Icelandic horses – Virtual mate selection Java servlet - Explanations
This servlet computes BLUP indices on the potential offspring of the selected parents. BLUP
indices are computed as the mean of the BLUP indices of the parents for the 17 following traits:
1) Height at withers
2) Mane and tail
3) Slow toelt
6) Neck, withers and shoulders
7) Back and hindquarters
9) Leg quality
10) Leg stance (Correctness of legs)
16) General impression (Form under rider)
Weighted aggregate indices are computed for: 18) Conformation, 19) Riding ability and 20) Total
The methods used for computation of the BLUP indices are described in Árnason, et al. (2006).
The data is from WorldFengur the global database on Icelandic horses.
The relationship (R) between the selected parents is computed by a recursive method of ter Heijden
et al. (1977). The diagonal elements of the numerator relationship matrix (A) had been calculated
in advance by the algorithm of Sigurdsson (Sigurdsson & Árnason, 1995). The inbreeding
coefficient (F) of the offspring is half the coefficient of relationship (R). Both the coefficients are
multiplied by 100 and expressed as %.
The average inbreeding coefficient in the population is about 2,8%. Close inbreeding should be
avoided as risk for genetic defects in the offspring increases with increased inbreeding (inbreeding
depression). Matings resulting in inbreeding coefficients above 5% should be avoided. Inbreeding
coefficients between 5% and 7% will be shown in purple colour and inbreeding coefficients above
7% are shown in red colour to indicate that this mating is not to be recommended.
Colour genotype probabilities
The basic colour variation of the Icelandic horse can be explained by segregation of alleles (genes)
in 8 known loci (Adalsteinsson, 2001). One more loci for “splashed white” is known in Icelandic
horses, but the registration of that colour is lacking in the data and is therefore ignored. Five other
colour loci are known in horses, but have not been found or confirmed in Icelandic horses.
Thorvaldsson (2004) has shown that the black smoky colour (glóbrúnn) is caused by the C
on a black basic colour. The same gene causes palomino when acting on chestnut basic colour and
buckskin when acting on bay basic colour. In many cases carrier of the C
gene are registered as
black. Thorvaldsson (2004) concludes that the dilutional effects of the C
gene on black basic
colour depends on an interaction with other genes, probably the same genes that control the strength
of the black colour which may vary from pale black (2200) to dark black (2700). In similar way
the yellowness of the palomino and buckskin colour probably depends on the strength of the
chestnut and bay colour, respectively. The champagne gene which was claimed to exist in
Tennessee Walking Horses by Sponenberg & Bowling (1996) is therefore almost certainly not to be
found in the Icelandic horse population according to the results of Thorvaldsson (2004).
Combination of A (agouti) gene, C
gene, D (dun) gene and E (extension) gene results in dark
yellow colour (often followed by light mane and tail). These horses should actually be registered as
bucskin (5???) but are often registered as palomino (4???) instead. For more thorough description
of coat colour genetics in horses, see Adalsteinsson (2001) or Bowling (1996, 2000).
Overview on the colour genes and their effects are given in the linked explanatory table.
The estimated gene frequency and the corresonding genotype frequency in the eight color loci in the
Icelandic horse population is shown in a special linked table here.
The estimates are based on the assumptions that the gene and genotype frequencies are in Hardy –
Weinberg equilibrium, i. e. not seriously affected by selection, mutation or migration (Falconer,
Coat colour of the horses registered in WorldFengur is given by a 4 digit code which was developed
by Stefánsdóttir (1991). Detailed description of the colour code is to be found on
The servlet show possible genotypes of the parents and of the offspring. The possible genotypes of
the parent have been computed in advance by the Genotype Elimination Algorithm (GEA) of Lange
(1997). The algorithm works iteratively and eliminates all genotypes of the individuals which are
incompatible with the phenotype of the individual or some of the related animals. By convention a
capital letter stands for the dominant allele in each locus while the small letter denotes the recessive
allele. The sign “-” means that the allele type can not be determined. When both alleles in a pair
are the same (e.g. EE or ee) the animal is said to be homozygous in that locus and when the pair has
different alleles (e.g. Ee) the animal is heterozygos. For further details see e.g. Bowling (1996).
The resulting possible genotypes of the parents from the run of the GEA program were used to
assign “phenotypic” code to the animals in the following manner: (e.g. for E-locus) 1 = ee, 2 = Ee
(or eE), 3 = EE, 4 = E-, 5 = e- ,and 9 = -- (genotype completely unknown). The computer program
for calculation of genotype probabilities was obtained from Kerr and Kinghorn (1996) and used to
compute the genotype probabilities shown in three rows in the tables for Sire and Dam. The
method calculates the conditional probability that the individual animal has a certain genotype
given all the data. The genotype probabilities are given with maximum of 3 significant decimals.
Genotype probabilities are given on the scale 0.0 to 1.0 (100%). All loci are assumed unlinked in
the calculations, even though some of the loci are known to be linked (located on the same
(cc) causes strong dilution of the coat colour towards almost an albino like state
(pseudo-albino, cremello, perlino; Icel.: hvítingi). Since many breeders like to avoid mating of two
carriers of this gene, a positive probability of this genotype cc is warned for by a red colour in the
table showing the genotype probabilities of the offspring. Homozygous RnRn (RR) are claimed to
be lethal (at least in some breeds) and mating of two carriers of this gene is recommended to be
avoided. A positive probability for the genotype state RR is coloured red in the table. Horses which
are homozygote in the silver colour gene (ZZ) tend to have poor vision and have even been found to
be nearly blind in some cases. Therefore mating of two carriers of the Z gene should definitely be
avoided and a positive probability for the genotype state ZZ is also coloured red in the table.
Possible coat colours of the progeny and probabilities of each colour type
Finally the the possible phenotypes associated with the offspring's genotypes are illustrated by
pictures. The conditional probabilities of each phenotype given the three possible genotypes in
each locus are calculated for the offspring. These conditional probabilities are shown in Árnason
and Sigurdsson (1998). The conditional probabilities of the additional colours Gray, Tobiano and
Roan, which may act independently on any basic colour are given separately.
The pictures are used in this servlet by the permission of WorldFengur. The photographer is late
Mr Fridthjófur Thorkelsson, who's contribution is greatly acknowledged.
Adalsteinsson, Stefán, 2001. Íslenski hesturinn – litir og erfdir. Ormstunga, Reykjavík, Iceland.
Árnason, Thorvaldur & Sigurdsson, Ágúst, 1998. A computing procedure for estimating genotype
probabilities at eight individual colour loci in the Icelandic toelter horse. 49
Annual Meeting of
the European Association for Animal Production, Poland, 24-27 August 1998.
Bowling, A.T., 1996. Horse Genetics. CAB International. Wallingford, Oxon, UK.
Bowling, A.T., 2000. The Genetics of the Horse. Chapter 3. CAB International. Wallingford,
Falconer. D.S., 1989. Introduction to Quantitative Genetics. Longman Scientific & Technical,
Harlow, Essex, UK. & John Wiley & Sons, Inc., New York, USA.
Kerr, R.J. & Kinghorn, B.P., 1996. An efficient algorithm for segregation analysis in large
populations. J. Anim. Breed. Genet. 113:457-469.
Lange, K., 1997. Mathematical and Statistical Methods for Genetic Analysis. Springer-Verlag,
New York, USA.
Sigurdsson, Ágúst & Árnason, Thorvaldur, 1995. Predicting genetic trend by uni- and multitrait
models. Acta Agric. Scand. Sect. A, Animal Sci. 45:1-10.
Stefánsdóttir, Gudrún, Jóhanna, 1991. Litaerfdir hrossa og könnun á tídni lita íslenskra hrossa.
B.Sci. Thesis, Agricultural University of Iceland. Hvanneyri, Iceland.
Ter Heijden, E., Chesnais, J.P., Hickman, C.G., 1977. An efficient method of computing the
numerator relationship matrix and its inverse matrix with inbreeding for large sets of animals.
Theor. Appl. Genet. 49:237-241.
Thorvaldsson, Gudni, 2004. Eru til kampavínslitir í íslenska hrossastofninum? Report RALA
027/BU-004. Agricultural University of Iceland, Hvanneyri, Iceland.