50 M.C.G. Davies Morel, V. Gunnarsson Animal Reproduction Science 64 2000 49–64
Thorkelsson, 1991; Palsson, 1996. The Icelandic horse stands 135 cm high with an es- timated worldwide population of 160,000 80,000 within Iceland. The horses are largely
managed extensively, stallions running out with their groups of mares under semi feral con- ditions during the summer. Some horses which are in training are kept in during the winter
period and some of the more popular stallions cover mares in an in-hand mating system at the beginning of the season, before being turned out to pasture with mares. The breeding
season of the Icelandic horse is from May to September with a peak from May to July Dyrmundsson, 1994. Very limited information is available on the breeding performance
of the Icelandic horse, let alone the effect that management practices may have. One re- port, however, suggests that fertility rates may be as high as 82.1 Hugason et al., 1985,
which compared to the fertility rates reported for other equine populations, for example, Thoroughbred 53–77 Sullivan et al., 1975; Merkt et al., 1979; Bowen, 1990; McDowell
et al., 1992, Ponies 35–82 Day, 1939; Hugason et al., 1985; Bristol, 1987; Garrot and Taylor, 1990, Heavy horses 59 Day, 1939 and Light horses 52 Day, 1939 is very
good. However, this figure arose from work carried out on a limited number of mares, and the results appear to exclude barren mares in some instances, hence, suggesting a possible
over estimation of fertility Hugason et al., 1985.
An official registration system for Icelandic horses has been in operation for many years but is limited in the information it records regarding reproductive performance. The fer-
tility of Icelandic stallions, though believed to be high, is felt by many breeders to have declined in recent years with a greater incidence of subfertile stallions. This survey, there-
fore, specifically aims to provide an overall figure for the fertility of Icelandic stallions. It is also hoped to provide information on the effect that various parameters may have on fertility
rates. Only with such basic information is it possible to analyse the current fertility rates of Icelandic horses and to organise breeding management in a way best suited to maintaining
good fertility. Such information will also create a starting point for further research and development work in the field. More generally this research, on an unimproved, genetically
isolated group of ponies, rather than the more usually used horses, may also help eluci- date some of the differences that exist between horses and ponies; the differing effects of
environment and the effects of man’s intensive management and selective breeding.
2. Materials and methods
A survey was carried out in 1995 with the co-operation of Icelandic horse breeding associations and individual stallion owners, whose stallions were standing at stud throughout
Iceland. In total, 27 stallions were used in the study, covering a total of 1590 mares. The stallions were selected according to the following criteria:
1. Those used by the local breed association. This allowed the use of routine paperwork
and minimised any differences between mare groups, as all mares had been accepted for breeding by the breed associations.
2. Those aged between 5 and 18, to minimise any age effect. 3. Those expected to cover not less than 40 mares during the season. In order to minimise
any effect of workload, as only stallions used to their optimum sexual capacity were included.
M.C.G. Davies Morel, V. Gunnarsson Animal Reproduction Science 64 2000 49–64 51
4. Those used during all three breeding periods, to allow direct comparisons to be made between in hand and pasture mating and between periods.
No changes were made to the normal extensive management of the stallions during the period of the survey. A standard report form was completed for each stallion request-
ing information on general stallion condition, environmental influences or changes that might have affected the stallion, weather conditions, feeding, exercise or training and
pasture. Information concerning the mares covered was also requested which included mare age, colour, breeding status, body condition along with the result of pregnancy di-
agnosis. Foaling rates were obtained in the spring of 1996 from the breeding association records.
Icelandic stallions are bred in three distinct periods during a breeding season. The exact start date and length of each period is determined by the local breed association andor the
stallion owner or leasee. In general, the periods are similar in length, on average 5 weeks. During period 1, stallions are used for covering in hand or run out at pasture with mares,
during periods 2 and 3 all stallions are out at pasture with a different group of mares for each period. During periods of pasture breeding mares are all turned out together on the
same day with the stallion being released immediately after. He remains with that group of mares until the end of the period. In all cases, each pasture is only used for one group each
summer.
2.1. Statistical evaluation Statistical evaluation was carried out on data for 27 Icelandic stallions. The calculation
for fertility rate FR was based on the result of foal or no foal for each mare that was covered. For mares covered in hand the FR is measured per cycle, whereas for mares at
pasture it is measured per period. As illustrated in Table 2, the range of values for adjusted fertility rates lies between
41.5 and 88.0, therefore, ANOVA is appropriate for statistical analysis of the results. When comparison within a factor or an interaction was required, contrasts among means least
squares means were tested, assuming a normal distribution of the trait in question Snedecor and Cochran, 1980.
The following factors were taken into account when statistically evaluating FR: 1. Factors number of observations:
a stallion 27; b period 3; c training level of stallion 3; d reproductive status of mare RS 3; e body condition of mare 3; f colour of mare 2.
2. Covariates: a age of stallion; b age of mare; c group size number of mares within a period;
d length of a period days. Statistical analysis was carried out by analysis of variance ANOVA-GLM in a statistical
program Minitab MINITAB — release 10, Minitab Inc., 1994. Due to the large numbers of factors and covariates and interactions between some of the factors it was decided, after trial
and error, that in order to simplify the analysis and account for these interactions and yet still achieve a meaningful analysis, to adjust FR according to one of the following models. The
only interaction included in these was stallion × RS as after trying numerous interactions in the models, this was the only one that proved significant. ANOVA and regression analysis
52 M.C.G. Davies Morel, V. Gunnarsson Animal Reproduction Science 64 2000 49–64
Table 1 Results for analysis of variance by models 1–4 for overall fertility rate for 27 Icelandic stallions in the breeding
season 1995 Model
Total R
2
Sourcecovariate d.f.
P-value Significance level R
2
a Significance of factors and covariates in the models No. 1 14.9
Age of mare 1
0.000
∗ ∗ ∗
1.0 Age of mare
2
1 0.000
∗ ∗ ∗
1.0 Stallion
26 0.000
∗ ∗ ∗
3.8 Reproduction status
2 0.309
NS Stallion × RS
52 0.000
∗ ∗ ∗
6.1 Colour group
1 0.977
NS Body condition
2 0.928
NS No. 2 3.6
Age of stallion 1
0.720 NS
Age of stallion
2
1 0.937
NS Group size
1 0.005
∗ ∗
1.0 Length of period
1 0.008
∗ ∗
1.0 Level of training
2 0.043
∗
1.0 Period
2 0.106
NS No. 3 relevant Breeding method
1 0.046
∗
1.0 No. 4 relevant Number of coverings
3 0.472
NS b Covariate coefficients
Covariate coefficient S.D. No. 1
Age of mare 0.038s
0.011 Age of mare
2
− 0.0017
0.00042 No. 2
Age of mare 0.034
0.010 Age of mare
2
− 0.0016
0.00041 Age of stallion
− 0.0093
0.026 Age of stallion
2
0.000096 0.0012
Group size of mare within period −
0.0062 0.0022
Length days of period 0.0047
0.0018
was then performed, as appropriate, to obtain the adjusted FR. Many more models could have been considered but for simplicity only those most likely to result in significance have
been given. These two models were as follows: Model 1:
Fertility rate = stallion × RS + colour of mare + body condition of mare + age of mare + age of mare
2
. Stallion × RS refers to the interaction between these two variables.
Model 2: Fertility rate = training level + period + age of mare + age of mare
2
+ age of stallion +
age of stallion
2
+ group size + length of period.
Interactions, when logical, were tried in the models but excluded if not statistically signif- icant. The conclusion was to include only the interaction between the stallion and the mare’s
reproductive status. Age of stallion was included in model 2 as even though in Table 1 it appears as N, it is well documented that FR is affected by stallion age. Though an attempt
M.C.G. Davies Morel, V. Gunnarsson Animal Reproduction Science 64 2000 49–64 53
was made to minimise this effect of age by the selection of stallions in the age range 5–18 years, a limited effect was observed in this work Section 3.1.7.
For the analysis of the effect the method of breeding had on stallion fertility two further models were used to specifically look at: the effect of the two different methods of covering
model 3 and the effect of the number of coverings model 4. To compare the two different methods of covering in hand versus pasture data was used
from the first period only as no in hand mating was practised in periods 2 and 3. Data from all 27 stallions was available but only 416 mares were covered during the first period. The
same statistical procedure was used as detailed above with the breeding method included as a factor in the following model:
Model 3: Fertility rate = breeding method + training level of stallion + reproductive status of
mare + body condition of mare + colour of mare + age of mare + age of mare
2
+ age of
stallion + age of stallion
2
+ group size + length of period.
A further evaluation of the method of hand covering was carried out to see if the number of times that a mare was covered per cycle had an effect. Data from 17 stallions used to
mate in hand 258 mares were available for this analysis. The same statistical procedure was used as detailed above with the number of coverings 1, 2, 3 or 3 included as a factor in
the following model:
Model 4: Fertility rate = number of coverings + training level of stallion + reproductive status of
mare + body condition of mare + colour of mare + age of mare + age of mare
2
+ age of
stallion + age of stallion
2
+ group size + length of period.
3. Results