Ontogeny of suckling behaviour in cattle-a model for several mammalian species ?

Suckling behaviour of nine beef cow-calf-pairs (Hereford and Charolais) and five dairy cow-calf-pair's (Polish Holstein-Friesian) was video-recorded from day one pp to day 70 (beef) or 82 (dairy) post partum (pp) regularly. Cows were only suckled during that time and milk yield could not be recorded. Bout was defined as all activities of the calf from the first contact between the mouth of the calf at a specific teat and the last contact at this teat, i.e., as a small sub-unit of a meal. Number of suckled teats, uniformity of relative shares of suckled teats, number of bouts, average bout and relative number of bout-teat-repetitions described the course of ontogeny of suckling behaviour. The first model included the effects breed, cow (breed), and day pp as regression. The analysis revealed the following: (i) number of teats, uniformity of relative shares between suckled teats, and number of bouts increased with increasing day pp. (ii) Average bout, relative number of bout-teat repetitions and mealbout-relationship decreased with increasing day pp. (iii) The same developments could be observed in both breeds, however, significantly slower in the dairy breed. The second model included the effects breed, cow (breed) and average bout as regression, representing milk yield. Breed changed from a very powerful into an almost negligible effect in that second analysis. The implications of the results for a wider group of mammalian species are discussed.


Introduction
Calf suckling behaviour was recorded within the framework of developing a physiological milk harvest process.Substantial differences were observed between suckling behaviour shortly after birth and a few weeks later.Therefore the ontogeny of suckling behaviour was in the focus of the research reported here.In a pre-study MAYNTZ (1996) proposed the ontogeny of suckling behaviour of Bos taurus being completed when the behaviour did not change qualitatively any more.The sequence of events listed below was essential for the course of ontogeny in that proposal.Those events were: (i) a bout pattern reflecting the milk flow in the gland (MAYNTZ and COSTA, 1998) had developed (for details see MAYNTZ et al., 2006); (ii) regular and frequent teat changes occurred between the suckled teats, (iii) all teats were suckled during a suckling meal (referred to as "meal" in the following), and (iv) the teats were suckled equally.Data were collected to describe the ontogeny of suckling behaviour in cattle.The variables evaluated here, and the description of ontogeny resulting from that, were the final steps of a long process.The length of that process resulted from the two following facts: Only marginal hints about the beginning but not the course of ontogeny could be taken from the literature (e.g.SELMAN et al., 1970;EDWARDS and BROOM, 1982;LANGHOLZ et al., 1987;VENTORP and MICHANEK, 1991;SCHÄFFER et al., 1999) and the research consumed an unforeseeable amount of time.The long period of investigation itself resulted from the methods chosen.A lot of time had to be spent to record 193 voluntary suckling meals or 70 hours videotape, to translate the videotapes into quantitative data-files and to test different graphical translation of the data-files into figures showing the main characteristics of a suckling meal at one glance.Understanding biological processes by viewing (LORENZ, 1959) is an accepted form of gaining knowledge; however, only if proper testing of hypotheses follows -as we would state.In all preliminary evaluations, the beef and dairy breeds differed significantly only concerning quantitative aspects of ontogeny and the dairy breed animals consumed a much longer period of time to reach the end of ontogeny of suckling behaviour described above.Viewing of the meal-graphs gave the impression, that the ending period of the dairy meals resembled the starting period of beef meals.Milk yield was higher in the dairy breed and milk yield generally should be higher in the starting period because the big number of short bouts during the long after-stimulation belongs to the ending period of a meal (MAYNTZ and COSTA, 1998).Thus the idea arose that milk yield could be a "common currency" between the two breeds.MAYNTZ and COSTA (1998) had defined a bout as the period of "... all activities of the calf from the first contact between the mouth of the calf at a specific teat and the last contact at this teat, if there is suckling between these two events" i.e., as a small sub-unit of a meal.They showed e.g., that a pharmacological blockage of ejection resulted in very small bouts throughout an entire meal or that a pharmacologically induced additional ejection of residual milk during the partially non-nutritive afterstimulation (LIDFORS et al., 1994) resulted in a consecutive short series of prolonged bouts.Thus, longer bouts indicate more milk in the corresponding cistern and vice versa.It had not been feasible to record milk yield when the data analysed here were collected, but average bout length could represent the individual differences in milk yield.Therefore the description of the course of ontogeny given here falls into two parts: The evaluation before and after the hypothesis that average bout length representing milk yield was the common independent variable determining the course of ontogeny across breeds and individuals.
Thus the objectives of this study have been: (i) To establish the differences between the two breeds, and consequently (ii) to describe the course of the ontogeny of suckling behaviour along with lifetime within each breed and (iii) finally to repeat that description along with average bout across the breeds.

Animals
Nine beef calf-dam-pairs (CDP), eight Hereford and one Charolais, were randomly selected from a herd at the Research Station of Ecological Agriculture and Breeding of Endangered Animals in Popielno, Poland.A recording period from March 28 to April 8 was agreed before the onset of recording.CDP, the calves of which were born during this period were selected.One additional Hereford calf was born March 27 and two further on March 26 (number 21-23 in Table 3) and were incorporated in the data set.The lactation number of the dams thereby selected varied between one and five.Additional five CDP (Polish Holstein-Friesian) were selected from another herd at the same Research station in a corresponding manner two years later.One additional Holstein calf was born one day before the agreed observation period (number 31 in Table 3) and was incorporated in the data set.That dairy herd averaged about 5500 litres per year on a grass-dominated diet during the year before the observations.The lactation number of the dams thereby selected varied between three and four.

Management
The calving stable of the beef breed was equipped with single cubicles (5.5 m 2 ) with a 5 to 10 cm thick straw mattress.(For further details see MAYNTZ et al., 2006).The selected CDP were kept together with further 54 Hereford-CDP on a remote, forest-enclosed pasture from 0800 to 1800 during the third observation period.Calves and dams were kept in adjacent corrals separated by bars because calves were offered shelter and concentrates in their corral during night.(For further details see MAYNTZ et al., 2006).The dairy cows gave birth in a conventional tied-up stall.The selected CDP were taken to a separate nearby pasture immediately before or after (one CDP) calving.During the following recording period, the six CDP stayed together on pasture separated from the rest of the herd.The dairy cows were not milked during observation.Thus breed was partially confounded with management.Therefore the special "breedmanagement-interaction" analysed here is meant in the following when speaking about the effect "breed".

Recording
Meals of the 12 selected beef cow-calf-pairs were observed during three periods, which were separated by 19 and 20 days.Recording was carried out during days of life 1 to 13, 21 to 36 and 60 to 80.The calves and their mothers stayed in the stable during the first and second observation periods.The beef cows were marked with large numbers on their flanks and thus could be selected as focal animals during the third observation period.A single observer worked during the three observation periods.During the first and second observation period, the observer slowly patrolled the aisles between 5.00 and 20.00.The camera was started and the observer quickly chose a position for it on the cow's side opposite to the calf and close to the floor when a meal seemed to be initiated.Recording was carried out continuously until a few minutes after the voluntary separation of cow and calf.During the pasture period the observer followed the herd to the pasture.He focused one CDP per day and recorded a voluntary meal between 12.00 and 18.00.Rules for start and end of recording were equivalent to those in the stable periods.Only one of two observers was available at a time for recording the dairy cows.One meal was recorded for each single CDP every day during the first week of life.Thereafter, one meal per week was recorded at the end of each week.These later recordings were made during a three-day window at the end of a week.Recording continued roughly 11 weeks pp.

Data handling
A time code was copied on the videotapes, showing hours, minutes, seconds and frames (25 frames / second) thus allowing different activities to be separated with an accuracy of 0.04 seconds.The tapes were read meal-by-meal concerning activities and places of the calf's mouth.The end of one activity coincided with the start of a new one.The activities are listed below.(i) Suckling, i.e., teat stretched in the mouth and sucking movements of the lower jaw, (ii) stimulate, i.e., lick, bite, massage, very slight push, take teat in mouth, (iii) transport of mouth, (iv) push, i.e., retraction of the head from the teat base and a consecutive strong push towards the teats base.The initial retraction of the head is missing in the "very slight push" mentioned above.(v) Strip, i.e. lengthening the teat considerably and stripping it from base to tip between the upper jaw and the front teats without sucking movements, (vi) calf not at udder, (vii) cow moves and (viii) cow kicks.The activities "push" and "stimulate" appeared in two versions each, with or with out the teat in the mouth.All activities could be combined, except with "calf not at udder".The following places of the calf's head were noted together with the activities: at (i) a specific teat, e.g., front right; (ii) between two specific teats, e.g., between front right and hind right; (iii) udder, i.e., at almost equal distance from all teats; and (iv) cow's body.Duration of every activity was calculated and a time scale created by cumulative addition of all activities.Later the bout durations at every teat were transferred to a column for each corresponding teat at its starting point on the time scale.The standard area-graph of the Excel program was chosen to visualise time scale and bouts of each teat.The main weakness of that graphical routine is that it does not allow free scaling of x-and y-axel, and thus is not fit for easy graphical comparisons.Its main advantage is that a graph can be produced very fast.

Dependent variables
During data handling, the questions listed below arose.An example of such a question can be seen in Fig. 1.At first that meal was a two-teat-meal (hind right and front right) with complete meal structure.Apparently the amount of milk suckled from these two quarters did not satisfy the calf.Thus it addressed a third teat (hind left) close to the end of after-stimulation of the first two quarters.The cistern of that late addressed quarter was full because stimulation had elicited an ejection also there (ZAKS, 1962).Consequently the first bouts at that third quarter were very long.From that graph the idea for the question arose when the suckled teats were addressed for the first time during a meal (see (ii) below).The questions mentioned above were: (i) How many teats were suckled during a meal?(ii) When was a specific teat addressed for the first time during a meal?(iii) How uniformly were the teats suckled?(iv) How many bouts were counted?(v) How long was the average bout?(vi) How often was a bout followed immediately by another one at the same teat?(vii) How long was the meal?(viii) How much time was spent on suckling or on non-suckling activities?(ix) How long was the longest bout?(x) When was a complete meal structure, i.e., including at least pre-stimulation, ejection and after-stimulation recorded for the first time?The variables described below were developed to answer these questions.
Variables describing the use of teats.(i) Number of suckled teats, (ii) relative entrance of first, second, third and fourth suckled teat into the meal i.e., time until suckling at that specific teat started for the first time in % of meal, (iii) uniformity of relative shares of suckled teats.Uniformity was calculated in the following way: (a) the standard deviation between the relative shares (sum of bouts at a specific teat in % of all bouts) of all four teats, i.e., including the not-suckled ones was calculated.(Perfect uniformity, i.e., 25 % suckling at every teat: 25, 25, 25, 25%, s = 0. Totally missing uniformity, i.e., only suckling at one teat: 100, 0, 0, 0%, s = 50).(b) The actual standard deviation was multiplied by 2 (Perfect uniformity: 0 x 2 = 0. Totally missing uniformity: 50 x 2 = 100).Thus any distribution of relative suckling time between the four quarters was expressed on a gliding scale between 0 and 100.(c) The scale was turned by subtracting the actual figure from 100 (Perfect uniformity: 100 -0 = 100.Totally missing uniformity: 100 -100 = 0).The third step was done exclusively to ease reading by linking perfect uniformity to 100 and totally missing uniformity to 0, (iv) number of bouts.(v) average bout, being sum of bouts divided by number of bouts and (vi) relative number of bout-teat-repetitions.The graphs mentioned above had revealed that mostly younger calves returned to the teat just suckled after a completed bout, but that this behaviour disappeared gradually.Because the different meals included a vide variety of bouts, the number of bout-teat-repetitions had to be expressed in % of all bouts.
Variables describing the meal as a whole.(i) Meal length, (ii) meal-bout-relationship (= meal length / number of bouts) expressing how much time of a meal was spent mainly on suckling (small figures) or also on not-suckling activities (high figures) and (iii) longest bout.
Qualitative variables indicating a complete meal structure.(i) Day pp, at which the occurrence of a complete meal structure could be recorded for the first time and (ii) number of suckled teats during that meal.

Models and levels of probability
The model for testing differences between breeds included the following effects: breed, cow (breed), and day pp as regression.The model using average bout instead of day pp included the following effects: breed, cow (breed), and average bout as regression.Data were analysed by GLM procedure of SAS (SAS/STAT, 1989).
Least square means were calculated for the variables significantly effected by the applied model.Contrasts between least square means were compared with Dunnets ttest.Only contrasts between the fixed effect breed are shown.
The composition of the data set and the number of replications were not planned to falsify a priori established hypotheses.Therefore the actual probability levels are given, however, not those ≥ 0.1.

Results
Tables 1 summarises the results of analysis of variance including the effects breed, cow (breed), and day pp as regression.Table 2 gives the contrasts between the breeds.Table 3 gives the day pp, at which a complete structure of a meal could be recorded for the first time (on average day 4 pp for the beef and day 12.3 pp for the dairy breed) and the number of suckled teats at the corresponding meal (2.92 and 1.83 respectively).Table 4 gives the regression equations for the beef and table 5 for the dairy breed with day pp as independent variable.Figure 2 demonstrates the relationship between the relative entrance of the different teats into a suckling meal and day pp.The analysis with day pp as main effect (Tables1 and 2) or independent variable (Tables 4 and 5) revealed the following: (i) Number of teats, uniformity of relative shares between suckled teats, and number of bouts increased with increasing day pp.
(ii) Average bout, relative number of bout-teat repetitions and meal-bout-relationship decreased with increasing day pp.(iii) An additionally suckled teat was not suckled Linear regressions with day pp as independent variable (x) for beef breed (Lineare Regressionen mit Tag pp als breeds concerning average bout indicate a close correlation between average bout and milk yield.Table 6 gives the results of analysis of variance of the complete meal data including the effects breed, cow (breed), and average bout as regression.
Table 7 gives the regression equations across the breeds with average bout as independent variable.

Day pp as main effect and differences between breeds
The differences between breeds can be quantified with the help of the regression equations in Tables 4 and 5: All four teats would be suckled from day 52 in the beef and from day 71 pp onwards in the dairy breed and a very high uniformity (about 90) of suckling activities between the teats would be reached at about day 58 and 90 pp respectively.After day 52 or 90 pp mentioned above the description of an average meal would be: (i) The first and second teat in the dairy and the first three teats in the beef breed would be addressed as fast as possible at the beginning of a meal.Only for the fourth teat a small delay could be seen (about 9 % of meal in the beef, respectively16 % in the dairy breed).(ii) A meal would include 139 (beef) and 63 bouts (dairy) with and average length of about 4.8 or 6 seconds respectively.(iii) No bout would be repeated at the same teat immediately, likewise almost no none-suckling activities would be seen in either breed.(iv) A meal would last for about 11 (beef) and 10 (dairy) minutes, (v) but the longest bout would differ substantially with about 24 second in the beef and 52.5 seconds in the dairy breed.An explanation for these differences between the breeds could be, that the initial oversecretion of the mother (MAYNTZ and SENDER, 2006) is bigger in the dairy breed and that therefore different periods of time were needed to level out secretion of the mother and need of the offspring.
That hypothesis demands that the mechanisms acting before levelling out were similar in both breeds.Assuming that this hypothesis cannot be falsified, we would see the same processes in e.g., wild species but during a much shorter period of time compared to the beef breed reported here.

A new approach: Average bout as main effect
The hypothesis with different levels of initial over-secretion together with spontaneous impressions from the meal graphs led to the idea to replace day pp by milk yield as main effect.Average bout instead of milk yield had to be used because recording of milk yield had not been feasible.It was striking that the significance for the effect breed decreased, when average bout instead of day pp was used in an otherwise similar model (Table 6).Comparing Tables 1 and 6 shows that breed changed from a very powerful into an almost negligible effect in the second analysis.Supposing that milk yield cannot be rejected in future tests as common factor between breeds, precise predictions are possible on the basis of that hypothesis.(i) All variables describing the end of ontogeny of suckling behaviour, especially the number of suckled teats, the uniformity of relative shares of suckled teats and the relative number of bout-teat-repetitions would reach the final stage earlier the lower the milk yield of the mother and vice versa.This alternative hypothesis is supported by the results of PASSILLÉ and RUSHEN (2006).(ii) Within that hypothesis it would not matter, why the milk yield of a specific mother is low or high: wild life or husbandry, first or later lactations (STEINHARDT and THIELSCHER, 2005), good or poor energy supply etc.All would influence the length and course of the ontogeny of suckling behaviour of the actual offspring via their effect on the actual milk yield of the mother.This hypothesis could contribute to further diminishing the black box between genetic potential, behaviour and phenotypic outcome of lactation (STEINHARDT and THIELSCHER, 2004;STEINHARDT and THIELSCHER, 2005).However, contributions of behaviour to different phenotypic outcome of lactation due to sex and/or birth weight of the calf could be explained by including the FIL-hypotheses (WILDE et al., 1988).
(For further arguments see MAYNTZ and SENDER, 2006).With the help of the regressions in table 7 we could describe the quantitative course of ontogeny of suckling behaviour in cattle across the breeds and individuals in the following way: The shorter the average bout, (i) the higher the number of suckled teat; however, 4 teats would hardly be suckled, instead the maximum would be about 3.95; (ii) the higher the uniformity of suckling activities between the teats; however, it would reach only a maximum of 87; (iii) the higher the number of bouts; however, almost never higher than 90; (iv) the lower the relative number of bout repetitions, however, immediate bout repetition at the same teat would not disappear completely; (v) the longer the meal; however, not longer than about nine minutes; (vi) the less not-suckling activities and (vii) the shorter the longest bout; however, not longer than about 12 seconds.Most of the quantifications above agreed well with the data analysed here, the third and the fifth, however, less.Probably these discrepancies resulted from: (i) the size of the data set analysed here might have been too small to fit second degree regressions; (ii) average bout is only an indirect measure for milk yield; (iii) an offspring might need a minimum number of sucking movements to realise that the corresponding cistern is empty, resulting in that a bout cannot fall below a lower threshold; (iv) the impression of the degree of filling of a cistern might also be subjected to experience and memory and therefore might not be based on the actual status of the cistern exclusively and (v) a minor part of variance might be effected by day pp.

General aspects
The study presented here allows to hypothesise that species, which share the anatomical and endocrinological characteristics listed below, show the same principal course of ontogeny of suckling behaviour as Bos taurus because they all have to reduce the initial over-secretion of the mother (MAYNTZ and SENDER, 2006) without drying of single mammary glands.The characteristics mentioned above are: (i) the need of a pre-stimulation (ZAKS, 1962) before ejection of milk, (ii) a relative middle size milk duct area (CROSS, 1977), where the milk is not under capillarity (referred to as "cistern" in the following), the volume of which is 20 to 40 % of the total milk duct volume (BRUCKMAIER et al., 1994;DEWHURST and KNIGHT, 1994), (iii) the cistern being more of cavern type (WIRZ, 1913), (iv) a relationship between number of offspring and lactating mammary glands greater than one and (v) the number of active glands being constant within and between lactations even if the number of nursed offspring varies.Keeping several glands lactating demands that offspring suckle from several glands regularly.The cavern-like cistern weakens the relation between the inflow into that cistern and the flow from the alveoli and in all those small ducts, where the milk is under capillarity resulting from squeezing (true alveoli alone) respectively relaxing (small ducts, BRUCKMAIER, 2005).Only milk in the cistern is available for harvest.An outflow from the cistern larger than the inflow into it makes waiting for refilling of a cistern profitable for offspring (MAYNTZ and COSTA, 1998).Waiting for refilling a single cistern is optimised when several cisterns are emptied in a sequence.Several mammalian taxa share these characteristics.
Recording average bouts could be feasible in husbandry and in zoos.In zoos it would be much more worthwhile than the conventionally recorded "lock on-" (e.g., TAVOLGA and ESSAPIAN, 1957;MCBRIDE and KRITZLER, 1951) or "nippleattachment-time" (e.g., MENDL and PAUL, 1989), because its close correlation to individual milk yield.Average bout would also help to differ better between nutritive and non-nutritive phases in a meal (e.g., WOLFF, 1968;LIDFORS et al., 1994;TANAKA, 1997) and thus to estimate milk supply of mammalian offspring in human care better.It remains a challenge to define "milk yield" in relation to "need" to make full use of that approach.
Analysing the ontogeny of suckling behaviour under the aspect of milk yield revealed another aspect.Cattle offspring were apparently born with a very simple rule: "look for milk, where you found it last".The development of every variable describing the use of teats did not reject that hypothesis.A calf started to address other teats not before the amount of milk harvested from the first addressed ones fell undoubtedly short.The strength of that rule can be judged from regression equations for relative number of bout-teat-repetitions in tables 4, 5 and 7. Starting from frequent repetitions, it takes a considerable number of days pp (64 in the beef and 78 in the milk breed) to get almost rid of theses often vain attempts to get milk from "where you found it last".
The big teacher in nature, i.e., shortage of resources, made a complicated behavioural pattern eventually emerge from a simple behavioural rule.

Table 1
Results of analysis of variance for the model breed, cow (breed), and day pp as regression (Resultate der Varianzanalyse mit den Varianzquellen Rasse, Mutter (Rasse), und Tag pp als Regression)

Table 2
Contrasts between least square means (LSM) of beef and dairy breed for variables significantly influenced by breed in the model breed, cow (breed), and day pp as regression (Kontraste zwischen least square means (LSM) von Fleisch-und Milchrindern für die Variablen, die im Modell Rasse, Mutter (Rasse), und Tag pp als

Table 3
First occurrence of a complete meal structure and number of suckled teats during the corresponding meal (Erstes Erscheinen einer kompletten Mahlstruktur und die Zahl der besaugten Zitzen während dieses Mahls)

Table 6
Main results of analysis of variance for the model breed, cow (breed), and average bout as regression (Hauptresultate der Varianzanalyse mit den Varianzquellen Rasse, Mutter (Rasse) und durchschnittliche Bout-