001 to 2009. In addition, periodic BCS measurements were recorded on all cows (1 to 5 scale). The herds operated seasonal-calving pasture-based production systems, with 67.5% of the cows calving in February or March. All data manipulation and analyses were carried out using the R statistical programming language (R Core Team, 2015). Service data included animal tag number, lactation number, calving and service dates, service number, service sire and pregnancy outcome. All services were performed by artificial insemination. The outcome in the original data was defined within the database as being successful if the last service within an individual lactation had a subsequent calving date, regardless of duration. All other services were defined within the database as negative. The results clearly indicate that cows that are too thin (≤2.75) or too fat (≥4.00) at calving are at risk of reduced reproductive performance. There was little discrimination between cows within the range 3.00 to 3.75 in terms of submission or conception related variables. Nevertheless, taking all of the data together, the objectives of minimizing BCS loss after calving and maximising subsequent reproductive performance during the breeding period would suggest that a target BCS at calving of 3.25 would be optimum, with no cows thinner than 3.00 and no cows fatter than 3.50. The results provide practical findings that can be incorporated in dairy cow nutritional and reproductive management.

Task 4.4 (France)
Lactation and reproduction are concomitant functions in dairy cows and in competition for resources. The present study aimed to quantitatively review the existing literature to clarify the implication of milk production and body reserves at each step of the reproduction process. Inclusion criteria for the studies were: comparison of at least 2 treatments and reporting of both reproduction and production performance. The final database consisted of 275 treatment groups from 75 articles. Data investigation showed that the only investigable relationships were between commencement of luteal activity (C-LA), days from calving to first observed estrus (COE1), conception rate at first service (CRAI1), overall pregnancy rate (PR), milk yield and body condition scores (BCS; converted to the 0-5 points scale). The results showed that C-LA was not related to milk yield and that the relationship between C-LA and BCS at calving was quadratic. Although COE1 is an indicator of C-LA, no relationship was identified between any of the BCS parameters and COE1. However, for each additional kg of milk yield produced at both peak and over the initial 14 weeks of lactation, COE1 was delayed by 1.1 days. In this meta-analysis, CRAI1 was affected by both milk yield and BCS. In addition, CRAI1 was reduced by 2.0 % (of inseminations) and 2.2 % for each additional kg of milk yield at peak and at service, respectively. CRAI1 was increased by 38.2 % and 22.0 % for each additional unit of BCS at service and at nadir, respectively. Finally, no relationship between milk yield and PR was identified. PR was increased by 42.8 % (of cows) and 16.8 % for each additional unit of BCS at calving and at nadir respectively. Postpartum cyclicity of dairy cows is mainly affected by BCS at calving, whereas estrus expression is mainly affected by milk yield and fertility is affected by both BCS and milk yield. Strategies adjusting feeding level, milking frequency and dry period length to target a BCS of 3.10 and limiting BCS loss and peak milk yield could be an effective way to improve reproduction. Even when target BCS is achieved, a high milk yield strategy will require strong attention on estrus expression to detect ovulations and ensure that high PR is achieved. On the other hand, mitigating the strong genetic selection on milk yield and selecting dairy cows for functional traits such as fertility and higher BCS would enable genetic improvement of reproduction performance.

Task 4.5 (Ireland)
Annual variation in pasture growth rate has a major effect on grass availability for grazing dairy cows, especially at the onset of lactation in early spring. The objective was to determine the effect of imposing acute (2 weeks) or chronic (6 weeks) periods of different levels of pasture allowance on indicators of metabolic health and fertility in early lactation dairy cows. Dairy cows (n = 96) were randomly assigned to one of four daily herbage allowances (DHA) of 60%, 80%, 100% and 120% of intake capacity for either 2 or 6 weeks (12 cows per treatment) during early lactation for 2 consecutive years. No supplemental concentrates were fed during the experimental period. Blood was collected once weekly during the study to determine circulating concentrations of glucose, non-esterified fatty acids (NEFA), β-hydroxybutyrate (β-HBA), insulin and Insulin-like growth factor-I (IGF-I). In Year 2 of the study, liver biopsies were collected from a subset of cows assigned to the 60% DHA for 2 weeks, 60% DHA for 6 weeks and the 100% DHA for 6 week treatments at experimental week 0, 2 and 6. Reverse-transcription quantitative PCR (RT-qPCR) analysis was used to determine the mRNA abundance of 24 target genes related to energy metabolism. Milk samples were collected three times per week to determine progesterone concentrations, and these data were used to calculate interval from calving to resumption of cyclicity and oestrous cycle characteristics. Neck collars were fitted to all animals before parturition to record activity data and determine oestrus intensity and duration.
There was no effect on BCS, blood metabolites/hormones or gene expression for the 2 weeks DHA duration. Mean plasma glucose, β-HBA and insulin concentrations were not affected by the 6 weeks DHA. There was a significant effect of 6 week DHA on circulating concentrations of NEFA (0.64±0.09 0.65±0.08 0.52±0.10 and 0.55±0.09 mmol/l, P < 0.05) and on IGF-I (71.9±5.09 82.2±5.08 85.7±5.5 and 82.9±5.9 ng/ml, P <0.05) for DHA 60%, 80%, 100% and 120%, respectively. At week 6, cows assigned to the 60% DHA for 6 weeks had increased mean expression of G6PC, PC, CPT1A, ACSL, IGF-I and decreased mean expression of FASN and ACOC. The 60% DHA treatment for 6 weeks duration caused increased expression of IGFBP2 at week 6. We conclude that imposing short periods (2 weeks) of restricted DHA had no effects on metabolic health in early lactation dairy cows. After 6 weeks of low DHA a more pronounced effect was observed, but even at the most restricted DHA treatment, the cows displayed co-ordinated adaptation to cope with reduced feed intake.
There was little effect of DHA on any of the variables determined from the milk P4 results (P > 0.05). The 120% DHA treatment resulted in more luteal phases before 60 days postpartum than the 80% DHA treatment (P = 0.04). Neither intensity nor duration of oestrus were affected by acute or chronic restrictions in DHA (P > 0.1). There was no significant effect of DHA on incidence of silent heats or heats without ovulation. From phenotypic fertility measures, there was little effect of DHA on reproductive performance, although, the 6-week 80% DHA treatment had a tendency for shorter mating start date to conception interval than the 6-week 100% DHA treatment (P = 0.08). A significant effect of DHA for the 6 week duration on calving interval was also detected. The 80% DHA cows had a significantly shorter calving interval than the 100% DHA (P = 0.03). Based on the conditions of this study, we conclude that imposing periods of restricted DHA had no severe effects on oestrous cycle characteristics, oestrous behaviour or overall reproductive performance.

Overall conclusions from WP4
The dairy cow feeding systems are widely divergent between different EU countries. Our results indicate that cows are capable of orchestrating a co-ordinated response to changes in nutrition, resulting in relatively minor changes in fertility phenotypes. In general, higher input feeding (or more generous pasture allowances) results in more favourable indicators of metabolic status, greater BCS and greater milk production, but consistent effects on resumption of cyclicity, oestrous cycle characteristics, uterine health and phenotypic fertility performance were not observed. On the other hand, the breed of cow did have an important effect on many of the phenotypes examined, including BCS, energy balance, oestrous cycle characteristics, and phenotypic fertility performance. Within a particular geographic region, the optimum feeding system will be dictated by many factors that are not directly under the farmers’ control (e.g. climate, soil type, milk price etc.). For each feeding system, the long-term breeding objective should be to genetically select cows that will have improved productivity within that feeding system.

WP5 Multi-site demonstration of reproductive management tools
The main objective of this WP was to develop tools that integrate knowledge created in other work packages (especially WP1), and that can be used by the farmer for decision support regarding fertility management. Being in control of reproductive management is a pre-requisite for crafting a cattle production system that maximizes lifetime productive efficiency in a balanced and sustainable manner. Achieving this has not been easy given the recent decline in fertility and reduced expression of estrous behavior in modern cattle. However, this project through a winning combination of disciplines has been creating improved decision support tools for innovative reproductive management strategies.
The objectives of WP5 were to:
1) Deploy, and evaluate, on commercial farms decision support tools that will allow the farmer to better target reproductive actions at individual cow and herd level, according to the livestock system he/she finds himself in.
2) Demonstrate the feasibility of transferring new fertility bio-markers into an in-line on-farm measuring system.

Fertility tools were available before, but they were based mainly on behavioral oestrus, which has been proven to decline in case of pure oestrus behavior. The tools built in this WP rely both on real-time on-farm progesterone measurements, and on predictive models.
WP 5 has provided three tools that can directly be used by farmers:

Tool for insemination worth predictor (IWP)
The first tool, the IWP is based on results from WP1. It combines progesterone profile features and cow traits in order to provide the farmer with a real-time probability of insemination success. The tool runs within the Herd Navigator system and provides a probability of insemination success for each detected oestrus. This tool will aid the farmer in deciding whether or not an oestrus is worth inseminating.
In order to be able to incorporate additional variables to the equation for insemination success, their effect on insemination success needed to be determined. Using literature study, potential predictors for probability of AI success were selected. These were: progesterone profile slope, length and height; milk yield; parity; oestrus number; insemination of previous oestrus; time between oestrus and insemination.
The relationship between the remaining potential predictors and insemination success was visualized in two steps: 1) an a priori curve was fitted to the predictor effect shape observed in the visualization plots; 2) the predicted and observed data were plotted in the same plot to examine differences between observed and predicted values.

After evaluation of the predictors the final model for IWP was built. Predictors were added to the model one by one. The base model was corrected for timing of oestrus (which affects insemination success, but is not related to the cow or her progesterone profile), and parity.
Variables were retained in the model if they improved the overall likelihood of the model. In the end, cycle length, milk yield, cycle height and insemination during the previous oestrus were found to be good predictors, and were used to build the final model. Model building was performed using a stepwise approach, adding the predictors one by one in different order. The model that had the smallest distance based on likelihood estimations was selected, and included respectively milk yield, cycle length, cycle height and previously inseminated oestrus.
It contains correction for timing and parity, cycle length of the cycle ending at that oestrus, milk yield, cycle height and whether the previous oestrus was inseminated as predictors.
The equation has been built from Herd Navigator progesterone data and has proven to be robust on farms with an average milk yield around 9500kg.

Next, it was used to predict insemination success in the independent test dataset. Data was collected from ten high yielding farms to test the robustness of the equation. Not only the final model but also all of the steps in the original model development were tested to ensure that no individual predictor in the original model was found to not contribute significantly when fit on the test dataset.
There were no redundant or overestimating variables in the equation when used on the test dataset, and if the probability of insemination success as estimated by the equation was high, the chances of this insemination being successful were significantly higher than it being unsuccessful. This indicates that the equation is robust for use on other farms. This equation will aid reproductive management on farm by allowing farmers to decide whether to inseminate to a given oestrus, and thus to optimize their insemination strategy on a cow by cow basis.
The equation for IWP is built into the Herd Navigator software and a paper about this research has been submitted to Journal of Dairy Science (December 2016).

Tool for reproductive management timing optimizer (RMTO)
The second tool, the RMTO is also based on results from WP1 and it uses simulations on cow and herd level to predict the effect of a change in fertility management on a given farm.
The RMTO tool will allow the end users (farmers and/or technicians depending on the country or the context considered) to evaluate, by simulation, the consequences on reproductive performances of different reproductive management options such as different oestrus detection methods, insemination methods (including sexed or crossbreed semen), and voluntary waiting period. Based on the RPM model developed in the WP1, the RMTO tool was developed using a cut-down version of this model called “RPM lite”, which matches both the available on-farm information and the required model response time. In WP5, a user-friendly interface was developed for a web-based product that allows users to design farm-tailored reproductive management protocols. Due to breeding systems diversity, a key component of the design was to ensure that the tool should be applicable in different farming systems and management strategies and farming systems (pasture vs. in stable, year-round calving vs. a seasonal calving system, etc.) In fine, the RMTO tool will help farmers to take the right decisions about their reproductive management from a technical and biological point of view (economy not currently included in the outputs).

In order to develop a tool that fit well with the field needs, it was decided to plan, as a preliminary step, discussions with farmers in France and in Sweden about their herd management, their main problems in the reproduction field and also to ask for available inputs they may have and what does an ideal RMTO tool look like. Focus groups were organized in Sweden and in France to maximize diversity in breed and herd management strategies.
In summary, the time is always the limiting factor on farms and therefore, farmers prefer a « quick and dirty » tool both in Sweden and in France. However, there was a consensus among audited persons about the importance and needs in terms of advice to farmers and benchmarks: challenging financial climate, larger herds, technologies development, increasing amount of data to integrate. Their expectations in the reproduction field are mainly focused on the following topics:
- Timing of insemination: when do I inseminate the cow?
- Start of insemination: what is the ideal voluntary waiting period?
- Semen type: what is the optimal use of sexed semen in my herd?
- Herd management: how many heifers should I breed?

Moreover, technicians favour a tool where they can compare long-term practices, while farmers prefer a day-to-day support tool for short-term management of their herd. A local RMTO software to be installed on a laptop was preferred to an online access tool because of some areas without internet connection.
Based on information previously cited, a prototype of RMTO tool was developed with regular virtual feedback meetings.
The RMTO tool was presented in Sweden and in France to some end-users to ask for their feedback about the prototype. The feedback showed that the overall global satisfaction score of end users is 4.25/10 and several remarks were made. Globally, these feedbacks showed that this 1st prototype is a very good idea that fits with farmers’ needs but that is not fully satisfactory at the moment. To be used in field, the prototype needs to be improved to be more robust (avoid bugs), easy to use (more pleasant graphical interface and faster simulation) and to be stronger validated in terms of outputs values. In a longer perspective, the models and the RMTO tool would integrate more information (economy, labour, food production) for a global advice to farmer taking into account all the aspects of a farming system.
These comments are now taken into consideration. By the end of the PROLIFIC project the RMTO prototype was placed on a restricted access part of the websites to disseminate the software to well-focused end-users and, by this way, to continue to get some complementary feedback and go forward through its optimization beyond the project. The final version is being prepared for publication on the Allice and Växa websites.

Progesterone based benchmarking tool (PBT)
The third tool, the progesterone benchmarking tool (PBT), provides farmers with key figures for commencement of luteal activity (CLA) and the number of open days (OD). Calculation of PBT key figures in multiple farms allows the farmers to benchmark against each other.
The key figures are based on progesterone data from the Herd Navigators system and the farmers can easily see how they are performing when compared to other farms, or whether they reach the goals they have set.
The key figures have been presented to Herd Navigator farmers and their advisors in the Netherlands. The feedback from the users has been: “Easier to understand the degree of homogeneity”. From an advice perspective it has been easier to identify lack of performance with the CLA and OD and to identify potential reasons behind the lack of performance and suggest intervention.
Next to the creation of the benchmark end user evaluation, a study was performed with actual farm data to define risk factors for delayed CLA and /or increased number of OD. This analyze has been made by combining data from the Swedish cow database and the Herd Navigator data. These results will be very important for both farmer and advisor in the follow-up when a meaningful difference is found between the farm result and the benchmark. The farm specific values indicates itself when the farm deviates from the benchmark, but the results from part two will allow the farmer and/or advisor to focus on possible interventions.
The development of PBT and identification of the risk factors will continue after the end of the PROLIFIC project.

Stage-gate protocol for adapting new biomarkers
Apart from the tool described above, a stage-gate model was designed to describe the process for introduction of new biomarkers into an automatically in-line and on-farm measurement system. This involves the development of a stage-gate protocol that can be used to evaluate the compatibility, precision, and repeatability characteristics of a biomarker when transferred from a laboratory-based to an on-farm technology. The final objective of this process was to develop and deliver a solution that creates both values for the customer and long-term profitability for the company. By following the stage-gate model it should be possible to deliver a mature solution and a prepared line organization.

Thus, all the deliverables and milestones of WP5 were achieved in a timely manner and with outreach to relevant stakeholders.
Potential Impact:
WP1 Multilevel integration and modelling of reproductive performances at different scales
The main practical implications from WP1 are that the project has produced the building blocks and tools for driving exploratory and prospective studies. These are:

A list of scenarios available for prospective studies
Beyond its interest for guiding the modelling and predicting works done in WP1, such a list should be considered as an interesting background for forthcoming prospective studies. Indeed it provides a shared view of the main reproductive strategies that are applied in European dairy systems and identifies the most important factors that experts considered as levers to improve reproductive efficiency in each type of production systems.

Knowledge capitalization, progress in systemic representation of reproductive physiology and of mechanisms whereby animal factors play on subfertility in dairy cows.
Two reproductive physiology models are now available from WP1. One focuses on the oestrus cycle level and is particularly relevant to help identifying what groups of genes are involved in altering reproductive hormones patterns. The other model is at the cow level and enables to quantify deviations in reproductive performance that can be expected when an animal factor changes separately or in combination with others. These two models both rely on systemic modelling and on explicit representations of biological mechanisms involved in hormones dynamics. Their development required an important work of integrating knowledge both from literature and expertise and they can be considered as forms of knowledge capitalization. The on-going collaboration with reproductive physiologists and systemic modellers has fulfilled one of the primary aims of a research model in that it has permitted a substantially improved understanding of the linkages between different components of this complex physiological process. It has also drawn attention to the existing gaps in knowledge, with notably the identification of the lack of information on which reproductive physiology mechanisms are impacted, and to what extent, by known disruptors of reproductive performance such as negative energy balance and increased production intensity. Accordingly, these models are well placed to generate different hypotheses on these linkages and thereby contribute to future research work to elucidate the links between performance and reproduction.
Three main applied perspectives can be related to the development of these models.
- The first one relates to benchmarking as models are relevant to provide references of the expected reproductive success we could expect in a virtual cow where no management rules interfere on cows’ lifespan. Such models are tools to explore the biological components of the interindividual variability in reproductive traits and allow to quantify the expected deviations if we change something in the characteristics of cows such as their milk potential or their nutritional status.
- The second applied perspective is that such models are relevant tools to test research hypotheses on biological mechanisms involved in subfertility and should be considered as alternative or at least complementary ways to animal testing.
- At the end, Reproductive Physiology Models have a real potential in teaching as they enable to show how reproductive hormones jointly vary with time according to the physiological status of cows and to visualize on what dynamics animal factors will play. Partners involved in the PROLIFIC project who have teaching activities in veterinarian or animal sciences have agreed that simplified versions of these models should be very useful to support reproduction teaching.

Decision support tools
A decision support tool (IWP) was developed in WP5 from the statistical work achieved in WP1 from progesterone data sets. This tool is now available to determine the best oestrus to inseminate. Such a tool is supposed to give farmers the opportunity to decrease the number of inseminations per pregnancy and to better control the reproductive timing of high yielding animals by adjusting their voluntary waiting period. This tool is currently based on a prediction equation that relies on progesterone profiles features and is thus dedicated to farms equipped with the Herd Navigator© device.

Among the two simulation tools developed at the herd level, one (RMTO derived from RPM_herd) is dedicated to technicians and is to be used for advice on farm. It will help technicians simulating and illustrating what could be the impacts of changes in reproductive management (oestrus detection method, voluntary waiting period, seasonal calving...) and/or animal factors on the herd performances (milk yield, demography, fertility). This tool is thus interesting to help farmers redesigning the reproductive strategy of their herd by comparing deviations in performances that could be expected when something changes in reproductive management or in animal characteristics.
The second tool (SimHerd coupled with Adam) remains a tool for researchers and helps comparing reproductive management scenarios based on the analysis of deviations in technical, environmental and economical performances they are expected to lead to (results from simulations). This tool appears to be quite relevant for prospective studies to help stakeholders to explore and then outline what could be the relevant guidelines that could be to prioritize and support to improve European dairy systems’ breeding and reproductive strategies and increase their sustainability.

In summary, Work package 1 has improved our understanding of reproductive physiology and the interaction between the animal’s reproductive ability, her production potential and nutritional environment, as affected by the prevailing reproductive management. This quantitative understanding of the “fundamentals” of reproductive performance is of substantial value to a range of stakeholders. Indeed, WP5 has clearly demonstrated both the appetite for and the potential of applications that build on WP1 models. It is already envisaged that these demonstration tools will be developed further and used for on-farm advice, evaluating the technical and economic consequences of possible future changes in reproductive and nutritional management. This will help farmers in deciding which management strategy best suits their farm. The models developed in WP1 can also readily form the basis of much larger scale simulations to test the impacts of different genetic selection strategies and the use of different reproductive technologies such as sexed semen, etc. Ultimately such simulations could be used to optimize the trade-off between production intensity and productive longevity within different regional production environments, and thereby contribute to an improved long-term efficiency and sustainability of the dairy industry.

WP2 Molecular approached from refined phenotypes
So far, only a little part of the results obtained from WP2 has been published already with full manuscripts but results have been disseminated in the scientific community at time of scientific congresses (International Embryo Transfer Society, International Congress on Animal Reproduction) and meetings (COST action Epiconcept) through abstracts. This concerns mainly results from in vitro models which are presented in Deliverable 2.5. A lot of results issued from the in vivo experiments based on molecular analyses in fat tissue, oocytes and granulosa cells, and endometrium will follow. This will be the source of numerous scientific publications based on each type of tissue which are on the way/will be prepared. Integration is needed before the identification of specific markers will be done and their interest for possible applications can be evaluated. However, all the data sets are /will be stored and possibly reused in future projects as bases for comparisons and genetic studies for instance in relation with reproductive health. As an example, the results from D2.5 have been used in connection with those from WP3 to define a list of genes potentially interesting for GWAS studies. Some of the data sets generated under PROLIFIC are already in use in continuation studies. Future publications will be put on line on the Webb site (until December 2017) and will go on to refer to PROLIFIC until all results have been valorised.

Potential impacts: First results from D 2.3 issued from the work on fat tissue changes around calving in relation with Negative Energy Balance (NEB) are of a particular interest and deserve dissemination to a larger audience with more technical attendance (3R in France...). The identification of markers from pre-calving fat tissue which may be predictive of the amplitude of NEB during lactation may be useful to identify animals more likely to suffer from NEB and associated metabolic disorders. More work is planned to investigate if such characteristics could be detected earlier in life, giving means to farmers to deal better with the management of this type of animals. Such criteria could be used as well as a phenotypic marker for genetic studies/selection. The opportunity of developing a simple test allowing the detection of the markers is under discussion.
The results from D 2.5 mentioned above are also of a particular interest, as for the first time they establish the link between the effects of pathogens and specific targets necessary for the establishment of pregnancy. Some of these and their possible use as markers for diagnostic tools of persistent inflammation of the genital tract are under investigation in a research project funded by the Swedish agency FORMAS aiming to study their use from in vivo conditions. The information is probably of a translational value and may be disseminated to a larger scientific audience i.e. not only in the research groups working with Livestock.

The research conducted and generated results through WP2 are of a broad interest and have societal implications far wider than those directed to fertility and animal productions. The approach used gives new and deep insights in the mechanisms being at the origin of metabolic or infectious diseases. Such information, especially the epigenetic data produced on different types of reproductive tissue (fat / uterine cells) may give keys to understand why some of the individuals are more sensitive to disease. As these mechanisms are often well/partially conserved at the molecular level, the results are of potential interest for a wide range of species. Also due to similar impacts of the factors involved and altered mechanisms in other types of tissues, the interest of the information is not limited to the field of reproduction and fertility. Our results, especially those issued from LPS experiments from D2.5 illustrate that many different functions are disturbed by a single molecule establishing a possible link between different diseases which may have a common origin. For instance the increased proliferative phenotype induced in endometrial cells and the down-regulation of a set of molecules in relation with cell adhesion and increased migration are not only susceptible to perturb fertility as these changes are fundaments of the deviation of cell properties found in other diseases such as cancer.

WP3 From genomics to selection
In WP3 a database of animals with progesterone phenotypes and SNP genotypes was created. The data consist of two major sources and can only be used within the PROLIFIC project. For further use the owners of the data have to be contacted. Source 1: EU project Robust Milk, data belongs to individual project partners (TEAGASC, SLU, SRUC, DLO). Source 2: Dutch commercial farms equipped with a herd navigator, data belongs to individual farmers.
Sequence data of 14 bulls generated by WP3 was added to the 1000 Bull Genomes project and is available for members of the 1000 Bull Genomes consortium. Membership can be acquired by sharing cattle sequence s, for conditions see
Single SNP GWAS results (p-values) for endocrine fertility traits have been published in Journal of Dairy Science and have been shared with other PROLIFIC WPs. Can be shared with other researchers (e.g. for meta-analysis, or comparison with other omics results) or (breeding) companies. Backsolved SNP-effects from genomic prediction of commencement of luteal activity can be made available for breeding companies to estimate genomic breeding values of animals without progesterone records.

In January 2014 an ITN (International Training Network) proposal was submitted to the EU. It was a joint effort between Prolific and Fecund partners to get EU-funding for PhD students within a strong network of interdisciplinary scientists working on cattle reproduction. Unfortunately it was not selected, the original proposal could be polished and resubmitted if partners are still interested, a good coordinator is needed. SLU has proposals to continue research on cattle fertility (contact Britt Berglund).

The results from WP3 show that heritable endocrine traits can be defined using milk progesterone profiles (here from in-line recording on milk samples using the Herd Navigator).
Finding causal variants for fertility is difficult, even with whole-genome sequence data, but it will be important for gene-editing if that will be the future. The problem with whole-genome sequence in cattle is the large extension of linkage disequilibrium which is beneficial for many genetic applications, but complicates fine-mapping of causal variants. Better sequence data, imputation, as well as an improved reference genome will be beneficial, but particular study designs to break down linkage disequilibrium are essential.

Breeding for fertility is currently applied using traditional traits based on calving and insemination records (e.g. calving interval, non-return rate, calving to first service). These traits are easy to measure population wide and are important for the economic breeding goal to improve fertility. At this moment measuring progesterone profiles on the same scale is too costly. However, a reference population of cows with such profiles like the database created in Prolific can be used for genomic prediction of breeding values for such traits. The impact of such genomic breeding values for endocrine traits on the selection response for fertility is dependent on the breeding goal definition. If a similar (and genetically correlated) trait can be measured on the whole population there is no benefit from the endocrine trait (e.g. commencement of luteal activity versus calving to first service or calving to first heat). Hence easier/cheaper to measure traits (e.g. from activity sensors) might be of less quality, but the quantity compensates for that.

WP4 Innovations in farm nutritional management to optimize cow fertility
The results of the complimentary breed and nutrition studies conducted in Ireland, France and Sweden will allow dairy farmers to make informed decisions regarding the impact of nutritional management (farm stocking rate, pasture allowance, dietary energy density) and breed/strain of dairy cow on production and reproduction. These are important sources of variation in cow productivity, and also farm profitability, and hence will underpin the competitiveness of dairy industries in EU countries.
In Ireland, the majority of dairy farms are seasonal calving, with most cows calving in late winter/early spring months to synchronize herd feed requirements with seasonal variation in pasture supply. The key driver of profitability is milk production per unit of land rather than milk production per cow. Hence, in the post-quota environment, farmers will increase stocking rate to increase milk production per unit of land. The experiment examined the consequences of increasing stocking rate on detailed fertility phenotypes using the two main breeds of cow in Ireland (Holstein-Friesian and Holstein-Friesian x Jersey crossbreds). The key findings were that with well managed pasture, increasing stocking rate from 2.5 cows per hectare to 3.3 cows per hectare (32% increase in carrying capacity), there was no effect on resumption of cyclicity, uterine health status, blood indicators of metabolic status or phenotypic fertility performance. Hence, it is feasible for farms that have well managed swards and high genetic merit cows (based on Irish Economic Breeding Index) to increase stocking rate, without having adverse effects on cow reproduction. This has important implications for potential herd expansion on Irish dairy farms.

The experiment conducted in France aimed to quantify the effect of genetic characteristics (breeds and genetic merit for production traits) and feeding systems on the ability of dairy cows to ensure reproduction. Our results suggested that:
- Selecting dairy cows for fat and protein contents instead of milk yield may not be the solution to maintain productive performance and to improve reproductive performance.
- It is possible, through genetic strategies, to improve some components of the reproductive process while impairing others. This showed that ovarian cyclicity, oestrus expression, and fertility are partly genetically disentangled. This may explain why the inclusion of reproductive traits in the breeding goal did not improve observed reproductive performance as expected. Detailed genomic index for each reproductive trait are necessary to improve reproduction of dairy cows and customize breeding goals to the requirements of different farming systems.
- Although they induced very contrasting responses in milk production and body reserve management, feeding system was not a limitation to dairy cows’ ability to ensure reproduction. This showed dairy cows have strong abilities to adapt to diverse environments.
- Other strategies that enable management of body condition and milk yield at key moments for the success of reproduction should be further investigated.

In Sweden, it is normal for cows to spend much of the year in confinement housing, and hence year round calving predominates. Hence, the key questions for dairy farmers revolve around the best breed of cow and the dietary energy density to optimise cow productivity and reproduction. It was observed that Swedish Red cows have lower milk production than Holstein cows, and maintain better energy status in early lactation, likely reflecting better reproductive potential when cows are inseminated.

One of the features of pasture-based systems is uncertainty regarding medium term (weeks to months) weather patterns. This is especially true in spring, when herd feed demand is rapidly increasing as cows calve, but prolonged cold weather could markedly reduce grass growth. Imposing an acute (2 week) or chronic (6 week) period of pasture restriction had modest effects on indicators of bioenergetics status (BCS change, blood metabolites, hepatic gene expression), but did not affect indicators of fertility (oestrous cyclicity, oestrous behaviour) or phenotypic fertility performance (submission rates, pregnancy rates). Again, this is important information for a farmer to have when faced with a pasture supply shortage, and allows flexibility in the nutritional management of pasture-based cows during periods of inadequate pasture growth.

After calving, it is difficult to increase cow BCS through nutritional management. As a result, it is vital that cows are managed to attain the optimal BCS at calving. Our analysis indicated that a BCS of 3.25 at calving (1 to 5 scale) is a practical trade-off between minimising subsequent BCS loss, and achieving high fertility performance at subsequent breeding. This is important information for farmers to incorporate into their herd nutritional management, as it sets the cows up for good production and reproduction performance with low BCS loss and hence low incidence of metabolic disorders.

WP5 Multi-site demonstration of reproductive management tools
The main impacts from WP5 in this project has been to produce tools that can be used by the farmer for decision support regarding fertility management.

Tool for insemination worth predictor (IWP)
A decision support tool for an insemination worth predictor (IWP) has been developed in WP5 based on the results found in WP1. IWP is a tool that could be incorporated into any reliable reproductive monitoring tool that measure progesterone in milk.
For demonstration in WP5, IWP is incorporated in the Herd Navigator software. Herd Navigator can, among other measures, determine the level of progesterone in milk. Selected milk samples from an AMS (milk robot) are used to build a progesterone profile for a cow. The system is able to detect increase and decrease in progesterone, and provides the farmer with an alarm when oestrus is approaching. Today the farmer is as a standard SOP advised to inseminate the cows two days after the alarm, the moment when ovulation is due to happen. Using progesterone profile features (e.g. slope and peak), combined with cow traits (e.g. milk yield) IWP is able to provide the farmer with a probability of insemination success between 0 and 1. Studies have shown that cows that have a higher probability value are more likely to get pregnant after insemination (P <0.05). This equation can be implemented on farms that monitor progesterone, and can support the farmer in deciding when to inseminate an oestrus. This will move the focus away from the current paradigm associated with poorer oestrus detection, where each detected oestrus is automatically inseminated regardless of the quality of that oestrus, to the situation of near perfect oestrus detection where the question is: which oestrus is worth inseminating?
WP5 has, combined with WP1, improved our understanding of progesterone profiles and how they can better be used as an on-farm indicator for fertility status. Given that progesterone is 1) easy to measure, 2) follows a fixed pattern during oestrus cycles, and 3) is almost as reliable in milk as in blood, it is one of the most promising indicators for fertility status we have. With the rapid increase of precision farming techniques, the availability of progesterone measures is likely to increase. With the work done on the IWP, this work package has not only described differences in shapes between profiles (Blavy et al. Overview of progesterone profiles in dairy cows, Theriogenology 86.4 (2016): 1061:1071), but also linked the differences in shape to probability of conception after insemination (A paper was written and has been submitted to Journal of Dairy Science (December 2016)). This information will prove to be useful for all future farms with available progesterone measurements.

Tool for reproductive management timing optimizer (RMTO)
The tool RMTO can be used to predict the benefits of a change in fertility management over time. Since it has no fixed management rules, and the rules need to be defined by the farmer in the beginning of the simulation, the model can be used in farm systems of different sizes, with different fertility management, operating in different circumstances. In fine, the RMTO tool will help farmers to take the right decisions about their reproductive management from a technical point of view (economy not currently included in the outputs). Furthermore, this quality of the model makes it not only very useful for use on farm, but also for use in science.
Based on the RPM model developed in WP1, the RMTO tool was developed using a cut-down version of this model called “RPM lite”, which matches both the available on-farm information and the required model response times. In WP5, a user-friendly interface was developed for a web-based product that allows users to design farm-tailored reproductive management protocols. Due to breeding systems diversity, a key component of this design was to ensure that the tool should be applicable in different farming systems, with different management strategies.
For example: one farm might be interested in increasing the longevity of dairy cows, or their lifetime production. Using field studies to find the best methods to achieve this is nearly impossible, given the time span of the trial needed. The RPM herd model and, linked, RMTO graphical user interface GUI can use multiple simulations to find promising strategies within a few hours. The model is not only able to compare many farms, it is also able to do so under different circumstances and/or farming systems. Right now the model is ready for use in the field, and will soon be published on the websites of Växa and Allice.

Progesterone-based benchmarking tool (PBT)
PBT is a tool that could be incorporated into any reliable reproductive monitoring tool. For demonstration in WP5, PBT uses progesterone data from Herd Navigator. Based on the progesterone profiles of the cows of a specific farm, this tool provides the farmers with a benchmark for commencement of luteal activity (CLA) and number of open days (OD) in their herd. The tool helps to provide insight into the fertility status on the farm and make it possible for the farmers to benchmark up against each other. PBT is currently in commercial evaluation by the industry. This tool can in the future be used to help farmers and advisors get insight on the status on the farm, and to select appropriate corrective measures. In order to be able to take corrective measures, it is necessary to know the underlying causes for the delayed CLA and/or increased number of OD. Therefore, a study is has been conducted on six Swedish Herd Navigator herds, where data was collected both from the Swedish cow database and Herd Navigator. These data are used to find underlying causes for delayed commencement of luteal activity and/or increased number of open days. Results of this study will be published, and thus be available for the whole scientific community. This newly created knowledge will help support not only Herd Navigator farmers, but the whole dairy industry.

Beside for the three tools described above a stage-gate model has been developed to fit into a company process and will make it possible to run a smooth process for introduction of new biomarkers into the marked in an automatic in-line and on-farm measurement system.

WP6 Outreach
PROLIFIC results have been disseminated to scientific community. For that PROLIFIC partners have published the project results in peer-reviewed scientific journals and have participated in relevant conferences, symposia or seminars to present the project and its results. There were more than 80 communications during the project such as oral presentations, posters, press releases, distribution of the project flyers, presentation of master thesis, and public defense of PhD thesis. Also PROLIFIC have jointly organized with FECUND EU FP7 project an International Conference, as a one-day satellite workshop of ESDAR 2016, for communication of main results to scientific community and stakeholders. This final event was a success with attendees from 22 countries (mainly Europe, and also the United States of America, New Zealand, Iran and Turkey). The follow-up was done through the documentation (all presentations, etc.) made available on PROLIFIC website.

About technology transfer to stakeholders, most of the results have been identified as not worth to be protected. However, the Progesterone Benchmarking Tool (PBT) developed under the project would potentially lead to a patent. Dissemination of information describing the decision support tools developed in the demonstration workpackage are described under WP5 activity.

About communication with the general public, the main vehicle has been the project website. Brochure was aimed at attracting attention of players potentially interested in the outcomes of the project and it was available also at website. A newsletter with Highlights from research WPs was prepared and was made available at the PROLIFIC web. It meant to report the development of the project to the “primary” circle of stakeholders interested in following the progress of the project. All information from Final Conference was compiled and sent to stakeholders participating in last PROLIFIC annual meeting and was also incorporated into prolific website for making it publicly available.

Further results from experiments performed under PROLIFIC project will come after the end of the project, specifically from the 4 RTD workpackages and at least 20 peer-reviewed scientific papers are under preparation (list in the third periodic report) and will acknowledge the project. PROLIFIC will deliver gain knowledge using different communications means, and the public website will remain active, even after the end of the project to keep communicating the results obtained after the end of the project.
List of Websites:
1.5 Project public website and contact

Dr. Joëlle Dupont
Institut National de la Recherche Agronomique (INRA)
INRA Tours
37380 Nouzilly, France
Phone: +33 (0)2 47 42 77 89
Fax : +33 (0)2 42 42 77 43

Ms. Elodie TAN (Project Manager)
INRA Transfert
3, rue de Pondichéry
F-75015 Paris, France
Phone: +33 (0)1 76 21 62 02

1.6 Project logo
See attachment