at SciVerse ScienceDirect

Theriogenology 79 (2013) 331–343

Contents lists available

Theriogenology

journal homepage: www.ther io journal .com

Data-derived reference profiles with corepresentation of progesterone,estradiol, LH, and FSH dynamics during the bovine estrous cycle

O. Martin a,b,*, N.C. Friggens a,b, J. Dupont c, P. Salvetti d, S. Freret c, C. Rame c, S. Elis c, J. Gatien d,C. Disenhaus e,f, F. Blanc g,h

a INRA, UMR0791 Modélisation Systémique Appliquée aux Ruminants, Paris, FrancebAgroParisTech, UMR Modélisation Systémique Appliquée aux Ruminants, Paris, Francec INRA, UMR085 Physiologie de la Reproduction et des Comportements, Nouzilly, FrancedUNCEIA, Department R&D, Maisons-Alfort, Francee INRA, UMR 348 Physiologie, Environnement et Génétique pour l’Animal et les Systèmes d’Elevage, Saint-Gilles, FrancefAGROCAMPUS OUEST, UMR1348 Physiologie, Environnement et Génétique pour l’Animal et les Systèmes d’Elevage, Saint-Gilles, FrancegClermont Université, VetAgroSup, UMR Herbivores, Clermond-Ferrand, Franceh INRA,UMR1213 Herbivores, Saint-Gènes-Champanelle, France

a r t i c l e i n f o

Article history:Received 4 May 2012Received in revised form 21 September 2012Accepted 29 September 2012

Keywords:CattleEstrousLHFSHProgesteroneEstradiol

* Corresponding author. Tel.: þ33 (0) 1 44 08 1808 18 53

E-mail address: olivier.martin@agroparistech.fr (

0093-691X/$ – see front matter � 2013 Elsevier Inchttp://dx.doi.org/10.1016/j.theriogenology.2012.09.0

a b s t r a c t

Subfertility in cows is often associated with alterations in the hormonal patterns involvedin the regulation of the estrous cycle. Reference profiles are needed to ground modelingprojects aimed at describing these alterations and to develop tools for detecting abnormaldynamics. Various schematic views of LH, FSH, progesterone (P4) and estradiol (E2)patterns have been published but with no clear indication of the extent to which they arederived from real data. The objective of this study was to generate standard profiles for themain reproductive hormones that can be proposed as reliable references to represent thenormal dynamics of these hormones over the estrous cycle. A database of hormonalprofiles was compiled with 40, 23, 33, and 34 profiles for LH, FSH, E2, and P4, respectively,derived from publications in which changes over time of at least three of these fourhormones, including LH, were reported. These profiles were digitalized and standardizedover the time throughout the estrous cycle, considering the interval between twosuccessive LH surges to be 21 days. After this standardization on the x-axis, a trans-formation on the y-axis was performed to center the profiles around their commondynamics. For each hormone, the reference profile was then considered to be the medianof the adjusted profiles. Quartiles were reported to account for the time evolution of thevariability around each reference profile. The reference profiles obtained showed that theprocedure used was satisfactory for extracting the overall changes over time of LH, P4, andE2. Results were less satisfactory for FSH, because of a higher variability observed betweenthe original profiles in our database. The corepresentation of the reference profiles, i.e.,when depicted together on the same scale, emphasizes the interplay between thesehormones more precisely than most of the schematic views available in literature. Thesedata-derived profiles can be considered to be generic and useful for benchmarking thenormal dynamics of gonadotrophins and steroid hormones over the estrous cycle in cow.

� 2013 Elsevier Inc. All rights reserved.

12; fax: þ33 (0) 1 44

O. Martin).

. All rights reserved.25

1. Introduction

Many studies have reported a decrease in the fertilityof dairy cows in the past few decades [1–3] and pointed

O. Martin et al. / Theriogenology 79 (2013) 331–343332

out that subfertility might involve various components ofthe reproductive process such as failure to recover post-partum ovarian activity, poor estrous expression, defec-tive oogenesis, decreased embryo survival, and alteredendometrium functions. All of these problems are asso-ciated with alterations of the hormonal patterns involvedin the regulation of the estrous cycle [4,5]. In this context,the question arises as to what is a normal hormonalprofile in cows?

Numerous studies considered this question mainlywith respect to progesterone (P4) patterns that were re-ported to be indicators of the reproductive status of thecow [6–10]. Data concerning E2, LH, and FSH are much lessfrequently measured on a sufficient number of cowscompared with progesterone data, and they are oftenstudied only during short periods in the cycle [11]. Theexceptions are studies dealing with the development ofsynchronization protocols [12] but, in this case, theprofiles have been artificially modified. So far, there havebeen no data-derived attempts to characterize referenceprofiles throughout the estrous cycle of reproductivehormones such as E2, LH, and FSH.

Using P4 profiles, studies have suggested rules to clas-sify estrous cycles as normal versus altered [8,13]. Usually,these rules rely on the definition of threshold values thatconcern the length of luteal phase and are often deter-mined by visual assessment [10]. Thresholds retained toclassify profiles often differ between studies partly becauseof biological specificities (e.g., suckling vs. dairy cows) butalso because of differences between the hormone assays orthe hormone monitoring methods used (milk vs. bloodsamples, number of measurements per week, chemicalanalyses, etc.) [14–16]. Further, despite the interdepen-dence between reproductive hormones, equivalent rulesets do not exist for the other reproductive hormones. Thus,it is difficult to compare profiles from different experimentsand difficult to fully describe distortions in reproductivehormone profiles.

Such a difficulty is emphasized if we consider thatvariation over time of P4 concentrations influences embryosurvival and might explain fertility problems [17]. Conse-quently, characterizing estrous profiles based only on thelength of luteal phases might be not precise enough. Meieret al. [18] and Gorzecka et al. [19] have suggested definingparameters that account for the variation over time(dynamics) of the hormonal concentrations. Consequently,in order to develop benchmarking reference profiles it willbe necessary to consider the dynamics of reproductivehormones, i.e., to capture the parameters which describethe variability in the shape of hormones profiles. Thanks tomathematical models, some progress has been made toquantify the rise and decline in P4 concentrations over theluteal phase [20,21] and this offers potential referenceparameters for P4 dynamics in milk. However, this has notbeen extended to the other major reproductive hormones,mainly because of the difficulty in measuring plasma and/or milk concentrations of gonadotrophins and E2 over thewhole cycle and over a large number of animals. Littlequantitative information is available for the bovine speciesspecifically addressing the simultaneous time course forchanges in gonadotrophins and ovarian hormone

concentrations [22]. To date, only schematic views havebeen reported to describe the overlapping patterns of themain reproductive hormones [23–25]. However suchschematic representations cannot be said to provide validquantitative reference profiles, especially because theyprovide no representation of the possible variability in theshape of profiles. This is a major limitation for the devel-opment of models that propose to interpret and predict thedynamics of the reproductive cycle based on a representa-tion of the physiological mechanisms involved [26,27].Calibration of these dynamic models relies on input curvesthat can be considered as references for nonalteredhormonal dynamics [28]. A representation of the biologicalvariability observed around reference dynamics is neededfor prediction.

The objective of the present work was to generate data-derived standard profiles for the main hormones involvedin the ovarian cycle (LH, FSH, P4, and E2) that can beconsidered as a reference to represent their codynamic andto evaluate their variability over the estrous cycle.

2. Materials and methods

2.1. Data collection

A database was compiled with 22 studies on hormonesof the estrous cycle in dairy and beef cows [29–50] pub-lished between 1973 and 2010. The articles selected to beincluded in the database each contained a descriptionthrough the complete or partial estrous cycle of the timechanges of plasma concentrations of LH (showinga preovulatory surge) associated with at least two of thethree other main hormones involved in the estrous cycle:FSH, E2, and P4. In the present study, a profile denotesa time series of one hormone over the estrous cycle anda data set denotes a series of at least three hormonalprofiles out of LH, FSH, E2, and P4 for the same individual orgroup mean. When articles were on the influence ofa specific hormonal treatment on hormonal profiles[34,35], only the control group was considered. Data wereextracted from tables [38,42] or digitized from figures re-ported in articles using Windig [51]. Some data sets couldnot be included, for example, when the LH profile did notshow a distinct surge [49] or when profiles belonged toshort or long cycles [47]. This led to a collection of 40 datasets (from one to four data sets per publication) including40, 23, 33, and 34 profiles for LH, FSH, E2, and P4respectively.

2.2. Data standardization

Hormonal plasma concentration units were homoge-nized to ng/mL for LH, FSH, and P4, and to pg/mL forE2 using, when necessary, the following molar weights :LH (C1014H1609N297O294S27), 23530.59 g/mol; FSH(C975H1513N267O304S26), 22673.11 g/mol; P4 (C21H30O2),314.47 g/mol; and E2 (C18H24O2), 272.39 g/mol.

Even though the average cycle duration varies from 21days to approximately 23 days for Holstein cows [9,52,53],data were standardized to provide information on thedynamic hormonal pattern within a standard estrous cycle

Table 1References included in the database.

Reference Animals Time period (days)LH peak at t ¼ 0

Hormone concentration unit

Data seta Breed Parity FSH LH P4 E2 Estrous cycle

Bleach et al., 2001 [48] 2 (2) British Friesian P �5.8 to 10.6 ng/mL ng/mL ng/mL pg/mL SBloch et al., 2006 [47] 1 (35) Holstein P, M �9.2 to 20.0 - ng/mL ng/mL pg/mL SBlödow et al., 1990 [49] 4 (1) Holstein P �3.0 to 21.1 - mg/L nmol/L pmol/L SBousquet, 1989 [46] 1 (?) Gobra ? 18.3 to 21.3 - ng/mL ng/mL pg/mL SBryner et al., 1990 [45] 2 (15) Crossbred beef M �1.7 to 16.4 ng/mL ng/mL ng/mL pg/mL NChenault et al., 1975 [44] 1 (6) Dairy P, M 15.0 to 23.3 - ng/mL ng/mL pg/mL NCooke et al., 1997 [43] 2 (6) Crossbred beef P �1.6 to 12.7 ng/mL ng/mL ng/mL pg/mL N, SCupp et al., 1995 [42] 1 (18) Composite beef M 4.0 to 19.0 ng/mL ng/mL ng/mL pg/mL SDesaulniers et al., 1995 [41] 3 (6) Holstein P, M �6.7 to 8.7 ng/mL ng/mL ng/mL pg/mL SDieleman et al., 1986 [50] 3 (3 to 8) Dutch-Friesian P, M 4.3 to 28.9 - mg/L nmol/L pmol/L NDobson, 1978 [39] 1 (6) Friesan P �0.9 to 1.2 ng/mL ng/mL pg/mL NDobson and Kamonpatana,

1986 [40]1 (?) Dairy ? 0.5 to 25.2 - ng/mL ng/mL pg/mL N

Echternkamp and Hansel,1973 [38]

1 (35) Holstein M �6.0 to 29.0 - ng/mL ng/mL pg/mL N

Eilts and Paccamonti,2004 [37]

1 (?) Dairy cattle ? �7.0 to 28.0 ng/mL ng/mL ng/mL pg/mL N

Evans et al., 1994 [36] 2 (3) Hereford P 2.0 to 23.1 ng/mL ng/mL ng/mL pg/mL NGong et al., 1996 [35] 2 (7) Hereford x Friesan P 1.0 to 43.8 ng/mL ng/mL ng/mL - SKaneko et al., 1992 [34] 2 (5) Japanese brown M �4.6 to 7.0 ng/mL ng/mL ng/mL pg/mL SMalhi et al., 2005 [33] 2 (6 or 7) Crossbred Hereford P, M �1.5 to 21.0 ng/mL ng/mL ng/mL pg/mL SRamirez-Godinez et al.,

1982 [32]2 (5) Polled Hereford M �4.0 to 10.2 ng/mL ng/mL ng/mL - N

Schams et al., 1978 [31] 2 (2) Braunvieh M �15.4 to 41.7 ng/mL ng/mL ng/mL - NVan Eerdenburg, 2010 [30] 1 (?) Dairy ? �2.8 to 25.3 - mg/L nmol/L pmol/mL NWolfenson et al., 2004 [29] 2 (10) Holstein P, M �1.8 to 24.9 - ng/mL ng/mL pg/mL S

Abbreviations: E2, estradiol; M, multiparous; N, natural estrous cycle; P, primiparous; P4, progesterone; S, estrous cycle following a synchronizationtreatment (induced luteal regression).

a Number of data sets per reference and number of animals per profile in parenthesis; data set is a pool of hormonal profiles for the same individual orgroup in the study.

O. Martin et al. / Theriogenology 79 (2013) 331–343 333

of 21 days to be consistent between breeds. The originaltime t, ranging from �15 days to 44 days, with t ¼ 0 cor-responding to the first reported LH peak, was standardizedto the standard time t* using the following adjustment ateach time point: t* ¼ t21 þ D, where t21 ¼ t , 21/L and L isthe estrous cycle length given in the article, derived fromthe interval between LH peaks in the figures, or otherwiseassumed to be equal to 21 days, and D ¼ þ21 if t21 < 0.5,D ¼ �21 if t21 � 21.5, D ¼ �42 if t21 � 42.5, and D ¼ 0otherwise. Segments of the original profile curves werethus adjusted to fall within the range of time t* in therange 0.5 to 21.5.

To derive the reference profile of each hormone fromthe database of profiles that showed variability in both theabsolute concentrations and in the amplitude of theprofiles, original data Yi (denoting LH, FSH, P4, or E2 fordata set i) were transformed to standardized values Yi*using: Yi* ¼ bi þ si , Yi, where bi and si are basal and scalingtransformation coefficients, respectively. The couple (bi, si)was estimated for each profile of each data set i and cor-responded to the transformation that minimized thedeviation of each profile Yi* from the profile YM, defined asthe median of the original profiles. This procedure wasperformed with Scilab [54] and the minimization criteriafor each profile was the sum of the squares of the differ-ences between Yi* and YM over the estrous cycle. Because ofthe shifts between times at which original data wereavailable, the median YM was computed on linearly inter-polated original profiles.

2.3. Derivation of reference profiles

Using the standardized profiles Yi*, derived from theprocedure described above, the reference profile wascalculated as the median YM* of Yi*, and the variability ofthis reference described by the first and third quartiles ofYi*.

To represent these reference profiles for each hormoneon the same graph, the median YM* of the standardizedprofiles Yi* was scaled to lie in the range 0.1 to 1 (using: 1–0.9 , [YM* � M] , [M � m] where M and m are respectivelythe maximum and minimum value of YM* over the estrouscycle).

This set of reference profiles was graphicallycompared with classic schematic views of the bovineestrous cycle gathered from the literature (seeSupplementary Material). These profiles were digitizedfrom figures, standardized to the standard time t* andscaled to lie in the range 0.1 to 1 as described above.Additionally, six data-derived profiles of P4 described inFriggens et al. [7] and Meier et al. [18] were graphicallycompared with the reference profile of P4 proposed inthis work. In these reports, the time t ¼ 0 was the onsetof estrus, which approximately corresponded to 24 hoursafter the time of LH surge defining t* ¼ 0. These profileswere thus shifted by 1 day to be comparable ona consistent time scale. The three profiles of P4 fromFriggens et al. [7] correspond to the 25%, 50%, and 75%quartiles of the distribution of milk P4 data representing

Fig. 1. Original profiles collected from literature during one estrus cycle of (A) lutenizing hormone (LH, N ¼ 22 profiles), (B) follicle stimulating hormone (FSH,N ¼ 13), (C) progesterone (P4, N ¼ 20), and (D) estradiol (E2, N ¼ 18).

O. Martin et al. / Theriogenology 79 (2013) 331–343334

121 lactations from 380 dairy cows. The three profiles ofP4 from Meier et al. [18] correspond to the peak-like(profile reaching indistinguishable peak), flat top-like

(profile reaching a plateau), and structured-like profiles(exhibiting a wave-like pattern) identified by clusteringof plasma data from 27 Holstein dairy cows.

O. Martin et al. / Theriogenology 79 (2013) 331–343 335

3. Results and discussion

3.1. Description of the literature database

The purpose of this study was to derive referenceprofiles for the dynamics of P4, E2, LH, and FSH duringa normal estrous cycle in cows. Although there existsa number of schematic representations of such profiles[23–25], we wanted our reference profiles to be based onreal data, permitting us to obtain at the same time-relevantestimates of the normal variability of such profiles. It wasalso important that these reference profiles were derivedfrom a reasonably wide range of sources to confer a degreeof generalizability to the profiles.

The 40 data sets included in the database (Table 1)contained 18 natural cycles (i.e., cycles not preceded byestrous synchronization) and 22 cycles after an estrousthat had been synchronized. The data sets representedgroup mean (N ¼ 36, two to 35 animals) or individual cow(N ¼ 4) hormone profiles. Animals were from dairy (N ¼26) or beef (N ¼ 14) breeds, and either primiparous (N ¼17), multiparous (N ¼ 14), both (N ¼ 5), or of unknownparity (N ¼ 4). All of the profiles of LH, FSH, P4, and E2from the original publications are shown in Figure 1 andsummary statistics are given in Table 2. The length of theestrous cycle, defined as the interval between twosuccessive LH peaks, ranged from 20.2 to 21.7 in the fivepublications in which it was reported and was otherwiseconsidered equal to 21 days.

As shown in Figure 1, therewas a substantial variation inthe absolute concentration and amplitude of the profiles.This is not surprising because these data come fromdifferent studies and thus different laboratories with all thepotential differences in methods and calibration that thisimplies (see 1. Introduction). This would clearly bea disadvantage if this study was of a comparative nature(e.g., to quantify differences between breeds in reproduc-tive hormone profiles). However, the purpose of this study

Table 2Quantitative description of the untransformed data and transformationcoefficients.

LH(ng/mL)

FSH(ng/mL)

P4(ng/mL)

E2(pg/mL)

DataNumber of references 22 13 20 18Number of data sets 40 23 34 33

Quartiles25% 0.90 0.76 0.81 3.0250% 1.47 12.14 2.73 5.2075% 2.97 40.78 5.81 9.27

Transformation coefficientsa

Basal (bi) coefficient quartiles25% 0.06 0.79 �0.08 2.4250% 0.82 0.87 0.31 3.2875% 1.39 0.91 0.53 3.89

Scaling (si) coefficient quartiles25% 0.26 0.08 0.68 0.3650% 0.60 0.14 0.88 0.4875% 1.69 0.22 1.24 0.65

Abbreviations: E2, estradiol; P4, progesterone.a In the case of FSH, transformationwas performed on data expressed in

percent of the median of each profile.

is to derive reference profiles of these hormones that arebroadly representative of the normal cow population. Inthis context, using data from a wide range of sources is anadvantage, contributing to the generalizability of theprofiles obtained.

3.2. Data standardization

To remove source-specific effects assumed to be largelybecause of methodological issues, the hormonal concen-trations were recalibrated using a procedure to adjust theconcentrations for a given profile according to the devia-tion of that profile from the median profile of the originalprofiles (Fig. 1; Table 1). The median was used rather thanthe mean because the median is more robust to outliersand thus was assumed to better represent the profile ofthe normal population. The resultant standardizedprofiles are shown in Figure 2 over the standard 21-dayestrous cycle.

In the case of FSH, original data range from 0.39 to127.53 ng/mL and are thus shown on a logarithmic scale inFigure 1. Three different subpopulations of original profilescan be seen that differ in orders of magnitude of 1, 10, and100. These differences of level in FSH plasma concentrationare known to be dependent on the assay type chosen andthe standard used [55]. Considering this, the method failedto extract a common profile, mainly because the medianused to tighten profiles was unrealistic. It was thus neces-sary to apply a preliminary rescaling of the original FSHdata. To keep the approach as simple as possible, originaldata for each profile were divided by the median of eachprofile. Afterward, the transformation of FSH was appliedon data sets sharing the same median, arbitrarily rescaledto one.

The procedure used involved adjustment to both thetime and the concentration scales to avoid deformationsin the resulting reference profiles that would haveoccurred by simply averaging the profiles. The comparisonof the original profiles with first the standardized profilesand then with the median profiles indicates that ourapproach was successfully achieved for each hormone(Figs. 1–3).

It is important to state that the reference profiles pre-sented in the present study are representative of the overalldynamic of these hormones during one estrous cycle.However, the standardization procedure used does notallow the capture of the temporal variability that isobserved within individual profiles of FSH, and to a lesserextent E2 (Fig. 2). It is well established that FSH concen-trations are strongly associated with follicular waves [56],whose number per estrous cycle might vary between cows[57]. The variability in the number of follicular waves isclearly apparent in some of the individual profiles of FSH(Fig. 1). Capturing these features would require a far moresophisticated time alignment procedure than the inter-LHpeak interval adjustment used in this study. Ideally, itwould require data in which not only were the 4 hormonesmeasured but also follicular growth (ultrasound of theovaries) to pin point the start and end of follicular waves.Alternatively, it would be possible to identify these pointsfrom a time-series analysis of the wave patterns observed

Fig. 2. Standardized profiles during one estrus cycle of (A) lutenizing hormone (LH, N ¼ 22 profiles), (B) follicle stimulating hormone (FSH, N ¼ 13),(C) progesterone (P4, N ¼ 20), and (D) estradiol (E2, N ¼ 18). Original profiles are adjusted to lie within (0.5–21.5) days from LH peak, standardized for a 21-dayinterval between LH peaks and standardized around the original median profile (see text for standardization procedure).

O. Martin et al. / Theriogenology 79 (2013) 331–343336

in the FSH data. Our preliminary investigation consideringsuch a data analysis method revealed that this was notfeasible with the available data.

The 25% and 75% quartiles of the distributions of thebasal (bi) and scaling (si) transformation coefficientsestimates are given in Table 2. The basal transformation

Fig. 3. Reference profiles, for a 21-day interval between LH peaks, duplicated over two successive estrus cycles of (A) lutenizing hormone (LH), (B) folliclestimulating hormone (FSH), (C) progesterone (P4), and (D) estradiol (E2), defined as the median (in bold) of the standardized profiles shown with the 25% and75% quartiles of the standardized profiles (in gray) (see text for standardization procedure).

O. Martin et al. / Theriogenology 79 (2013) 331–343 337

coefficients were used to correct for differences in theoverall level of original profiles and corresponded toa vertical shift, i.e., an additive value in the unit of the

hormone concentration. The basal transformation effectwas very small for FSH, LH, and P4 and mainly concernedE2. The scaling transformation coefficients were used to

Fig. 4. Scaled reference profiles for a 21-day interval between LH peaks, duplicated over two successive estrus cycles of lutenizing hormone (LH, thin black curve),follicle stimulating hormone (FSH, bold light gray curve), progesterone (P4, thin gray curve), and estradiol (E2, bold dark gray curve), defined as the median of thestandardized profiles scaled to a 0.1 to 1 y-axis (see text for standardization procedure).

O. Martin et al. / Theriogenology 79 (2013) 331–343338

correct for differences in the amplitude of variation oforiginal profiles and corresponded to a vertical flattening(si < 1) or stretching (si > 1). Results showed that thetransformation coefficients for LH, P4, FSH, and E2 arequite distinct. Values observed for E2 and FSH are notablylower than 1 (0.48 and 0.14 respectively) and evokeda moderate and an important flattening of originalprofiles. This result is related to the heterogeneity in thedynamics of the original profiles for these two hormones,and is consistent with the diversity of the number offollicular waves. In the case of LH and P4, estimates of thescaling transformation coefficient are distributed around1, meaning that solutions were found to tighten theoriginal profiles around a common dynamic, either byflattening or stretching the original profiles. These resultsallow us to qualify the variability in the hormonedynamics that appeared to relate mainly to: the ampli-tude for P4 and LH, the shape itself for FSH, and bothamplitude and shape for E2. This is consistent with thefunctional lifespan of the associated reproductive struc-tures giving rise to these dynamics. The shape of FSH isindirectly linked to the follicular dynamics, via the para-crine feedback of E2 and inhibin on the anterior pituitarygland. In turn, FSH modulates the rhythm of follicularwaves. The FSH dynamic is thus a highly distortablesignal. The LH dynamic is a quasi-stationary signalinterrupted by tremendous peaks, that shape most of thepattern. The amplitude and shape for E2 are directlylinked to the number and stages of follicles and thenumber of follicular waves, making the E2 dynamics

rather irregular. Though the P4 dynamic is directly linkedto the regular development of the corpus luteum, itsamplitude is directly linked to the systemic clearancewhich can vary considerably from one individual toanother.

In this work, the unit considered for the adjustmentprocedure was the profile and not the data set or the study.All profiles were thus considered as being members ofa common collection of curves, regardless of their source.Because the focus of this study was to capture a commondynamic pattern for each hormone, our procedure did notpreserve the variability between profiles. Thus, the intra-study variability that we assumed to be mainly because ofbiological factors, and the interstudy variability that weassumed to be largely because of differences in experi-mental designs andmeasurementmethods, were voluntaryremoved from our reference profiles to preserve only thevariability pertaining to the dynamics. Despite a sizeablebody of literature that describes the effects of factors suchas energy balance [58–60], milk yield [9,61,62], metritis[63], or endometritis [64–66] on reproductive performance,few studies reported their effects on the reproductivehormone profiles. With respect to P4, metritis was shownto modify the shape of the profile, with a tendency towardlower P4, and flatter luteal phase profiles [19]. However, theeffect of metritis only accounted for 10% of the observedvariability in profile shapes [19]. In that study, breed andparity had no significant effect on the shape of the P4profile during estrous cycles although the length of lutealphases was shown to be affected by breed [9,19].

Fig. 5. Reference profiles of progesterone (P4) on a 0.1 to 1 scale, for a 21-day interval between LH peaks, duplicated over two successive estrus cycles (bold linewith 25%–75% interquartile area in gray) shown together with the three P4 profiles (thin gray lines) redrawn from (A) Friggens et al. (2008) [7] and (B) Meier et al.(2009) [18] (see text for procedure of reference derivation and literature references).

O. Martin et al. / Theriogenology 79 (2013) 331–343 339

3.3. Reference profiles

The median and quartiles of the standardized profilesare presented in Figure 3, inwhich the 21-day dynamics areduplicated over two successive cycles to highlight thetransition around the LH surge. These profiles show thatthe standardization procedure was satisfactory forsynthesizing the overall dynamic of these four hormonesduring one estrous cycle.

Several key time points (in days from LH peak) can beextracted from these reference profiles. The nadir of LHoccurs on day 12 (t* ¼ 12.5) and the LH surge (arbitrarydefined here as LH > 0.2; scale, 0.1–1) occurs during days20 to 22 (t* in the range 20.1–21.6). The FSH surge globallyoccurs during days 20 to 23 and is characterized by a firstpeak on day 21 (t*¼ 21) with an exact synchronizationwiththe LH peak, followed by a second peak on day 22 (t* ¼21.7). A second peak of FSH is not always reported in theliterature, probably because of infrequent sampling. It hasbeen well established that the synchronous surges of LH

and FSH are involved in ovulation and development of thecorpus luteum [67], but understanding is still lacking aboutthe function of the second peak of FSH. In ewes and cows itwas described to occur 4 to 24 hours after the LH peak, i.e.,around the time of ovulation [68–70], but the cause of thispeak has not been determined. The nadir of P4 level and theonset of the rise in P4 occur on Day 1 (t* ¼ 0.8). Themaximal value of P4 is reached on Day 14 (t*¼ 14.0) and P4is higher than 90% of the maximumvalue from Day 11 to 16(t* in the range 11.4–16.4). The decrease in P4 pertaining toluteal regression starts after Day 14 and is established onDay 17. The profile of E2 clearly shows a small surgebetween Days 2 to 7 with a peak on Day 5 (t* ¼ 5.1). Thenadir of E2 is thereafter reached on Day 11 (t* ¼ 10.7) andthe preovulatory E2 surge occurs globally fromDay 17 to 22with a peak value on Day 21 (t* ¼ 20.7). The intervalbetween peaks of E2 and LH thus equals 8 to 9 hours, whichis consistent with the result reported by Walters et al. [68]who observed the highest E2 concentrations 6 to 8 hoursbefore the LH surge.

Fig. 6. Reference profiles, for a 21-day interval between LH peaks, duplicated over two successive estrus cycles of (A) lutenizing hormone (LH), (B) folliclestimulating hormone (FSH), (C) progesterone (P4), and (D) estradiol (E2) shown together with literature schematic profiles on a 0.1 to 1 y-axis (see text forprocedure of reference derivation and literature references).

O. Martin et al. / Theriogenology 79 (2013) 331–343340

These are, to our knowledge, the first reference profilesfor these hormones in cattle that have been calculated froma broad range of observed data in which each data source

provided profiles of at least three of the four hormones inquestion. The fact that each study measured at least threeof the hormones allows these reference profiles to be used

O. Martin et al. / Theriogenology 79 (2013) 331–343 341

not only independently but also to reference the correlatedchanges in these four hormones. The medians of the stan-dardized profiles scaled to 0.1 to 1 for LH, FSH, P4, and E2are presented together in Figure 4 (duplication of the 21-day dynamics over two successive cycles). It is interestingto note that the pronounced increase in E2 occurs duringthe decline from peak of P4 and that this is followed by theincrease in FSH and LH. Plasma E2 concentrations areelevated before the dual surge of LH and FSH, but theconcentration of this steroid has declined before the secondsurge of FSH. There is no doubt today that the preovulatoryLH surge is related to previous E2 secretion in domesticanimals. An increase in plasma E2 precedes LH surge innatural conditions.

Accepting that these profiles can be considered asreference profiles, at least until larger data sets that havemeasured these four reproductive hormones becomeavailable, provides a useful framework against which tocompare particular profiles. In this context, it is clear thatthere is a need to be able to quantify the effects of potentialperturbing factors on reproductive physiology, and thisrequires a reference for normal estrous cycles. For example,Royal et al. [1,71,72] examined the effects of genetic selec-tion for milk production, and the associated performancechanges, on fertility as indexed by changes in the types ofP4 profile relative to a base population measured in the1970s. Because of recent technological advances, P4 profilescan now be routinely measured on commercial farms [73],and thus offer the opportunity to carry out herd leveldiagnosis of reproductive performance and management.In order to benchmark the profiles of a particular herd thereis a need for a reference profile. This study provides a firstbasis for developing such benchmarks. However, it shouldbe noted that the present profiles reflect the particularconditions of the studies fromwhich they came and as suchshould not be assumed to represent all conditions and allpopulations.

Even though the references profiles we obtained fromthe standardization procedure appear quite relevant froman expert point of view, the question arises of validation.This is not an easy task because the number of articlesreporting multiple profiles is limited. The consequence isthat we could not reserve a subset of data to provideindependent validation. Although we could have carriedout cross-validation, this is only a limited form of validationespecially because of the inherent scale differencesbetween individual studies. It is, however, possible tocompare single hormonal profiles, and in particular P4profiles. A number of studies have analyzed P4 profilesfrom measurements in milk, on relatively large numbers ofanimals [1,7,74]. In the study of Friggens et al. [7], theaverage profile of 120 estrous cycles that precededsuccessful inseminationwas presented. Figure 5 shows thisprofile, derived frommilk P4 measurements, plotted on thesame scale as the median adjusted P4 profile obtained fromthe present study (based on plasma measures). A very closeagreement between the two can be observed. In Figure 5,the peak-like, flat top-like, and structured-like P4 profilesidentified by clustering of plasma data from 27 Holstein-Friesian dairy cows by Meier et al. [18] are also presented.The variability between these shape profiles is close to the

interquartile variability around our reference profile. Thisprovides validation, for P4, that the median profile in thisstudy can be considered as a useful reference curve, forinstance to calibrate or validate models. With respect to theother reproductive hormones, no reliable, independent,data-derived profiles were found in the literature we haveanalyzed, so that the only comparison we made is withclassic schematic views of the bovine estrous cycle gath-ered from the literature (see Supplementary Material).These schematic profiles (Fig. 6) appear to be quite differentwhen they are plotted on a same figure with similar level(y-axis) and time scales. Such variability reveals that whendrawing a representation of reproductive hormone profiles,authors might have different views to account for thedynamics and simplify them. The statistical comparisonbetween such schematic views and our reference profiles isnot possible because there would be no way of ascertainingwhether the difference was significant. To use referenceprofiles to illustrate the dynamics of reproductivehormones over the bovine estrous cycle, we consider thatone would be forced to conclude that the data-drivenprofile is the most valid.

There is today a need to be able to predict how thereference profile will be deformed or altered according tothe known factors that can perturb reproductive perfor-mance [75]. This is necessary if the farm advisor wishesto interpret any discrepancies between an observed anda benchmark profile. Being able to predict quantitativechanges in the dynamics of reproductive hormonesaccording to various perturbing factors implies devel-oping simulation models of reproductive physiology. Thisneed has been first recognized for humans [76–78] andnow for farm animals [27,79]. Initiatives to do this arenow under way [26,28]. However, validating andextending such models requires robust reference profiles.For example, Selgrade et al. [78] clearly showed how thedata sets used for estimating the parameters of theirmodel for hormonal control of the menstrual cycle mightinfluence the ability of the model to simulate normal andaltered cycles in women. With this study, we believe wehave taken a first step toward providing such referenceprofiles for cows.

4. Conclusion

The data-derived reference profiles presented in thisstudy give a coherent representation of the main repro-ductive hormone dynamics when compared with most ofthe schematic views available in literature and to P4median profiles or models resulting from the analysis oflarge data sets. These data-derived profiles can then beconsidered to be generic and useful for the development ofbenchmarking procedures of the normal dynamics ofgonadotrophins and steroids hormones over the estrouscycle in cow. Furthermore, they provide reference profilesto be used in modeling approaches.

Acknowledgments

The authors thank the division of Animal Physiology andLivestock Systems of INRA for supporting this study.

O. Martin et al. / Theriogenology 79 (2013) 331–343342

Appendix A. Supplementary data

Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.theriogenology.2012.09.025.

References

[1] Royal MD, Darwash AO, Flint APF, Webb R, Woolliams JA,Lamming GE. Declining fertility in dairy cattle: changes in traditionaland endocrine parameters of fertility. Anim Sci 2000;70:487–501.

[2] Lucy MC. Reproductive loss in high-producing dairy cattle: wherewill it end? J Dairy Sci 2001;84:1277–93.

[3] Dobson H, Walker SL, Morris MJ, Routly JE, Smith RF. Why is itgetting more difficult to successfully artificially inseminate dairycows? Animal 2008;2:1104–11.

[4] Kesler DJ, Garverick HA, Bierschwal CJ, Elmore RG, Youngquist RS.Reproductive hormones associated with normal and abnormalchanges in ovarian follicles in postpartum dairy cows. J Dairy Sci1979;62:1290–6.

[5] Båge R, Gustafsson H, Larsson B, Forsberg M, Rodríguez-Martínez H.Repeat breeding in dairy heifers: follicular dynamics and estrouscycle characteristics in relation to sexual hormone patterns. Ther-iogenology 2002;57:2257–69.

[6] Lamming GE, Darwash AO. The use of milk progesterone profiles tocharacterise components of subfertility in milked dairy cows. AnimReprod Sci 1998;52:175–90.

[7] Friggens NC, Bjerring M, Ridder C, Højsgaard S, Larsen T. Improveddetection of reproductive status in dairy cows using milk proges-terone measurements. Reprod Domest Anim 2008;43(Suppl 2):113–21.

[8] Cutullic E, Delaby L, Gallard Y, Disenhaus C. Dairy cows’ reproduc-tive response to feeding level differs according to the reproductivestage and the breed. Animal 2011;5:731–40.

[9] Disenhaus C, Cutullic E, Blanc F, Agabriel J. Breed comparison of postpartum ovarian activity in cows. J Dairy Sci 2009;92(Suppl 1):498.

[10] Mann GE, Keatinge R, Hunter M, Hedley BA, Lamming GE. The use ofmilk progesterone to monitor reproductive function in beef sucklercows. Anim Reprod Sci 2005;88:169–77.

[11] Ginther O, Bergfelt D, Kulick L, Kot K. Pulsatility of systemic FSH andLH concentrations during follicular-wave development in cattle.Theriogenology 1998;50:507–19.

[12] Stevenson JS. Progesterone, follicular, and estrual responses toprogesterone-based estrus and ovulation synchronization protocolsat five stages of the estrous cycle. J Dairy Sci 2008;91:4640–50.

[13] Bulman DC, Lamming GE. Milk progesterone levels in relation toconception, repeat breeding and factors influencing acyclicity indairy cows. J Reprod Fertil 1978;54:447–58.

[14] Nebel RL, Whittier WD, Cassell BG, Britt JH. Comparison of on-farmlaboratory milk progesterone assays for identifying errors indetection of estrus and diagnosis of pregnancy. J Dairy Sci 1987;70:1471–6.

[15] Colazo MG, Ambrose DJ, Kastelic JP, Small JA. Comparison of 2enzyme immunoassays and a radioimmunoassay for measurementof progesterone concentrations in bovine plasma, skim milk, andwhole milk. Can J Vet Res 2008;72:32–6.

[16] Posthuma-Trumpie GA, van Amerongen A, Korf J, van Berkel WJH.Perspectives for on-site monitoring of progesterone. Trends Bio-technol 2009;27:652–60.

[17] Mann GE, Payne JH, Lamming GE. Hormonal regulation of oxytocin-induced prostaglandin F2alpha secretion by the bovine and ovineuterus in vivo. Domest Anim Endocrin 2001;21:127–41.

[18] Meier S, Roche JR, Kolver ES, Boston RC. A compartmental modeldescribing changes in progesterone concentrations during the oes-trous cycle. J Dairy Res 2009;76:249–56.

[19] Gorzecka J, Callesen H, Pedersen KM, Friggens NC. The relationshipbetween postpartum vaginal discharge symptoms and progesteroneprofile characteristics in lactating dairy cows in Denmark. Ther-iogenology 2011;75:1016–28.

[20] Friggens NC, Chagunda MGG. Prediction of the reproductive statusof cattle on the basis of milk progesterone measures: modeldescription. Theriogenology 2005;64:155–90.

[21] Meier S, Roche JR, Kolver ES, Verkerk GA, Boston RC. Comparingsubpopulations of plasma progesterone using cluster analyses. JDairy Sci 2009;92:1460–8.

[22] Bergfeld EG, Kojima FN, Cupp AS, Wehrman ME, Peters KE,Mariscal V, et al. Changing dose of progesterone results in sudden

changes in frequency of luteinizing hormone pulses and secretion of17 beta-estradiol in bovine females. Biol Reprod 1996;54:546–53.

[23] Peters AR, Ball PJH. Reproduction in cattle. Second Edition. Oxford,Blackwell Science; 1995.

[24] Forde N, Beltman ME, Lonergan P, Diskin M, Roche JF, Crowe MA.Oestrous cycles in Bos taurus cattle. Anim Reprod Sci 2011;124:163–9.

[25] Perry G. The bovine oestrus cycle. South Dakota State University,Cooperative Extension Service, USDA. 2004. Available at: http://pubstorage.sdstate.edu/AgBio_Publications/articles/FS921A.pdf.

[26] Boer HMT, Stötzel C, Röblitz S, Deuflhard P, Veerkamp RF,Woelders H. A simple mathematical model of the bovine estrouscycle: follicle development and endocrine interactions. J Theor Biol2011;278:20–31.

[27] Vetharaniam I, Peterson AJ, McNatty KP, Soboleva TK. Modellingfemale reproductive function in farmed animals. Anim Reprod Sci2010;122:164–73.

[28] Boer HMT, Röblitz S, Stötzel C, Veerkamp RF, Kemp B, Woelders H.Mechanisms regulating follicle wave patterns in the bovine estrouscycle investigated with a mathematical model. J Dairy Sci 2011;94:5987–6000.

[29] Wolfenson D, Inbar G, Roth Z, Kaim M, Bloch A, Braw-Tal R. Follic-ular dynamics and concentrations of steroids and gonadotropins inlactating cows and nulliparous heifers. Theriogenology 2004;62:1042–55.

[30] Van Eerdenburg F. Oestrous detection and fertility management. In:Valdmann A, editor. Reproductive challenges facing today’s dairyindustry. Estonia, Tartu: Estonian University of Life SciencesSummer School, Estonian Institute of Veterinary Medicine andAnimal Science; May 2010. p. 24–8.

[31] Schams D, Schallenberger E, Menzer C, Stangl J, Zottmeier K,Hoffmann B, et al. Profiles of LH, FSH and progesterone in post-partum dairy cows and their relationship to the commencement ofcyclic functions. Theriogenology 1978;10:453–68.

[32] Ramirez-Godinez JA, Kiracofe GH, Schalles RR, Niswender GD.Endocrine patterns in the postpartum beef cow associated withweaning: a comparison of the short and subsequent normal cycles. JAnim Sci 1982;55:153–8.

[33] Malhi PS, Adams GP, Singh J. Bovine model for the study ofreproductive aging in women: follicular, luteal, and endocrinecharacteristics. Biol Reprod 2005;73:45–53.

[34] Kaneko H, Watanabe G, Taya K, Sasamoto S. Changes in peripherallevels of bioactive and immunoreactive inhibin, estradiol-17 beta,progesterone, luteinizinghormone, and follicle-stimulating hormoneassociated with follicular development in cows induced to super-ovulate with equine chorionic gonad. Biol Reprod 1992;47:76–82.

[35] Gong JG, Campbell BK, Bramley TA, Gutierrez CG, Peters AR, Webb R.Suppression in the secretion of follicle-stimulating hormone andluteinizing hormone, and ovarian follicle development in heiferscontinuously infused with a gonadotropin-releasing hormoneagonist. Biol Reprod 1996;55:68–74.

[36] Evans AC, Adams GP, Rawlings NC. Endocrine and ovarian follicularchanges leading up to the first ovulation in prepubertal heifers. JReprod Fertil 1994;100:187–94.

[37] Eilts B, Paccamonti D. The bovine estrous cycle. Comparative Ther-iogenology, Louisiana State University, Department of VeterinaryClinical Sciences, VMED 5361. 2004. Available at: http://www.vetmed.lsu.edu/eiltslotus/theriogenology-5361/the_bovine_estrous_cycle.htm.

[38] Echternkamp SE, Hansel W. Concurrent changes in bovine plasmahormone levels prior to and during the first postpartum estrouscycle. J Anim Sci 1973;37:1362–70.

[39] Dobson H. Plasma gonadotrophins and oestradiol during oestrus inthe cow. Reproduction 1978;52:51–3.

[40] Dobson H, Kamonpatana M. A review of female cattle reproductionwith special reference to a comparison between buffaloes, cows andzebu. J Reprod Fertil 1986;77:1–36.

[41] Desaulniers DM, Lussier JG, Goff AK, Bousquet D, Guilbault LA.Follicular development and reproductive endocrinology duringa synchronized estrous cycle in heifers and mature cows displayingcontrasting superovulatory responses. Domest Anim Endocrin1995;12:117–31.

[42] Cupp AS, Stumpf TT, Kojima FN, Werth LA, Wolfe MW, Roberson MS,et al. Secretion of gonadotrophins change during the luteal phase ofthe bovine oestrous cycle in the absence of corresponding changes inprogesterone or 17ß-oestradiol. Anim Reprod Sci 1995;37:109–19.

[43] Cooke DJ, Crowe MA, Roche JF. Circulating FSH isoform patternsduring recurrent increases in FSH throughout the oestrous cycle ofheifers. J Reprod Fertil 1997;110:339–45.

O. Martin et al. / Theriogenology 79 (2013) 331–343 343

[44] Chenault JR, Thatcher WW, Kalra PS, Abrams RM, Wilcox CJ. Tran-sitory changes in plasma progestins, estradiol, and luteinizinghormone approaching ovulation in the bovine. J Dairy Sci 1975;58:709–17.

[45] Bryner RW, Garcia-Winder M, Lewis PE, Inskeep EK, Butcher RL.Changes in hormonal profiles during the estrous cycle in oldlactating beef cows. Domest Anim Endocrin 1990;7:181–9.

[46] Bousquet D. Aspect hormonal du cycle chez la vache. In: Mieuxmaîtriser la reproduction des espèces domestiques par le transfertd’embryons. Senegal, Dakar: Sommet de la Francophonie, JournéesScientifiques; May 2–11, 1989.

[47] Bloch A, Folman Y, Kaim M, Roth Z, Braw-Tal R, Wolfenson D.Endocrine alterations associated with extended time intervalbetween estrus and ovulation in high-yield dairy cows. J Dairy Sci2006;89:4694–702.

[48] Bleach EC, Glencross RG, Feist SA, Groome NP, Knight PG. Plasmainhibin A in heifers: relationship with follicle dynamics, gonadotro-pins, and steroids during the estrous cycle and after treatment withbovine follicular fluid. Biol Reprod 2001;64:743–52.

[49] Blödow G, Rommel P, Kanitz E, Kitzig M, Rodewald S, Schneider F.Investigations on the blood plasma hormone profile in heifersfollowing synchronization of the cycle with PGF2a. Arch Exp Vet-erinarmed 1990;44:599–610.

[50] Dieleman SJ, Bevers MM, Van tol H TM, Willemse AH. Peripheralplasma concentrations of oestradiol, progesterone, cortisol, LH andprolactin during the oestrous cycle in the cow, with emphasis onthe peri-oestrous period. Anim Reprod Sci 1986;10:275–92.

[51] Lovy D. Windig data digitizer. 1996. Available at: http://www.unige.ch/sciences/chifi/cpb/windig.html.

[52] Sartori R, Haughian JM, Shaver RD, Rosa GJM, Wiltbank MC. Compar-ison of ovarian function and circulating steroids in estrous cycles ofHolstein heifers and lactating cows. J Dairy Sci 2004;87:905–20.

[53] Sakaguchi M, Sasamoto Y, Suzuki T, Takahashi Y, Yamada Y. Post-partum ovarian follicular dynamics and estrous activity in lactatingdairy cows. J Dairy Sci 2004;87:2114–21.

[54] Consortium Scilab. Digiteo. Scilab: Logiciel open source gratuit decalcul numérique, (OS, Version 5.3.3). 2011. Available at: http://www.scilab.org.

[55] Rose MP, Gaines Das RE, Balen AH. Definition and measurement offollicle stimulating hormone. Endocr Rev 2000;21:5–22.

[56] Schams D, Schallenberger E, Hoffmann B, Karg H. The oestrous cycleof the cow: hormonal parameters and time relationships concerningoestrus, ovulation, and electrical resistance of the vaginal mucus.Acta Endocrinol 1977;86:180–92.

[57] Adams GP, Jaiswal R, Singh J, Malhi P. Progress in understandingovarian follicular dynamics in cattle. Theriogenology 2008;69:72–80.

[58] Beever DE. The impact of controlled nutrition during the dry periodon dairy cow health, fertility and performance. Anim Reprod Sci2006;96:212–26.

[59] Garnsworthy PC, Lock A, Mann GE, Sinclair KD, Webb R. Nutrition,metabolism, and fertility in dairy cows: 1. Dietary energy sourceand ovarian function. J Dairy Sci 2008;91:3814–23.

[60] Roche JF. The effect of nutritional management of the dairy cow onreproductive efficiency. Anim Reprod Sci 2006;96:282–96.

[61] Buckley F, O’Sullivan K, Mee JF, Evans RD, Dillon P. Relation-ships among milk yield, body condition, cow weight, andreproduction in spring-calved Holstein-Friesians. J Dairy Sci2003;86:2308–19.

[62] Wathes DC, Fenwick M, Cheng Z, Bourne N, Llewellyn S, Morris DG,et al. Influence of negative energy balance on cyclicity and fertilityin the high producing dairy cow. Theriogenology 2007;68(Suppl 1):S232–41.

[63] Williams EJ, Sibley K, Miller AN, Lane EA, Fishwick J, Nash DM, et al.The effect of Escherichia coli lipopolysaccharide and tumournecrosis factor alpha on ovarian function. Am J Reprod Immunol2008;60:462–73.

[64] LeBlanc SJ, Duffield TF, Leslie KE, Bateman KG, Keefe GP, Walton JS,et al. Defining and diagnosing postpartum clinical endometritis andits impact on reproductive performance in dairy cows. J Dairy Sci2002;85:2223–36.

[65] McDougall S, Hussein H, Aberdein D, Buckle K, Roche J, Burke C,et al. Relationships between cytology, bacteriology and vaginaldischarge scores and reproductive performance in dairy cattle.Theriogenology 2011;76:229–40.

[66] McDougall S, Macaulay R, Compton C. Association betweenendometritis diagnosis using a novel intravaginal device andreproductive performance in dairy cattle. Anim Reprod Sci 2007;99:9–23.

[67] Schwartz NB, McCormack CE. Reproduction: gonadal function andits regulation. Annu Rev Physiol 1972;34:425–72.

[68] Walters DL, Schams D, Schallenberger E. Pulsatile secretion ofgonadotrophins, ovarian steroids and ovarian oxytocin during theluteal phase of the oestrous cycle in the cow. J Reprod Fertil 1984;71:479–91.

[69] Schams D, Karg H. Radioimmunologic determination of LH in bovineserum, with special attention to the estrous cycle. Acta Endocrinol1969;61:96–103.

[70] Kiddy CA, Odell WA. Radioimmunoassay of blood LH at estrus andovulation in cattle [Abstract]. J Anim Sci 1969;29:192.

[71] Royal MD, Flint APF, Woolliams JA. Genetic and phenotypic rela-tionships among endocrine and traditional fertility traits andproduction traits in Holstein-Friesian dairy cows. J Dairy Sci 2002;85:958–67.

[72] Royal MD, Pryce JE, Woolliams JA, Flint APF. The genetic relationshipbetween commencement of luteal activity and calving interval,body condition score, production, and linear type traits in Holstein-Friesian dairy cattle. J Dairy Sci 2002;85:3071–80.

[73] Friggens NC, Løvendahl P. The potential of on-farm fertility profiles: in-line proges-terone and activity measurements. In: Royal MD, Friggens NC, Smith RF, editors.Fertilityindairycows:bridgingthegaps.BSAS,Cambridge,UK:CambridgeUniversityPress; 2008. p. 72–8.

[74] Petersson KJ, Berglund B, Strandberg E, Gustafsson H, Flint APF,Woolliams JA, et al. Genetic analysis of postpartum measures ofluteal activity in dairy cows. J Dairy Sci 2007;90:427–34.

[75] Friggens NC, Disenhaus C, Petit HV. Nutritional sub-fertility in thedairy cow: towards improved reproductive management througha better biological understanding. Animal 2010;4:1197–213.

[76] Reinecke I, Deuflhard P. A complex mathematical model of thehuman menstrual cycle. J Theor Biol 2007;247:303–30.

[77] Selgrade JF, Schlosser PM. A model for the production of ovarianhormones during the menstrual cycle. Fields Inst Commun 1999;21:429–46.

[78] Selgrade JF, Harris LA, Pasteur RD. A model for hormonal control ofthe menstrual cycle: structural consistency but sensitivity withregard to data. J Theor Biol 2009;260:572–80.

[79] Blanc F, Martin GB, Bocquier F. Modelling reproduction in farmanimals: a review. Reprod Fertil Dev 2001;13:337–53.

Online Supplementary MaterialSchematic views of reproductive hormones profiles (n ¼ 11) redrawn from references cited below

Reference of original schematic view Digitized and rescaled schematic view

Advanced Animal Technology Ltd., Hamilton, New Zealandhttp://www.aat.co.nz/images/graph123.gif

Ennuyer M 2000. Follicular growth in cows: practical aspectsfor management of breeding.Point Vétérinaire 31, 9-15.

O. Martin et al. / Theriogenology 79 (2013) 331–343343.e1

(continued )

Reference of original schematic view Digitized and rescaled schematic view

Forde, N., Beltman, M.E., Lonergan, P., Diskin, M., Roche, J.F. &Crowe, M. a. (2011) Oestrous cycles in Bos taurus cattle. Animalreproduction science, 124, 163-9.

Learning Reproduction in Farm Animals - Cow and Mare Estrous Cycles.Oklahoma State University. Adapted from Ginther, O.J. 1992. 2nd EditionReproductive Biology of the Mare. p288http://animalsciences.missouri.edu/reprod/Notes/estrous/estrous.htm

Ownby, 2001 Oklahoma State University College of Veterinary Medicinehttp://instruction.cvhs.okstate.edu/histology/fr/HiFRp15.htm

Chapter 5 The Estrous Cycle - by Choong-Saeng ParkGyeongsang National University (South Korea)http://nongae.gsnu.ac.kr/wcspark/teaching/chap5.html

(continued on next page)

O. Martin et al. / Theriogenology 79 (2013) 331–343 343.e2

(continued )

Reference of original schematic view Digitized and rescaled schematic view

New Bolton Center Field Service Department, University of PennsylvaniaSchool of Veterinary Medicine’s Computer Aided Learning programhttp://cal.vet.upenn.edu/projects/fieldservice/Dairy/REPRO/estfolwv.htm

THE BOVINE ESTROUS CYCLE, 2004 - George Perry, Extension BeefReproduction Management Specialist, South Dakota State University,Cooperative Extension Service, USDAhttp://agbiopubs.sdstate.edu/articles/FS921A.pdf

Peters A and Lamming E 1983. Hormone patterns and reproduction incattle. In Practice 5, 153-158.

Ponsart, 2003. Le cycle oestral. BTIA 110, 20-22. D’après GAMEPI,CD Rom AFC

O. Martin et al. / Theriogenology 79 (2013) 331–343343.e3

(continued )

Reference of original schematic view Digitized and rescaled schematic view

Lee Rinehart, 2009: Dairy Production on Pasture: An Introduction toGrass-Based and Seasonal Dairying, Publication of ATTRA, NationalSustainable Agriculture Information Service, Graph courtesy of Universityof Missouri Extensionhttp://attra.ncat.org/attra-pub/PDF/grassbaseddairy.pdf

O. Martin et al. / Theriogenology 79 (2013) 331–343 343.e4