Back to Topics Sorted by Author  Back to Topics Sorted by Program

Changes in Social Behaviour of Mares Pre and Post Partum Compared to Behaviour of Non Pregnant Herd Mates in Semi-feral Mixed Herds of Icelandic Horses
Machteld C van Dierendonck,1 Hrefna Sigurjónsdóttir,2and Anna G Thorhallsdóttir3
1Tolnegenweg 39, 3776 PT Stroe, The Netherlands
Voice +31.342.401.502    Fax +31.342.475.656
2Iceland University of Education, 105 Reykjavik, Iceland
3The Agricultural University at Hvanneyri, Iceland


In modern horse management, there is a tendency toward more natural housing and grouping of horses. The rationale, other than welfare considerations, is the positive relationship between the prevention and treatment of stereotypies and husbandry: providing more natural social contact and freedom of movement (McGreevy et al, 1995; Lebelt, 1998; McBride and Long, 200. Often these domestic pasture groups contain both mares and castrated males (geldings). Domestic social grouping, whether it is organised as a herd or not, is different from the typical natural forms of social organisation (i.e. harems, multiple male bands and bachelor bands), upon which most of the research literature is based. Domestic groups are often larger than natural social groups. It is important to investigate how social grouping influences the behaviour of subgroups in such a domestic setting in order to be able to define the needs and basic requirements of horses in such a management system in the future.

In an “unnatural” mix of semi-free ranging Icelandic mares and geldings of all age classes, we measured some changes in behaviour of foaling mares from preparturient state to post parturient state with non pregnant (barren) herd mates as contemporary controls. Estep et al (1993) investigated changes in social behaviour of mares in relation to foaling. Their article gives a good overview of the literature concerning this subject. With that review combined with the literature since 1993, the following hypotheses concerning changes in behaviour of foaling mares pre and post partum were tested in the work reported here:

a)      affiliative behaviours would decrease following parturition

b)      frequency of close spatial proximity to herd mates (other than own foal) would decrease following parturition

c)      rank would increase following parturition

d)      aggressive behaviour would increase following parturition

e)      time budgets would include less social interaction and more vigilance following parturition


            Subjects. Pastured groups were studied in each of two years, 1997 and 1999. In 1997 a group of 34 animals of 1 year-old and older were studied. This included 12 pregnant mares ages 9 to 19 years, 5 barren (non pregnant) mares ages 10 to 20 with suckling yearlings (4 intact colts and 1 filly), 2 adult geldings ages 9 and 10 years, and 10 sub-adults (5 geldings ages 2 to 4 years and 4 females ages 2 to 3 years). Eight foals (2 males and 6 females) were born, 5 actual births were observed. One foal died.

In 1999 there were 31 animals of 1 year-old and older at the start. These included 14 pregnant mares ages 11 to 22 years; 4 barren adult mares ages 6 to 24 years; 4 mature geldings ages 11 to 18 years; and 9 sub-adults including 1 yearling colt, 4 geldings 3 to 4 years of age, and 4 females ages 2 to 4 years. Eighteen horses including 11 adult mares were the same as in 1997. Eight of the animals from two other farms, including 4 of the breeding mares, were new to the group. This was done to introduce unfamiliarity. Fourteen foals (5 colts and 9 fillies) were born in 1999, 13 birth' were observed. One foal died.

The complete herd (60-70animals) was free-ranging year round in a mountainous area exceeding 200 hectares. Supplementary silage was provided between December and May.

Observations. The observations took place for 24 hours a day from May 6 – June 11 in 1997 and from May 3 – June 13 in 1999. During observation periods, the selected groups were held within an 8-hectare sub enclosure. Observation methods consisted of 15-minute focal animal observation of adult mares (1997), alternated by free sampling on sexual, affiliative, dominance and intervention behaviours and half hour spatial sampling in 1997 (within 2 horse length). In 1997 the 12 pregnant mares and 5 barren mares were observed as focal animals. For analysis, the adult pregnant mares were treated as two separate individuals: (1) pregnant and (2) post partum.

Data summary and statistics. The unequal presentation of the entities, due to the different times the foals were born, required calibration matrices to be constructed to recalculate the data back to comparable frequencies per 24 hours per individual or dyad per period. The barren mares served as control. Post priori at each foaling, a barren mare was randomly chosen and her data were also analysed separately from that date onwards. When these mares served as control for the pregnant mares they were called barren-preg and when serving as control for post partum mares they are called barren-post.

Data collection and analysis were done using The Observer®, Systat® and MatMan® software. Statistics included for matrix correlation the partial rowwise Tau (t) KR permutation test (2000 permutations) controlling for structural zero's (Devries 1993); Wilcoxon Signed Rank test (WSR-Z) for 2 related samples (ie pregnant and post partum mares), Mann Whitney U (MWU-Z) for the independent samples and Friedman for k related samples (Sokal and Rohlf, 1981; Siegel and Castellan, 1988). Rank orders were constructed as described in Van Dierendonck et al (1995) and are presented in Sigurjónsdóttir, et al (manuscript submitted).


            In many observed factors, post partum behaviour was different from pre partum as well as from that of the barren mares. Most (15 out of 18) mares left the core of the group by 50-100 meters for parturition. They tended to return to the periphery of the herd within hours after birth. 

Allogrooming frequency and distribution. The allogrooming frequency dropped significantly in both years after parturition: (pregnant vs post partum 1997: WSR-Z = -2.37, n = 7, p = 0.018; 1999: WSR-Z = -3.18, n = 13, p = 0.001). However, the pregnant mares did not differ in frequency from the barren-preg mares in any year (1997: MWU-Z = -1.35, n = 19, p = 0.196; 1999: MWU-Z = -1.76, n = 27, p = 0.085). While the allogroom frequency of post partum mares was also lower in relation to the barren-post mares: (1997: MWU-Z = -3.14, n = 19, p = 0.001; 1999: MWU- = -3.29, n = 27, p = 0.001). There was a tendency to lower the allogroom frequency ongoing the research (barren-preg vs barren-post 1997: WSR-Z = -2.03, n = 5, p = 0.043; 1999: WSR-Z = -1.83, n = 5, p = 0.068). The total individual frequencies for the mares did not differ significantly between the years, when in the same reproductive status (pregnant or barren): (WSR-Z = -1.60, n = 9, p = 0.110). Thus the years are comparable.

In each year, allogrooming occurred more often among mares of the same reproductive status (pregnant, post partum or barren) than between them: 1997: Partial Kendall rowwise correlation (Tau rw;XY.Z) = 0.16, n = 24, p = 0.0005; 1999: Tau rw;XY.Z = 0.14, n = 32, p = 0.0005).

Close proximity frequency and distribution. In line with the allogrooming results, spatially the mares formed subgroups according to their reproductive status (measured in 1997 only, Tau rw;XY.Z = 0.36, n = 24, p = 0.000). Thus there were spatially distinct groups of pregnant mares, post partum mares and barren mares. The post partum mares did not change their frequency of close proximity compared to their pregnant state (MWU-Z = -1.36, n = 7, p = 0.173). But, not surprisingly, they redirected it more to their newborn foal: when the dam-foal frequency was artificially set to "0", there was a clear drop in frequency in relation to other herd members than their new born foal (MWU-Z = -2.20, n = 7, p = 0.028).

Dominance. In the over all rank orders with all individuals >1 year, the (next) older mares took the top places. If the adult mares were divided into 5-year age classes, the results strongly suggested that the top ranking mares came from the second oldest age class. In a regression analyses a second-degree curve fitted the data better (and significantly) compared to a linear correlation of rank vs age for mares, (quadratic regression 1997: F = 3.78, 14 df, p<0.05; 1999: F = 12.28, 15 df, p<0.001). It explained a lot of the variation in 1999 (R2 = 0.621), while in 1997 only a third of the variation was explained by the age class (R2 = 0.351). Only in 1997 was there some indication that the birth of a foal would increase the dam's rank. There was no indication of a higher frequency of dominance-related behaviours of post partum mares compared to the pre partum or barren mares. The rank position was remarkably consistent over the years. From the 1997 to the 1999 observations, only 2 adult mares had exchanged position.

Time budget. A total of 374 hours and 33 minutes of time budget behaviours were recorded in 1997 in 1220 focal animal observations. For the time budget the observed behaviours were categorized into six groups: eat, rest, stand vigilant, move, socialize, miscellaneous. The data were analysed per number of days since a foal was born.

Table 1. Overall time budget in %, by reproductive status group















Post partum














Post partum mares spent smaller proportions of time eating and in social interaction and a spent greater proportion of time vigilant than the barren mares. The barren mares smaller percentages of time resting and vigilant than other mares. They had a greater percentage of time eating and, interestingly, in social interaction (2% to 4%) and in miscellaneous behaviour (which included sexual behaviour). The percentage of time spent vigilant by post partum mares in the first 24 hours after parturition, was significantly greater than that of both pregnant and barren mares on these same days (MSR-Z = -2.20, n = 5, p = 0.043 and MWU-Z = -2.73, n = 14, p = 0.004). There was still a significant difference between percentages of time spent vigilant (MSR-Z = -2.20, n = 7, p = 0.028) and eating (MSR-Z = -2.03, n = 7, p = 0.043) for post partum mares on the day of parturition compared with the seventh day after. After two weeks this difference was smaller, but still detectable.

Figure 1. Mean allogrooming frequency of pregnant, post partum, and barren mares in 1997 and 1999


In this study, pregnant, post partum, and barren mares were kept together with all other sex-age groups (except a mature stallion). The hypotheses put forward above were supported except those in relation to dominance (c, d). In the research of Estep et al (1993) where only foaling mares were together, the post partum mares also allogroomed less frequently and did not rise in rank as was the same in this study. The Estep mares also had stable rank orders, for those that could be constructed.

In the Estep study preferred associates changed following parturition, which was attributed to the increased distance to other mares in general. The mares in our study did not change their allogrooming partners neither post partum nor over the years. The mutual grooming partners are not always the same as those with whom they associated spatially (Kimura, 1998), but in this study often they were (see Sigurjonsdottir et al, this proceedings): there was a significant relationship between these two association measures for with all animals involved (Sigurjónsdóttir et al, submitted).

The Icelandic mares did not return to their pre partum rates of allogrooming within 5 weeks of parturition, while the Estep mares returned to their pre partum rates 3-4 weeks after birth. The lower frequencies of the barren mares when serving as post controls is not unlikely due to the end of the moulting period.

In general, the horses in our study seemed to be more socially stable, with longer lasting dyadic relations, than the Estep study mares. Since allogrooming seems to be the most important factor determining positive social relations between mares (Feh and Mazieres, 1993; Kimura, 1998; this study) it was analysed more in depth for the mares. In this study, allogrooming partners were more often of the same reproductive status, which could be explained by their spatial distribution: post partum they mingled together with their newborn foals. It likely that the explanation for this is that they have a common goal of strengthening the bond with their own foal, while also keeping their newborn foals away from the sometimes hectic interactions, especially of the sub-adults, in the centre of the herd. This would be consistent with the bonding theory of Klingel (1969). A factor contributing to aggregations of pregnant mares may be that also they are not as agile as the nonpregnant mares and also may avoid the more hectic centre of the herd.

The time budget of the post partum mares was most deviating in a higher proportion of vigilant behaviour. Higher proportion vigilance was still recorded after one week. This is not necessary according to Klingel’s bonding theory, so it could serve another function.

From this research we can learn that the presence of other sex-age classes in a domestic herd does not seem to alter the described aspects of post partum behaviour of mares. However, horse keepers should consider providing ample space. It seems unnecessary to remove mares for parturition.


Devries H (1993) The rowwise correlation between 2 proximity matrices and the partial rowwise correlation. Psychometrika 58(1): 53-69.

Estep DQ, Crowell-Davis SL, Earlcostello SA, Beatey SA (1993) Changes in the social-behavior of drafthorse (Equus caballus) mares coincident with foaling. Appl Anim Behav Sci 35(3): 199-213.

Feh C, Mazieres J de (1993) Grooming at a preferred site reduces heart rate in horses. Anim Behav 46: 1191-1194.

Kimura R (1998) Mutual grooming and preferred associate relationships in a band of free-ranging horses. Appl Anim Behav Sci 59(4): 265-276.

Klingel H (1969) Reproduction in plains zebra, Equus burchelli boehmii: behaviour and ecological factors. J Reprod Fertil Suppl 6: 339-345.

Lebelt D (1998) Stereotypic behaviour in the horse: general aspects and therapeutic approaches. Praktische Tierarzt 79: 28-32.

Mcbride SD, Long L (2001) Management of horses showing stereotypic behaviour, owner perception and the implications for welfare. Vet Rec 148(26): 799-802.

McGreevy PD, Cripps PJ, French NP, Green LE, Nicol CJ (1995) Management factors associated with stereotypic and redirected behaviour in the thoroughbred horse. Eq Vet J 27(2): 86-91.

Sigurjónsdóttir H, Vandierendonck M, Thorhallsdottir AG (submitted 2001) Social structure in mixed herds of Icelandic horses: Kinship matters.

Siegel S, Castellan NJ (1988) Non Parametric Statistics for the Behavioural Sciences. McGraw-Hill, Singapore.

Sokal RR, Rohlf FJ (1981) Biometry. Freeman and Co, London

Vandierendonck MC, Devries H, Schilder MBH (1995) An analysis of dominance, its behavioural parameters and possible determinants in a herd of Icelandic horses in captivity. Neth J of Zool 45(3-4): 362-385.